• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

MSI & ASUS Prep 4GHz RAM AGESA v1.2 BIOS - slow memory can fclk itself

Soldato
Joined
26 Sep 2010
Posts
7,146
Location
Stoke-on-Trent
There's been talk about a 2GHz fabric since the XT refresh last year, and that talk was bolstered by some rumouring about part of Zen 3's redesign was to explicitly make a 2GHz fabric more realistically possible, not just super golden sample chips. I guess AMD had to make sure the hardware was capable of it before pushing out software updates to enable it.
 
Associate
Joined
10 Jul 2009
Posts
1,559
Location
London
My computer stopped running 4 DIMMS at all recently. Yet magically the third set I tried worked. I hope this release brings some stability.

indeed, Asrock BIOS would code 92 if I set RAM to 3600Mhz, speed which worked with 3900x before fine. Code 92 was not even related to RAM
 
Soldato
Joined
18 Oct 2002
Posts
10,950
Location
Bristol
All else equal (inc timings) what benefits are people seeing from clocking infinity fabric and RAM at 2GHz instead of a more typical 1800MHz? It would be a ~11% increase in bandwidth, less if timings had to be slackened. But is that actually humanly noticeable in any real workloads?
 
Associate
Joined
10 Jul 2009
Posts
1,559
Location
London
All else equal (inc timings) what benefits are people seeing from clocking infinity fabric and RAM at 2GHz instead of a more typical 1800MHz? It would be a ~11% increase in bandwidth, less if timings had to be slackened. But is that actually humanly noticeable in any real workloads?

benefit beyond 3600Mhz/1800Mhz IF is only seen in memory benchmarks. There is no substantial performance increase with speeds above 3600Mhz. Considering that you have to be extremely lucky and have quite expensive memory kit in order to hit speeds around 4000Mhz, going after them is just for enthusiasts. You won't lose anything in real world staying with 3600Mhz
 
Associate
OP
Joined
22 Oct 2012
Posts
1,086
All else equal (inc timings) what benefits are people seeing from clocking infinity fabric and RAM at 2GHz instead of a more typical 1800MHz? It would be a ~11% increase in bandwidth, less if timings had to be slackened. But is that actually humanly noticeable in any real workloads?
There's a good analysis Premiumbuilds did that showed ~3800 MHz did give benefit but performance actually dropped after that, even with a 1:1 fclk to speed ratio. However I'd expect the drop off to shift to 4 GHz plus with AGESA 1.2 and onwards.

Ryzen-Zen-3-RAM-Timespy-CPU-score-vs-RAM-Speed-1024x696.png


Ryzen-Zen-3-RAM-Ram-Speed-vs-FPS-5800X-Raibow-6-1024x687.png


Source: https://premiumbuilds.com/features/zen-3-ram-speeds-benchmark-analysis/
 
Associate
Joined
13 Sep 2013
Posts
1,662
Location
Aberdeen
Tried the Asus one. No joy running 3800 on 5950x but weirdly 3833 will boot to win but makes the system so slow its unworkable. Seems a few report the same issue.

I'll stick to 3733, won't see any difference going higher anyway imo.
 
Soldato
Joined
15 Oct 2019
Posts
11,656
Location
Uk
All else equal (inc timings) what benefits are people seeing from clocking infinity fabric and RAM at 2GHz instead of a more typical 1800MHz? It would be a ~11% increase in bandwidth, less if timings had to be slackened. But is that actually humanly noticeable in any real workloads?
According to 1usmus going over 1900 results in packet throttling and higher latency.
 
Soldato
Joined
29 May 2005
Posts
4,896
MSI has 1.2.0.0 on x570 boards for some time - since 12th Jan.

Still all beta bios tho. Doesn’t give me the confidence it is stable. Although I really like to try it out.
 
Associate
Joined
28 Oct 2009
Posts
2,307
Location
Dark Side of the Moon
I'm actually thinking of dropping down to 3600 with CL14 from 3800 CL16.

CL14 @ 3600: 7.7 ns latency
CL16 @ 3800: 8.4 ns latency

So lower speed but tighter timings looks faster. When I get a couple of hours to test I will give it a go. My sticks look reasonable (not as good as the 16g kit I sold here) but I think I should be able to get a decent result.
 
Back
Top Bottom