Discussion Ryzen 3000 series benchmark thread ** Open **

Page 9 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

majord

Senior member
Jul 26, 2015
433
523
136
Are there any comprehensive OC tests about yet? Wondering if PBO + offset undervolt is the key (again)
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
That's odd. Greg at Science Studio tried a 3900X in an Asrock B350 motherboard. He didn't do a whole lot of testing, but apparently it worked. He's gonna try an A320 board next.

That video is great in so many ways. He flashed his UEFI in a thunderstorm. It satisfies my morbid curiosity. Bring on A320!

Get ready for software fixes which will likely impact ST and light threaded performance.

Surprise surprise.
 

moinmoin

Diamond Member
Jun 1, 2017
4,933
7,619
136
Latency between cores:
D-4Hx1-VUAAL_Ao.png

D-4Hx19UIAE_S69.png

D-4Hx2BVAAAdGUr.jpg

D-4H6TDUYAEUe_7.jpg

Source

Takeaway: Inner-CCX latency is still the best. Inter-CCX/Inter-CCD latency is now more stable and (potentially, likely depending on the not mentioned memory speed) slightly lower than Intel's mesh. If this is as scalable as it looks this spells well for both Threadripper 3 and Epyc 2.
 

IEC

Elite Member
Super Moderator
Jun 10, 2004
14,323
4,904
136
Oh my.
AnandTech said:
(7/8): We've noticed large frequency boost behaviour changes with new motherboard firmware that was released on launch day (7/7). We are currently re-running all our test suite numbers and updating the article with the new data soon as appplicable. For further details please read here.

AnandTech said:
We ran our original review numbers with the latest available firmware for the MSI MEG X570 ACE motherboard last week (Version 7C35v11). On Saturday the 6th MSI had shared with us a notice about a new version coming out, which became available to download to us on Sunday the 7th, the launch day and date of publication of the review.


We’ve had more time to investigate the new firmware, and have discovered extremely large changes in the behaviour of the frequency boosting algorithm. The new firmware (Version 7C35v12) for the motherboard contains AMD’s new ComboPI1.0.0.3.a firmware.


We discovered the following direct measureable effects between the two firmware versions:




We notice a significant change in the CPU’s boosting behaviour, now boosting to higher frequencies, and particularly at a faster rate from idle, more correctly matching AMD’s described intended boost behaviour and latency.


We’re currently in the process of re-running all our suite numbers and updating the article where necessary to reflect the new frequency behaviour.

Just looking at those boost clock differences, that could potentially completely close or exceed the gap vs the 9900K in more than a few games.
 

cortexa99

Senior member
Jul 2, 2018
318
505
136
I do wonder why AMD release a thing that isn't technically done, as for how Zen2 looks like now maybe the release date should be postponed a bit until they get BIOS fully functioned. Now the reviews that not showing best Zen2 have been all over the world, why AMD's marketing strategy often screws things up as dumb as this way.....:rolleyes:

EDIT: from the chart above looks like the all core turbo @ 4.2-4.3 for old BIOS while 4.5 the new one. That's......funny(out of words to describe this situation)
 
  • Like
Reactions: Gikaseixas

TheGiant

Senior member
Jun 12, 2017
748
353
106
another review- a good one with stock and finally some oc numbers

what really interests me the power to get 9900K 5GHz performance- looks like we wont see it because the r3 8C simply wont reach that level of performance

the numbers non oced (mighty efficiency of 3700x) https://www.purepc.pl/procesory/tes...7_3700x_premiera_architektury_zen_2?page=0,27 and OC
https://www.purepc.pl/procesory/tes...7_3700x_premiera_architektury_zen_2?page=0,52

what seems tricky here and is not visible on the first look in comparison that the 3700X needs 4,4GHz (and a little more) to achieve stock unchecked (power wise) 9900K performance
but comparing power numbers of stock power unchecked 9900K and oced 3700x to 4,4GHz looks not very well for the 7nm r3 part- in high perf area it is of the same efficiency power wise as the 14nm 9900k- I must say I expected more

but I didnt see a test of undervolted 3700X- at 4GHz all core the r3 looks monster efficient, much much better than 9900K
 

TheGiant

Senior member
Jun 12, 2017
748
353
106
Oh my.




