4 ms·

All they've got left at this point is "I want the best gaming performance and literally nothing else matters". For the average person maybe that's all they've g

All they've got left at this point is "I want the best gaming performance and literally nothing else matters". For the average person maybe that's all they've got left. But technically their are still ahead in some interesting ways when we look at x86 extensions: By decreasing order of impact: AVX 512 TSX SGX Their next architecture could bring bfloat16 too but I wonder how useful that would be on a CPU instead of on a GPU. AVX512 and TSX can give Intel a BIG performance advantage in some niche (but foundational) applications

LargoLasskhyfv 6y ago
By that metric alone you could also wait/hope/wish for Via Centaur Cha Ncore which i commented about @ https://news.ycombinator.com/item?id=22716893 https://news.ycombinator.com/item?id=22716893 in another context.
madengr 6y ago
I use some engineering software that uses Intel Math Kernel Libraries. It is 30% faster on an Intel CPU, versus AMD with equivalent speed/cores.
- jiggawatts 6y ago
  Is it a 30% perf/$ advantage though? In most comparisons I've seen Intel lost horrendously even with AVX-512 code compared to EPYC 2 when price was factored in. Doubly so if the cost of electricity and cooling is included, because the watts/core for AMD is much lower than anything Intel has.
  - threeseed 6y ago
    You can see my 10980xe Geekbench 4/5 results here: https://browser.geekbench.com/user/threeseed https://browser.geekbench.com/user/threeseed Compare against 3950x and 3960x and you will see it's closer to the 3960x for quite a bit cheaper and with 6 less cores. And that's not entirely AVX-512 code. Also power consumption is a little overblown since all CPUs scale their frequencies so even though mine runs at 5GHz it will spend most of the day below 2GHz or so.
- LargoLasskhyfv 6y ago
  Are you aware of the workaround for the mkl to not fall back to SSE on AMD? [1] https://old.reddit.com/r/MachineLearning/comments/f2pbvz/discussion_workaround_for_mkl_on_amd/ https://old.reddit.com/r/MachineLearning/comments/f2pbvz/dis...
  - madengr 6y ago
    No, thank you! I am going to try this.
- gameswithgo 6y ago
  that is because intel won’t fix a bug that doesn’t let the amd cpus use avx
  - FartyMcFarter 6y ago
    It's not a bug, there's a well documented history of Intel artificially crippling performance on non-Intel chips. They've done this with their compiler (and got sued for it), libraries, and they've even introduced code into glibc to do that.

LargoLasskhyfv 6y ago

By that metric alone you could also wait/hope/wish for Via Centaur Cha Ncore which i commented about @ https://news.ycombinator.com/item?id=22716893 https://news.ycombinator.com/item?id=22716893 in another context.

madengr 6y ago

I use some engineering software that uses Intel Math Kernel Libraries. It is 30% faster on an Intel CPU, versus AMD with equivalent speed/cores.

jiggawatts 6y ago
Is it a 30% perf/$ advantage though? In most comparisons I've seen Intel lost horrendously even with AVX-512 code compared to EPYC 2 when price was factored in. Doubly so if the cost of electricity and cooling is included, because the watts/core for AMD is much lower than anything Intel has.
- threeseed 6y ago
  You can see my 10980xe Geekbench 4/5 results here: https://browser.geekbench.com/user/threeseed https://browser.geekbench.com/user/threeseed Compare against 3950x and 3960x and you will see it's closer to the 3960x for quite a bit cheaper and with 6 less cores. And that's not entirely AVX-512 code. Also power consumption is a little overblown since all CPUs scale their frequencies so even though mine runs at 5GHz it will spend most of the day below 2GHz or so.
LargoLasskhyfv 6y ago
Are you aware of the workaround for the mkl to not fall back to SSE on AMD? [1] https://old.reddit.com/r/MachineLearning/comments/f2pbvz/discussion_workaround_for_mkl_on_amd/ https://old.reddit.com/r/MachineLearning/comments/f2pbvz/dis...
- madengr 6y ago
  No, thank you! I am going to try this.
gameswithgo 6y ago
that is because intel won’t fix a bug that doesn’t let the amd cpus use avx
- FartyMcFarter 6y ago
  It's not a bug, there's a well documented history of Intel artificially crippling performance on non-Intel chips. They've done this with their compiler (and got sued for it), libraries, and they've even introduced code into glibc to do that.