Comments Locked

45 Comments

Back to Article

  • Grebuloner - Wednesday, March 19, 2014 - link

    If X99 can be what X79 was supposed to be, I might upgrade my SNB-E earlier than the Skylake-E I was planning on. I am disappointed on the consumer core count...don't want to pay for a 2P Xeon when I don't need the validation, etc., just the cores.
  • extide - Wednesday, March 19, 2014 - link

    Get an E5 16xx v2 then, they go to 8 cores and are available right now.
  • f0d - Wednesday, March 19, 2014 - link

    cant overclock it though
  • buffhr - Thursday, March 20, 2014 - link

    Actually the E5 1650v2 has an unlocked multiplier same as 3930k c2 just with ECC support.
  • jmke - Thursday, March 20, 2014 - link

    why would you want to overclock the CPU? ;-)
  • ddriver - Thursday, March 20, 2014 - link

    Why would you not?
  • ImSpartacus - Saturday, March 22, 2014 - link

    Because it has excellent performance & thermal characteristics out of the box.

    Overclocking the CPU requires superior cooling or settling with a warmer/louder chassis. If you're overclocking your GPU (which actually needs it), then you don't want to waste cooling potential on your CPU.
  • sephirotic - Sunday, March 30, 2014 - link

    Moderate and small ovverclocks ALLWAYS is always nice performance-price wise. Is liky buying a more expensive processor. multi core processors have poor performance for single core applications and low clock by default. Overclocking mitigates that.
    Talking about 8 cores, the superior cooling is not really that much since the purpose is to increase single and double threaded applications, not multithread where you can limit the overclock. For only 2~3 cores stressed, a 30% overclock would hardly increase thermal load to noisy leves if you use any 12cm aftermaket coolers. Intel's stock cooler is noisier in full load than any cheap 30 usd 12cm tower heatsink. The extra heat dissipated would hardly impacts in the case air sorrounding the GPU. specially with a tower heatsink that directs the heat generated directly out of the case, differently from the low profile cooler intel uses.
    So sorry to say that, but you clearly isn´t a overclocker and got all the points and drawbacks wrong.
  • Bohefus1 - Wednesday, August 6, 2014 - link

    you clearly don't have a grasp of the english language
  • BMNify - Thursday, March 20, 2014 - link

    more cores without AVX2 or better are not worth the money today for most generic uses, such as x264 UHD encoding etc
  • f0d - Thursday, March 20, 2014 - link

    actually handbrake scales pretty much linearly with more cores - i have tested it with my 3930k
  • BMNify - Thursday, March 20, 2014 - link

    yes it does for the old AVX SIMD portion , but compare that to the later AVX2 optimizations and you get up to 70% ish better data throughput for a given 1080P/UHD highest visual quality encode, you should try using a generic FFMPEG/AVCONV bat script as that gives you better throughput when tuned over the handbrake GUI...
  • twtech - Friday, March 28, 2014 - link

    It's a chicken-and-egg problem. Somebody has to put a larger number of cores in the hands of end-users so that development tools and software will have a increasing reason to catch up.
  • f0d - Wednesday, March 19, 2014 - link

    i was also disappointed at how many cores came out on consumer x79 - sure there was the ability to use xeons but you cant oc them so my 6 core 5.0ghz 3930k would have been just as fast or even faster than any 8 core xeon (for encoding videos which is what i mostly need the cores for)
  • Samus - Wednesday, May 14, 2014 - link

    Maybe this will finally be a legit reason to upgrade my x58, i7 975, which coincidently was the first DDR3 platform, and triple-channel at that.
  • munim - Wednesday, March 19, 2014 - link

    Would X99 be coming out too early for Sata Express to be implemented?
  • JMcGrath - Sunday, April 13, 2014 - link

    I have been looking for a definitive answer to this very same question myself - or even if (or when) X79 will receive SATA Express.

    However, I am going to go out on a limb and bet on the fact it WILL be included.

    Considering Z87 Boards are already seeing SATA Express models (search ASUS Z87 DELUXE/SATA Express), and add the fact that even consumer boards such as H97 - as well as Z97/Z99 - will feature SATA Express, I see no reason X99 would not!

    Whether X99 has *native* support or not remains to be seen, although this may be a feature left for (perhaps intentionally) FLEX IO.

    Either way, I am not too worried about it. I had put a hold on my new build going all the way back to a 3930K which I exchanged for a 4930K waiting and hoping for implementation on the X79 boards (which unfortunately seems will not come). Although I have finally decided to just proceed with my build on a high end X79 motherboard (still deciding between the RIVE Black, X79 Dark, P9X79-WS, etc).