Just looking at those boost clock differences, that could potentially completely close or exceed the gap vs the 9900K in more than a few games.
yep
really looking forward to it
and if it does, man AMD screwed this release....promoting the r39x as the best gaming CPU and doing this
....they never learn
 
  • Like
Reactions: HurleyBird

dnavas

Senior member
Feb 25, 2017
355
190
116
yep
really looking forward to it
and if it does, man AMD screwed this release....promoting the r39x as the best gaming CPU and doing this
....they never learn

That might be a bit of an exaggeration, but it's true that they seem to have had sufficient time.... Regardless, I would like to see what power consumption and temps look like after this change. I can't imagine AMD was just burning electrons as a hold-back at lower freqs.... AFAIK, the anand review noted at-limit tdp behavior. Purely selfishly, power consumption is going to directly affect what's possible in TR absent custom cooling.
 

Asterox

Golden Member
May 15, 2012
1,026
1,775
136
That's odd. Greg at Science Studio tried a 3900X in an Asrock B350 motherboard. He didn't do a whole lot of testing, but apparently it worked. He's gonna try an A320 board next.


What you expect, on stock it is expected to work without any problems.How is that possible, well simple VRM is more then enough for stock CPU everyday normal work= it was tested by Asrock.

Here is one very bad MSI example, VRM-s are about same classe as on Asrock motherboard."But this motherboard didnt get suport not even for very power hungry R5 3600 or R7 3700X 65W CPU-s.":hushed:

https://www.msi.com/Motherboard/support/B350-TOMAHAWK-ARCTIC

"Look at this mighty Gigabyte motherboard", also have suport for R9 3900X.

https://www.gigabyte.com/Motherboard/GA-AB350M-HD3-rev-1x#support-cpu
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
I'm somewhat surprised by this as it seems overall IPC has definitely caught up to Skylake levels, or maybe even exceeded it slightly. Yet gaming is a bit behind. Does AMD need to get their latencies down even lower to finally beat Intel at gaming?

Why is it a surprise? The Intel parts have a clock speed advantage.
 

Gideon

Golden Member
Nov 27, 2007
1,608
3,569
136
What you expect, on stock it is expected to work without any problems.How is that possible, well simple VRM is more then enough for stock CPU everyday normal work= it was tested by Asrock.

Here is one very bad MSI example, VRM-s are about same classe as on Asrock motherboard."But this motherboard didnt get suport not even for very power hungry R5 3600 or R7 3700X 65W CPU-s.":hushed:

https://www.msi.com/Motherboard/support/B350-TOMAHAWK-ARCTIC

Yeah I'm actually on the very-same motherboard (non-artic) as it was one of the most sold cheap B350 mobos of the day. Until yesterday they had a BETA bios up, but it was removed as people couldn't actually boot any 3xxx series up with it. Therefore I don't think this is *only* connected to VRMs.

Overall, still very unlucky considering, that pretty much all the other B350 and even A320 boards did get working beta bioses:

https://www.msi.com/blog/the-latest-bios-for-amd-300-400-series-motherboard

I hope they will release a new working beta at some point. Yes the VRMs are crap, but certainly not worse than, say, A320M BAZOOKA, that did get the update.
 

lixlax

Member
Nov 6, 2014
183
150
116
That's odd. Greg at Science Studio tried a 3900X in an Asrock B350 motherboard. He didn't do a whole lot of testing, but apparently it worked. He's gonna try an A320 board next.


Not bad, not bad. But the final boss would still be the 3950X on an A320 board.

Has anyone reviewed the chipset fan behaviour on various X570 boards? I think before the launch MSI were the only ones to state that their chipset fan is temperature controlled and can also have manual profiles in the BIOS.
 

linkgoron

Platinum Member
Mar 9, 2005
2,286
810
136
Oh my.




Just looking at those boost clock differences, that could potentially completely close or exceed the gap vs the 9900K in more than a few games.

Looks like this 7/7 release was rushed/too early for both Navi and Ryzen 3.
 
  • Like
Reactions: Brunnis

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
Just looking at those boost clock differences, that could potentially completely close or exceed the gap vs the 9900K in more than a few games.

This is only going to impact performance by 1-3% in average. Clock speeds in games should be pretty consistent and its the same reason why SpeedShift didn't show any changes outside of very bursty applications.
 

