• What
    is this?
    You've landed on the AMD Portal on AnandTech. This section is sponsored by AMD. It features a collection of all of our independent AMD content, as well as Tweets & News from AMD directly. AMD will also be running a couple of huge giveaways here so check back for those.
    PRESENTED BY
POST A COMMENT

44 Comments

Back to Article

  • Taft12 - Tuesday, October 25, 2011 - link

    Trinity should make its first appearance in Q1'12, full availability being in Q2'12.

    So, we'll get Trinity APUs (and a new socket, bleh) likely before we can even buy the "good" desktop Llanos (A8-3800 or A6-3600).

    I like desktop Llano, but it's hard to recommend when it becomes a dead end so fast. It's like the bad old days of Intel (more recently LGA1156)
    Reply
  • Snoop - Tuesday, October 25, 2011 - link

    I wonder if they are going to be able to get the power down for the mobile market? I cannot imagine a Bulldozer derivative with decent performance without crazy power usage. Reply
  • Roland00Address - Tuesday, October 25, 2011 - link

    The architecture itself is not power hungry, it is the fact Global Foundries 32nm process is still unmature and thus the chips run hot for there is a lot of leakage and the AMD is using a high voltage for the chips (for a high voltage Llano or Bullzdozer chip is still a possible chip sale for AMD)

    There is no doubt that AMD is going to get better performance per watt out of Bulldozer and Bulldozer deritiatives such as Piledriver in the future. Remember how bad Phenom was when it first came out, Agena sucked but it was improved with Deneb and the third improvement Thuban and Stars isn't that bad either. Problem is that AMD competition (Intel and ARM) is a moving target, they have to improve their product faster than everybody else improves theirs.
    Reply
  • Snoop - Tuesday, October 25, 2011 - link

    I am far from cpu architect but long pipelines, super high transistor counts, with low per clock performance and a reliance on high clock speeds seems to translate to high power usage. How is the GF process going to change this part of the architecture? Reply
  • niva - Tuesday, October 25, 2011 - link

    GF won't change the architecture, it would be up to AMD to change that at a later time. However, once GF gets the manufacturing to a better quality you can reduce the voltage, or increase the clocks for the same heat output.

    Architecture changes (like the pipeline, or the single FPU per 2 integer cores) are made for reasons. I'm not happy with the results of the choices they made for the architecture with Bulldozer, maybe in the long term it will turn out to be the right direction but right now, it doesn't seem to be so.

    Super high transistor count isn't necessarily a bad thing, CPUs are becoming more and more complicated, expect this trend to continue not just for AMD but any company which makes processors (including ARM). Everything we expect out of these machines is becoming more complicated which drives up the transistor count.
    Reply
  • Penti - Tuesday, October 25, 2011 - link

    Dude the cache does use as many transistors as a whole Nehalem or SandyBridge cpu/processor. That's obviously a mistake, not one they can do on mobile chips like the mobile version of Piledriver though so it ought to be a quite different overall architecture/product/cpu. We'll have to see. Reply
  • Roland00Address - Wednesday, October 26, 2011 - link

    Four Cpu Modules (8 cores)+L2 Cache=852 million transistors
    L3 Cache=405 million transistors
    Everything Else (I/O, DDR3 memory controller, Logic and Routing)=800 million transistors.

    So for Trinity AMD is going to only have two cpu modules so at least 426 million transistors. There will be less everything else since AMD won't be doing any multiple cpu needing interconnects. It is rumored trinity will have a 7000 series gpu based on a 4D arranagement with 480 shaders. Well a turks gpu (480 shaders, hd6670) uses 716 million transistors.

    So we are talking a billion transistor chip lets hope they can keep it closer to 1 billion and not 2 billion like bulldozer is.
    Reply
  • DanNeely - Tuesday, October 25, 2011 - link

    You're confusing fundamentally bad, with Intel botched the implementation here. IBMs power 6 CPUs also used very long pipelines compensated with high clockrates (upto 5ghz shipping parts in 08) quite successfully in thier Power6 architecture.