    Even with 2 (possibly 3) GTX 780 Titan Black's, or Flagship Maxwell GPU - 40 PCIe 3.0 lanes is more than enough to support 2, 3 or even 4 top GPUs and a SATA Express Controller!! Unless you are planning on running more than 4 SATA Express SSDs there would be no issue, at which point you would run into CPU limitations before On-board or discrete controller issues.
  • Scabies - Wednesday, March 19, 2014 - link

    Ahh, finally DDR4!
  • iwod - Wednesday, March 19, 2014 - link

    Haswell-E will be new Xeon E3 with 8 Core?
    i.e Broadwell wont be availbe on Server / Desktop ( Except that one with Iris Pro )?

    Slight confused here.
  • Kevin G - Thursday, March 20, 2014 - link

    Haswell-E will be the basis for Xeon E5 v3 that should ship by the end of the year. It will use socket 2011v3 which has support for DDR4.

    Broadwell on socket 1150 will be the basis for the next generation of Xeon E3 v4 chips due earlier next year.
  • iwod - Thursday, March 20, 2014 - link

    Ok I am confused ( I haven't been following Desktop CPU for a long time )
    Did Intel ever had a LGA 2011 i7 Desktop Chip? Because i couldn't not find any on Intel Ark website. They were all 1155. Even the 3rd Generation.
    If yes, then why is this 8 Core new to Desktop. Since on server side there are already 8 Core IvyBridge E5 v2 based on Socket 2011 as well.
    Broadwell is delayed on Desktop. Mobile coming first. So the 2015 year of Skylake we will see a Desktop Broadwell?
    Um this is all a bit strange, at least the way things work are no longer predictable with the Tick Tock timeframe.
  • f0d - Thursday, March 20, 2014 - link

    "Did Intel ever had a LGA 2011 i7 Desktop Chip?" yes there was many 3820 3930k 3960x 4820k 4930k 4960x

    "If yes, then why is this 8 Core new to Desktop" because they were all 4 and 6 core cpus
  • BMNify - Thursday, March 20, 2014 - link

    "The shift to DDR4 memory has been expected for a number of quarters. "

    actually its clear that while DDR4 will give you a potential 1.5x to 2x better ram data throughput, it all seems far to little far to late now, i still cant see the masses of industrial cloud installations etc wanting to pay for the limited benefit when its clear for a long time now that ratified HMC (hybrid memory cube) at lower power and 17x's the data throughput are also almost here in bulk manufacture, it just takes the orders to the foundries to make it happen...

    ""More than 100 developer and adopter members of the hybrid memory cube consortium (HMCC) this week announced is has reached consensus for the global standard that will deliver a much-anticipated, disruptive memory computing solution. Developed in only 17 months"

    "More than 100 leading technology companies from Asia, Japan, Europe and the U.S. have joined the effort, including Altera, ARM, Cray, Fujitsu, GlobalFoundries, HP, IBM, Marvell, Micron Technology, National Instruments, Open-Silicon, Samsung, SK Hynix, ST Microelectronics, Teradyne and Xilinx. Continued collaborations within the consortium could ultimately facilitate new uses in HPC, networking, energy, wireless communications, transportation, security and other semiconductor applications."
  • jasonelmore - Thursday, March 20, 2014 - link

    all those companies dont matter unless intel is on the list. they have to bake support into their platforms, or you will see hardly any adoption given ARM's market share in the server market
  • BMNify - Thursday, March 20, 2014 - link

    you seem unaware that in fact Intel where one of the original developers of HMC at the fall Intel Developer Forum in September 2011.

    customers like Intel and AMD are not consortium members, it is generally thought that they opted not to be involved in the open standard in order to develop their own way of using the technology. Intel demonstrated a prototype HMC device during the fall Intel Developer Forum in September 2011, deeming it the fastest and most efficient DRAM ever built..."

    http://www.i-micronews.com/upload/image/PG/compare...
  • BMNify - Thursday, March 20, 2014 - link

    http://blogs.intel.com/intellabs/2011/09/15/hmc/

    Reinventing DRAM with the Hybrid Memory Cube
    By Bryan Casper on September 15, 2011
    "Today, Intel CTO Justin Rattner is demonstrating the Hybrid Memory Cube, the fastest and most efficient Dynamic Random Access Memory (DRAM) ever built.

    I want to give you some background on how and why we collaborated with Micron on this new memory technology. One of my research passions is helping to design computers to be faster and more energy efficient.

