• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)

Page 462 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Abwx said:
I m saying that SPEC Povray score difference between Zen 4 and RPL is about the same as Computerbase.

So you ll have to assume that AVX2 bring no perfs enhancement in Povray for RPL, otherwise it wouldnt be faster in SPEC (wrt to Zen 4) by the same amount than when using AVX2, that s basically what you re saying.

Or that Geekerwan numbers are flawed eventually, wich i already pointed as being a possibility.
Click to expand...

You can draw absolutely no conclusions about a test largely comparing scalar floating-point performance (because of having no vector intrinsics) based on performance involving vector intrinsics. Even in the "AVX2-less" builds that reviewers have used, it's still using AVX.

SPEC does not use AVX (except what the compiler can emit itself.) You cannot draw the kinds of comparisons you're talking about. I don't find Geekerwan's result suspicious, just meaningless.
 
Abwx said:
I m saying that SPEC Povray score difference between Zen 4 and RPL is about the same as Computerbase.

So you ll have to assume that AVX2 bring no perfs enhancement in Povray for RPL, otherwise it wouldnt be faster in SPEC (wrt to Zen 4) by the same amount than when using AVX2, that s basically what you re saying.

Or that Geekerwan numbers are flawed eventually, wich i already pointed as being a possibility.
Click to expand...

Edit: @SarahKerrigan beat me to it.
 
eek2121 said:
Zen 5 CPU-Z single core benchmark is supposedly 910. Will share source soon.

https://twitter.com/x/status/1795457854213865927

EDIT: Sorry,, I put the twitter link below, had to get to a desktop. I wanted to link the original baidu post, but have had issues.

That is for the engineering sample of Zen 5. For reference I get around 740-750 on my stock 7950X, so about 21% higher. Not bad.

Oh and if this was already posted, apologies. Sometimes this thread moves quite quickly and it can be hard to keep up.
Click to expand...
100% sure this screenshot is fake. Looks like they just messed around with the CPU ID on a Zen 4 part.
 
SarahKerrigan said:
You can draw absolutely no conclusions about a test largely comparing scalar floating-point performance (because of having no vector intrinsics) based on performance involving vector intrinsics. Even in the "AVX2-less" builds that reviewers have used, it's still using AVX.

SPEC does not use AVX (except what the compiler can emit itself.) You cannot draw the kinds of comparisons you're talking about.
Click to expand...

So now you re telling me that when the 7950X is deprived of AVX it lose more perf than RPL deprived of AVX, knowing that the 7950X is surely faster when both CPUs are limited to AVX, that s quite a stretched assumption given that you have no numbers to validate this point, just assumptions.

SarahKerrigan said:
I don't find Geekerwan's result suspicious, just meaningless.
Click to expand...

Whatever is meaningless is forcibly suspicious.
 
Abwx said:
So now you re telling me that when the 7950X is deprived of AVX it lose more perf than RPL deprived of AVX, knowing that the 7950X is surely faster when both CPUs are limited to AVX, that s quite a stretched assumption given that you have no numbers to validate this point, just assumptions.
Click to expand...

Except there are tests showing as such, namely tests within SPECfp. You have provided zero evidence to the contrary, so. . .

Abwx said:
Whatever is meaningless is forcibly suspicious.
Click to expand...

What? I can make all kinds of meaningless statements that are absolutely true. Something being meaningless does not force suspicion.
 
Abwx said:
Spec was routinely used with optimised code using AVX on official submissions, that it is cheatinng or not, and geekerwan s numbers are suspicipous in this respect when
it come to Povray.

CPU2017 Integer Rate Result: Dell Inc. PowerEdge R760 (Intel Xeon Platinum 8480+)

CINT2017 result for PowerEdge R760 (Intel Xeon Platinum 8480+); SPECrate2017_int_base: 904; SPECrate2017_int_peak: 931
www.spec.org
Click to expand...

You don't know what you're looking at.

All those flags can control is whether the compiler attempts to autovectorize to AVX2. GCC contains flags to do something similar. So does AMD's AOCC. It means absolutely nothing about the Povray AVX2 fastpath. Zilch.

Please stop. You clearly have never touched SPEC yourself and don't know how it works.
 
SarahKerrigan said:
Please stop. You clearly have never touched SPEC yourself and don't know how it works.
Click to expand...
Curious question: if someone compiles POVRAY with that option turned on and uses it on Intel CPU but runs the binary compiled with that option omitted on the Ryzen, will the Ryzen suffer in the performance comparison?
 
igor_kavinski said:
Curious question: if someone compiles POVRAY with that option turned on and uses it on Intel CPU but runs the binary compiled with that option omitted on the Ryzen, will the Ryzen suffer in the performance comparison?
Click to expand...

Nobody but Intel compiles with ICC. Anandtech doesn't. Geekerwan doesn't. Running it on Ryzen would likely be a little slower but not significantly - ICC generates decent code for AMD CPUs these days (but is itself increasingly irrelevant because gcc and clang have gotten really good.)
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top