AMD's Future Chips & SoC's: News, Info & Rumours.

Page 81 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
Alleged AMD EPYC ‘Rome’ 7nm Based 64 Core Processor Performance Leaks Out – Scores an Incredible 12,500 Points in Cinebench Multi-Tasking Benchmark
https://wccftech.com/amd-epyc-rome-7nm-64-core-cpu-performance-benchmark-leak/
 
Aparently linux has no problem to keep all 32 cores well feed.
https://www.phoronix.com/scan.php?page=article&item=amd-linux-2990wx&num=4
https://www.techspot.com/review/1683-linux-vs-windows-threadripper-vs-core-i9/

That said, feeding 32 cores with only 4 memory channels is a handicap.
https://www.pcworld.com/article/3298859/components-processors/how-memory-bandwidth-is-killing-amds-32-core-threadripper-performance.html

And don't forget that the 16 core part has higher frequencies, so if your workload is not heavily threaded fewer but faster cores are the way to go.
See for instance Threadripper 2990WX vs i9-7980XE
ryzen_threadripper_2990wx_cinebench_thread_scaling_percentage-100767443-orig.jpg

 


I also hear that the next generation might be moving away from NUMA. This would be a good move to help solve RAM and performance issues with more than 32 cores moving forward.
 
Zen 2 engineering sample leak
8 cores, 16 threads, base clock of 4.0 GHz and a boost clock of 4.5 GHz
https://wccftech.com/amd-zen-2-ryzen-8-core-16-thread-cpu-leak/

Being an ES, take the data with a grain of salt. Final products can have different core counts and different clocks.
 
According to EETimes, TSMC has taped out their first N7+ design that will use EUV in mass production next year, and is on track to begin risk production of 5nm next April, ans start mass production in 2020.

https://www.eetimes.com/document.asp?doc_id=1333827&page_number=1

In the end I think that GF droping 7nm was very positive for AMD. As of today it's TSMC who has a significant lead over everybody else.

Intel server processors on 12 nm process (sorry I meant improved 10 nm 😀) aren't expected till 2020 2H, if TSMC remains on track AMD could be on TSMC 5nm by 2021.
 
If it goes well with Risk and then Ramp up is just the natural next step, 2021 for 5nm seems really feasible.

I'm surprised they are already getting into 5nm, to be fair. I thought they would've taken a bit longer.

Cheers!
 
Cray Is Back to Selling AMD-Powered Supercomputers

Cray announced it has sold a CS500 cluster powered by AMD EPYC processors to the Hass F1 Team, which will use the system to run aerodynamic simulations for its Formula One racecars.

https://www.top500.org/news/cray-is-back-to-selling-amd-powered-supercomputers/
 
the likelihood that i would switch to Intel after owning Ryzen is equal to 0 even if you gave me a million dollars i would stay with Ryzen.

Down below our my scores all stock with a H150i and 3200mhz memory

https://browser.geekbench.com/v4/cpu/10289606?fbclid=IwAR3b5zOLeQ9zAoKB_ufsEwbCLgAOM0dn4Fnf0MjpS-N29trQYR22YiqE2TM

https://www.3dmark.com/3dm/29531720?fbclid=IwAR1yrUr_b5z_XYrDR1V1UhpTesGG_FNYo_QYlREQ0hTioPwLOHvW1wjjqTY

My PC feels so snappy just like it did with a 4790K! Zen 2 and Zen 3 is coming and i know they will provide IPC improvements looking at 9900K benchmarks i have to say with a 15% clock speed advantage is it really that impressive when its only 20% faster then a 2700X?

I say no and i hope Amd does everything to improve their IPC if i owned that company i wouldn't spend one penny on increasing frequency i would put everything towards a 6 core CCX and IPC improvements!
 
No one wants to do it, because Intel has a disadvantage in mixed workloads that put all cores into activity. IF is better than Ring over 4 cores and Intel knows it. I'm even willing to say that Intel is still able to do so well because of how good their IMC is and how fast the Ring is as well. You lower the Ring speed and it will show.

Cheers!
 




I mean in several benchmarks i have compiled i'm willing to say that its close if not in Amd's favor by a few percentage points overall though Intel is 5% or so ahead in IPC but that number is so low and Amd is rumored to improve that number by 10%-15% with Zen 2.

Be extremely funny if Amd kept increasing IPC while intel was the one trying to race for the clock speed. Funny mainly over physics.

Also for a million dollars i guess i would say i would switch to Intel and just keep my 2700X
 


After all these years of price inflation with Intel holding a monopoly in high performance server chips, I would expect everyone to take on a percentage of AMD products, if for no other reason than to attempt to leverage better rates from Intel.
 
Some rumors and speculation about AMD EPYC II

https://semiaccurate.com/2018/10/29/more-details-about-amds-rome-cpu-leak/

https://www.youtube.com/watch?v=K4xctJOa6bQ&t=
 


64 cores 32MB Cache(double the 16MB Zen currently has) Simply put, more cache more performance. I hope the desktop parts are made out of these dies. 8 cores 32MB of cache would be pretty sweet@7nm. I've held off on upgrading until these guys get released.
 


Remember that cache access speeds go down as size increases; you can kill performance if you make the cache too large.
 
AMD continues to get design wins in HPC.

Cray revealed yesterday (october 30) the details of its next-gen supercomputing architecture, Shasta, selected to be the next flagship system at NERSC. The National Energy Research Scientific Computing Center, has chosen a Cray Shasta supercomputer for its NERSC-9 system, slated for delivery in late 2020. The system will feature AMD Epyc processors and Nvidia GPUs offering a combined peak performance of ~100 petaflops.

https://www.hpcwire.com/2018/10/30/cray-unveils-shasta-lands-nersc-9-contract/

Besides that, there are rumors that AMD might have gotten an Exascale design win in the context of the CORAL-2 request for proposals
More on CORAL-2 RFP: https://www.nextplatform.com/2018/04/09/bidders-off-and-running-after-1-8-billion-doe-exascale-supercomputer-deals/

A few days ago I already mentioned that
Cray announced it has sold a CS500 cluster powered by AMD EPYC processors to the Hass F1 Team, which will use the system to run aerodynamic simulations for its Formula One racecars.
https://www.top500.org/news/cray-is-back-to-selling-amd...
Click to expand...

Seems that AMD EPYC has a bright future.
 
More on NERSC-9 AMD based supercomputer at Top500 News
https://www.top500.org/news/crays-next-generation-supercomputer-headed-to-berkeley-lab-in-2020/

The system will be a mix of CPU-only nodes and GPU-accelerated nodes, with more of the former than the latter. The GPUs will be sourced from NVIDIA, and will most like be the company’s next-generation Tesla offering; AMD will be providing the CPUs, presumably based on its third-generation EPYC processors (codenamed Milan). The CPU-only nodes will be dual-socket servers, while the NVIDIA-accelerated nodes will have a single CPU and four GPUs.

Perlmutter represents the lab’s NERSC-9 machine, which as recently as 2017 was going to deliver over 100 petaflops of capacity. That suggests the NERSC originally had another design in mind, perhaps, like Cori, based on Intel silicon. Since Intel dropped its Xeon Phi processor line and is way behind schedule even getting its 10nm Xeon CPUs into the field, it’s possible the lab had to turn to AMD and NVIDIA to meet the system’s planned deployment date.
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom