Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AMD Ryzen price and release date revealed

pcworld.com

Post reply on HN
Print view View original

31–40 of 139 posts

Re: AMD Ryzen price and release date revealed

#31

Post by philipkglass »

gcp wrote: post #17
gnipgnip wrote: post #7

I wish people benchmark GEMM performance for all of us math folk.

Ah, but what should they benchmark GEMM with? MKL? I don't think so, it takes the worst codepath if the chip isn't GenuineIntel. OpenBLAS? Doesn't have Ryzen support. Doesn't even recognize some AMD cores. ACML? Wasn't updated since 2013 or so. This is a serious question. What are you supposed to do if you need a GEMM kernel for Ryzen? I sure hope AMD puts out updated ACML libraries real soon .

I'd like to see benchmarks with OpenBLAS. Unlike ACML it's under active development, unlike MKL it won't deliberately screw AMD performance, and it offers "pretty good" performance across every environment I've tried it in. Good enough that it's not worth paying for MKL, not worth going through ATLAS's self-tuning routine, not worth changing my build scripts to use vecLib under OS X. If OpenBLAS currently runs poorly on Ryzen I hope Ryzen will get some development love, because I kind of hate using ATLAS and at this point it would take a major advantage to tempt me back away from open source components.

Re: AMD Ryzen price and release date revealed

#32

Post by bryanlarsen »

cat199 wrote: post #26
mtgx wrote: post #21

Jim Keller deserves a frigging statue in front of the AMD HQ, if he doesn't have one there already. I'm not even kidding. His efforts shouldn't be easily forgotten, and it could serve to inspire new generations of AMD engineers. Beyond the product quality itself, I think AMD has had a pretty smart launch strategy by releasing the CPU chips first to show that it can beat "Intel's best". But they really need to start f…

From what I recall of their recent marketing material, it seemed like like they are looking to corner the market on 64bit ARM servers and also have some hybrid amd64/ARM chip plans to help people wanting to take advantage of ARM in datacenters migrate.. but I'm no expert and this is based on a recollection of marketing materials from a few months back - 64bit arm could be a huge thing in large scale light-duty comput…

AMD K12[1] is their full-custom ARM64 core. Originally scheduled for 2016, it was delayed to focus on Zen, probably the right move.

It's still scheduled for 2017, but there hasn't been much noise about it for a long time so I suspect it's been delayed, as they're still focusing on Zen & Vega.

Zen & K12 probably share a lot of elements, so if Zen is awesome, K12 probably will be too.

1: https://en.wikipedia.org/wiki/AMD_K12

Re: AMD Ryzen price and release date revealed

#33

Post by deepnotderp »

gcp wrote: post #15
deepnotderp wrote: post #11

Earlier quoted context omitted.

Iirc gemm and float performance isn't very good, since amd made a decision to reduce math performance in order to decrease latency.

It's not so much reduced math performance (it's twice faster than Bulldozer!), it's that Intel chips as of Haswell have dual 256-bit FPUs per core. If you use them, the Intel chips downclock (sometimes severely) in order not to violate their TDP, but the dual FPUs are still there, and it's still a win for GEMM. I can see why AMD didn't follow along here, but it could be a factor in some small spaces - when you need G…

Wow, who uses 256-bit floats?

Re: AMD Ryzen price and release date revealed

#34

Post by arca_vorago »

This is really exciting, as I have been waiting to build a new system with the new AMD gear. Lots of people also don't hear about it, but I am super-excited to see the new server class CPU's. I built a quad opteron 6380 system and have been in love with them ever since, but they weren't perfect and had some issues I hope are fixed with this new line.

Re: AMD Ryzen price and release date revealed

#35

Post by jsheard »

deepnotderp wrote: post #33
gcp wrote: post #15

Earlier quoted context omitted.

