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

Page 754 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
H

Hitman928

Diamond Member
Apr 15, 2012
6,755
12,501
136
  • Like
  • Love
  • Wow
Reactions: Tlh97, Mopetar, IEC and 4 others
DisEnchantment

DisEnchantment

Golden Member
Mar 3, 2017
1,779
6,798
136
DisEnchantment said:
Did you try the MSI BIOS for memory timing optimization?

Also when will you post the results with PBO+CO + Mem tuning on Linux running GB6? :p
Click to expand...

igor_kavinski said:
Somebody's custom cooled Zen 5: https://browser.geekbench.com/v6/cpu/7040992

:p

That's more like it! https://browser.geekbench.com/v6/cpu/compare/7040992?baseline=7302203

vs. his 7950X3D: https://browser.geekbench.com/v6/cpu/compare/7040992?baseline=4531828
Click to expand...

I knew he was on it.
Hopefully he has a second SSD to boot Linux too. :D

Hitman928 said:
Looks like it is running at 5.95 GHz max boost so custom cooled and overclocked beyond even the default Fmax.
Click to expand...
He mentioned he is only using PBO + CO, but this is ES, so would be interesting if he can do this on retail.
 
  • Like
Reactions: Tlh97, Elfear and igor_kavinski
I

igor_kavinski

Lifer
Jul 27, 2020
28,174
19,218
146
vs. the 7950X3D, I can see that:

The V-cache has an effect and it reduces the lead of Zen 5 in many tests.

PDF Renderer, Clang, Asset compression, Background Blur and Raytracer MT tests don't care about cache so Zen 5 trounces the X3D pretty hard in these subtests.

Horizon Detection MT loves cache, enabling the 7950X3D to pull ahead of Zen 5.

Cache or no cache, Object Remover subtest is oblivious to whether it's Zen 4 or Zen 5.
 
  • Like
Reactions: Tlh97, Mopetar and lightmanek
Jan Olšan

Jan Olšan

Senior member
Jan 12, 2017
626
1,262
136
Geddagod said:
Because Zen 3 also took the ST crown for Intel for a while, and because people are also waiting on ARL for a generational comparison?
Also, a score of ~3460 would put it ~15% faster than the 13900k. Don't think ARL would be too far behind tbh.
Click to expand...
Some people kinda tend to make an opinion and then don't pay that much attention to the empiric evidence. Perhaps the "5%" people just looked at games and decided it sucks so when the GB scores show ~15% jump they ignore it or decided that's outlier or something.

When I looked at Davind Huang's test of RAI 300, it looked like Intel could overtake Zen 5 in mobile in GB6 and GB5 single core if Lion COve IPC goes up by exactly 14% in the tests. I'm not sure about desktop ARL × R9 9950X though.

Geddagod said:
The microcode patch doesn't impact longer ST tests much. It's been released for a couple days IIRC.
Click to expand...

Didn't Geekbench pause and start its loads? I think that was exactly the scenario where Intel could be affected because the VID cap fix could affect the speed of boost ramping. Continuous 1T load could be unaffected, bursty loads affected. Also, if TVB is fixed to actually not allow the highest clocks over 70C temperature as it is supposed to be, then that could worsen Intel's 1T score. The effect of Inte default settings is a separate thing, actual 253W PL2 likely only affects MT score, but the settings were touching other things too...
 
Last edited:
  • Like
Reactions: Tlh97, exquisitechar, Geddagod and 3 others
H

Hitman928

Diamond Member
Apr 15, 2012
6,755
12,501
136
Jan Olšan said:
Some people kinda tend to make an opinion and then don't pay that much attention to the empiric evidence. Perhaps the "5%" people just looked at games and decided it sucks so when the GB scores show ~15% jump they ignore it or deiced that's outlier or something.

When I looked at Davind Huang's test of RAI 300, it looked like Intel could overtake Zen 5 in mobile in GB6 and GB5 single core if Lion COve IPC goes up by exactly 14% in the tests. I'm not sure about desktop ARL × R9 9950X though.



