Slashdot Mirror

← Back to Stories (view on slashdot.org)

AMD's Latest Server Compute GPU Packs In 32GB of Memory

Posted by timothy on from the absurdly-cheap-by-some-metrics dept.

Deathspawner writes: Following-up on the release of 12GB and 16GB FirePro compute cards last fall, AMD has just announced a brand-new top-end: the 32GB FirePro S9170. Targeted at DGEMM computation, the S9170 sets a new record for GPU memory on a single card, and does so without a dual-GPU design. Architecturally, the S9170 is similar to the S9150, but is clocked a bit faster, and is set to cost about the same as well, at between $3,000~$4,000. While AMD's recent desktop Radeon launch might have left a bit to be desired, the company has proven with its S9170 that it's still able to push boundaries.

46 comments

  1. Finally by blackt0wer · · Score: 3, Funny

    This will cut my rendering time of Hentai down dramatically.

  2. First 3x Posts are Retarded by Anonymous Coward · · Score: 2, Funny

    1) You're old... We get it!
    2) 2013 called, they want their fad back
    3) "Render" Hentai? Unless you count encoding DVD rips in some lossy codec to be "rendering": it really won't. Do you have some fucked up 3D animation hentai running on OpenGL or DirectX?

    1. Re:First 3x Posts are Retarded by Anonymous Coward · · Score: 0

      RapeLay or BattleRaper2?

    2. Re:First 3x Posts are Retarded by wolrahnaes · · Score: 1

      Do you have some fucked up 3D animation hentai running on OpenGL or DirectX?

      In a world where we have VR headsets there's a possibility their answer is yes.

      --
      I used to get high on life, but I developed a tolerance. Now I need something stronger.
  3. So what happened to the comments lately? by Anonymous Coward · · Score: 1

    Starting at about the time that the "read the X comments" link was removed, I noticed a big drop in the number and quality of comments and moderation. Am I the only one?

    Also, what happened to polls? I've only seen one since they removed the sidebar. Have those been killed off?

    1. Re:So what happened to the comments lately? by blackt0wer · · Score: 1

      Slashdot has decided that this website shall be used as a marketing tool to further their own existence rather than a site that is "News for Nerds" as originally intended. Therefore, it's a tragic consequence that some users will feel it's not worth writing insightful messages 100% of the time. Besides, there are only a few production houses where these cards will really be able to be used, and they've already got their orders places, so for the rest of us, these cards are merely measuring sticks by which we realize how quickly our formerly "new" cards are becoming more obsolete.

    2. Re:So what happened to the comments lately? by Anonymous Coward · · Score: 1, Informative

      The quality went down some time ago. All the dumb conspiracy nuts, and the women-hating cave-men, made this a place that smarter people make fun of, rather than participating in.

    3. Re:So what happened to the comments lately? by 0100010001010011 · · Score: 2

      Rather than improving the site and taking on refugees from Reddit they decided to double down on terrible design and appeal to the worst of them.

      It's like Italy deciding to turn away all the moderates and only deciding to accept

      Meanwhile those of us that have been around Slashdot since 2000 have pretty much thrown our hands up.

      I wish I had enough knowledge and freetime to rewrite INN with moderation. Usenet 2.0. It seems that the Eternal September has hit the web.

    4. Re:So what happened to the comments lately? by LWATCDR · · Score: 1

      "Meanwhile those of us that have been around Slashdot since 2000 have pretty much thrown our hands up."
      Imagine how I feel.
      I have to agree with you. I really want a new old slashdot but I doubt that the changes I would make would make everyone happy.

      --
      See my blog http://ilovecookes.blogspot.com/ for light hearted technical information.
    5. Re:So what happened to the comments lately? by Anonymous Coward · · Score: 0

      The quality went down some time ago. All the dumb conspiracy nuts, and the women-hating cave-men, made this a place that smarter people make fun of, rather than participating in.

      What do you mean? Karmashock's still here! Oh wait, you said the smarter ones left, not the others.

  4. First Five Posts are Retarded by Anonymous Coward · · Score: 0, Funny

    1) You're old... We get it!
    2) 2013 called, they want their fad back
    3) "Render" Hentai? Unless you count encoding DVD rips in some lossy codec to be "rendering": it really won't. Do you have some fucked up 3D animation hentai running on OpenGL or DirectX?
    4) Pointless analysis of the first three posts
    5) This post

  5. Re:Viva la beta! by jellomizer · · Score: 2

    640kili-Terabytes should be good. That way we can buffer over a full year of 8k video.

    --
    If something is so important that you feel the need to post it on the internet... It probably isn't that important.
  6. What are these used for? by faway · · Score: 1

    Excuse me. For what are these used?

    1. Re:What are these used for? by faway · · Score: 1

      I expected Hollywood is the main user but I know I don't know for sure. Really? Hhentai makers. By the way, AFAIK all hentai sucks unless you have a counterexample.

    2. Re:What are these used for? by AqD · · Score: 3

      These are just huge waste of money. You can use the same money to buy several of gaming cards - each of them is way faster than workstation card for many types of operations. Also since GPU computation is highly parallel it makes no point to have one super card instead of many cheap cards combined together.

