AMD Ryzen (Summit Ridge) Benchmarks Thread (use new thread)

[sirmo]

I work for one of the firms you mentioned. The reality in these corps is, decisions aren't made by superstar engineers but by their management and the execs. Most of whom be non-technical, businessy, snakeoil salesmen.

Excellent decisions worth an easy billions in profit get canned. Viable, cheaper solutions get overlooked. In-house excellency gets buried. Products far ahead of time don't even get a chance. All without explanation to the superstar engineers.

That's the reality of international corporations ESPECIALLY when things are not going well. Corporate politics is absolutely crazy.

Engineering/Science =! Business/Politics.

Trust me, it's not the engineering I doubt.

That's always implied. I agree about the executives (been an engeneeir for 20+ years).

I do feel like Lisa Su and Mark Peppermaster are different though, they are both engineers. I get that vibe from them. There was a panel with Mark and Jim on youtube while back.

And I really like how Lisa took a moment to thank her people when she unveiled Zen few months back. https://youtu.be/fBHYdoSYMZk?t=3450
Those words seem genuine.

Dirk Meyer was also an engeneeir and Bulldozer came out under his leadership.. so YMMV. But I do hope AMD comes back to being competitive again. It would be great for the market and us enthusiasts as a whole.

 

[frozentundra123456]

That there is a 4C/8T Zen at all to test at this point is surprising to me. I was expecting Zen to be 8C only for a little while, at least.
If there are already 4C Zen chips being tested, then I am puzzled by the lack of a release.

Yea, makes me suspicious of the validity of the "leak", actually. By all accounts Zen was only supposed to be 8 core.

 

[CHADBOGA]

Let me remind you this quote
“It is the first time in a very long time that we engineers have been given the total freedom to build a processor from scratch and do the best we can do,” said Suzanne Plummer, a director of design engineering at AMD and also a veteran Austin chip engineer, who heads development of a “Zen”-based processor, in an interview with MyStatesman.

When you give total freedom to a team of experiences engineers, the result can't disappoint you.

Total freedom is nice and all, but level of resourcing matters a lot too.

 

[LTC8K6]

It seems, you missed the 4C8T ES on Zauba months ago.

Are we talking about the APU version of Zen, or just plain Zen, regarding 4C/8T?

 

[bjt2]

And you would expect both cores to be load store limited in that workload, which is why i kind think all these port counting things get a little silly.

If you have a calculation involving only FADD and/or FMUL, probabily you will be memory limited. A FDIV suffice to be not memory limited anymore. And if you use SIMD, probabily you need also pack and shuffle instructions. You will be memory limited only if the memory instructions are about 50% and there are only few low latency/high throughput instructions (max 3-4 FADD/FMUL/FMAC/PACK/SHUFFLE instructions each load or store).
Actual code is much more complicated that vector add or mul...

 

[bjt2]

Well, the processors are there with their optimization manuals. I don't know what kind of "secret sauce" is still missing...

There are things like branch prediction algorithms that can't be deduced even from optimization manuals... And Zen optimization manuals are not out yet.
Plus, we here take as given that Zen will have lower IPC even than haswell... But we have seen only a bench here: blender...

 

[bjt2]

In science, hypothesis can be pretty absurd without empirical test results.

Meaning, we will have to test the chip TO THEN see if it has any weak points, Achilles Heel, glass jaws. In the past 11 years, on paper, things have nearly always looked FAR better than they turned out in reality.

Only one major weakness is needed to make this chip a failure. That doesn't have to be uarch. Cache and branch prediction are make or break but it can also be the power, clocks or yields.

Sent from HTC 10

These are the culprits... The engine is solid... We must see the controller (branch prediction) and the feeding unit (caches)... Historically INTEL was better (way better) in both... Let's see if AMD has cathed up...

 

[Dresdenboy]

Are we talking about the APU version of Zen, or just plain Zen, regarding 4C/8T?

It says "Zeppelin" like for the 8C variant (same die).
https://www.zauba.com/import-4c-8t-am4-hs-code.html

 

[LTC8K6]

It says "Zeppelin" like for the 8C variant (same die).
https://www.zauba.com/import-4c-8t-am4-hs-code.html

Zeppelin was supposed to be the server type APU, wasn't it?