    A portion of my creative energy over my career has been to improve the interconnect within computer systems so that communication between the microprocessor, DRAM, storage and peripherals is faster and lower power with each successive generation. In other words, I’m an I/O guy.

    One of the biggest impediments to scaling the performance of servers and data centers is the available bandwidth to memory and the associated cost. As the number of individual processing units (“cores”) on a microprocessor increases, the need to feed the cores with more memory data expands proportionally.

    Legacy DDR-style of DRAM main memory isn’t going to cut it for much of the future high-end systems. Being an I/O researcher, my initial efforts to solve the memory bandwidth problem were focused exclusively on the I/O to improve the circuits, connectors and wires that help to form the connection between the microprocessor and memory...."
  • iwod - Thursday, March 20, 2014 - link

    Well iGPU is heavily bottlenecked by Memory Bandwidth as well. So why Isn't Intel bringing this to Market instead?
  • BMNify - Thursday, March 20, 2014 - link

    good question, perhaps anand can ask them
  • BMNify - Thursday, March 20, 2014 - link

    you also seem unaware as regards "they have to bake support into their platforms, or you will see hardly any adoption given ARM's market share in the server market" that in fact many of the best server kit is in fact now using ARM SoC IP NoC (Network On a Chip), CoreLink CCN-508 that can deliver up to 1.6 terabits of sustained usable system bandwidth per second with a peak bandwidth of 2 terabits per second (256 GigaBytes/s)

    its also interesting to note that with the defined "short-reach (SR) and ultra short-reach (USR) interconnection across physical layers (PHYs) for applications requiring tightly coupled or close-proximity memory support " and the generic current CoreLink CCN-508 that can deliver up to 1.6 terabits of sustained usable system bandwidth per second with a peak bandwidth of 2 terabits per second (256 GigaBytes/s) at processor speeds scaling all the way up to 32 processor cores total..... at least on paper

    Intel's Knights Landing Xeon Phi due in 2015 with its antiquated QPI interconnect and its expected ultra short-reach (USR) interconnection only up to 500MB/s data throughput seems a little/lot short on real data throughput by then, Intel seem to need to go to Wide IO/HMC and a far better ultra short-reach interconnect ASAP it seems....
  • Khato - Thursday, March 20, 2014 - link

    So many errors I have to wonder if it's all intentional?

    So, what are some examples of these 'best server kit' that use ARM's CoreLink CCN-508? Which was only announced late last year by the way, hence why there are no products using it yet. Second, it's an on-chip interconnect and hence that peak bandwidth isn't anything special.

    As for Knight's Landing, QPI is an external interconnect use for communication between packages through common PCB traces and capable of over 25 GB/s. And you know that Knight's Corner already has a peak memory bandwidth of 352 GB/s, right? Care to take a guess at the sustained bandwidth its on-chip interconnect is capable of? I'd imagine not since you want to pretend that ARM's CCN-508 is something awesome and revolutionary.
  • dstarr3 - Thursday, March 20, 2014 - link

    I have the sneaking suspicion that X99 is going to crammed full of new interfaces and technologies that haven't quite reached maturity yet...
  • georgekn3mp - Thursday, March 20, 2014 - link

    Wow and I thought my SB-E 3930k with 16GB Quad-channel 1866 DDR-3 RAM was fast...

    well at least I could re-purpose my Raided SSD's and the dual GTX670 with an X99 board...but then just the Has-E CPU and DDR4 and mobo would cost as much as the dual GPU and pair of SSD cost me in 2012. I would hope for more than a 10% increase overall...

    But going from 40-50 GB/s memory bandwidth to whatever Quad-DDR4 has PLUS two more cores (4 more threads) would make a bigger difference than the upgrade from SB-E to IB-E that I skipped?
  • chizow - Thursday, March 20, 2014 - link

    Too little too late for me for this E platform, got tired of waiting for a worthy successor to X58 that was on par with Intel's Mainstream Performance platform and it never came, so I caved and built two Z87/Haswell rigs instead. Haswell-E and X99 would've been great if it came last year, but I got tired of Intel dragging their feet.
  • semo - Monday, April 21, 2014 - link

    You got tired of waiting on Intel so you bought Intel products instead? Well that's the reality. Back to the bad old days of pre Athlon 64. At least they are still shrinking the nodes for now but it is a shame that there is almost no competition in the home CPU and chipset market
  • azazel1024 - Thursday, March 20, 2014 - link

    I hope that means that Haswell-E and Broadwell-E will be 6/8 core chips and not a single 8 core extreme chip, a 6 core and then a quad core or two on the low end. I'd really like something markedly faster than what I am running now for heavily multithreaded tasks (Lightroom, Photoshop, Handbrake). My biggest issue is, I can't justify shelling out a $300-400+ premium over high end quad core processor just to get a hexacore.

