Where are Athlon II X4s in the graphs ? Where's Phenom II X2 BE ?
Where is a 785G mobo roundup ? I am still hearing only i5/i7/P55. This is frustrating. You are not keeping up with your name. Drop the Intel hype and do something for the normal people that try to build computers on a budget.
Sorry if this has been asked or commented before (I haven't read the full 11 pages).
I build workstations for the animation and video industry and I factory-overclock them. Bloomfield has been very well received by my customers and I'm really happy with it. Now I'm studying Lynnfield and from what I've read, I don't feel quite comfortable with how Lynnfield is overclocked. Here are my two main issues:
1. Anand mentioned in past articles that overclocking Lynnfield would imply overclocking the PCIe bus, since the controller is integrated. How does this affect graphics and other cards? I'm not talking only about gaming cards but also Quadros or RAID controller cards.
2. Also, it seems Lynnfield OC needs voltage tweaking. This sounds not as nice as Bloomfield stock voltage overclocking, but what are the real consequences and drawbacks (if any) of voltage rising?
Thanks in advance for answering these two issues and thanks to the Anandtech's staff for such in depth articles.
It's been going on for a while now but the price increase for minimal performance increase is getting pretty silly. Back during the <1000MHz days a 100MHz bump was nothing to sneeze at, and even during the 1-2GHz days a 100-200MHz bump wasn't that bad. But honestly they (both AMD and INTEL) have gotten rediculous with their gouging of the higher end. Honestly ~130MHz difference between the 870 and the 860?!?
Their only saving grace (for stock clockers, or very moderate OC'ers) is the higher turbo levels of the 870, but again in most situations (that is those that do not task 3-4 cores simultaneously) the clock difference between the 860 and 870 is <150MHz, which on a ~3.5GHz core is virtually nil.
They have used pricing like this in the past, just that there were usually several options in the more sane ($300 and less) range before the final few clock bumps were disproportionately expensive.
When I bought my current E6600 it was a little over $300. The E6700 was over $500 for a 266MHz bump, so technically 10% but still nothing to write home about for the money. If the 870 were unlocked that price might be justified, as of now I agree with you that it is not so much.
Athlon X4: 300m transistors, no L3 cache, performs about 90% as fast as a Phenom X4
Phenom X4: 758m transistors, 8mb L3 cache (or is it 6?)
Does anyone think AMD isnt getting their money's worth out of the 458 million transistors used on the Phenom II to provide the L3 cache? I mean, more than double the manufacturing complexity for a small increase in performance?
Anand, the review shows overclocking results for the i7 860 with the
retail cooler, but I doubt anyone who intends to oc these CPUs will
stick with the retail cooler.
There is much comment on how the 920 compares to the 860. If I understand
correctly from earlier articles, the 920 should have an advantage
vs. the 860 if both are hard oc'd (ie. Turbo off) to aid in those
situations where the user is definitely going to be used all 4 cores,
in my case video encoding.
Thus, what I'd really like to know is, how does the 920 compare
to the 860 when both are oc'd using a _good_ air cooler such as the
Thermalright U120E or Cogage TRUE Spirit? And is there any real
difference in the maximum achievable overclocks when each chip if
oc'd using a good cooler? (only referring to air cooling here)
Thus, two questions, both aimed at tasks which do use all 4 cores
(video encoding, rendering, etc.):
1. If the 860 and 920 are oc'd to the same clock, presumably with
Turbo off being the most sensible setting, which is faster? How
does power consumption differ? (wrt to total power used for a given
task)
2. How high can each chip be oc'd using a good cooler? Does the 920
have a usefully higher limit? What difference does it make to the
test results?
I've still not found a site which has done this comparison. For me,
the comparison data of 860 vs. 920 at stock speeds with Turbo active
is interesting, but not useful.
And of course it would be most intriguing to see how the other
CPUs fit into this scenario when oc'd using a good cooler, ie. the
750, 975 and the various AMDs. Given the very low cost of the entry
AMD quad-core, maybe that would be a more productive platform?
My original train of thought before Lynnfield came out was that an
oc'd 920 for video encoding was worth the extra cost because it
would complete a task much sooner and thus overall use less power.
The results compared to the best AMD CPUs at the time bore this out
nicely. But with the 860 added to the mix, I can't work out whether
the 920 would attain a better oc than the 860 with a good cooler,
and/or be faster anyway when oc'd & Turbo off, and thus be a better
choice. Charts show the 860 often doing better, but that's at stock
speed. Other charts with oc'd results set the CPUs at the same speed
using the retail cooler, whereas I'd be trying to push the 920 as
high as I could within reason using a good cooler.
So, would it be possible please for you to compare the various CPUs
when overclocked to their best sensible extent using a good cooler,
not the supplied retail cooler?
For example, elsewhere I read of people attaining 4.2GHz for the 920
with little difficulty.
"1. If the 860 and 920 are oc'd to the same clock, presumably with
Turbo off being the most sensible setting, which is faster? How
does power consumption differ? (wrt to total power used for a given
task)
I've still not found a site which has done this comparison. For me,
the comparison data of 860 vs. 920 at stock speeds with Turbo active
is interesting, but not useful."
It has already been done with the i7 870, a CPU that has shown to be marginally faster than the i7 860 in most tests. What follows are benchmarks where the i5 750, i7 870, and i7 920 are run at the 750/920 common clock speed of 2.66ghz and the very attainable turbo speed of the i7 870 of 3.2ghz, no turbo in all cases.
The difference in nearly all gaming benchmarks, using settings that takes the GPU out of the mix as a bottleneck, all at the same clock speeds, are within a very very tight percentage range, at the very most a 10% spread. With Crysis and Far Cry 2 it is closer to a 1% spread. Multimedia benchmarks have a similar or slightly larger spread but nothing I would consider significant, and certainly not in all cases.
Most things being the same (at least in games which is what I mainly use my PC for anyways, work and everything else are done on my OS X machines), the LGA 1156 CPUs are more appealing to me due to their lower net cost and comparable performance levels relative to consumer level LGA 1366 CPUs and boards.
Some comments. Intel claims having split the market in Highend, Midrange and Lowend. 'Lynnfield' XEONs inherited the IMC from the highend X5570 type DP-XEON CPUs, outperforming the LGA1366 single-socket XEON W35XX CPUs either by the usability of RDIMMs (up to 8GB per module!) and speed (DDR3-1333 oficiallay supported with 1 DIMM per channel). And, not to mention, the higher NB clock of 2.40 GHz instead of 2.13 GHz.
Several tests with highly parallel code for numerical modelling showed, that a XEON W3550 outperforms a XEON 3470, IF we lift NB clock to 2.66 GHz as used for the W3580 type AND also use DDR3-1333 memory. Well, if someone is looking for a small midrange fast server box also utilising GPGPU AND maybe a fast SAS RAID card (PCIe 2.0), one will be better with X58-Tylersburg platform than 3410 Ibex-Peak. But now look at Intels harassment: High end still is top, also in price, low end (Lynnfield) outperforms technically (even IMC) midrange, but midrange still offers advantages, but why spending the same money for a W3550 XEON if someone can obtain more memory (32 GB instead of 24 GB, even if 32 GB is slower) and a more sophisticated IMC? Well, it may be looking to sharp on some technical details, but when XEON X5570 was introduced Q3/2009, this is the same time when W3550 at D0 stepping was introduced, the modern IMC technology has already been market-ready and is now also realised in Lynnfield. Why do we not see LGA1366 mid-range XEON CPUs with all these benefits? The modular design should simply allow those changes. I still prefere the LGA1366 three-channel memory solution because I definitely do have advantages but I'm not willing to spend more than 1000 EUROs for a W/X55XX CPU nor want I spend money for a outdated 'midrange' CPU. And speaking of the NOW, Lynnfield does have some disadvantages, so it is not an alternative to a LGA1366 platform.
I do not understand this policy, there is no clean and straight forward thread.
Thanks for the reference! Alas, their choice of oc speed is way too
low to be useful in this regard. I would have though at least 3.8 to
be a sensible target, but I guess they wanted a clock that all the
CPUs including the Ph2 could use. What I'd like to know is, assuming
with a good air cooler both the 860 and 920 can run at 4GHz stable,
which is faster for tasks that do use all 4 cores? (ie. Turbo off)
Also, given the use of a good air cooler, what's the highest sensible
oc each CPU can attain? From all I've read, the 920 D0 should be able
to reach higher oc levels (4.2 to 4.3, vs. around 4.0 for the 860?),
but on the other hand the larger power consumption increase with the
920 might mean the 860 is a better buy if the performance is within a
reasonable range. Atm there's no data for this.
Heh, if I was rich I'd buy both systems and report the differences. :D
Check the other article Anand just posted, seems that with both at 3.8GHz and almost everything equal (except uncore) the Bloomfield is ~3% faster on average. Which doesn't address the question of max OC, but as that is always going to vary by individual processor anyway, they would have to get enough processors to try and get an average for each type.
Ian - No problem. The new i5 and i7 CPUs have just *barely* been out long enough for standard reviews to be published. I'm actually surprised that we got non-turbo "apples-to-apples" benchmarks as it stands. I reckon that we'll see overclocking articles and benchmarks on them soon enough and that your questions will be answered.
I've been searching around for updated release information on SATA-3. According to articles on several sites on the web posted around January of this year, SATA-3 was supposed to be available mid-year. It's not out yet, or if it is I can't find any companies/motherboards that support it.
I'd really like to see an examination of Lynnfield vs. Bloomfield performance using SATA-3, if/when it becomes available on either or both platforms.
No. Look at how virtually any MP-aware OS has done scheduling for decades: try to spread the load equitably over CPU's. That yields the best absolute throughput, and all other things being equal, it always will.
But things are not so simple today. The equation is more complex, as throughput/power scaling is non-linear as it tends to come in quantum steps--especially the first step (CPU off vs. CPU on even at minimal clock rate).
Only relatively recently have OS's considered throughput/watt--or more specifically, the relationship (often non-linear) introduced by systems that can dynamically enable/disable processing elements--as a factor in the scheduler's decisions.
In short: "Grouping threads" has nothing to do with whether the OS "sucks at multithreading" (all decent MP-aware OS's are nominally equivalent), but whether it accounts for power/throughput in its scheduling decisions.
I know the likelihood of anyone reading this post this deep in the comments are slim and the likelihood of it doing any good are even slimmer but I have to point this out.
Power Consumption benchmarks are only useful for telling us how much cooling we'll need for a given chip. In which case telling us total system power consumption is rather useless. If that is the intent here, some thought should be put into figuring out a way of at the very least accurately estimating power consumption of the chip.
If the reason for showing system power consumption is to give us some idea of how much it'll cost us to run that particular setup (or how much damage you're doing to the environment if you're of that mindset), then the benchmarks really should be measuring energy consumption. It's like the old days when marketing was done based on CPU frequency, that was only telling us half the picture.