    At the moment most of AMDs troubles appear to be due to GF's 32nm process sucking; although it will probably be a year or two before we know if the process was also masking architectural problems.
    Reply
  • Zoomer - Wednesday, October 26, 2011 - link

    The pipeline isn't that long; it's just longer than before. More stages is a legit strategy to split the work up more, perhaps relieving a bottleneck.

    Low per clock performance? Per thread, yes. But the overall threaded performance goes up since more threads can run. They had to go 2B transistors for 8 cores; that's because they were first and foremost targeting servers and HPC. These have higher margins than the piddly desktop market, unfortunately.
    Reply
  • jeremyshaw - Wednesday, October 26, 2011 - link

    2B transistors is about equal to two Thubans... you could have 12 cores, and better per-core performance, too :p Hell, even comparible power consumption, despite being on 45nm... Reply
  • frozentundra123456 - Tuesday, October 25, 2011 - link

    I dont really understand the blame for Bulldozers power usage being laid on GF. I mean didnt they talk to each other during development?? Didnt AMD know what GF could and could not do?? Should they not have designed a chip that took into account the capabilities of GF??

    Since AMD has no fabs of its own, it would seem paramount that they know what GF can do and design a chip that GF can produce properly. Just seems like there is some sort of disconnect between AMD and GF. Or GF is just plain incompetent.

    And to me it signals trouble for AMD if AMD/GF cannot even design and produce an effecient 32 nm chip when intel has been doing it for what, almost 2 years, and is ready to launch 22 nm.
    Reply
  • DanNeely - Tuesday, October 25, 2011 - link

    They knew what GF thought it would be capable of doing (and most of the major details were locked in place several years ago). Unfortunately for AMD and for us, GF's 32nm process still has a lot of bugs to be worked out. Cumulatively these are resulting in lower than expected yields, higher than expected leakage currents, and slower than expected clock rates. The net result is that bulldozer is launching in its current underwhelming state. Reply
  • Roland00Address - Wednesday, October 26, 2011 - link

    But you forgot one last thing, having to use a high voltage on the chip due to the low yields. A high voltage means higher power consumption. Reply
  • xxtypersxx - Tuesday, October 25, 2011 - link

    Depending on when AMD discovered they would miss IPC and clock speed targets with BD, it makes sense that they couldn't correct it within the program schedule due to the amount of design work already done. One can only hope however that when they did come to this realization, Piledriver was early enough in that they could implement the changes in that design. If Piledriver is to Bulldozer was Phenom II was to Phenom I then AMD could be in a pretty good place.

    The socket change is disheartening though and makes FM1 AMD's 1156. I just hope it came about as a result of adding hypertransport or something and not because Trinity draws too much power for current FM1 boards...
    Reply
  • MySchizoBuddy - Tuesday, October 25, 2011 - link

    When will they we coming out with an APU for the HPC market. 10,000 APUs sound better than 10,000 CPUs and 10,000 GPUs. Reply
  • Hubb1e - Tuesday, October 25, 2011 - link

    "Bulldozer's poor single-threaded performance might cause the performance upgrades to be limited to multithreaded tasks though"

    Not likely. The only reason Bulldozer is able to post any gains at all is because it can run 8 threads vs 4 threads for Phenom II. The Piledriver core in Trinity can only do 4 threads vs 4 threads in Llano. It's a 2 module chip vs a 4 core chip. Either their performance preview is overstated (best case scenario vs average scenario) or they've been able to fix the major issues with Bulldozer. Then again, Llano had pretty pathetic core speed so maybe a 4.2 ghz 2 module Piledriver could come close.

    One other thing that worries me about Trinity is the HUGE # of transistors in Bulldozer. If they move that over to Piledriver without optimizations then this chip could be too big even when they take out the unnecessary L3 cache. God I hope graphics can save AMD until their next architecture comes out...
    Reply
  • basket687 - Tuesday, October 25, 2011 - link

    +1
    Even if both have similar multi threaded performance then in that case Trinty is likely to have slightly better single threaded performance than llano because of the "module" approach.
    Reply
  • eastyy - Tuesday, October 25, 2011 - link

    another socket change is annoying wish the new am3+ processors the fx series had a built in apu for the same price that would have made them more attractive Reply
  • formulav8 - Tuesday, October 25, 2011 - link

    It may be able to run on FM1 boards. It could be similar to how the old AM2 boards worked with Ph1/Ph2 but couldn't give it its full performance benefits since HTT could only run @ 1000mhz instead of 3600-4000mhz. Reply
  • haukionkannel - Tuesday, October 25, 2011 - link

    Well Llano has 4 real cores and Trinity have two 1.5 cores... And still Trinity is expected to perform better than Llano. Interesting indeed. There has to be huge ghz boost, or they is coming some serious upgrades in architecture.

