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

AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel

zdnet.com

Post reply on HN
Print view View original

21–30 of 165 posts

Re: AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel

#21

Post by arcticbull »

exar0815 wrote: post #9
guelo wrote: post #4

Why is AMD making these chips with 14nm transistors instead of using TSMC's 7nm stuff? AMD is only catching up to Intel because Intel's 10nm has been delayed so much, but if Intel can ever figure out their 10nm process they'll be in the lead again.

AFAIK is Threadripper 2 12nm. I guess its because they can deliver very high speed with that process, without cannibalizing the already overtaxed TSMC-7nm Lines. Threadripper-users dont care that much about energy efficiency - and its a small market, which gets cannibalized by the 3950X in the new generation.

It's 7nm, similar to Rome.

Re: AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel

#22

Post by randogogogo »

guelo wrote: post #4

Why is AMD making these chips with 14nm transistors instead of using TSMC's 7nm stuff? AMD is only catching up to Intel because Intel's 10nm has been delayed so much, but if Intel can ever figure out their 10nm process they'll be in the lead again.

This can be a bit confusing, basically some parts within the CPU package 7nm dies and other parts are 14nm dies. The chiplets that are the CPU core are the prior, interconnects (or the I/O die) is the latter.

Re: AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel

#23

Post by ogrisel »

Symmetry wrote: post #2

AMD is in a really good position right now. Intel still has the lead in low idle power which is good in laptops. Ryzen lets different cores have different max frequencies so if your code is single threaded and your operating system isn't the newest that could be a reason to go Intel. Likewise if its single threaded and can take advantage of AVX-512 or the Intel Math Kernel Library. But otherwise?

The BLAS & Lapack subset of the API of the Intel Math Kernel Library (MKL) is very well implemented in open source projects such as OpenBLAS and BLIS:

https://github.com/flame/blis

Both are well optimized for AMD CPUs.

Re: AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel

#24

Post by vj44 »

ogrisel wrote: post #23
Symmetry wrote: post #2

AMD is in a really good position right now. Intel still has the lead in low idle power which is good in laptops. Ryzen lets different cores have different max frequencies so if your code is single threaded and your operating system isn't the newest that could be a reason to go Intel. Likewise if its single threaded and can take advantage of AVX-512 or the Intel Math Kernel Library. But otherwise?

The BLAS & Lapack subset of the API of the Intel Math Kernel Library (MKL) is very well implemented in open source projects such as OpenBLAS and BLIS: https://github.com/flame/blis Both are well optimized for AMD CPUs.

I work in this space... and let's just say that MKL is definitely NOT well optimized for AMD's chips. You'll be lucky to get 10-20% efficiency. Nevermind openblas.

Re: AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel

#25

Post by dragontamer »

Symmetry wrote: post #2

AMD is in a really good position right now. Intel still has the lead in low idle power which is good in laptops. Ryzen lets different cores have different max frequencies so if your code is single threaded and your operating system isn't the newest that could be a reason to go Intel. Likewise if its single threaded and can take advantage of AVX-512 or the Intel Math Kernel Library. But otherwise?

The main selling point for Intel isn't speed anymore.

Intel still has superior performance counters and debugging features. Mozilla's rr (Record and Replay framework) only works on Intel for example, and Intel vTune is a very good tool. AVX512 is also an advantage, as you've noted.

There are other instruction set advantages: I think Intel has faster division / modulus operator, and also has single-clock pext / pdep (useful in chess programs).

For most people however, who might not be using those tools, I'd argue that AMD's offerings are superior.

Re: AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel

#26

Post by DuskStar »

arcticbull wrote: post #19
guelo wrote: post #4

Why is AMD making these chips with 14nm transistors instead of using TSMC's 7nm stuff? AMD is only catching up to Intel because Intel's 10nm has been delayed so much, but if Intel can ever figure out their 10nm process they'll be in the lead again.

Their biggest win over Intel was using 14nm IO bridge to connect 7nm CPU cores ("chiplets") in the Ryzen 3000 era. They use 14nm for IO where there's little benefit to the process shrink so they can save the lower-yielding 7nm process for where it counts: cores.

And I think you're missing the biggest bit, there - each set of 8 cores is its own die. That means that if they have a yield of 95% for that one die, they'll have a yield of 95% for a 64 core monster, too. Compare with Intel, where octupling the die area means their yields would go from 95% to 0.95^8 = 66%. (And if they had 50% yields to start, Intel would end up with 0.4% yields after octupling - AMD would still have 50% total)

Re: AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel

#27

Post by localhost »

vj44 wrote: post #24
ogrisel wrote: post #23

Earlier quoted context omitted.

The BLAS & Lapack subset of the API of the Intel Math Kernel Library (MKL) is very well implemented in open source projects such as OpenBLAS and BLIS: https://github.com/flame/blis Both are well optimized for AMD CPUs.

I work in this space... and let's just say that MKL is definitely NOT well optimized for AMD's chips. You'll be lucky to get 10-20% efficiency. Nevermind openblas.

Intel optimizes these libraries down to the stepping level of the CPUs. So not surprising if they are not optimized at all for AMD

Re: AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel

#28

Post by exar0815 »

arcticbull wrote: post #21
exar0815 wrote: post #9

Earlier quoted context omitted.

AFAIK is Threadripper 2 12nm. I guess its because they can deliver very high speed with that process, without cannibalizing the already overtaxed TSMC-7nm Lines. Threadripper-users dont care that much about energy efficiency - and its a small market, which gets cannibalized by the 3950X in the new generation.

It's 7nm, similar to Rome.

Threadripper 2 was 12nm, Threadripper 14. The new TR3 will be 7nm

Re: AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel

#29

Post by marmaduke »

vj44 wrote: post #24
ogrisel wrote: post #23

Earlier quoted context omitted.

The BLAS & Lapack subset of the API of the Intel Math Kernel Library (MKL) is very well implemented in open source projects such as OpenBLAS and BLIS: https://github.com/flame/blis Both are well optimized for AMD CPUs.

I work in this space... and let's just say that MKL is definitely NOT well optimized for AMD's chips. You'll be lucky to get 10-20% efficiency. Nevermind openblas.

Parent said OpenBLAS and Blis, not MKL, are optimized for AMD.

Re: AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel

#30

Post by marmaduke »

astrodust wrote: post #17
GiorgioG wrote: post #14

I have the original Threadripper 1950X overclocked to 4.1ghz and I’ve been pretty happy with it. As a developer I can’t find much reason to upgrade 2 years after its release. Single core improvements (from AMD or Intel) aren’t game-changing. More cores won’t do anything for me (.NET Core, angular, etc) I’ve considered switching back to intel for their 5ghz processors but based on benchmarks, I wouldn’t see anything b…

If you ever get into Docker/Kubernetes, being able to spin up a local cluster for testing is a pretty big deal. Having that running in the background and not impacting your usual workload is huge. I've been testing a Kubernetes cluster built on 32-core AMD servers and it's unreal the workloads you can throw at them. I'm used to 4 or 8-core Xeon chips and this is a whole different game.

How much of that is core count though? I've used minikube several times and it didn't seem to matter much how many cores it had (beyond 4)
Post reply on HN