Discussion Ryzen 3000 series benchmark thread ** Open **

Page 23 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Feb 23, 2017
636
578
106
A bit naff that their most popular B450 isn't getting the 32MB BIOS; no Gaming Pro Carbon AC.
 

Abwx

Diamond Member
Apr 2, 2011
9,006
376
126
B450 is considered low cost, so no surprise that they are cutting corners.

Beside AMD stated that they are going to disable PCIe4 support in pre X570 MBs in their next AGESA, so big bios memory capacity is possiby not required for those chipsets, most annoying is that those MSI MBs do not support Bristol Ridge, not a big loss, but still...
 
Feb 23, 2017
636
578
106
I know, but not the one I want.
Its annoying because you know that when 4th gen comes out that the Max versions will get support, but the originals won't.
 

Ajay

Diamond Member
Jan 8, 2001
5,308
261
136
I know, but not the one I want.
Its annoying because you know that when 4th gen comes out that the Max versions will get support, but the originals won't.
Keeping socket compatibility is apparently harder than it would seem to be :(
 

moinmoin

Senior member
Jun 1, 2017
915
402
106

Abwx

Diamond Member
Apr 2, 2011
9,006
376
126
This is interesting since these 12nm APUs apparently use the same finer grained frequency management that the 7nm CPUs use, with the same result that the max turbo frequency is seldom reached.
According to this review the all core turbo is as high as 4.02GHz with the CPU being within its rated TDP, that s something like 300Mhz better than the 2400G at same power.

SC turbo seems to boost to 4.196GHz , all in all that s a welcomed improvement, definitly 12nm will be long lived for low cost products.


Max with all core turbo is on AIDA :




 

Ajay

Diamond Member
Jan 8, 2001
5,308
261
136
According to this review the all core turbo is as high as 4.02GHz with the CPU being within its rated TDP, that s something like 300Mhz better than the 2400G at same power.

SC turbo seems to boost to 4.196GHz , all in all that s a welcomed improvement, definitly 12nm will be long lived for low cost products.
That’s a nice improvement! Rats, my mother's computer died a couple of months too soon. I had to build with a 2400g :(
 

Ajay

Diamond Member
Jan 8, 2001
5,308
261
136
When mfg's cheap out and don't think ahead, WRT BIOS chip capacity, then sure, yes.
Makes me wonder if AMD had set guidelines and the manufacturers saved money instead; or if AMD didn’t provide enough support.
 

Kenmitch

Diamond Member
Oct 10, 1999
7,362
282
126
Makes me wonder if AMD had set guidelines and the manufacturers saved money instead; or if AMD didn’t provide enough support.
Not sure, but I'd imagine they had a minimum and suggested set of guidelines.
 

IEC

Super Moderator
Super Moderator
Jun 10, 2004
13,737
870
136
The image shows he hit 70°C Tmax. For water cooling with 360mm radiator with 6x good fans in push-pull configuration that is very much a believable and attainable temperature. Just slightly worse than what I would get on my delidded 8700K OC.

I stay in the 70s (°C) on the same benchmark with a max OC on a 3900X 12c/24t (double the chiplets, 50% more active cores). That's with a *high* ambient of nearly 27°C (80F). He's probably in the high 60s/low 70s °F ambient. Ambient temps don't make a substantial difference when you have great airflow. If it makes a difference in the OC attainable it's probably 25MHz at the best given temperature scaling noticed by der8auer and others. You really need to drop another 20°C below ambient (read: extremely impractical) to make another step up in maximum frequency achievable.
 

ClockHound

Golden Member
Nov 27, 2007
1,051
70
106
"AMD has done a fantastic job here overall, and we're very aware this is the start to the end of our company in general. As both AMD and Intel optimize their binning process more and more, overclocking will not be possible as CPUs will boost themselves on their own to the highest clocks possible. "

That's terrible for SL. But worse for this forum - forced to become: CPUs and NO Overclocking!
 
Apr 27, 2000
12,759
1,563
126
They can take my overclocking from my cold, dead hands!

Wait, they're already cold and dead. Hmmmm. Not a good motto for a zombie.
 

CHADBOGA

Golden Member
Mar 31, 2009
1,843
111
136
So do we have an idea for what the likely top speeds that Ryzen 3000 models will boost to for :

One core
Two cores
Four cores
Six cores
Eight cores etc?

Are certain models not even reaching the turbo speed for even one core, at any time? :oops:
 

coercitiv

Diamond Member
Jan 24, 2014
3,359
887
136
Are certain models not even reaching the turbo speed for even one core, at any time? :oops:
Think of it this way:
  • AMD made a boost algorithm that takes all relevant parameters into account in order to take boost clocks to their limit: power, current, temperatures (and probably voltage as a byproduct of silicon quality & temps).
  • MB makers probably did what they do best and overshot voltage - either directly or as a indirect consequence of aggressive LLC.
Now, what happens when a chip is fed with more voltage than it needs to? Power and temps go up, fast. And what happens to boost algorithms designed to operate around silicon temperature (probably hot spots too)? It underperforms.

Case in point from my own experience: i7 8700 on Z370M board from MSI. The "Auto" settings are so aggressive in terms of voltage that I can see up to 1.35V for single-core operation. Power draw is so high that under my silent cooling a single thread of Prime 95 small FFT can bring a core close to TjMAX before it migrates to another physical core (90+C). Setting CPU voltage in BIOS to ~1.2V drops temps in the same scenario to around 70C. (Noctua NH-D14 with low RPM fans).

I would wait until BIOS versions are sorted out, until people get to test and understand those tests before calling out AMD on boost clocks (for now we can call them out for bad launch BIOS support).
 
Last edited:

Similar threads



ASK THE COMMUNITY

TRENDING THREADS