AMD Rome EPYC Zen2 reviews

Pretty impressive, really positive reception everywhere.

I really don't know much about their intended markets, how easily does a server farm or w/e switch platforms?
 
Pretty impressive, really positive reception everywhere.

I really don't know much about their intended markets, how easily does a server farm or w/e switch platforms?

Virtualization just needs a new cluster then migrate. Can't mix and match Intel and AMD in the same cluster and expect HA to work.

Windows and linux bare metal builds should be pretty straight forward.
 
Most of my server experience is with multi-database transaction servers, and I gotta say 128 threads and 128 lanes is pretty stonkin'. I can remember when we got a Sequent server with 12 486DX/25's and a 45-drive SCSI cluster (9 x RAID 5 for a total of 1 gigabyte of storage), and we thought we were hot shit.

P.S. And we were.
 
Last edited:
Last edited:
This is a big enough difference that I see AMD getting back to that glorious 2006-era 25% server market share from Intel inside three years.


Even worse, the 50% higher performance/watt probably means any momentum ARM server chips currently have is suddenly dead in the water.
 
Last edited:
That's an impressive processor. I'll never have need of one, but wow.
Competition is good. We will see what the future holds and what will trickle down to consumers eventually.

Imagine water-cooling on that Epyc tho. That I'd like to see.
 
No one has any love for this review? I think one of the best in terms of analysis. https://www.servethehome.com/amd-epyc-7002-series-rome-delivers-a-knockout/
Had a laugh at Anadtech doing a single core shootout with server processors.


Some of my favorite quotes:

CI/CD Use
... massive caches, more memory bandwidth, and more cores mean that the AMD EPYC 7002 series does more than keep up. It is not just an Intel Xeon Platinum competitor. Instead, the question is whether AMD has a 2:1 consolidation ratio over Intel Xeon Platinum.

7-Zip Performance
AMD still has the better platform, but we think here the Xeon Platinum 8280 can be very competitive with an 80% discount off of list price.

OpenSSL Performance
We think Intel is still very competitive here with Xeon Platinum 8280 street pricing in the $1000-1200 range.
(note the 8280 lists for $13000, so a typical 60% discount still puts it at $7800)

GROMACS w/AVX-512
The Intel Xeon Platinum 8280 system was using 40% more power than the AMD EPYC 7742 system here.

...

Even with AVX-512 and better optimizations, the Intel Xeon chips are about on par with their AMD counterparts, yet use more power to deliver similar performance. We look at this as more of a worst-case scenario for EPYC 7002 and it is still competitive.

Re: one of the test systems used
We are also not allowed to name because Intel put pressure on the OEM who built it to have AMD not disclose this information, despite said OEM having their logo emblazoned all over the system. Yes, Intel is going to that level of competitive pressure on its industry partners ahead of AMD’s launch.

Performance/W
In every dual-socket test, we are seeing the AMD EPYC 7742 out-perform the dual Intel Xeon Platinum 8280 configuration sometimes by ~2x, and yet the maximum power is lower.

Looks like a solid win on almost all fronts to me.
 
"Patrick Moorhead, founder of Moor Insights & Strategy, said he estimates that AMD took “low single-digit” market share from Intel with the first generation of EPYC server chips and that he expects that share to expand with the second generation announced Wednesday.

In terms of performance, Moorhead said the new AMD chips are better than Intel’s chips at some kinds of computing work but may lag on others. He said that new Intel chip features for machine learning tasks and new Intel memory technology being with customers such as German software firm SAP SE (SAPG.DE) could give Intel an advantage in those areas.

Moorhead said in a research note that most every large business he talks to “wants more competition in the space to accelerate innovation and lower costs. With that said, none of these customers would adopt AMD if it didn’t have some advantages.”

https://www.reuters.com/article/us-amd-alphabet-idUSKCN1UX2KL
 
.....In terms of performance, Moorhead said the new AMD chips are better than Intel’s chips at some kinds of computing work but may lag on others....

Wow, just wow. I don't think I found anything Intel was faster at in any of the benchmarks besides memory latency, and that is not really an application benchmark.
From what I have seen, Rome destroys Intels top dog by almost 2x while using LESS power.
 
Wow, just wow. I don't think I found anything Intel was faster at in any of the benchmarks besides memory latency, and that is not really an application benchmark.
From what I have seen, Rome destroys Intels top dog by almost 2x while using LESS power.

It's a strange quote: why would he claim they are better at some things, but then speculate that they may lag on others? Either you have the performance data or you don't.
 
If anyone wants even more performance comparisons, Phoronix has a short new article: https://www.phoronix.com/scan.php?page=news_item&px=AMD-EPYC-7742-2P-Summary

Phoronix said:
As explained yesterday, the only areas where Cascadelake (green-ish line) is beating AMD's Rome is with their own MKL-DNN software (in some of the sub-tests at least), in MariaDB/MySQL strangely but that is being further investigated, and then when testing single-threaded workloads individually without fully taxing them in parallel as would be common in a production server.
 
Last edited:
Pretty nice when a single AMD CPU bests a pair of the best Intel makes (in most benchmarks).
 
  • Like
Reactions: GHRTW
like this
Had a laugh at Anadtech doing a single core shootout with server processors.