Idle benchmarks won't differ at all between power and energy consumption measurements, but everything off-idle WILL differ. If you tell us that CPU1 uses 100W while at full load converting a video and CPU2 uses 120W it makes people think that CPU1 is 20% more efficient, but that may not necessarily be the case. If CPU1 takes 12 minutes to convert that video and CPU2 takes only 10, then they're equally efficient because they have used the same amount of energy to perform the same task.
What should be done with these non-idle benchmarks is to perform a specific task and measure how much energy was consumed in the process. Report it in watt-seconds or kilowatt-hours or however you want, but just report it in units of energy.
Also, while we're discussing changing to energy consumption, it should also be noted that some of what we do on computers takes a fixed amount of time regardless of computer power (e.g. watching videos/movies, something that almost everyone does) so doing a full-load test on something of that nature would be impossible/useless. But it does present the interesting scenario that, for example playing a Blu-Ray movie, will create varying levels of load (and in turn power usage) on CPUs of different performance and thus despite the fact that it'll always run for the same amount of time there will be a reason to measure the energy consumption of such a scenario.
Strongly supported. OddTSi makes a good case which I don't think is countered by GourdFM's answer. Users can be educated on this issue, methinks. It will make a better basis for making decisions regarding green computing.
I certainly understand your point of view, but most desktop users will not be concerned with task-oriented energy usage. For nearly all users you will either be in an idle situation (office work/web browsing) or a load situation (gaming)[1] for almost your entire computing session. Normal users don't sit there waiting for a task to complete and then immediately turn off the computer. Even for task oriented computing (e.g. programming, video encoding, rendering) where the user will be waiting for a task to complete before doing more useful work, the reality is that the user will be in one of those two states for the majority of their time over the year. In that sense knowing idle and load power draw is enough for most people.
If you were concerned about energy efficiency of distributed computing (e.g. folding@home) your interest with task-oriented energy efficiency would be a valid concern when making purchasing decisions for infrastructure. If you are so concerned about completing the most tasks for the least amount of money on the desktop, your expenses are more likely to be dominated by salaries than energy expenditures or hardware costs. YMMV.
The cost of air conditioning can also complicate the situtation beyond naive energy usage of the computer system. If all you care about is energy usage do you pick a system that will complete the task using 1000 Wh in 4 hours or one that will complete the task using 1005 Wh in 6 hours? If it's a hot summer afternoon and you're in Death Valley you are going to expend a lot more than 5 Wh cooling the room with the former system than the latter. If it's a chilly winter's morning and you're in Siberia the first system will look a lot more attractive.
---
Footnote [1]: Actually gaming isn't so much an ideal load situation anymore since multi-core processors became common. That may change as developers become more adept at exploiting the power of multiple cores.
> Report it in watt-seconds or kilowatt-hours or however you want, but just report it in units of energy.
In other words, you want a measurement in joules for the amount of energy consumed for a task. Techreport.com has such a measurement in their CPU reviews, at least for the Cinebench task.
Call me crazy, but,
85Watts of idle power consumption is crazy!
That's like my Core2Duo Laptop processing @ full power,while transmitting and receiving Wifi @ full speed and range, and HD fully active doing random R/W's; all while the screen's set to maximum brightness.
That's 85 Watts!
While a Corei7 does not seem to do anything at that power consumption.
It, all by itself needs all that power for staying alive! That's just plain crazy!
I asked a similar question in one of the other articles, so pardon me if this sounds repetitive.
According to the Turbo charts, the slowest Turbo speed is higher than the stock speed. Why is that? For example, why not just make the 750 a stock GHz of 2.8 GHz instead of 2.66GHz?
> According to the Turbo charts, the slowest Turbo speed is higher than the stock speed. Why is that? For example, why not just make the 750 a stock GHz of 2.8 GHz instead of 2.66GHz?
To minimize power consumption when the core is active (C0 or C1 states) but under less than full load. (There are also other ways to manage it via, e.g., P-states, S-states, etc. but that's another subject.)
Requisite bad car analogy :) You want the lowest idle RPM that is smooth and responsive; anything more wastes resources and energy.
I have a I860 right here.
I checked with CPU-Z.
Even if i run indigorenderer with 8 threads for an hour, it runs with x22.
So either i am the victim of a lucky mixup at intels chip packaging plant and got an 870, or something lese is wrong... (or right. As its runing perfectly fine)
I'm guessing it's because if you turn turbo OFF, it would be 2.66.
I'm wondering though, if you need 2 cores on the 860, does it shut off 2 cores and use 2 physical ones or does it shut off 3 cores and use 1 physical plus 1 HT core?
Despite your love affair for this chip, it's a solution in search of a problem.
It's clearly inferior to the Bloomfield. Despite running at higher clock speeds, sometimes a lot, it actually loses to the i7 920. Overclock them, and there's just no comparison. The i7 920 is better. No one with any knowledge of computers would buy the i7 860. They'd get the real deal, the i7 920. This pertains even moreso for the i7 870. Basically, the Lynnfield is an idiot's procesesor, except for the i5 750.
If you can't afford a Bloomfield, that's really your best choice. Except, like I mentioned in a previous post, this is a Celeron, without the platform. If they coupled this with an IGP, you'd have something that would sell. The i5 750 is still not without appeal with a discrete card, but, then, most of the market likes IGPs. And if you know something, and have some money, you're not going to get the brain-damaged Lynnfield. You'll get the Bloomfield.
It's not a mystery, really. The mystery is why it would even sell marginally well. I think once they couple it with a decent IGP, it will really take off though. Until then, I think they'll be lucky to settle for mediocrity.
The Athlon stole the show. No one needs a brain-damaged version of a better chip, unless it breaks into a new market with price. Arguably the i5 750 did, kind of. Clearly the Athlon did. I think that's going to generate more real excitement, if less motherboard pictorals.
Incorrect, there are already benchmarks out there where the i5 720, i7 870, and i7 920 are all underclocked to 2.66ghz (the speed of the i7 920) and overclocked to 3.2ghz (a very attainable turbo speed by the i7 870 and i5 720). The difference in nearly all gaming benchmarks, using settings that takes the GPU out of the mix as a bottleneck, all at the same clock speeds, are within a very very tight percentage range, at the very most a 10% spread (with Crysis and Far Cry 2 it is closer to a 1% spread).
I think the lower price of motherboards makes the LGA 1156 CPUs very very attractive. I don't see any reason to get an LGA 1366 board unless you really want to futureproof yourself for six and eight core CPUs. That said, I don't see a point; most games still use single, maybe dual cores, and upgrading CPUs within a motherboard cycle almost never happens for me. By the time it is time to put together a new PC (average every two years for me) there is an entirely new ecosystem of CPUs, motherboards, and RAM that I need to get into and I end up keeping almost nothing from the old rig.
Wouldn't it serve you better by creating your own website, instead of attacking Anand's articles and playing second fiddle here? You seem to think you know better than Anand does, so why don't you enlighten the rest of us with your better/correct knowledge at your own site? Or has that been done already?
I'm currently in the process of putting together a new system and the choice between the i5 750 and the i7 860 is very hard. The 920 isn't even an option for me; The 860 outperforms the 920 in most scenarios, and when the 920 comes ahead, it's less than the margin of error.
Are you looking at the same graphs I am? The ones that show the 860 performing better, or at worst identically, to the 920 over and over? For the same CPU price and lower motherboard price? This isn't a Celeron. This isn't something you pair up with an IGP. This is the current generations upper-end bang-for-the-buck champion.
Add in the 750, and I see no reason to get a 920. Two-thirds the price, with most of the benchmarks showing performance parity, sometimes a little less. Again, with a lower motherboard cost as well.
If anything, the 920 is the solution searching for a problem. The 860 just took over the job of the 920, except for a few cases. The 750 offers almost the same performance, but at an even lower cost.
These chips aren't brain-damaged; They just took out the cancerous tumors. Lower cost, lower power, and equal performance. The 860 stole the 920's thunder, and the 750 gives us a very complete mid-range.
Ok, let's do your way.
In the magic lands of Ecualia, there was a magic blacksmith cyclope, he forged the Core i7 860 such way it would be better than BloomField.
the story is even worst than that:
somewhere at satan clara's intel's HQ the evil orcs at intel decided that free overclocking was not profitable so they decided to create the turbo crap story, and call it a 'feature' so this way they would use all available overclocking headroom in the cpu and begin charging for it.
so they are killing free overclocking for intel users.
now you have to pay for overclocking because lynnfield turbo overclocking is consuming all the overclocking ability of the cpu
now intel can make this turbo overclocking more aggressive in some cpus and charge more for it.
but there is another problem because intel's orcs betrayed the evil nvidia warlock who sweared revenge and now intel cpus are underperforming when paired with an nvida graphics card.
Not really... Just look at the numbers. The i5 750 is outperforming the PII 965 in almost every case. And when the 965 pulls ahead, it's not enough to justify the increase in price and huge increase in power draw.
> Not really... Just look at the numbers. The i5 750 is outperforming the PII 965 in almost every case
... when given room to Turbo. When not given room to Turbo, then the winner is...? Based on the limited number of Turbo off benchmarks available, my theory is that the i5 750 will fall below the 965 with no room to Turbo. I'm not completely convinced yet that the i5 750 is the definitive winner over the 965. CPU comparison's aren't that clearcut anymore with the variableness of Turbo mode.
On a side note, I do agree the price tag of the 965 is certainly too high, even if the 965 is hypothetically better than the 750 under heavy load. Maybe AMD is fighting the 750 with the slightly lower priced PII 955, which anyone can multiplier bump into a 965.
How many of you trolls are there? Turbo is on by default, so the end user is going to have the advantage of it all the time. The only reason benchmarks were run without Turbo is for the three of you out there who are going to turn it off... no wait, you're not even going to buy it in the first place because you're just here to troll.
"No one with any knowledge of computers would buy the i7 860. They'd get the real deal, the i7 920." Unfortunately, there are 30W that beg to differ. No one with a sane mind would pass over the 860 so easily.
I'd say the 750 and 860 are both sweet spots, but for different budgets. They both are amazing performers for their price segment. After motherboard prices come down a tad more, there will be a pretty big gap between the 1156 and 1366. I really don't see the 920 lasting much longer in that kind of situation. Even the 940 is a little less attractive because the performance gains for the amount spent are really lacking when you go above the 870.
this benchmarks were taken with turbo overclocking on, so the lynnfield is overclocked at least 600 mhz, is illegal to say these are stock speed results, and compare with phenom 2 at stock speed.
is unfair and biased.
These are stock results in terms of this is how the processor comes as stock - with turbo enabled.
If they were overclocking the CPU outside of what is warrantied and allowed by Intel, THEN it would be unfair, but if the CPU is sold with the capability available and enabled to overclock itself, then it is not cheating or "illegal" to say that it's stock.
If you really want to be amused, then feel free to go back to the 9800XT days of ATI, who are now owned by AMD.
Back in those days, the Radeon 9800XT (made by the now AMD owned ATI) used to overclock itself, from a base clock of 412MHz up to 440MHz if possible.
ATI (now owned by AMD) have already participated in this "illegal" automatic overclocking 'war' and now you say it's biased when Intel use a clever technology to improve performance.
Personally I think it's a great feature, although what should really be done is an examination of its usefulness.
Take some i5/i7 systems, put them in regular cases with stock and aftermarket heatsinks on them, and alter the environment in which they are used to see how good the turbo feature is when it's not a (presumably) open lab environment such as seen at Anandtech.
That sort of suggestion from someone who claims "illegal" benchmark results might be more helpful than claiming it's "illegal" or "unfair".
Is it illegal or unfair to benchmark an ATI card with DX10.1 and an NV card with only DX10 if the DX10.1 codepath in a game does nothing more than improve AA performance? No, it's not, it's taking advantage of a feature that only one side has implemented. To take away from what that side has done would be stupid. Deliberately crippling someone to prevent their potential from being show is stupid. Maybe we should put the best basketball players in wheelchairs so they can't perform as well as normal?
You kinda wonder when he will realize he's wasting all his time and attention on a frakking computer chip. "illegal"? Illegal with regards to what law?
In the zorro's world, processors can only be a fixed speed. Since Intel's newest dynamically changes he thinks it's somehow cheating.
I wonder if the zorro thinks mobile CPU power saving features are cheating when they throttle down to save energy?
Just because a CPU can now change thier speed based on needs does not mean it's cheating. Better get used to it as this is the future of multicore CPUs.
You already made a similar comment in the other article. It's not overclocked, you're a moron. It's the same as saying that it underclocks as it uses more cores. That's the stock speed, which varies. Overclocking occurs when users modify the speeds beyond the specifications.
I don't be too much of a PITA, but surely the 'real' overclocking frequencies for the 860 should be of the 21*BCLK variety, no? Because it's stock rated speed is 21*133, its overclocked-at-stock-voltage-with-turbo speed should be 21*150 instead of 22*150. Looks to me as there's some kind of turbo going on on the CPUZ screenshots. Or am I missing something?
It is hardly surprising they are not selling well.
Because, Nehalem didn't make as Big Jump as C2D in performance.
It is expensive, not only the chip itself. But the platform. Pentium 4 to C2D doesn't require new memory. And in some cases doesn't require new Motherboard as well.
Athlon Quad Core is Cheap. Consumers cares about Cores, not threads.
Economy doesn't allow to spend money upgrading on what is already working perfectly.
SSD offer much better value for money in terms of upgrade and investment.
No Integrated Graphics for Nehalem yet. ( Money need to spend on Graphics. )
Personally i am waiting for Sandy Bridge ( or even Ivy Bridge for FMA )
It doesn't?
My socket 478 P4 3.0C would like to disagree with you, considering that it used DDR ram. Practically all of the later P4's were s775, using DDR2 boards.
I wanted to upgrade - I bought a new board and new ram to go along with it.
1. new proc
2. new mobo
3. new memory
4. new gpu
5. new psu
You were required to pick up 1-5 new parts. When I upgraded, I needed all 5.
Stating about saving money, then saying an SSD offers much better value for money in terms of upgrade and investment. WTF. As of right now, it's probably the worst choice, unless you already have the best parts available.
I think P55 may be the problem - without USB 3.0, PCIe 3.0 SATA 6gbs
If all were included it would be a must have - even 1 of the 3 would make it very tempting.
Perhaps they could release them like "Draft n"... That may get sales moving.
I'm not so sure this is fact. Other rumors say, Intel aims for 1 million shipped chips by the end of 2009. Thats not exactly shabby, bad or "not so good". Intel, historically, has a pretty good grasp of what is possible due to their very close ties with their partners.
Then again, this could also be just a rumor without much truth to it. We'll have to see whos right.
But I already know that you're right with some of what you said. I was not impressed with i7 (LGA1366). I was waiting for i5/i7 (LGA1156). And while I think it can be a good product for a relatively fair price (depending on which model you take), I cant justify ditching my Q6600 for it just yet. The best reason would be power consumption and turbo.
Instead, I went with a SSD and got more improvement out of that than any other upgrade could possibly deliver.
The 32nm parts may be interesting again. If not, I'll just wait for sandy bridge. Or maybe even AMDs bulldozer. My current system will take me there easily...
Well, it will defiantly sell well in terms of OEM market. Since they will sell the same amount of PC, and Intel will be pushing Lynfield into their throat anyway.
I wont even called that Sales, it is more like tax on those OEM makers.
I think it wont sell well in terms of Retail market.
Yes. It is Ivy Beidge ( the shrink of Sandy Bridge ) for FMA.
It was supposed to be for Sandy Bridge, but some changes delay it to Ivy Bridge. So unless they have postponed it AGAIN. it should be out with Ivy Bridge.
I know the basic archetecture of Lynnfield is superior to Bloomfield, but you are not using 6gb of tri-channel memory for the i920. That is where the i920 really shines. Is there a reason that you are not testing with 6gb of ram with the 920 other than apples to apples testing that needs to be done? Just curious.
lynnfield has another gigantic bottleneck which is the dmi bus speed of only 2GBps, phenom 2 hypertransport speed is 41.6 GB/s which means that is 20 times faster than lynnfield when communicating with the chipset,that shows why phenom 2 is better than lynnfield. this is going to be a real problem in the next future.
You realize that in EVERY benchmark, the i7 860 was running at higher clock speeds than the i7 920. Sometimes by a lot, with turbo mode. Also, Anand uses inferior memory for the i7 920, to try to 'prove' the validity of the brain-damaged P55 platform.
Despite his bad attempt, the i7 920 STILL outperformed it. If you clock them at the same rate, with the same uncore, it's only ugly for the Lynnfield.
It's not superior. Well, in performance. It's got nice power characteristics, and it's cheaper to implement. But, your remark is purely idiotic.
Lynnfields are categorizes as mainstream, therefore no matter how advanced their architecture is compared to Bloomfield, it won't out perform it. Intel purposely did this and you should know this by now, but I guess you don't care as your too busy sucking your own c0ck.
"Your the kind of man that can be use as a blueprint to build a idiot".
Sorry. I didn't mean to say anything that would warrant an "idiot" label. I have just been reading from Anand that the Lynnfield core is better than the Bloomfield. In all his testing he never uses 6gb of ram in his tests. I understand that he wants to measure the cpus on a level playing field, but when you put 6gb of pc1600 ram on an i920 those scores increase considerably. From what I understand that is something that the Lynnfield cannot achieve. Was just wondering if Anand could throw that (i920 w/6gb) into his charts. It seems that almost all Bloomfield owners are gonna be running 6gb (3x 2gb) and not 4gb. The i920 uses that extra bandwidth and it truly performs better when so equipped. I hope that makes more sense.
Anand does indeed use 1066 RAm for the i920 and 1333 for the i860/870/750, but those are the numbers intel specifies them at as you can see on intel.com, now the problem is that people don't use it with such RAM but with 1600 or higher, and also that if you compare the 2 it would be nice of you also compared them with at least equal speed RAM, and when you would use setup that the average guy that builds his own system then you'd get completely different results.
So I guess anand is reviewing for businesses who take what they get on the cheap, pre-made systems with 'stock speed' RAM
"If you clock them at the same rate, with the same uncore, it's only ugly for the Lynnfield. "
That is not the case. Actually, overclocking uncore speeds makes very little if any difference unless your application of choice is SuperPi or one of the older 3DMark benches where you can realize some measurable differences (wow, look I gained 0.84 seconds in SuperPI 32M and 27 points in 3DMark06) in the results.
Anyway, Lynnfield will clock both uncore and memory significantly higher than Bloomfield so I have a hard time understanding your comments about this subject. If you do not like Lynnfield, that is fine, but the continued comments about inferior memory being used or uncore rates or turbo modes (which the 920 is running by the way and is inferior to the 860s turbo mode) is really wearing thin at this point.
I already showed the results with DDR3-1600 C7 and nothing changes at DDR3-2000 C7, except VTT/VDimm is much lower on Lynnfield than Bloomfield for equal memory clocks.
For a daily platform, I would take the P55/860 over the X58/920 any day of the week. It simply performs better in most cases and uses significantly less power to do it. In fact, if based on just SOHO computer usage and typical gaming scenarios, I would take the 790FX/965BE over the X58/920. For just a SOHO non-gaming system, I am going 785G/Athlon II all the way.
For benchmarking, I will go with the X58 platform, but even then you have to ask yourself why. Unless you have been provided with cherry components and have a limitless supply of LN2 for setting records, there is not much point in using this platform now unless you are in the workstation arena where future processor upgrades will make a difference.
The only other advantage is in multi-gpu gaming, where the X58 will make a difference in the benchmarks. However, you will not notice the difference between the two platforms in blind A/B gaming comparisons. I know, I tried it on a few users who own X58 platforms. ;) Once again, the vast majority of gamers do not run CF/SLI so even this small advantage is a moot point for most, especially when you consider the performance of the upcoming GPU releases.
The X58/920 consumes significantly more power and performs about the same if not worse at times than the P55/860. So, unless you have two very specific needs for X58, there are great alternatives available from both Intel and AMD. That is the crux of our message in these articles.
Gary,
you have mentioned a few times that Lynnfield uses significantly less power, and obviously at stock settings that is true. But if additional voltage is needed for overclocking, that advantage would seem to disappear. Do you have any tests done to show both platforms at higher clocks? Or how about both at their maximum clock with turbo still enabled, to see if you can save some power while still getting a similar end clockspeed when needed?
We will have an article shortly on power consumption (static) with Lynnfield and Bloomfield overclocked. Just to cut to the chase, Lynnfield wins by a clear amount, even with slightly higher voltages on the CPU.
Would you care to prove any of this? It looks to me on your own bad benchmarks, you run the Lynnfield at higher voltages and higher uncore.
Where is your proof the Lynnfield can clock the uncore higher? Why did you clock them differently then? Where is your data supporting this?
Your remark about turbo was beneath even you. I never complained about turbo modes, except in the context that people here will overclock, and make them irrelevant. I have said repeatedly, against the bashers, it's a nice technology. But, it doesn't show the architectural differences, which a lot of us were curious about. Did you forget you are supposed to be 'tech' site, not PC Magazine?
I agree with you about power. I've mentioned that a few times. I really like the lower power of the Lynnfield. It's a big advantage. I also would rather use an Athlon based machine for a daily machine, especially with the IGP. I've stated all this. But, for my compiles, I'd rather have an overclocked 920 than a brain-damaged Lynnfield. The same for games.
I'm not too crazy about Anand's benchmarks with the uncore of the Bloomfield running faster. Did you see how there were real differences in the results? 3.5% is not a little when you've only changed the CPU. Part of that is because of the uncore, so it's not really fair to the Lynnfield. I'm glad to finally see you guys showing a difference, but, really, it would have been better with uncore at the same rate.
The memory, it would appear, he's using for the Bloomfield is 1066, that's inferior to 1333.
