Recent content by JF_Aidan_Pryde

  1. J

    Nvidia Volta Speculation Thread

    With the new info, I think I've finally managed to do the math to get to Xavier's 30 tera ops. Basically it's a sum of tensor ops, FP32 ops, and DLA ops. Initially Xavier was 20 TOPs @ 20 watts. Then NVIDIA bumped it to 30 TOPs @ 30 watts. Basically a 50% power bump. So I bumped the clocks by...
  2. J

    NVIDIA Tegra Architecture

    Some indirect measurements of Google's TPU from Google's paper on neural machine translation. They note that because of workload mix / CPU-GPU transfer not being optimal, GPU is not performing optimally in these measurements.
  3. J

    NVIDIA Tegra Architecture

    https://blogs.nvidia.com/blog/2016/09/28/xavier This is a pretty crazy chip: 7 billion transistors 512 cores 20 Tera ops 16nm FF 20 watts Due end of 2017 NVIDIA claims this one chip backs all the power of a Drive PX 2 computer (2 x Parker + 2 x Parker). I haven't figured out how this is...
  4. J

    Pascal FP16/FP32/INT8 Performance

    Seems like Nvidia is being really smart about how they segment their products: – If you want high DP perf, must buy GP100 – GP100 is also the best card for training since it's the only desktop card that does 2x FP16. Assume FP16 is good for training but INT8 isn't. – If you want really fast...
  5. J

    Pascal FP16/FP32/INT8 Performance

    Hey guys, Wanted to start a clean thread on since it's a little hard to navigate the uber threads. Just want to get a clean record for the various instruction rates of various Pascal based parts. NVIDIA has provided a sparse matrix of information. Of note is the restriction of FP16 on consumer...
  6. J

    Nvidia Pascal Speculation Thread

    Where are you getting that from? Last year CES slides showed Tegra X1 having 30 images/sec for AlexNet. This year's slides show 450 images/sec for Titan X. That's 15x delta, on par with FP32 perf delta.
  7. J

    Nvidia Pascal Speculation Thread

    But it doesn't add up even with "deep learning ops". PX 2 has ~3x more "deep learning OPS" vs. Titan X but it has ~6x more AlexNet performance. There'a another magical "2x" something going on.
  8. J

    Nvidia Pascal Speculation Thread

    Dumb question—how is it that Drive PX 2 has 6x the ImageNet performance of Titan X when it only has ~3x the "deep learning ops" performance? On which actual hardware performance metric (ram, flops, dl-flops) does it have 6x over Titan X?
  9. J

    Finance website with earnings charts?

    I would like to see a historical chart of earnings. I don't see that option on Google finance.
  10. J

    Finance website with earnings charts?

    Plenty of websites have good charts of stock price but where can I get an overlay of earnings? Or margins? Any recommended software/websites that can display a good range of fundamental information? Thanks.
  11. J

    Why AA doesn't get the respect it probably deserves

    The AA debate began in the Voodoo 5 / Geforce 2 era. Back then, the typical playable AA resolution was 1024 x 768 on a 17" CRT. I think it should be realised that AA is not as desparately needed as back then. The pixel density of today's good LCD panels is about 3 times as a 17" CRT at 1024 x...
  12. J

    The most entertaining essay I've ever read

    I've read it twice and it still makes me laugh with unrivalled smugness. Bertrand Russell's "In Praise of Idleness": http://www.zpub.com/notes/idle.html
  13. J

    Ethics as it pertains to returning broken stuff

    Technically, he did a bad, illegal thing. But everyone knows law does not equate to ethics. But let's just say that he did the unethical thing. The second question in this thread relates to the response that he god as a result of what he did. Personally I think this thing is pretty trivial...
  14. J

    3dm07

    You go girl! (I want to go to Crete for my next holiday)
  15. J

    Activision's 360 games not running in full 720p?

    What a mess. Makes you wonder if it was really worth the trouble of taking the ROP and putting it on an eDRAM daughter die.
Back
Top