But relevant. I mean, if you can have 50 VMs running "ok" vs. 40 VMs running pretty darn fast. I know as a "lab" user, feeling "fast" is important to me in addition to scale.
 
But relevant. I mean, if you can have 50 VMs running "ok" vs. 40 VMs running pretty darn fast. I know as a "lab" user, feeling "fast" is important to me in addition to scale.

Well considering Zen2's IPC is generally slightly above intel's at the moment, It is expected to have the same or better single core performance per clock speed.
For VMs unless you're doing CPU intesive stuff, storage IO latency/speed is more important for responsiveness anyways at least in my limited experience.
 
It's a strange quote: why would he claim they are better at some things, but then speculate that they may lag on others? Either you have the performance data or you don't.

They are speculating because they haven't been able to find the specific benchmark to make it true yet, while there is proof AMD is faster in others :).
 
Last edited:
It's a strange quote: why would he claim they are better at some things, but then speculate that they may lag on others? Either you have the performance data or you don't.

Also...Moorhead needs more time to divest of his intel holdings...
 
mhw98x.gif


If those don't boost their next quarter earnings, I don't know what will. Total Decimation at half the price
 
Goes twice as fast, 40% cheaper.

Yet apparently this would be impossible for them to do to nVidia because business.

But still, imagine nVidia getting this colossally murdered. A card fully twice the speed of a 2080ti for the price of a 2070. That's how far up shit creek Intel is right now.
 
Ok, so I rarely keep up with these launches, as I'm still on Haswell and happy. But I have a question.

I only read the Phoroix article, but it seems that the Intel 8280 chip is 28/56 and the AMD 7742 is 64/128? Am I reading that right?

If that's the case why is everyone surprised it is beating the Intel chip? At first I was unsure if my numbers were correct, but I checked the first page opening notes and it states the AND chip beats Intel considerably in both cores and threads.

Am I missing something?

Don't get me wrong, the power usage being fairly close on a chip with more than twice the cores and threads made me lol.
 
Ok, so I rarely keep up with these launches, as I'm still on Haswell and happy. But I have a question.

I only read the Phoroix article, but it seems that the Intel 8280 chip is 28/56 and the AMD 7742 is 64/128? Am I reading that right?

If that's the case why is everyone surprised it is beating the Intel chip? At first I was unsure if my numbers were correct, but I checked the first page opening notes and it states the AND chip beats Intel considerably in both cores and threads.

Am I missing something?

Don't get me wrong, the power usage being fairly close on a chip with more than twice the cores and threads made me lol.
it's two of the intel cpus, vs one epyc. Still more threads on the epyc system, but only a few.

Edit: well, two epycs, but I dunno if it's for all tests. would have to read the whole article
 
Intel would have been broken up because of Monopoly laws but AMD just saved them :)
 
Thanks! I knew I was missing something for sure.
Just looked again, it is two epycs, so maybe still lopsided. I'd have to finish the article to understand better, but probably a combination of price, and power, besides the sheer number of threads and pcie lanes.
 
Just looked again, it is two epycs, so maybe still lopsided. I'd have to finish the article to understand better, but probably a combination of price, and power, besides the sheer number of threads and pcie lanes.

Ok, so maybe my initial question still stands. Outside of price, how can we expect a system with more cache, more cores, more threads, 8 channel DDR4 AND a faster DDR4 ability to NOT best it?
 
I only read the Phoroix article, but it seems that the Intel 8280 chip is 28/56 and the AMD 7742 is 64/128? Am I reading that right?
yes
If that's the case why is everyone surprised it is beating the Intel chip?
AMD just launched a processor that is a lot cheaper than intel's high end, outperforms intel's high end in multithreaded workloads, and does it with about the same amount of power.

It is surprising that AMD managed to do that. Although I guess we still have to see if they can supply the market with those prices.
 
Ok, so maybe my initial question still stands. Outside of price, how can we expect a system with more cache, more cores, more threads, 8 channel DDR4 AND a faster DDR4 ability to NOT best it?
Well, I don't know how many people are surprised, but I imagine there are some who are happy to know that it beats it, even if it does take double the threads to do it, if it still uses less power to get there. Wouldn't you be, if that was part of your bottom line?
 
Well, I don't know how many people are surprised, but I imagine there are some who are happy to know that it beats it, even if it does take double the threads to do it, if it still uses less power to get there. Wouldn't you be, if that was part of your bottom line?

I definitely mentioned that the cost makes it a way better option. Power as well.

I just felt that everyone in here was going nuts over something that would be impossible to be worse.
 
I definitely mentioned that the cost makes it a way better option. Power as well.

I just felt that everyone in here was going nuts over something that would be impossible to be worse.
It's definitely not impossible, though I agree it's very unlikely considering Epyc's historical performance, and generally good thread scaling.
 
This is a big enough difference that I see AMD getting back to that glorious 2006-era 25% server market share from Intel inside three years.


Even worse, the 50% higher performance/watt probably means any momentum ARM server chips currently have is suddenly dead in the water.
You are right, it is starting to feel like that again, and AMD is definitely the only company truly keeping x86-64 alive, thriving, and innovative at this point.
What is Intel doing? Oh right, those continuously growing security holes making their processors feel less like x86-64 and more like Swiss cheese.
 
Back
Top