I would be curious about your video claims. They might be right, but what would be really interesting to try to identical cards, except for memory. My guess is, the inferior setup of the Lynnfield would manifest itself more with smaller memory cards, since you'd have to use PCIe more. I could be wrong, of course, it's just theory, but it would be interesting.
Maybe you guys should stop trying to twist benchmarks to make your point, and just run them to give information. That's the crux of my irritation. The Lynnfield is a good product for a lot of people, and I really like what AMD just did. But, when you do little horsecrap things to manipulate results to illustrate your opinion, that's just wrong. You're not nearly smart enough to think for everyone else; no one person is.
Considering most of it is in the existing reviews, how about you go back and read them for yourself? And then take his word for it. Or don't, it's your choice, and your loss if you don't.
No offense, but you're clearly an idiot. And that is clearly demonstrated by this comment, "Anand uses inferior memory for the i7 920, to try to 'prove' the validity of the brain-damaged P55 platform."
What was inferior about the Patriot Viper memory? Care to explain why? I personally think you can't defend that comment on any front, esp. considering it has been shown in testing to be an excellent overclocking selection for DDR3 memory, esp. for the price. It'll do 1600 speeds, albeit with using a CAS 8 setting, but for what it costs, it's excellent stuff.
Grow up, and when you finally move out of your parent's house, maybe you'll be mature enough to lose the teenage "I know-it-all" attitude you possess right now.
In previous tests Anand has used the fastest validated speed for each platform, so 1066 for Bloomfield and 1333 for Lynnfield. Not that it probably makes much difference in anything but synthetic benchmarks.
Anand, can't you ban this guy? You have to be tired of watching him come on here and verbally assault any person he doesn't agree with, including yourself. Do everyone a favor and toss him to the curb.
Of course the 860 ran at higher clock speeds, why would Anand underclock it? It was compared to the 920 because they share the same price point. That is until you add the motherboard, then the i7 is like 100$ more expensive.
And Anand summed it up nicely: the 1366 platform is now for people who need hexa-cores someday or who think SLI/Crossfire is reasonable.
Oh, and if the P55 Platform is "braindamaged", then apparently all major tech sites are in on the conspiracy.
I don't have a problem with the 860 running at higher clock speeds, but if the architecture were better, it would never lose to a processor running at lower clock speeds.
In short, the architecture is not clearly better. It's worse, the margin is the only thing worth discussing.
The basic architecture of Lynnfield is the same as Bloomfield. The differences are the topology of the platform (PCI on die instead of in the chipset, 2 mem controllers instead of 3, no QPI in Lynnfield). The cores are exactly the same, the cache is exactly the same.
I have yet to see any real world scenarios where triple channel memory "really shines." The inclusion or exclusion of a triple channel set up would account for variations of about 1% either way. In other words, less than the margin of error.
Most of the US doesn't live reasonable close to the 21 cities that have microcenters. In cause you haven't looked there really aren't that many locations so "on-going, still valid sale prices you can get today", really doesn't apply to the vast majority of the people in the country.
It is a valid price, but please add the disclaimer "If you have a microcenter nearby" instead of just pretending those are widely available prices. I plan on buying a processor when I help my parents move down near Philly next month, otherwise I (like most) don't have a Microcenter anywhere nearby.
Yeah with only 2% of the people able to get one at near that price. Quoting an obvious loss leader as valid pricing for those looking it nutty. Newegg or ZZF is a much better gauge of price.
Who cares, let him pay up. Nothing on P55 has made me regret getting that $200 D0 920. Nothing. Not even close. The OC, heat, and platform pricing advantages all failed to materialize.
The 920 is not a 2.66Mhz bloomfield. It is a 3.8GHz chip supporting the fullest featured consumer platform at the moment.
For online purchases, you may still have to factor in sales tax and shipping. There's also the cost of having to wait.
Between your current computer and the upgrade, what is the performance gap? For that gap, how much time could have been saved in your work? Because you saved time at work, how much $$ was saved?
Things can become complicated or we can just stick with the retail pricing at these stores and leave out the other factors.
The I7 860 is $290 at newegg with free shiping and no fees of any kind.In microcenter you have $230 + $20 tax + gas .The point is the differance is less than $40 with microcenter .You should always compare final price for the purpose of being fair. How long have you been waiting for I7 860 cpu to come out ? How much have you lost for that time ?So if you have been waiting 8 months how is 3 days going to be bad?If that is so bad how come you din't get 920 before?
But how many LOCs per fortnight can they process? Could you please provide your metrics in more understandable formats. At the very least I would like to see how many TuxRacer compiles per hogshead of cider (the good stuff, not the end-of-season stuff) we can expect.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
121 Comments
Back to Article
papapapapapapapababy - Monday, September 21, 2009 - link
THE stupid upgrading path.NO SATA 6 GB /s...
NO USB 3.0...
NO PCI Express 3.0...
NO THANKS. NEXT.
haplo602 - Monday, September 21, 2009 - link
Where are Athlon II X4s in the graphs ? Where's Phenom II X2 BE ?Where is a 785G mobo roundup ? I am still hearing only i5/i7/P55. This is frustrating. You are not keeping up with your name. Drop the Intel hype and do something for the normal people that try to build computers on a budget.
sicofante - Monday, September 21, 2009 - link
Sorry if this has been asked or commented before (I haven't read the full 11 pages).I build workstations for the animation and video industry and I factory-overclock them. Bloomfield has been very well received by my customers and I'm really happy with it. Now I'm studying Lynnfield and from what I've read, I don't feel quite comfortable with how Lynnfield is overclocked. Here are my two main issues:
1. Anand mentioned in past articles that overclocking Lynnfield would imply overclocking the PCIe bus, since the controller is integrated. How does this affect graphics and other cards? I'm not talking only about gaming cards but also Quadros or RAID controller cards.
2. Also, it seems Lynnfield OC needs voltage tweaking. This sounds not as nice as Bloomfield stock voltage overclocking, but what are the real consequences and drawbacks (if any) of voltage rising?
Thanks in advance for answering these two issues and thanks to the Anandtech's staff for such in depth articles.
ggathagan - Tuesday, September 22, 2009 - link
This seems to get lost frequently, but Lynnfield is all about the mid-tier market.Going down the Lynnfield road for the workstations you describe would be a BAAAAD idea.
Your industry is also one of the few that, in most reviewers opinions, has benefited from the triple-channel memory capabilities of the X58 platform.
7Enigma - Monday, September 21, 2009 - link
It's been going on for a while now but the price increase for minimal performance increase is getting pretty silly. Back during the <1000MHz days a 100MHz bump was nothing to sneeze at, and even during the 1-2GHz days a 100-200MHz bump wasn't that bad. But honestly they (both AMD and INTEL) have gotten rediculous with their gouging of the higher end. Honestly ~130MHz difference between the 870 and the 860?!?Their only saving grace (for stock clockers, or very moderate OC'ers) is the higher turbo levels of the 870, but again in most situations (that is those that do not task 3-4 cores simultaneously) the clock difference between the 860 and 870 is <150MHz, which on a ~3.5GHz core is virtually nil.
strikeback03 - Monday, September 21, 2009 - link
They have used pricing like this in the past, just that there were usually several options in the more sane ($300 and less) range before the final few clock bumps were disproportionately expensive.7Enigma - Monday, September 21, 2009 - link
But in the past the performance jump was greater. We're talking a (theoretical) 4% difference between the 870 and 860!strikeback03 - Monday, September 21, 2009 - link
When I bought my current E6600 it was a little over $300. The E6700 was over $500 for a 266MHz bump, so technically 10% but still nothing to write home about for the money. If the 870 were unlocked that price might be justified, as of now I agree with you that it is not so much.Proteusza - Monday, September 21, 2009 - link
Athlon X4: 300m transistors, no L3 cache, performs about 90% as fast as a Phenom X4Phenom X4: 758m transistors, 8mb L3 cache (or is it 6?)
Does anyone think AMD isnt getting their money's worth out of the 458 million transistors used on the Phenom II to provide the L3 cache? I mean, more than double the manufacturing complexity for a small increase in performance?
mapesdhs - Sunday, September 20, 2009 - link
Anand, the review shows overclocking results for the i7 860 with the
retail cooler, but I doubt anyone who intends to oc these CPUs will
stick with the retail cooler.
There is much comment on how the 920 compares to the 860. If I understand
correctly from earlier articles, the 920 should have an advantage
vs. the 860 if both are hard oc'd (ie. Turbo off) to aid in those
situations where the user is definitely going to be used all 4 cores,
in my case video encoding.
Thus, what I'd really like to know is, how does the 920 compare
to the 860 when both are oc'd using a _good_ air cooler such as the
Thermalright U120E or Cogage TRUE Spirit? And is there any real
difference in the maximum achievable overclocks when each chip if
oc'd using a good cooler? (only referring to air cooling here)
Thus, two questions, both aimed at tasks which do use all 4 cores
(video encoding, rendering, etc.):
1. If the 860 and 920 are oc'd to the same clock, presumably with
Turbo off being the most sensible setting, which is faster? How
does power consumption differ? (wrt to total power used for a given
task)
2. How high can each chip be oc'd using a good cooler? Does the 920
have a usefully higher limit? What difference does it make to the
test results?
I've still not found a site which has done this comparison. For me,
the comparison data of 860 vs. 920 at stock speeds with Turbo active
is interesting, but not useful.
And of course it would be most intriguing to see how the other
CPUs fit into this scenario when oc'd using a good cooler, ie. the
750, 975 and the various AMDs. Given the very low cost of the entry
AMD quad-core, maybe that would be a more productive platform?
My original train of thought before Lynnfield came out was that an
oc'd 920 for video encoding was worth the extra cost because it
would complete a task much sooner and thus overall use less power.
The results compared to the best AMD CPUs at the time bore this out
nicely. But with the 860 added to the mix, I can't work out whether
the 920 would attain a better oc than the 860 with a good cooler,
and/or be faster anyway when oc'd & Turbo off, and thus be a better
choice. Charts show the 860 often doing better, but that's at stock
speed. Other charts with oc'd results set the CPUs at the same speed
using the retail cooler, whereas I'd be trying to push the 920 as
high as I could within reason using a good cooler.
So, would it be possible please for you to compare the various CPUs
when overclocked to their best sensible extent using a good cooler,
not the supplied retail cooler?
For example, elsewhere I read of people attaining 4.2GHz for the 920
with little difficulty.
Ian.
KoolAidMan1 - Monday, September 21, 2009 - link
Ian - In response to:"1. If the 860 and 920 are oc'd to the same clock, presumably with
Turbo off being the most sensible setting, which is faster? How
does power consumption differ? (wrt to total power used for a given
task)
I've still not found a site which has done this comparison. For me,
the comparison data of 860 vs. 920 at stock speeds with Turbo active
is interesting, but not useful."