Asterox

Golden Member
May 15, 2012
1,026
1,775
136
yep
really looking forward to it
and if it does, man AMD screwed this release....promoting the r39x as the best gaming CPU and doing this
....they never learn

There is nothing to screw, even from first testing Ryzen 3000 is to much for Intel overpriced CPU jewels.

CPU lives very long time, or as i see AMD again has done some fine Sandbagging.
 
  • Like
Reactions: Space Tyrant

dlerious

Golden Member
Mar 4, 2004
1,769
717
136
Not bad, not bad. But the final boss would still be the 3950X on an A320 board.

Has anyone reviewed the chipset fan behaviour on various X570 boards? I think before the launch MSI were the only ones to state that their chipset fan is temperature controlled and can also have manual profiles in the BIOS.
Haven't seen much on motherboards yet. The focus has been on CPUs with navi reviews coming now. I'd like to see a few roundup reviews for motherboards, really interested in seeing how well passive cooling works with the Gigabyte Aorus Xtreme.
 

amrnuke

Golden Member
Apr 24, 2019
1,181
1,772
136
Looks like this 7/7 release was rushed/too early for both Navi and Ryzen 3.
And even then, the 3700X and 5700 look like an absolutely great combination for 1440p gaming.

FPS numbers at 1440p for the 5700 via TPU range from a low of 45.1 to a high of 138.9, with a median of above 80 fps.

At 1440p the CPU matters a little less, but even then, the 3700X is within 2.5% of the 9900K.
 

dzoni2k2

Member
Sep 30, 2009
153
198
116
This is only going to impact performance by 1-3% in average. Clock speeds in games should be pretty consistent and its the same reason why SpeedShift didn't show any changes outside of very bursty applications.

It took half a second to boost from idle. Half a second. And no, clock speed in games is far from consistent. It actually fluctuates wildly.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
It took half a second to boost from idle. Half a second. And no, clock speed in games is far from consistent. It actually fluctuates wildly.

Thermal based fluctuations exist, but that's different from what we're seeing here.

Some rare games might see a bigger benefit but mostly CPUs run at similar levels and don't really transition between idle and peak in games. 0.5 seconds is nothing when games are played for hours.

Update: AT review seems rushed that's all.

Look at the Computerbase.de review. They did a test with the driver where it recognizes UEFI CPPC2 and the latest Windows updates.

https://www.computerbase.de/2019-07/amd-ryzen-3000-test/2/

It gained 4-5% in Cinbench R15/R20, and POV-Ray ST.
 
Last edited:

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
3% is huge considering how low the delta between 3900X and 9900K is in gaming. We're also bound to get a few percent more from scheduler and memory optimizations. At this stage every percent matters (for public perception).

Look at my updated post right above yours.

Yea, I think AT review was just rushed. Their R15 CB score is equal to Computerbase's result without the driver.
 

lixlax

Member
Nov 6, 2014
183
150
116
Thermal based fluctuations exist, but that's different from what we're seeing here.

Some rare games might see a bigger benefit but mostly CPUs run at similar levels and don't really transition between idle and peak in games. 0.5 seconds is nothing when games are played for hours.

Update: AT review seems rushed that's all.

Look at the Computerbase.de review. They did a test with the driver where it recognizes UEFI CPPC2 and the latest Windows updates.

https://www.computerbase.de/2019-07/amd-ryzen-3000-test/2/

It gained 4-5% in Cinbench R15/R20, and POV-Ray ST.
Their results are basically in line with the ones that AMD had on their slides as well.
 

Abwx

Lifer
Apr 2, 2011
10,847
3,297
136
Yea, I think AT review was just rushed. Their R15 CB score is equal to Computerbase's result without the driver.

213/3168 in R15 ST/MT at Computerbase and 204/3090 for AT, granted the MT score is subject to more variability but the ST score show about 200MHz difference, this can be seen in AT frequency/time chart with both firmwares.

https://www.computerbase.de/2019-07/amd-ryzen-3000-test/3/#diagramm-test-cinebench-r15-multi

https://www.anandtech.com/show/14605/the-and-ryzen-3700x-3900x-review-raising-the-bar/9
 
  • Like
Reactions: lightmanek

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
213/3168 in R15 ST/MT at Computerbase and 204/3090 for AT,

MT doesn't gain from that, only ST.

213 is with the drivers. Computerbase got 204 without, which is identical to Anandtech.

I doubt we'd get 3% in games. Similar to MT results in rendering, we'd get nothing.