    Has anyone seen "low" speed buldoser tested? In the beginning low speed Phenoms were reasonable in their power usage, but leaked like a hell when pushed to extreme. It would be nice to know if bulldoser in less hunry with smaller speed or not?
    Reply
  • fic2 - Tuesday, October 25, 2011 - link

    Didn't Anand or someone here do a side-by-side of a bunch of chips clocked 1GHz? Would be nice to add BD to that list. And yeah, see how low the voltage can go and the power usage at that speed. Reply
  • silverblue - Wednesday, October 26, 2011 - link

    I'm not sure. I know somebody (Toms?) did a single core comparison a few months back, though. Reply
  • fic2 - Wednesday, October 26, 2011 - link

    That sounds like what I was thinking of. Reply
  • C300fans - Tuesday, October 25, 2011 - link

    We all need magic to boost up. If 4 threads Trinity can gain 20% improvement against Llano, it means 8 threads can gain 40%.. That would be a deal. Reply
  • Hubb1e - Tuesday, October 25, 2011 - link

    keep dreaming. Reports from AMD say that Piledriver is 10% more efficient than Bulldozer per core. The 20% performance gains probably come from clock speed. Llano had a really really pathetic 2.9 ghz clockspeed for 4 cores. Trinity could likely launch at 4.2 ghz on 2 modules and eventually scale to high 4ghz or even into the 5s if they get the process working. This plus the 10% gain in efficiency over Bulldozer could give it a pretty substantial single threaded edge over Llano making up for the fact that it's only a 2 module chip.

    And IMO mainstream customers are much better served by fast 2 core chips than slower 4 core chips. Your mom runs very few Apps that take advantage of the extra 2 cores on Llano making the 2 module approach a pretty decent strategy. I think Llano would have been a better mainstream chip if it had 400 stream processors, and only 2 cores at 3.7-3.8 ghz. Likely what happened with Llano is that they didn't reach high clocks because they let automated tools design the chip knowing it's just a stopgap until a new architecture.
    Reply
  • IlllI - Tuesday, October 25, 2011 - link

    "IPC will be higher
    Single threaded performance will be higher"
    Reply
  • Malih - Wednesday, October 26, 2011 - link

    yeah, that one really hit potential customer in the face. I doubt they'll deliver the planned 20%, maybe more like 20% in one specific application, and the rest of them are below that. Reply
  • Mishera - Wednesday, October 26, 2011 - link

    They've been pretty reliable on their projections at least for the past few years. Although with this whole "what's a core" thing there may be some confusion on what that 20% is.

    I'm just worried about the heat this thing is going to produce with a gpu attached.
    They got water-cooling for laptops?
    Reply
  • frozentundra123456 - Tuesday, October 25, 2011 - link

    Well here we go again with another chip for the forums to speculate and argue about, and more promises from AMD that they may or may not be able to back up. The threads from Bulldozer havent even died down yet!!

    I thought the best place for Llano was in the mobile market. Unless they can lower the power consumption, improve IPC, and reduce the die size, I am not sure that a trinity based APU will be an improvement over Llano. At least maybe they can get better graphics performance with a reasonable power envelope. The first Llano was not quite strong enough in either the graphics or CPU performance, although decent at both. The problem was Intel CPU performance was much better. AMD graphics was better, but still not quite good enough for the kind of gaming that I want to do.