It has already been done with the i7 870, a CPU that has shown to be marginally faster than the i7 860 in most tests. What follows are benchmarks where the i5 750, i7 870, and i7 920 are run at the 750/920 common clock speed of 2.66ghz and the very attainable turbo speed of the i7 870 of 3.2ghz, no turbo in all cases.
http://www.hardocp.com/article/2009/09/07/intel_ly...">http://www.hardocp.com/article/2009/09/...ntel_lyn...
The difference in nearly all gaming benchmarks, using settings that takes the GPU out of the mix as a bottleneck, all at the same clock speeds, are within a very very tight percentage range, at the very most a 10% spread. With Crysis and Far Cry 2 it is closer to a 1% spread. Multimedia benchmarks have a similar or slightly larger spread but nothing I would consider significant, and certainly not in all cases.
Most things being the same (at least in games which is what I mainly use my PC for anyways, work and everything else are done on my OS X machines), the LGA 1156 CPUs are more appealing to me due to their lower net cost and comparable performance levels relative to consumer level LGA 1366 CPUs and boards.
Eisenfaust - Monday, September 21, 2009 - link
Some comments. Intel claims having split the market in Highend, Midrange and Lowend. 'Lynnfield' XEONs inherited the IMC from the highend X5570 type DP-XEON CPUs, outperforming the LGA1366 single-socket XEON W35XX CPUs either by the usability of RDIMMs (up to 8GB per module!) and speed (DDR3-1333 oficiallay supported with 1 DIMM per channel). And, not to mention, the higher NB clock of 2.40 GHz instead of 2.13 GHz.Several tests with highly parallel code for numerical modelling showed, that a XEON W3550 outperforms a XEON 3470, IF we lift NB clock to 2.66 GHz as used for the W3580 type AND also use DDR3-1333 memory. Well, if someone is looking for a small midrange fast server box also utilising GPGPU AND maybe a fast SAS RAID card (PCIe 2.0), one will be better with X58-Tylersburg platform than 3410 Ibex-Peak. But now look at Intels harassment: High end still is top, also in price, low end (Lynnfield) outperforms technically (even IMC) midrange, but midrange still offers advantages, but why spending the same money for a W3550 XEON if someone can obtain more memory (32 GB instead of 24 GB, even if 32 GB is slower) and a more sophisticated IMC? Well, it may be looking to sharp on some technical details, but when XEON X5570 was introduced Q3/2009, this is the same time when W3550 at D0 stepping was introduced, the modern IMC technology has already been market-ready and is now also realised in Lynnfield. Why do we not see LGA1366 mid-range XEON CPUs with all these benefits? The modular design should simply allow those changes. I still prefere the LGA1366 three-channel memory solution because I definitely do have advantages but I'm not willing to spend more than 1000 EUROs for a W/X55XX CPU nor want I spend money for a outdated 'midrange' CPU. And speaking of the NOW, Lynnfield does have some disadvantages, so it is not an alternative to a LGA1366 platform.
I do not understand this policy, there is no clean and straight forward thread.
mapesdhs - Monday, September 21, 2009 - link
Thanks for the reference! Alas, their choice of oc speed is way too
low to be useful in this regard. I would have though at least 3.8 to
be a sensible target, but I guess they wanted a clock that all the
CPUs including the Ph2 could use. What I'd like to know is, assuming
with a good air cooler both the 860 and 920 can run at 4GHz stable,
which is faster for tasks that do use all 4 cores? (ie. Turbo off)
Also, given the use of a good air cooler, what's the highest sensible
oc each CPU can attain? From all I've read, the 920 D0 should be able
to reach higher oc levels (4.2 to 4.3, vs. around 4.0 for the 860?),
but on the other hand the larger power consumption increase with the
920 might mean the 860 is a better buy if the performance is within a
reasonable range. Atm there's no data for this.
Heh, if I was rich I'd buy both systems and report the differences. :D
Thanks for replying!!
Ian.
strikeback03 - Monday, September 21, 2009 - link
Check the other article Anand just posted, seems that with both at 3.8GHz and almost everything equal (except uncore) the Bloomfield is ~3% faster on average. Which doesn't address the question of max OC, but as that is always going to vary by individual processor anyway, they would have to get enough processors to try and get an average for each type.KoolAidMan1 - Monday, September 21, 2009 - link
Ian - No problem. The new i5 and i7 CPUs have just *barely* been out long enough for standard reviews to be published. I'm actually surprised that we got non-turbo "apples-to-apples" benchmarks as it stands. I reckon that we'll see overclocking articles and benchmarks on them soon enough and that your questions will be answered.MouseBTFH - Sunday, September 20, 2009 - link
I've been searching around for updated release information on SATA-3. According to articles on several sites on the web posted around January of this year, SATA-3 was supposed to be available mid-year. It's not out yet, or if it is I can't find any companies/motherboards that support it.I'd really like to see an examination of Lynnfield vs. Bloomfield performance using SATA-3, if/when it becomes available on either or both platforms.
Scheme - Monday, September 21, 2009 - link
The Marvel 6Gb controllers had significant problems and mobo manufacturers pulled them from their initial P55 designs.The Barracuda XT announcement explains where they're at right now.
Wwhat - Sunday, September 20, 2009 - link
The w7 does a better job at 'grouping threads' made me laugh, another way to put it would be 'it sucks at multithreading' then eh.has407 - Sunday, September 20, 2009 - link
No. Look at how virtually any MP-aware OS has done scheduling for decades: try to spread the load equitably over CPU's. That yields the best absolute throughput, and all other things being equal, it always will.But things are not so simple today. The equation is more complex, as throughput/power scaling is non-linear as it tends to come in quantum steps--especially the first step (CPU off vs. CPU on even at minimal clock rate).
Only relatively recently have OS's considered throughput/watt--or more specifically, the relationship (often non-linear) introduced by systems that can dynamically enable/disable processing elements--as a factor in the scheduler's decisions.
In short: "Grouping threads" has nothing to do with whether the OS "sucks at multithreading" (all decent MP-aware OS's are nominally equivalent), but whether it accounts for power/throughput in its scheduling decisions.
OddTSi - Saturday, September 19, 2009 - link
I know the likelihood of anyone reading this post this deep in the comments are slim and the likelihood of it doing any good are even slimmer but I have to point this out.Power Consumption benchmarks are only useful for telling us how much cooling we'll need for a given chip. In which case telling us total system power consumption is rather useless. If that is the intent here, some thought should be put into figuring out a way of at the very least accurately estimating power consumption of the chip.
If the reason for showing system power consumption is to give us some idea of how much it'll cost us to run that particular setup (or how much damage you're doing to the environment if you're of that mindset), then the benchmarks really should be measuring energy consumption. It's like the old days when marketing was done based on CPU frequency, that was only telling us half the picture.
Idle benchmarks won't differ at all between power and energy consumption measurements, but everything off-idle WILL differ. If you tell us that CPU1 uses 100W while at full load converting a video and CPU2 uses 120W it makes people think that CPU1 is 20% more efficient, but that may not necessarily be the case. If CPU1 takes 12 minutes to convert that video and CPU2 takes only 10, then they're equally efficient because they have used the same amount of energy to perform the same task.
What should be done with these non-idle benchmarks is to perform a specific task and measure how much energy was consumed in the process. Report it in watt-seconds or kilowatt-hours or however you want, but just report it in units of energy.
Also, while we're discussing changing to energy consumption, it should also be noted that some of what we do on computers takes a fixed amount of time regardless of computer power (e.g. watching videos/movies, something that almost everyone does) so doing a full-load test on something of that nature would be impossible/useless. But it does present the interesting scenario that, for example playing a Blu-Ray movie, will create varying levels of load (and in turn power usage) on CPUs of different performance and thus despite the fact that it'll always run for the same amount of time there will be a reason to measure the energy consumption of such a scenario.
georges1977 - Monday, September 21, 2009 - link
+1khaakon - Sunday, September 20, 2009 - link
Strongly supported. OddTSi makes a good case which I don't think is countered by GourdFM's answer. Users can be educated on this issue, methinks. It will make a better basis for making decisions regarding green computing.GourdFreeMan - Sunday, September 20, 2009 - link
I certainly understand your point of view, but most desktop users will not be concerned with task-oriented energy usage. For nearly all users you will either be in an idle situation (office work/web browsing) or a load situation (gaming)[1] for almost your entire computing session. Normal users don't sit there waiting for a task to complete and then immediately turn off the computer. Even for task oriented computing (e.g. programming, video encoding, rendering) where the user will be waiting for a task to complete before doing more useful work, the reality is that the user will be in one of those two states for the majority of their time over the year. In that sense knowing idle and load power draw is enough for most people.If you were concerned about energy efficiency of distributed computing (e.g. folding@home) your interest with task-oriented energy efficiency would be a valid concern when making purchasing decisions for infrastructure. If you are so concerned about completing the most tasks for the least amount of money on the desktop, your expenses are more likely to be dominated by salaries than energy expenditures or hardware costs. YMMV.
The cost of air conditioning can also complicate the situtation beyond naive energy usage of the computer system. If all you care about is energy usage do you pick a system that will complete the task using 1000 Wh in 4 hours or one that will complete the task using 1005 Wh in 6 hours? If it's a hot summer afternoon and you're in Death Valley you are going to expend a lot more than 5 Wh cooling the room with the former system than the latter. If it's a chilly winter's morning and you're in Siberia the first system will look a lot more attractive.
---
Footnote [1]: Actually gaming isn't so much an ideal load situation anymore since multi-core processors became common. That may change as developers become more adept at exploiting the power of multiple cores.
Eeqmcsq - Saturday, September 19, 2009 - link
> Report it in watt-seconds or kilowatt-hours or however you want, but just report it in units of energy.In other words, you want a measurement in joules for the amount of energy consumed for a task. Techreport.com has such a measurement in their CPU reviews, at least for the Cinebench task.
http://techreport.com/articles.x/17545/13">http://techreport.com/articles.x/17545/13
ProDigit - Saturday, September 19, 2009 - link
Call me crazy, but,85Watts of idle power consumption is crazy!
That's like my Core2Duo Laptop processing @ full power,while transmitting and receiving Wifi @ full speed and range, and HD fully active doing random R/W's; all while the screen's set to maximum brightness.
That's 85 Watts!
While a Corei7 does not seem to do anything at that power consumption.
It, all by itself needs all that power for staying alive! That's just plain crazy!
Anand Lal Shimpi - Saturday, September 19, 2009 - link
Keep in mind there's a much larger power supply and about 1.4 billion transistors of a GeForce GTX 280 idling in that power value as well :)Take care,
Anand
Eeqmcsq - Saturday, September 19, 2009 - link
I asked a similar question in one of the other articles, so pardon me if this sounds repetitive.According to the Turbo charts, the slowest Turbo speed is higher than the stock speed. Why is that? For example, why not just make the 750 a stock GHz of 2.8 GHz instead of 2.66GHz?
has407 - Saturday, September 19, 2009 - link
> According to the Turbo charts, the slowest Turbo speed is higher than the stock speed. Why is that? For example, why not just make the 750 a stock GHz of 2.8 GHz instead of 2.66GHz?To minimize power consumption when the core is active (C0 or C1 states) but under less than full load. (There are also other ways to manage it via, e.g., P-states, S-states, etc. but that's another subject.)