It's not so much reduced math performance (it's twice faster than Bulldozer!), it's that Intel chips as of Haswell have dual 256-bit FPUs per core. If you use them, the Intel chips downclock (sometimes severely) in order not to violate their TDP, but the dual FPUs are still there, and it's still a win for GEMM. I can see why AMD didn't follow along here, but it could be a factor in some small spaces - when you need G…

Wow, who uses 256-bit floats?

It doesn't handle 256-bit floats, in this context 256-bit means it can operate on eight 32-bit floats or four 64-bit floats at the same time.

Re: AMD Ryzen price and release date revealed

#36

Post by mizzack »

mtgx wrote: post #21

Jim Keller deserves a frigging statue in front of the AMD HQ, if he doesn't have one there already. I'm not even kidding. His efforts shouldn't be easily forgotten, and it could serve to inspire new generations of AMD engineers. Beyond the product quality itself, I think AMD has had a pretty smart launch strategy by releasing the CPU chips first to show that it can beat "Intel's best". But they really need to start f…

It'll be interesting to see how Vega and on-die HBM ties into all of this with Infinity Fabric. AMD has been laying this foundation for a while.

A "Zega" APU would be killer in notebooks as well as HPC...

Re: AMD Ryzen price and release date revealed

#38

Post by Cyph0n »

tracker1 wrote: post #27

I just have to say, I really hope that this is real. AMD has burned a lot of trust a few times in overzealously stating performance marks in their prior generations of CPUs. I ran AMD for several years until the Core-2 came out from Intel (currently very happy with an i7-4790K). If these comparisons are real, my next build may be AMD again.

Seems like you skipped over the FX series (Bulldozer, I believe). I got a FX-8320 a few years back, and my brother got a FX-6300. Both are capable processors and are MUCH cheaper than their counterparts (i5 and i3, respectively).

Heck, if I built a PC in the past year, I would have gone with a FX-8350 just because it's so damned cheap yet very performant. Once Zen is out, it's really a no-brainer for me.

Re: AMD Ryzen price and release date revealed

#39

Post by wmf »

mizzack wrote: post #36
mtgx wrote: post #21

Jim Keller deserves a frigging statue in front of the AMD HQ, if he doesn't have one there already. I'm not even kidding. His efforts shouldn't be easily forgotten, and it could serve to inspire new generations of AMD engineers. Beyond the product quality itself, I think AMD has had a pretty smart launch strategy by releasing the CPU chips first to show that it can beat "Intel's best". But they really need to start f…

It'll be interesting to see how Vega and on-die HBM ties into all of this with Infinity Fabric. AMD has been laying this foundation for a while. A "Zega" APU would be killer in notebooks as well as HPC...

I think Raven Ridge is what you're looking for.

Re: AMD Ryzen price and release date revealed

#40

Post by RussianCow »

Cyph0n wrote: post #38
tracker1 wrote: post #27

I just have to say, I really hope that this is real. AMD has burned a lot of trust a few times in overzealously stating performance marks in their prior generations of CPUs. I ran AMD for several years until the Core-2 came out from Intel (currently very happy with an i7-4790K). If these comparisons are real, my next build may be AMD again.

Seems like you skipped over the FX series (Bulldozer, I believe). I got a FX-8320 a few years back, and my brother got a FX-6300. Both are capable processors and are MUCH cheaper than their counterparts (i5 and i3, respectively). Heck, if I built a PC in the past year, I would have gone with a FX-8350 just because it's so damned cheap yet very performant. Once Zen is out, it's really a no-brainer for me.

The biggest problem with the FX series, IMO, was the deceptive marketing. The FX-8000 CPUs were marketed as being 8-core, when in reality every pair of cores shared an FPU and cache, so real-world performance was much closer to that of a quad-core for many workloads. Add on top of that the abysmal IPC (compared to Intel's offerings) and high power consumption, and there was very little reason to get an FX-8350 or similar over an i5.

(Yes, today the FX CPUs are much cheaper, but when they came out they were barely so. I know because I got an FX-8350 as soon as it came out and regretted it.)

Post reply on HN