    BTW, did anyone else think that the headline was a bit misleading?? Dont really see any "details" here, just a general overview and the projections from AMD. Lets hope they come closer to the mark than Bulldozer did.
    Reply
  • fic2 - Tuesday, October 25, 2011 - link

    Not to mention nothing new here. Pretty sure all of this has been known for a while now. Reply
  • Boogaloo - Wednesday, October 26, 2011 - link

    A 30% increase in integrated graphics performance is laughably low. Reply
  • silverblue - Wednesday, October 26, 2011 - link

    Do remember that Trinity is at the same process as Llano. You can't just magic up a massive performance boost if you're looking to keep die size sensible, and regardless, the memory interface can only be improved so far. Reply
  • Mugur - Wednesday, October 26, 2011 - link

    What if some of the performance issues will be corrected by simply using the memory controller at cpu speed, like Intel does, instead of the same 2-2.2 Ghz like it's now...? Reply
  • silverblue - Wednesday, October 26, 2011 - link

    It'd certainly reduce latency. I'd laugh if a good deal of Bulldozer's problems are solved by doing just that. Reply
  • arjuna1 - Wednesday, October 26, 2011 - link

    "In terms of speed, AMD is claiming up to 20% increase over Llano. Bulldozer's poor single-threaded performance might cause the performance upgrades to be limited to multithreaded tasks though, unless AMD can do magics with Piledriver (aka 2nd gen Bulldozer)"

    Yet another intel fanboi biased comment from from Kristian Vättö, hey Anand, hire a couple of AMD geek's, I mean, to even the scales, this is making your site look really bad.
    Reply
  • silverblue - Wednesday, October 26, 2011 - link

    There's nothing fanboyish about the remark. Piledriver and future iterations of the architecture are rumoured to bring 10-15% improvements on the CPU side, but that is in comparison to the first generation Bulldozer which appears to perform slower clock-for-clock than K10.5.

    Unless there's something easily fixable with Bulldozer, Piledriver isn't exactly going to set the world alight. In any case, there's limited space so you can't very well throw in a GPU that's an order of magnitude stronger than the one in the A8-series.

    A 20% improvement across the board would be nice.
    Reply
  • kenyee - Wednesday, October 26, 2011 - link

    Sorry, but I'm an AMD fan and have designed chips in a former life. I'd agree w/ that comment unfortunately. Given how BD is roughly the same speed as a Phenom II in single threaded apps, I don't see how it's going to be faster in Llano form. It'll be a dual core 4 thread system :-P Reply
  • krumme - Thursday, October 27, 2011 - link

    No you are not. You are not sorry, you are not an AMD fan and you havnt designed chips.
    Sorry.
    Reply
  • Kristian Vättö - Wednesday, October 26, 2011 - link

    The numbers are what they are, you can't change them. It's hard to praise AMD since they haven't been anything else but an underdog in the CPU segment lately. Reply
  • Anonymous Blowhard - Wednesday, October 26, 2011 - link

    Yeah, that "telling the truth, no matter how much it hurts" is really a bad approach.

    U MAD BRO?
    Reply
  • Soulkeeper - Wednesday, October 26, 2011 - link

    I just built a llano system

    Hopefully trinity will be a drop in replacement
    FM2 sounds like something that should wait for DDR4 to me
    Reply
  • starcaptor - Wednesday, October 26, 2011 - link

    This update better be "DA BOMB" if you know what I mean. Reply
  • TC2 - Sunday, November 06, 2011 - link

    a free of charge idea:
    instead of
    "Details on Trinity - AMD's Next Gen APU"
    you could name the article
    "Details on Trinity - AMD's Next Gen GOSSIP GENERATOR"
    Good luck :)
    Reply
  • tipoo - Wednesday, March 28, 2012 - link

    I know consoles aren't mentioned much here but I'm curious, a supposed insider recently told tech sites that the next Playstation will use an AMD x86-64 CPU as well as a Southern Islands GPU, so that could either be separate chips or something like Trinity. I wonder if it will use the Trinity APU (southern islands GPU with up to four Piledriver cores) or a separate CPU and GPU, with a SI GPU and the FX-Next-like processor with up to eight cores. The four core might outperform the Cell by magnitudes but they spent a lot of time convincing people that the Cell was amazing, and going from 1PPE plus 6 usable SPE cores to 4 would sound like a downgrade to some who know less about hardware. Reply

Log in

Don't have an account? Sign up now