Requisite bad car analogy :) You want the lowest idle RPM that is smooth and responsive; anything more wastes resources and energy.
imsabbel - Sunday, September 20, 2009 - link
Sorry, but it seems nobody is "getting" it.I have a I860 right here.
I checked with CPU-Z.
Even if i run indigorenderer with 8 threads for an hour, it runs with x22.
So either i am the victim of a lucky mixup at intels chip packaging plant and got an 870, or something lese is wrong... (or right. As its runing perfectly fine)
strikeback03 - Monday, September 21, 2009 - link
Crank up the temperature in the case though, and it might drop back to x21.They mentioned previously with the i7 920 that more often than not runs at 2.8GHz
has407 - Sunday, September 20, 2009 - link
No, there's nothing "wrong" except maybe your assumptions or math?22x 133MHz = 2.93GHz. That's exactly what an 860 should be at with 4 cores active under full load.
An 870 would idle at 2.93GHz; with 4 cores active and under full load you should see 24x 133MHz = 3.20GHz.
yacoub - Saturday, September 19, 2009 - link
Go learn about TDP.hulu - Saturday, September 19, 2009 - link
" According to the Turbo charts, the slowest Turbo speed is higher than the stock speed. Why is that? "It's not only how many cores you use but also what instructions are being executed that contributes to whether turbo is used.
TemjinGold - Saturday, September 19, 2009 - link
I'm guessing it's because if you turn turbo OFF, it would be 2.66.I'm wondering though, if you need 2 cores on the 860, does it shut off 2 cores and use 2 physical ones or does it shut off 3 cores and use 1 physical plus 1 HT core?
TA152H - Saturday, September 19, 2009 - link
Despite your love affair for this chip, it's a solution in search of a problem.It's clearly inferior to the Bloomfield. Despite running at higher clock speeds, sometimes a lot, it actually loses to the i7 920. Overclock them, and there's just no comparison. The i7 920 is better. No one with any knowledge of computers would buy the i7 860. They'd get the real deal, the i7 920. This pertains even moreso for the i7 870. Basically, the Lynnfield is an idiot's procesesor, except for the i5 750.
If you can't afford a Bloomfield, that's really your best choice. Except, like I mentioned in a previous post, this is a Celeron, without the platform. If they coupled this with an IGP, you'd have something that would sell. The i5 750 is still not without appeal with a discrete card, but, then, most of the market likes IGPs. And if you know something, and have some money, you're not going to get the brain-damaged Lynnfield. You'll get the Bloomfield.
It's not a mystery, really. The mystery is why it would even sell marginally well. I think once they couple it with a decent IGP, it will really take off though. Until then, I think they'll be lucky to settle for mediocrity.
The Athlon stole the show. No one needs a brain-damaged version of a better chip, unless it breaks into a new market with price. Arguably the i5 750 did, kind of. Clearly the Athlon did. I think that's going to generate more real excitement, if less motherboard pictorals.
KoolAidMan1 - Saturday, September 19, 2009 - link
Incorrect, there are already benchmarks out there where the i5 720, i7 870, and i7 920 are all underclocked to 2.66ghz (the speed of the i7 920) and overclocked to 3.2ghz (a very attainable turbo speed by the i7 870 and i5 720). The difference in nearly all gaming benchmarks, using settings that takes the GPU out of the mix as a bottleneck, all at the same clock speeds, are within a very very tight percentage range, at the very most a 10% spread (with Crysis and Far Cry 2 it is closer to a 1% spread).http://www.hardocp.com/article/2009/09/07/intel_ly...">http://www.hardocp.com/article/2009/09/...ntel_lyn...
I think the lower price of motherboards makes the LGA 1156 CPUs very very attractive. I don't see any reason to get an LGA 1366 board unless you really want to futureproof yourself for six and eight core CPUs. That said, I don't see a point; most games still use single, maybe dual cores, and upgrading CPUs within a motherboard cycle almost never happens for me. By the time it is time to put together a new PC (average every two years for me) there is an entirely new ecosystem of CPUs, motherboards, and RAM that I need to get into and I end up keeping almost nothing from the old rig.
So yeah, I don't really agree with you.
chrnochime - Saturday, September 19, 2009 - link
Wouldn't it serve you better by creating your own website, instead of attacking Anand's articles and playing second fiddle here? You seem to think you know better than Anand does, so why don't you enlighten the rest of us with your better/correct knowledge at your own site? Or has that been done already?Just saying...
jordanclock - Saturday, September 19, 2009 - link
I'm currently in the process of putting together a new system and the choice between the i5 750 and the i7 860 is very hard. The 920 isn't even an option for me; The 860 outperforms the 920 in most scenarios, and when the 920 comes ahead, it's less than the margin of error.Are you looking at the same graphs I am? The ones that show the 860 performing better, or at worst identically, to the 920 over and over? For the same CPU price and lower motherboard price? This isn't a Celeron. This isn't something you pair up with an IGP. This is the current generations upper-end bang-for-the-buck champion.
Add in the 750, and I see no reason to get a 920. Two-thirds the price, with most of the benchmarks showing performance parity, sometimes a little less. Again, with a lower motherboard cost as well.
If anything, the 920 is the solution searching for a problem. The 860 just took over the job of the 920, except for a few cases. The 750 offers almost the same performance, but at an even lower cost.
These chips aren't brain-damaged; They just took out the cancerous tumors. Lower cost, lower power, and equal performance. The 860 stole the 920's thunder, and the 750 gives us a very complete mid-range.
the zorro - Saturday, September 19, 2009 - link
if you go to newegg you can see that lynnfield is not selling, because is crippled, expensive and phenom 2 wipes and mops the floor with core i5 750.KoolAidMan1 - Saturday, September 19, 2009 - link
You are either deluded or trolling. The i5 720 both costs less and outperforms the Phenom II X4 965 BE in pretty much every chart.DigitalFreak - Saturday, September 19, 2009 - link
He's the troll formerly known as snakeoil. Just ignore him and he'll go away.the zorro - Saturday, September 19, 2009 - link
why, did you ran out of arguments?strikeback03 - Monday, September 21, 2009 - link
No, just following this proverb"Do not argue with a fool, those watching may not know the difference between you and he"
DigitalFreak - Saturday, September 19, 2009 - link
Arguing with someone who is clearly delusional is pointless. On the other hand, if you're mentally retarded then you have my sympathy.Troll Trolling - Saturday, September 19, 2009 - link
Ok, let's do your way.In the magic lands of Ecualia, there was a magic blacksmith cyclope, he forged the Core i7 860 such way it would be better than BloomField.
Beat my argument troll.
the zorro - Saturday, September 19, 2009 - link
the story is even worst than that:somewhere at satan clara's intel's HQ the evil orcs at intel decided that free overclocking was not profitable so they decided to create the turbo crap story, and call it a 'feature' so this way they would use all available overclocking headroom in the cpu and begin charging for it.
so they are killing free overclocking for intel users.
now you have to pay for overclocking because lynnfield turbo overclocking is consuming all the overclocking ability of the cpu
now intel can make this turbo overclocking more aggressive in some cpus and charge more for it.
but there is another problem because intel's orcs betrayed the evil nvidia warlock who sweared revenge and now intel cpus are underperforming when paired with an nvida graphics card.
SpaceRanger - Saturday, September 19, 2009 - link
We gotta get a "Do Not Feed The Trolls" sign for the end of every page posted here at Anandtech.jordanclock - Saturday, September 19, 2009 - link
Not really... Just look at the numbers. The i5 750 is outperforming the PII 965 in almost every case. And when the 965 pulls ahead, it's not enough to justify the increase in price and huge increase in power draw.Eeqmcsq - Saturday, September 19, 2009 - link
> Not really... Just look at the numbers. The i5 750 is outperforming the PII 965 in almost every case... when given room to Turbo. When not given room to Turbo, then the winner is...? Based on the limited number of Turbo off benchmarks available, my theory is that the i5 750 will fall below the 965 with no room to Turbo. I'm not completely convinced yet that the i5 750 is the definitive winner over the 965. CPU comparison's aren't that clearcut anymore with the variableness of Turbo mode.
On a side note, I do agree the price tag of the 965 is certainly too high, even if the 965 is hypothetically better than the 750 under heavy load. Maybe AMD is fighting the 750 with the slightly lower priced PII 955, which anyone can multiplier bump into a 965.
yacoub - Saturday, September 19, 2009 - link
How many of you trolls are there? Turbo is on by default, so the end user is going to have the advantage of it all the time. The only reason benchmarks were run without Turbo is for the three of you out there who are going to turn it off... no wait, you're not even going to buy it in the first place because you're just here to troll.vol7ron - Saturday, September 19, 2009 - link
"No one with any knowledge of computers would buy the i7 860. They'd get the real deal, the i7 920." Unfortunately, there are 30W that beg to differ. No one with a sane mind would pass over the 860 so easily.vol7ron - Saturday, September 19, 2009 - link
I want one :) I think the 860 is the sweet spot for price/performancejordanclock - Saturday, September 19, 2009 - link
I'd say the 750 and 860 are both sweet spots, but for different budgets. They both are amazing performers for their price segment. After motherboard prices come down a tad more, there will be a pretty big gap between the 1156 and 1366. I really don't see the 920 lasting much longer in that kind of situation. Even the 940 is a little less attractive because the performance gains for the amount spent are really lacking when you go above the 870.strikeback03 - Monday, September 21, 2009 - link
Has the 940 ever been attractive? Only for those who couldn't or wouldn't OC a 920.the zorro - Saturday, September 19, 2009 - link
this benchmarks were taken with turbo overclocking on, so the lynnfield is overclocked at least 600 mhz, is illegal to say these are stock speed results, and compare with phenom 2 at stock speed.is unfair and biased.
Lonyo - Saturday, September 19, 2009 - link
A new name for snakeoil?These are stock results in terms of this is how the processor comes as stock - with turbo enabled.
If they were overclocking the CPU outside of what is warrantied and allowed by Intel, THEN it would be unfair, but if the CPU is sold with the capability available and enabled to overclock itself, then it is not cheating or "illegal" to say that it's stock.
If you really want to be amused, then feel free to go back to the 9800XT days of ATI, who are now owned by AMD.
Back in those days, the Radeon 9800XT (made by the now AMD owned ATI) used to overclock itself, from a base clock of 412MHz up to 440MHz if possible.
ATI (now owned by AMD) have already participated in this "illegal" automatic overclocking 'war' and now you say it's biased when Intel use a clever technology to improve performance.
Personally I think it's a great feature, although what should really be done is an examination of its usefulness.
Take some i5/i7 systems, put them in regular cases with stock and aftermarket heatsinks on them, and alter the environment in which they are used to see how good the turbo feature is when it's not a (presumably) open lab environment such as seen at Anandtech.
That sort of suggestion from someone who claims "illegal" benchmark results might be more helpful than claiming it's "illegal" or "unfair".