    3. Re:What are these used for? by chuckymonkey · · Score: 3, Interesting

      Not necessarily, in scientific computing cards like this are important. The biggest problem with GPU computing in general is the time it takes to copy from main memory to GPU memory and back. It really makes GPU difficult to work with and generally the gains in parallelization don't really pay off considering the amount of time it takes to make those memory copies. Being able to load more into memory and have it stay there is a big deal.

      --
      "Some books contain the machinery required to create and sustain universes."-Tycho
    4. Re:What are these used for? by chuckymonkey · · Score: 1

      Virtual desktops, scientific computing, big render farms. There's all kinds of things, they're not really for home users though. I mean, you could buy one but you're unlikely to be able to use it effectively. This is targeted at a market and computation scale that's much much larger than most people work with. People who buy these don't buy one at a time, they buy them by the dozens.

      --
      "Some books contain the machinery required to create and sustain universes."-Tycho
    5. Re:What are these used for? by Anonymous Coward · · Score: 1

      Scientific applications that need to process what we in the business call "An unbelievable metric fuckton of linear algebra."

      A (now depressingly standard) billion-cell 3D CFD simulation = solving a minimum of 5 billion simultaneous nonlinear equations per timestep = solving probably some dozens of 5-billion-variable linear algebra problems per timestep since implicit methods are all the rage.

    6. Re:What are these used for? by Anonymous Coward · · Score: 1

      each of them is way faster than workstation card for many types of operations

      Many, but not all. There are a wide variety of computational uses for GPUs these days. Some are processing speed bound, some of bandwidth bound, others are memory bound. Adding a bunch of cards isn't going to help with some computations that require large intermediate buffers with a lot of cross dependence, as you'll just be stuck waiting for the same memory to be moved back and forth between cards, assuming it can even fit in them.

    7. Re:What are these used for? by Anonymous Coward · · Score: 1

      I don't believe the Hawaii chip used for this card has ECC protection for any of its shader array SRAMs. That may significantly reduce the usefulness of this product for high-volume installations (MTBF will likely be unacceptably low). Well, maybe one could make-do, but computations would have to be double-checked or otherwise qualified. A 32GB framebuffer is nifty, for sure, but definitely a niche product for the foreseeable future.

    8. Re:What are these used for? by Anonymous Coward · · Score: 0

      The GCN architecture supports ECC. I believe that the on-chip SRAMs use hardware ECC (9 bits stored for 8 bits data), but the main memory uses software ECC, at about a 15% storage capacity penalty.

      It is possible that ECC isn't implemented in the smaller chips, of course, but Hawaii/Grenada most definitely have it.

    9. Re: What are these used for? by AgNO3 · · Score: 1

      You dont deal with large data or making textures in say Mari.

      --
      OMG Ponies!!! with Glitter!!!! I miss Pink :-(
    10. Re:What are these used for? by Anonymous Coward · · Score: 0

      ECC isn't needed for scientific computing. The likelihood of getting a single error using non-ECC memory is already vanishingly low.

      ECC is only required for financial computing (where even a vanishingly low chance of an error that could cost millions of dollars is an unacceptably high risk) and in mission-critical industrial applications, where a single error might make something important catch fire.

      In scientific computing, errors in your calculations (should they ever appear) are generally detectable and tolerable. You'll be running the simulations many times with different parameters and different input data anyway, looking for and analysing patterns in your results.

      For use in render-farms, the same applies: you'll be rendering similar things in similar ways, many times, until it all looks about right. Errors in your calculations (should they ever appear) are detectable, tolerable and easily corrected.

  7. SETI@Home? by Insanity+Defense · · Score: 1

    So how fast does it run SETI@home?

    1. Re:SETI@Home? by Anonymous Coward · · Score: 0

      About as fast as it runs Crysis, so don't hope to find aliens anytime soon.

  8. DGEMM-Double precision GEneral Matrix Multiplicatn by Wargames · · Score: 3, Informative
    DGEMM:

    D = Double precision (as opposed to S = Single precision, C - Complex single precision, or Z - Complex Double precision)

    GE = GEneral, as opposed to HErmetian for example.

    M = Matrix

    M = Multiplication

    --
    -- Each tock of the Planck clock is a new world and here we are still life. --
  9. An APU with even 16GB of integrated memory by drinkypoo · · Score: 0

    An APU with even 16GB of integrated memory would be news, let alone 32. A video card with 32GB of memory is not. A GPU with 32GB of integrated memory would be news, but that is not what this is. This is a video card with 32GB of memory.

    Of course, AMD's own press release gets this wrong, but that's not excuse. It just means this is C&P bullshit.

    --
    "You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
    1. Re:An APU with even 16GB of integrated memory by Anonymous Coward · · Score: 1

      Putting 32GB of ram on a card with a 512bit memory bus is impressive (that's a lot of traces!) regardless of the fact that it isn't an APU. It is interesting to see this dual direction from AMD however. While they're releasing HBM and the associated "Fury" products the hard 4GB limit is obviously impacting competition and makes the newly released product non-competitive. It does seem like once they can improve the memory limit and get the cost down, that type of product would be a great addition to an APU, finally granting near-competition to the discrete options. Question is, do we wont to spend several hundred dollars for a part that likely won't be compatible with a new motherboard when the next improvement comes along.