 

[The Stilt]

Zeppelin is the name of the actual die design. The die is used in Summit Ridge, Snowy Owl & Naples. Raven is the name for the APU (die) design based on Zen µarchitecture.

 

[LTC8K6]

Ah, I see it is Wiki that has Zeppelin as a server APU, which is probably just wrong.

 

[Dresdenboy]

Ah, I see it is Wiki that has Zeppelin as a server APU, which is probably just wrong.

According to the leaked Fudzilla slide, there are Zeppelin dies next to some other dies (Greenland, HBM stacks) on an interposer.

 

[cdimauro]

There are things like branch prediction algorithms that can't be deduced even from optimization manuals...

That's why I'm warmly waiting for some test with emulators, compilers, etc.

And Zen optimization manuals are not out yet.

At least Intel's one is out.

Plus, we here take as given that Zen will have lower IPC even than haswell... But we have seen only a bench here: blender...

As I said many times, it's better to wait for a wider set of tests.

However I think that AMD chose Blender because it favored its micro-architecture. So, it's like an upper/best case. But'd like to see more tests even in MT area.

 

[KTE]

^They used 55% for EXC boost in their marketing slides and power efficiency something ridiculous (like 2.5x) based off that because it was the most they could find (CB), whereas the real average increase was 9-13%, as they later revealed.

Sent from HTC 10

 

[sirmo]

http://blogs.barrons.com/techtrader...uraged-by-renduchintalas-pragmatic-influence/

Intel president, is expecting Zen to disrupt the market until 10nm parts.

Our faith in 10 nm raising the bar for enthusiast PCs is why we see the threat presented by AMD’s Zen as being fairly manageable, with only short term disruption in 2017.

 

[Arachnotronic]

http://blogs.barrons.com/techtrader...uraged-by-renduchintalas-pragmatic-influence/

Intel president, is expecting Zen to disrupt the market until 10nm parts.

Intel's Murthy Renduchintala said no such thing.

Here's what he actually said, I can't believe I had to transcribe this.

First of all, I think competition is an absolutely necessary environment to get the best out of any engineering organization. Again, our attitude towards our competitors, whether it be AMD or anybody else is to treat them with the highest degree of respect. And that translates into an internal virtue of basically never taking your current position for granted.

So we continue to focus on delivering technology that we think represents the best we can possibly do, and if we focus on the best we can possibly do, then the competition will hopefully be appropriately taken care of in our roadmap going forward.

So for us, we're very focused on achieving our performance cadence, our performance attributes, in our new products going forward.

I have to say, I haven't seen enough of the Zen to calibrate its exact competitive level, but we're treating it like it's going to be a legitimate competitor and therefore we need to deliver our best to make sure we're able to deliver the business performance that we expect to deliver over the course of the next 12 months.

 

[sirmo]

Intel's Murthy Renduchintala said no such thing.

Here's what he actually said, I can't believe I had to transcribe this.

Wait, so did Tiernan Ray misquote him? I literally quoted from the original quote. Are you sure it's the same interview you're quoting, have a source?

 

[Arachnotronic]

Wait, so did Tiernan Ray misquote him? I literally quoted from the original quote. Are you sure it's the same interview you're quoting, have a source?

http://edge.media-server.com/m/p/v5rzwrhk

Morgan Stanley conference. Tiernan Ray does good work, this one's not on him. This is just a flaw with the analysis that Ray was quoting. The quote does not come from Murthy himself.

 

[sirmo]

http://edge.media-server.com/m/p/v5rzwrhk

Morgan Stanley conference. Tiernan Ray does good work, this one's not on him. This is just a flaw with the analysis that Ray was quoting. The quote does not come from Murthy himself.

He didn't even mention 10nm in that audio portion talking about AMD/Zen.. and that just seems like such an odd thing to throw in if it wasn't said at all. This could be something he said in a 1-1 chat.

 

[Arachnotronic]

He didn't even mention 10nm in that audio portion talking about AMD/Zen.. and that just seems like such an odd thing to throw in if it wasn't said at all. This could be something he said in a 1-1 chat.

He did mention 10-nanometer during the conference, but not specifically in the AMD/Zen portion. It's the bit where the analyst says that Intel originally focused on clock speeds, etc. but now desktop performance improvements haven't been as large. Murthy then says something along the lines of "all of the above" in terms of improving performance at the high end and improving battery life in thin-and-light devices.