    So I am hoping that means that the lowest end Haswell-E/Broadwell-E is going to be a Hexacore processor now...and in the same $300-400 price range that the current low end Ivy-E quad core processor is.

    I can justify $100-200 dearer price tag to increase my core count by 50%.
  • Ytterbium - Monday, June 9, 2014 - link

    Seems like there will be 6's at low end and 8 for extreme. The x820K was kind of pointless, worse performance on high end platform.
  • toyotabedzrock - Thursday, March 20, 2014 - link

    This seems like a badly thought out release plan. If your targeting enthusiasts you don't release a Cpu with last years design and the new mainstream design at the same time.
  • lilmoe - Thursday, March 20, 2014 - link

    I've yet to see the value in Intel's E-Series chips, at least not with *that* difference in price.
  • DnaAngel - Wednesday, May 21, 2014 - link

    They have their uses, especially if you run highly threaded applications. I never saw the justification in the top end E's , but the k versions are manageable at $500 ish. Like the IVB-E 4930k. If they stay true to the numbering scheme, the go to Haswell-E would be the 5930k.

    I dont mind paying $500ish for an 8 core 16 threaded CPU. Tie that in with DDR4 and X99 and 2 GTX 870's, but I don't plan on grabbing up the upgrade till Summer '15. Gives manufacturers time to refine boards (r2.0's) and DDR4 to stabilize in price with revisions to its speeds etc. Hopefully by next summer, Asus will of released a Rampage Black Edition X99 based Motherboard. Rampage V I presume. As an upgrade to the Rampage that will be available with Haswell E's launch.
  • Stuka87 - Thursday, March 20, 2014 - link

    Looks like I am going to push off building a new system a bit longer than. My aging gaming box plays the majority of things fine, so I can hold off a little longer.
  • GB-RS.com - Thursday, April 17, 2014 - link

    For those that cannot wait, I have posted the ultimate Dream Machine with every component you could think of all at drastically lowered prices if you should buy the combination. I am a bit behind in my posting here so I cannot put up an advertisement but if you visit the Rampantspeculation.com forum you can get a look. This deal is a gift to whoever it was made for. My wife and I are headed back to a work we started in Nepal in 1985 and this was a complete change in plans. So I sincerely hope that I will find the person who "needs" what I no longer "need" and would enjoy making someone's day very bright! Thanks and blessings.
  • I AM MY FATHERS SON - Monday, May 12, 2014 - link

    Garbage, garbage, garbage! Z87 was a real disappointment. Z97 looks even worse.

    GIVE ME BANDWIDTH!
  • DnaAngel - Wednesday, May 21, 2014 - link

    There wasn't really any new technologies for Z87 to grasp. Another reason why it pays to only do new builds every 3-5 years. The only reason one would need to get Z87 was if they are coming from a much older platform. Of course there are 2 sides to every argument and the elitist who upgrade yearly will have something to say, but for the rest of us, money doesn't grow on trees and we have to be more strategic in upgrade cycles. My rule of thumb is to save up for a decent build with some longevity in mind, build it, then save for the next one in 3-5 years. $5-$10 out of each paycheck put away, adds up pretty good over that 3-5 years. Essentially funding the next upgrade and easy on the wallet, instead of forking out all at one time.
  • vknyvz - Friday, August 15, 2014 - link

    It's time to upgrade, I am still on my 950 yup 1st gen. Been waiting for the 5th Gen but my motherboard is crashing 5 times a day :(.

    All I need to know will the x99 chipset will support 5th Gen CPUs or there will be a new chipset. Or x99 will support Broadwell but there will also be another chipset (now i wouldn't like not having the latest) because I gotta update Haswell-E my system is crashing.... but when comes out I will definitely drop in a Broadwell.
  • Insight - Wednesday, August 27, 2014 - link

    In my thinking, back in the day, updating to a new computer was due to rapidly changing tech introducing new things. From monochrome and a 640Kb maximum memory limit in the old personal computer days up to just a few years ago. For everything I do on a computer I don't see a need for a new one any time soon. I am on the X58chipset with the 1366 socket. No games require a new chipset that I know of. I don't do photoshop or any other kinds of things that require resources that exceed those of my system. In the grand scheme of things, if it becomes faster and cheaper I would make the jump. So far I don't see anything newer that is that much faster and certainly is not cheaper.

Log in

Don't have an account? Sign up now