    2. Re:An APU with even 16GB of integrated memory by mangobrain · · Score: 1

      I'm confused. How would a GPU with 32GB of "integrated memory" be news, but a GPU with 32GB of [non-integrated*] memory is not news? I'm not sure what you mean when you say "integrated memory". This is not the graphics half of an APU, it is a discrete card, and nowhere do the summary or the press release state otherwise. The term "compute GPU" just means it's targeted at computing workloads, not graphics workloads.

      What exactly is your complaint?

      * Not even sure what this means, but you seem to be contrasting "integrated memory" and "memory".

    3. Re:An APU with even 16GB of integrated memory by Anubis350 · · Score: 1

      My suspicion is the GP meant "on-die" not "integrated" Still a silly post though, for one thing I dont think, as you pointed out, that they know what a compute GPU is

      --
      "goodbye and hello, as always" ~Prince Corwin, from Zelazny's Amber series
    4. Re:An APU with even 16GB of integrated memory by wbr1 · · Score: 1

      This is a discrete card with 1 GPU and non-integrated memory. Not an APU. This is designed for rendering and other parallel compute tasks. Go google.. do you see any single GPU card that addresses this much RAM? I do not.

      --
      Silence is a state of mime.
  10. How exciting. by Anonymous Coward · · Score: 0

    Same SP/DP FLOPS as W9100.
    Same memory bandwidth as W9100.
    Same power consumption as W9100.
    Twice the memory size of W9100 (32GB from 32*8Gb vs. 16GB from 32*4Gb).

    On other words, "AMD pushing the boundaries" == "there's now mass availability of 8Gb GDDR5, slap it on a existing card design real quick"

    1. Re:How exciting. by Anonymous Coward · · Score: 0

      What would really be beneficial would be instead of of multi-Gigabyte DRAM, they implemented perhaps a GB of super-fast single-cycle SRAM instead over a couple of dies. Low latency, no cache required, optimal compute performance.

  11. DGEMM without ECC? by Anonymous Coward · · Score: 0

    Who does DGEMM without ECC? This is a professional line GPU designed for compute but it doesn't have ECC...

    1. Re:DGEMM without ECC? by Anonymous Coward · · Score: 0

      True

    2. Re:DGEMM without ECC? by Anonymous Coward · · Score: 2, Informative

      Like every other Hawaii based professional card, it does.

  12. Re:DGEMM-Double precision GEneral Matrix Multiplic by Anonymous Coward · · Score: 0

    Fortran's most lasting legacy: Function names which are impossible to understand without having the lookup table next to you.

  13. What puts "G" into "GPU"? by mi · · Score: 1

    Is it still a "Graphics Processing Unit", if it does not even offer any way to connect a display to it?

    Well, maybe, it just means "Ginormous" now...

    --
    In Soviet Washington the swamp drains you.
    1. Re:What puts "G" into "GPU"? by MrDoh! · · Score: 1

      If I can throw this in a machine with 2 SLI'd Titan X's for another 5 frames per second, it'll still sell.

      --
      Waiting for an amusing sig.
    2. Re:What puts "G" into "GPU"? by Anonymous Coward · · Score: 0

      It's still perfectly capable of working with graphics primitives like vertices & pixels, and of rasterising polygons and shaders into framebuffers, so yes, it's still a graphics processing unit. By the same token, a pair of GPUs that are used in Crossfire or SLI are still both GPUs when only one of them is connected to a monitor. The one that isn't connected to a monitor just has to hand its framebuffers off to another device before they can be displayed

      I realise you were being facetious, but is a PC still a 'personal' computer when it's just a server in a rack? Is a CD still a 'compact' disc, now that everything else makes a CD look bloody enormous? These aren't questions that should keep us up at night.

    3. Re:What puts "G" into "GPU"? by mi · · Score: 1

      Point is, this hardware has nothing to do with graphics — it is put into servers to aid specialized computations using chips originally developed for graphics. But they aren't used for graphics — not in this application. Thus "GPU" is a misnomer.

      --
      In Soviet Washington the swamp drains you.
  14. Re:DGEMM-Double precision GEneral Matrix Multiplic by Anonymous Coward · · Score: 0

    Of course OpenGL itself has supported this for ages (we use true 64-bit rendering for financial work). We generally use nVidia because AMD's (ie. ATI's) OpenGL has always been extremely buggy. AMD is just now catching up as usual with their piss-poor OGL implementation.

  15. Type Mismatch Error in 10 by stalky14 · · Score: 1

    The real question here is: when did "compute" go from being a verb to an adjective?

  16. Re:DGEMM-Double precision GEneral Matrix Multiplic by Anonymous Coward · · Score: 0

    Not true. Hermitian applies here as well. It's a transform that can be applied.