Is it illegal or unfair to benchmark an ATI card with DX10.1 and an NV card with only DX10 if the DX10.1 codepath in a game does nothing more than improve AA performance? No, it's not, it's taking advantage of a feature that only one side has implemented. To take away from what that side has done would be stupid. Deliberately crippling someone to prevent their potential from being show is stupid. Maybe we should put the best basketball players in wheelchairs so they can't perform as well as normal?
Chlorus - Saturday, September 19, 2009 - link
You kinda wonder when he will realize he's wasting all his time and attention on a frakking computer chip. "illegal"? Illegal with regards to what law?the zorro - Saturday, September 19, 2009 - link
the law that says:you will not steal.
Chlorus - Saturday, September 19, 2009 - link
HOW IS THAT FUCKING STEALING!? How is using a stock feature stealing? You are aware that AMD is planning to use the same feature to?DigitalFreak - Saturday, September 19, 2009 - link
If you guys would just ignore him and not reply to his posts, he'll go away.Griswold - Saturday, September 19, 2009 - link
Well, you didnt go away either...coolkev99 - Monday, September 21, 2009 - link
In the zorro's world, processors can only be a fixed speed. Since Intel's newest dynamically changes he thinks it's somehow cheating.I wonder if the zorro thinks mobile CPU power saving features are cheating when they throttle down to save energy?
Just because a CPU can now change thier speed based on needs does not mean it's cheating. Better get used to it as this is the future of multicore CPUs.
vol7ron - Saturday, September 19, 2009 - link
You already made a similar comment in the other article. It's not overclocked, you're a moron. It's the same as saying that it underclocks as it uses more cores. That's the stock speed, which varies. Overclocking occurs when users modify the speeds beyond the specifications.Get over yourself AMD fanboi.
Nich0 - Saturday, September 19, 2009 - link
I don't be too much of a PITA, but surely the 'real' overclocking frequencies for the 860 should be of the 21*BCLK variety, no? Because it's stock rated speed is 21*133, its overclocked-at-stock-voltage-with-turbo speed should be 21*150 instead of 22*150. Looks to me as there's some kind of turbo going on on the CPUZ screenshots. Or am I missing something?iwodo - Saturday, September 19, 2009 - link
It is hardly surprising they are not selling well.Because, Nehalem didn't make as Big Jump as C2D in performance.
It is expensive, not only the chip itself. But the platform. Pentium 4 to C2D doesn't require new memory. And in some cases doesn't require new Motherboard as well.
Athlon Quad Core is Cheap. Consumers cares about Cores, not threads.
Economy doesn't allow to spend money upgrading on what is already working perfectly.
SSD offer much better value for money in terms of upgrade and investment.
No Integrated Graphics for Nehalem yet. ( Money need to spend on Graphics. )
Personally i am waiting for Sandy Bridge ( or even Ivy Bridge for FMA )
zero2dash - Tuesday, September 22, 2009 - link
It doesn't?My socket 478 P4 3.0C would like to disagree with you, considering that it used DDR ram. Practically all of the later P4's were s775, using DDR2 boards.
I wanted to upgrade - I bought a new board and new ram to go along with it.
zero2dash - Tuesday, September 22, 2009 - link
Ack, frickin' quote didn't work.[quote]Pentium 4 to C2D doesn't require new memory.[/quote]
afkrotch - Tuesday, September 22, 2009 - link
The move from P4 to C2D can be huge or small.1. new proc
2. new mobo
3. new memory
4. new gpu
5. new psu
You were required to pick up 1-5 new parts. When I upgraded, I needed all 5.
Stating about saving money, then saying an SSD offers much better value for money in terms of upgrade and investment. WTF. As of right now, it's probably the worst choice, unless you already have the best parts available.
Dobs - Sunday, September 20, 2009 - link
I think P55 may be the problem - without USB 3.0, PCIe 3.0 SATA 6gbsIf all were included it would be a must have - even 1 of the 3 would make it very tempting.
Perhaps they could release them like "Draft n"... That may get sales moving.
Griswold - Saturday, September 19, 2009 - link
I'm not so sure this is fact. Other rumors say, Intel aims for 1 million shipped chips by the end of 2009. Thats not exactly shabby, bad or "not so good". Intel, historically, has a pretty good grasp of what is possible due to their very close ties with their partners.http://www.digitimes.com/news/a20090916PD219.html">http://www.digitimes.com/news/a20090916PD219.html
Then again, this could also be just a rumor without much truth to it. We'll have to see whos right.
But I already know that you're right with some of what you said. I was not impressed with i7 (LGA1366). I was waiting for i5/i7 (LGA1156). And while I think it can be a good product for a relatively fair price (depending on which model you take), I cant justify ditching my Q6600 for it just yet. The best reason would be power consumption and turbo.
Instead, I went with a SSD and got more improvement out of that than any other upgrade could possibly deliver.
The 32nm parts may be interesting again. If not, I'll just wait for sandy bridge. Or maybe even AMDs bulldozer. My current system will take me there easily...
iwodo - Sunday, September 20, 2009 - link
Well, it will defiantly sell well in terms of OEM market. Since they will sell the same amount of PC, and Intel will be pushing Lynfield into their throat anyway.I wont even called that Sales, it is more like tax on those OEM makers.
I think it wont sell well in terms of Retail market.
IntelUser2000 - Saturday, September 19, 2009 - link
Ivy Bridge is a shrink of Sandy Bridge to 22nm. It's Haswell that will have FMA.iwodo - Sunday, September 20, 2009 - link
Yes. It is Ivy Beidge ( the shrink of Sandy Bridge ) for FMA.It was supposed to be for Sandy Bridge, but some changes delay it to Ivy Bridge. So unless they have postponed it AGAIN. it should be out with Ivy Bridge.
bigboxes - Saturday, September 19, 2009 - link
I know the basic archetecture of Lynnfield is superior to Bloomfield, but you are not using 6gb of tri-channel memory for the i920. That is where the i920 really shines. Is there a reason that you are not testing with 6gb of ram with the 920 other than apples to apples testing that needs to be done? Just curious.the zorro - Saturday, September 19, 2009 - link
lynnfield has another gigantic bottleneck which is the dmi bus speed of only 2GBps, phenom 2 hypertransport speed is 41.6 GB/s which means that is 20 times faster than lynnfield when communicating with the chipset,that shows why phenom 2 is better than lynnfield. this is going to be a real problem in the next future.silverblue - Saturday, September 19, 2009 - link
Actually, as HT 3.0 is limited to 16-bit width on AMD desktop boards, it's half that.DigitalFreak - Saturday, September 19, 2009 - link
Next future? As opposed to the current future? I must thank you for all the laughs I get from reading your posts.TA152H - Saturday, September 19, 2009 - link
No offense, but you're clearly an idiot.You realize that in EVERY benchmark, the i7 860 was running at higher clock speeds than the i7 920. Sometimes by a lot, with turbo mode. Also, Anand uses inferior memory for the i7 920, to try to 'prove' the validity of the brain-damaged P55 platform.
Despite his bad attempt, the i7 920 STILL outperformed it. If you clock them at the same rate, with the same uncore, it's only ugly for the Lynnfield.
It's not superior. Well, in performance. It's got nice power characteristics, and it's cheaper to implement. But, your remark is purely idiotic.
Where do you get stuff like this from?
Etern205 - Sunday, September 20, 2009 - link
Lynnfields are categorizes as mainstream, therefore no matter how advanced their architecture is compared to Bloomfield, it won't out perform it. Intel purposely did this and you should know this by now, but I guess you don't care as your too busy sucking your own c0ck."Your the kind of man that can be use as a blueprint to build a idiot".
bigboxes - Sunday, September 20, 2009 - link
Sorry. I didn't mean to say anything that would warrant an "idiot" label. I have just been reading from Anand that the Lynnfield core is better than the Bloomfield. In all his testing he never uses 6gb of ram in his tests. I understand that he wants to measure the cpus on a level playing field, but when you put 6gb of pc1600 ram on an i920 those scores increase considerably. From what I understand that is something that the Lynnfield cannot achieve. Was just wondering if Anand could throw that (i920 w/6gb) into his charts. It seems that almost all Bloomfield owners are gonna be running 6gb (3x 2gb) and not 4gb. The i920 uses that extra bandwidth and it truly performs better when so equipped. I hope that makes more sense.Gary Key - Wednesday, September 23, 2009 - link
http://www.anandtech.com/mb/showdoc.aspx?i=3639">http://www.anandtech.com/mb/showdoc.aspx?i=3639 - 1600 C7 on both platforms, 6GB for X58 and 8GB for P55, it does not make a measurable difference in performance compared to the 4GB setups.Wwhat - Sunday, September 20, 2009 - link
Anand does indeed use 1066 RAm for the i920 and 1333 for the i860/870/750, but those are the numbers intel specifies them at as you can see on intel.com, now the problem is that people don't use it with such RAM but with 1600 or higher, and also that if you compare the 2 it would be nice of you also compared them with at least equal speed RAM, and when you would use setup that the average guy that builds his own system then you'd get completely different results.So I guess anand is reviewing for businesses who take what they get on the cheap, pre-made systems with 'stock speed' RAM
Gary Key - Saturday, September 19, 2009 - link
"If you clock them at the same rate, with the same uncore, it's only ugly for the Lynnfield. "That is not the case. Actually, overclocking uncore speeds makes very little if any difference unless your application of choice is SuperPi or one of the older 3DMark benches where you can realize some measurable differences (wow, look I gained 0.84 seconds in SuperPI 32M and 27 points in 3DMark06) in the results.
Anyway, Lynnfield will clock both uncore and memory significantly higher than Bloomfield so I have a hard time understanding your comments about this subject. If you do not like Lynnfield, that is fine, but the continued comments about inferior memory being used or uncore rates or turbo modes (which the 920 is running by the way and is inferior to the 860s turbo mode) is really wearing thin at this point.
I already showed the results with DDR3-1600 C7 and nothing changes at DDR3-2000 C7, except VTT/VDimm is much lower on Lynnfield than Bloomfield for equal memory clocks.
For a daily platform, I would take the P55/860 over the X58/920 any day of the week. It simply performs better in most cases and uses significantly less power to do it. In fact, if based on just SOHO computer usage and typical gaming scenarios, I would take the 790FX/965BE over the X58/920. For just a SOHO non-gaming system, I am going 785G/Athlon II all the way.
For benchmarking, I will go with the X58 platform, but even then you have to ask yourself why. Unless you have been provided with cherry components and have a limitless supply of LN2 for setting records, there is not much point in using this platform now unless you are in the workstation arena where future processor upgrades will make a difference.
The only other advantage is in multi-gpu gaming, where the X58 will make a difference in the benchmarks. However, you will not notice the difference between the two platforms in blind A/B gaming comparisons. I know, I tried it on a few users who own X58 platforms. ;) Once again, the vast majority of gamers do not run CF/SLI so even this small advantage is a moot point for most, especially when you consider the performance of the upcoming GPU releases.