 

[sirmo]

He did mention 10-nanometer during the conference, but not specifically in the AMD/Zen portion. It's the bit where the analyst says that Intel originally focused on clock speeds, etc. but now desktop performance improvements haven't been as large.

Right, the quote from Barron's specifically mentioned 10nm and Intel as it pertains to Zen, which is why I think these are different interviews.

 

[Arachnotronic]

Right, the quote from Barron's specifically mentioned 10nm and Intel as it pertains to Zen, which is why I think these are different interviews.

Nope, the analyst is referring to this interview (remember, the Barron's quote is quoting the analyst not Murthy himself). I think the analyst just "put two and two together," and in this case basically got 5.

Do you think that "Zen will disrupt the market" and "I haven't seen enough of the Zen to calibrate its exact competitive level, but we're treating it like it's going to be a legitimate competitor and therefore we need to deliver our best to make sure we're able to deliver the business performance that we expect to deliver over the course of the next 12 months" can really be reconciled?

10-nanometer desktop CPUs won't arrive until late 2018 at the very earliest, more likely in 2019. That's outside of the 12 month window that Murthy was talking about.

 

[sirmo]

Nope, the analyst is referring to this interview (remember, the Barron's quote is quoting the analyst not Murthy himself). I think the analyst just "put two and two together," and in this case basically got 5.

Do you think that "Zen will disrupt the market" and "I haven't seen enough of the Zen to calibrate its exact competitive level, but we're treating it like it's going to be a legitimate competitor and therefore we need to deliver our best to make sure we're able to deliver the business performance that we expect to deliver over the course of the next 12 months" can really be reconciled?

10-nanometer desktop CPUs won't arrive until late 2018 at the very earliest, more likely in 2019. That's outside of the 12 month window that Murthy was talking about.

Yeah the way the quote is written is unclear. It's hard to tell Moores words from Murthy's. Some stuff was clearly said by Murthy but how much of it, can't really say.

 

[guskline]

It won't be long now until we really see how ZEN performs. I was a long time faithful AMD follower but the hype before Bulldozer, the release of it then the onslaught of SandyBridge, Ivy Bridge Haswell, Skylake and now Kaby Lake has made it VERY hard to have much faith the AMD will get back into the cpu game.

Realistically, I doubt ZEN will be much faster than Haswell but will have to compete on price. Sorry for being so pessimistic but my hopes for AMD have been dashed so many times it gets very hard to believe they will truly be competitive on performance alone.

Nonetheless, owning a "multi-core" 5960x makes me wonder how close ZEN 8c/16t can come to it or even surpass it.

 

[sirmo]

It won't be long now until we really see how ZEN performs. I was a long time faithful AMD follower but the hype before Bulldozer, the release of it then the onslaught of SandyBridge, Ivy Bridge Haswell, Skylake and now Kaby Lake has made it VERY hard to have much faith the AMD will get back into the cpu game.

Realistically, I doubt ZEN will be much faster than Haswell but will have to compete on price. Sorry for being so pessimistic but my hopes for AMD have been dashed so many times it gets very hard to believe they will truly be competitive on performance alone.

Nonetheless, owning a "multi-core" 5960x makes me wonder how close ZEN 8c/16t can come to it or even surpass it.

I don't think you're being pessimistic. I think beating Haswell level performance would be a big win for AMD (a small miracle). It would actually offer an alternative to Intel in the mainstream. Personally I don't have any illusions that Zen will somehow win the performance crown. That would be too optimistic. I could maybe see it in a niche discipline or two, but overall I still expect Intel to have a lead. I think Zen will fall short on the clocks.

I am excited though, because if they deliver on their promise we will probably get cheaper manycores. I am also excited for the APUs. Because up until now, AMD APUs never realized their potential being held back by sub par CPU cores. And Zen could change that. Which means you might be able to pick up a laptop that can game decently for a bargain. And that's good for everyone.

As someone who's been stuck on Haswell because performance gains haven't been that impressive since.. I was looking at actually getting more cores, which I can use in stuff I do. And while I thought about splurging on a Broadwell-E, I am going to wait and see how Zen fares.

Zen getting close, could also be a new chapter AMD has been desperately needing to give them some much needed resources to continue development and perhaps one day actually even challenge Intel at the top.