Didn't Geekbench pause and start its loads? I think that was exactly the scenario where Intel could be affected because the VID cap fix could affect the speed of boost ramping. Continuous 1T load could be unaffected, bursty loads affected. Also, if TVB is fixed to actually not allow the highest clocks over 70C temperature as it is supposed to be, then that could worsen Intel's 1T score. The effect of Inte default settings is a separate thing, actual 253W PL2 likely only affects MT score, but the settings were touching other things too...
Click to expand...

Yeah, it looks like LNL should match, or slightly exceed, STX, at least the highest binned/TDP parts. On desktop, I think ARL will have a harder time catching up to Zen 5.
 
I

igor_kavinski

Lifer
Jul 27, 2020
28,174
19,218
146
H

Hitman928

Diamond Member
Apr 15, 2012
6,755
12,501
136
  • Like
Reactions: Tlh97, lightmanek and igor_kavinski
Jan Olšan

Jan Olšan

Senior member
Jan 12, 2017
626
1,262
136
  • Like
Reactions: igor_kavinski
G

gdansk

Diamond Member
Feb 8, 2011
4,796
8,098
136
Jan Olšan said:
2584-2588 results for GB5 (5.4.1) ST running at ~5700 MHz

Gigabyte Technology Co., Ltd. B650E AORUS PRO X USB4 - Geekbench

Benchmark results for a Gigabyte Technology Co., Ltd. B650E AORUS PRO X USB4 with an AMD Ryzen 9 9950X processor.
browser.geekbench.com browser.geekbench.com

Gigabyte Technology Co., Ltd. B650E AORUS PRO X USB4 - Geekbench

Benchmark results for a Gigabyte Technology Co., Ltd. B650E AORUS PRO X USB4 with an AMD Ryzen 9 9950X processor.
browser.geekbench.com browser.geekbench.com

Not DDR5-5600 / JEDEC I think?
Click to expand...
That's better. 15.4% faster than my stock 7950X 1T.
 
  • Like
Reactions: Tlh97, Racan and exquisitechar
Hail The Brain Slug

Hail The Brain Slug

Diamond Member
Oct 10, 2005
3,890
3,331
146
Hitman928 said:
Memory and power limits come much more into play with MT results, especially on GB5. I expect MT to improve as well with Zen 5, we just need to wait for proper, controlled testing.
Click to expand...
From what we've gathered so far, it seems the 9950X will scale well beyond the 230W stock PPT while the 7950X tends to stall out around 210-220W, so I can definitely see it being power constrained at stock settings.
 
  • Like
Reactions: Tlh97, lightmanek and Elfear
Det0x

Det0x

Golden Member
Sep 11, 2014
1,490
5,117
136
igor_kavinski said:
Somebody's custom cooled Zen 5: https://browser.geekbench.com/v6/cpu/7040992

:p

That's more like it! https://browser.geekbench.com/v6/cpu/compare/7040992?baseline=7302203

vs. his 7950X3D: https://browser.geekbench.com/v6/cpu/compare/7040992?baseline=4531828
Click to expand...
Have to remember the ES is just running a "baseline OC", in summertime with varm weather, i fully expect to beat it once i get my retail 9950X and get optimized :innocent:
And my 7950X3D you are comparing against is ranked #1 on all metrics on hwbot (really a golden sample)
1723573605801.png
Have been battling alpha AGESA gremlings for 2 months already, and just when things start to get sorted -> back to start with a new AGESA release candidate.
Currently the main problem all mb vendors are facing are high VDD (1.55+) limitations when running high memoryspeed.

btw "synced" 2200/8800MT/s have already been reached under SS 👍
(but it dont really help since we are so bottlenecked by FCLK)
 
Last edited:
  • Like
  • Wow
Reactions: Tlh97, exquisitechar, lightmanek and 4 others
Jan Olšan

Jan Olšan

Senior member
Jan 12, 2017
626
1,262
136
gdansk said:
That's better. 15.4% faster than my stock 7950X 1T.
Click to expand...
I recall 15ish-16ish % was what David Huang got for GB5 and GB6 ST in his preliminary Zen 5 (Strix) testing back in june. Unless I am mistaken, I remember noting how that was one thing that kinda landed almost at the official 16% number. Those weren't final results tho.
 
You must log in or register to reply here.

TRENDING THREADS

COMPANY
RESOURCES
FOLLOW
Top Bottom