The X58/920 consumes significantly more power and performs about the same if not worse at times than the P55/860. So, unless you have two very specific needs for X58, there are great alternatives available from both Intel and AMD. That is the crux of our message in these articles.
strikeback03 - Monday, September 21, 2009 - link
Gary,you have mentioned a few times that Lynnfield uses significantly less power, and obviously at stock settings that is true. But if additional voltage is needed for overclocking, that advantage would seem to disappear. Do you have any tests done to show both platforms at higher clocks? Or how about both at their maximum clock with turbo still enabled, to see if you can save some power while still getting a similar end clockspeed when needed?
Gary Key - Wednesday, September 23, 2009 - link
We will have an article shortly on power consumption (static) with Lynnfield and Bloomfield overclocked. Just to cut to the chase, Lynnfield wins by a clear amount, even with slightly higher voltages on the CPU.strikeback03 - Thursday, September 24, 2009 - link
Cool, thanks.TA152H - Saturday, September 19, 2009 - link
Gary,Would you care to prove any of this? It looks to me on your own bad benchmarks, you run the Lynnfield at higher voltages and higher uncore.
Where is your proof the Lynnfield can clock the uncore higher? Why did you clock them differently then? Where is your data supporting this?
Your remark about turbo was beneath even you. I never complained about turbo modes, except in the context that people here will overclock, and make them irrelevant. I have said repeatedly, against the bashers, it's a nice technology. But, it doesn't show the architectural differences, which a lot of us were curious about. Did you forget you are supposed to be 'tech' site, not PC Magazine?
I agree with you about power. I've mentioned that a few times. I really like the lower power of the Lynnfield. It's a big advantage. I also would rather use an Athlon based machine for a daily machine, especially with the IGP. I've stated all this. But, for my compiles, I'd rather have an overclocked 920 than a brain-damaged Lynnfield. The same for games.
I'm not too crazy about Anand's benchmarks with the uncore of the Bloomfield running faster. Did you see how there were real differences in the results? 3.5% is not a little when you've only changed the CPU. Part of that is because of the uncore, so it's not really fair to the Lynnfield. I'm glad to finally see you guys showing a difference, but, really, it would have been better with uncore at the same rate.
The memory, it would appear, he's using for the Bloomfield is 1066, that's inferior to 1333.
I would be curious about your video claims. They might be right, but what would be really interesting to try to identical cards, except for memory. My guess is, the inferior setup of the Lynnfield would manifest itself more with smaller memory cards, since you'd have to use PCIe more. I could be wrong, of course, it's just theory, but it would be interesting.
Maybe you guys should stop trying to twist benchmarks to make your point, and just run them to give information. That's the crux of my irritation. The Lynnfield is a good product for a lot of people, and I really like what AMD just did. But, when you do little horsecrap things to manipulate results to illustrate your opinion, that's just wrong. You're not nearly smart enough to think for everyone else; no one person is.
yacoub - Saturday, September 19, 2009 - link
"Gary, Would you care to prove any of this?"Considering most of it is in the existing reviews, how about you go back and read them for yourself? And then take his word for it. Or don't, it's your choice, and your loss if you don't.
Griswold - Saturday, September 19, 2009 - link
You can take this as an offense, you are the idiot here. Always been, always will be.C'DaleRider - Saturday, September 19, 2009 - link
No offense, but you're clearly an idiot. And that is clearly demonstrated by this comment, "Anand uses inferior memory for the i7 920, to try to 'prove' the validity of the brain-damaged P55 platform."What was inferior about the Patriot Viper memory? Care to explain why? I personally think you can't defend that comment on any front, esp. considering it has been shown in testing to be an excellent overclocking selection for DDR3 memory, esp. for the price. It'll do 1600 speeds, albeit with using a CAS 8 setting, but for what it costs, it's excellent stuff.
Grow up, and when you finally move out of your parent's house, maybe you'll be mature enough to lose the teenage "I know-it-all" attitude you possess right now.
strikeback03 - Monday, September 21, 2009 - link
In previous tests Anand has used the fastest validated speed for each platform, so 1066 for Bloomfield and 1333 for Lynnfield. Not that it probably makes much difference in anything but synthetic benchmarks.Scheme - Saturday, September 19, 2009 - link
Woah, did you forget to take your ritalin last night?mesiah - Saturday, September 19, 2009 - link
Anand, can't you ban this guy? You have to be tired of watching him come on here and verbally assault any person he doesn't agree with, including yourself. Do everyone a favor and toss him to the curb.jonup - Sunday, September 20, 2009 - link
Noooo! This would be too cruel. We need a joker to make us laugh from time to time.tim851 - Saturday, September 19, 2009 - link
Let me guess, you have an i7 920...?Of course the 860 ran at higher clock speeds, why would Anand underclock it? It was compared to the 920 because they share the same price point. That is until you add the motherboard, then the i7 is like 100$ more expensive.
And Anand summed it up nicely: the 1366 platform is now for people who need hexa-cores someday or who think SLI/Crossfire is reasonable.
Oh, and if the P55 Platform is "braindamaged", then apparently all major tech sites are in on the conspiracy.
jordanclock - Saturday, September 19, 2009 - link
What do you mean by "inferior memory?"And of course the 860 was running at a higher clock rate: That's how it is designed to run. Even without Turbo.
TA152H - Saturday, September 19, 2009 - link
I don't have a problem with the 860 running at higher clock speeds, but if the architecture were better, it would never lose to a processor running at lower clock speeds.In short, the architecture is not clearly better. It's worse, the margin is the only thing worth discussing.
TA152H - Saturday, September 19, 2009 - link
I should have said, i7 920 still outperformed it in a few benchmarks.Pity there isn't an edit.
JumpingJack - Saturday, September 19, 2009 - link
The basic architecture of Lynnfield is the same as Bloomfield. The differences are the topology of the platform (PCI on die instead of in the chipset, 2 mem controllers instead of 3, no QPI in Lynnfield). The cores are exactly the same, the cache is exactly the same.jordanclock - Saturday, September 19, 2009 - link
I have yet to see any real world scenarios where triple channel memory "really shines." The inclusion or exclusion of a triple channel set up would account for variations of about 1% either way. In other words, less than the margin of error.blyndy - Saturday, September 19, 2009 - link
Intel Core i7 920214 / $284 = 0.75 SYSmarks per $
Intel Core i7 870
233 / $562 = 0.41 SYSmarks per $
Intel Core i7 860
223 / $284 = 0.79 SYSmarks per $
Intel Core i5 750
217 / 196 = 1.11 SYSmarks per $
AMD Athlon II X4 620
147 / 99 = 1.48 SYSmarks per $
yacoub - Saturday, September 19, 2009 - link
Your prices are wrong. The 860 is $230, which makes it 0.97 SYSmarks per $.The 750 is $160, which means 1.36 SYSmarks per $ by your measurement.
johnsonx - Saturday, September 19, 2009 - link
I just love it when someone quotes some below cost, loss-leader sale price they heard about somewhere once to prove a value arguement.yacoub - Saturday, September 19, 2009 - link
You mean "on-going, still valid sale prices that you can get today".stanljl - Wednesday, September 23, 2009 - link
Most of the US doesn't live reasonable close to the 21 cities that have microcenters. In cause you haven't looked there really aren't that many locations so "on-going, still valid sale prices you can get today", really doesn't apply to the vast majority of the people in the country.strikeback03 - Monday, September 21, 2009 - link
It is a valid price, but please add the disclaimer "If you have a microcenter nearby" instead of just pretending those are widely available prices. I plan on buying a processor when I help my parents move down near Philly next month, otherwise I (like most) don't have a Microcenter anywhere nearby.formulav8 - Saturday, September 19, 2009 - link
Yeah with only 2% of the people able to get one at near that price. Quoting an obvious loss leader as valid pricing for those looking it nutty. Newegg or ZZF is a much better gauge of price.NA1NSXR - Saturday, September 19, 2009 - link
Who cares, let him pay up. Nothing on P55 has made me regret getting that $200 D0 920. Nothing. Not even close. The OC, heat, and platform pricing advantages all failed to materialize.The 920 is not a 2.66Mhz bloomfield. It is a 3.8GHz chip supporting the fullest featured consumer platform at the moment.
kilkennycat - Saturday, September 19, 2009 - link
Where? Where? Where? From a reputable supplier and with in-stock delivery, of course.Newegg and ZipZoomFly: 860 $299
850 $199
mgivler - Monday, September 21, 2009 - link
Microcenter, for in-store purchase. I purchased an i7 860 last week for $229. The i5 750 is cheaper, $159 seems right.marsspirit123 - Monday, September 21, 2009 - link
"Microcenter, for in-store purchase. I purchased an i7 860 last week for $229. The i5 750 is cheaper, $159 seems right."Yeah with 8.5 % tax that price is $250 + how much for gas?
afkrotch - Tuesday, September 22, 2009 - link
For online purchases, you may still have to factor in sales tax and shipping. There's also the cost of having to wait.Between your current computer and the upgrade, what is the performance gap? For that gap, how much time could have been saved in your work? Because you saved time at work, how much $$ was saved?
Things can become complicated or we can just stick with the retail pricing at these stores and leave out the other factors.
marsspirit123 - Tuesday, September 22, 2009 - link
The I7 860 is $290 at newegg with free shiping and no fees of any kind.In microcenter you have $230 + $20 tax + gas .The point is the differance is less than $40 with microcenter .You should always compare final price for the purpose of being fair. How long have you been waiting for I7 860 cpu to come out ? How much have you lost for that time ?So if you have been waiting 8 months how is 3 days going to be bad?If that is so bad how come you din't get 920 before?BlueBlazer - Saturday, September 19, 2009 - link
I think he's referring to this?http://www.microcenter.com/single_product_results....">http://www.microcenter.com/single_product_results....
Core i5 750 = $159.99
BlueBlazer - Saturday, September 19, 2009 - link
As well as this...http://www.microcenter.com/single_product_results....">http://www.microcenter.com/single_product_results....
Core i7 860 = $229.99
My +/- 1 cents..
Ninevah - Monday, September 21, 2009 - link
Or this, for that matter:http://www.microcenter.com/single_product_results....">http://www.microcenter.com/single_product_results....
Core i7 920 = $199.99
vol7ron - Saturday, September 19, 2009 - link
This doesn't even factor in the savings on Watts used.jordanclock - Saturday, September 19, 2009 - link
But how many LOCs per fortnight can they process? Could you please provide your metrics in more understandable formats. At the very least I would like to see how many TuxRacer compiles per hogshead of cider (the good stuff, not the end-of-season stuff) we can expect.BlueBlazer - Saturday, September 19, 2009 - link
Skewing the numbers? Try..Intel E5300
142 / 69.99 = 2.02 SYSmarks per $
BlueBlazer - Saturday, September 19, 2009 - link
Where in the review is the PII X4 620?howmoney - Friday, June 11, 2010 - link
best buy com hardware http://www.com-hardware.com