Drag to reposition cover

ASUS Pro WS W790E-SAGE SE + Intel Xeon Sapphire Rapids SP ES = works it

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

DHamov

Active Member
Jan 12, 2024
112
27
28
So the Q03J Asrock combination did not work. I freaked out without computer power so there is an Asus Sage again. When i ran it before i thought my memory overclock was working. I thought so because it showed in linux that the installed speed was 6000MT. The bios also shows it. And when i did the Aida64 Benchmark, i also had High scores, some of it a bit better than Joseph. It is not a competition but just a reference. AIDA64 shows. also memory speed as set in bios here in screenshot 6200MT. But HWinfo only showed 2400x2 so 4800. This bothered me and i did not know who to believe. So now i had the opportunity to test everything once more: one time with 6200 in the bios and one time with 4800 in the bios, everything else the same, to see if there was a difference.And to see if Aida64 or HWinfo was right? and...
AIDA64_MEM_SAGEQ03J_4800.PNG

aida64_6200_hw4800_multmax4700.PNG

So it seems that there is not really a measurable difference, and if there is a difference it is towards the 4800MT setting strange enough! So i think HWinfo is right, and my overclock is not really working. But for Joseph his overclock is showing in HWinfo, so his overclock works. However besides latency, i seem to have a bit faster bandwidth about 234 vs 265 +13% for reading. and similar for copy and write. Similar score made me believe that the overclock was working, before. Also in my own application, i think i was sloppy and compared this score with the one of the 5600 Fury pro sticks. However, now it seems that the overclock setting, is not really the cause of the performance.
My guess is that it is the difference is because of dual rank vs single rank. But i am not sure.

Kizune, Bert, anyone else... if you also have Servermemory running overclocked. Does it really show in HWinfo? Maybe you other than me, really succeeded in overclocking, or is it similar as here?

Another weird thing is this turbo max of 4700 for the cpu?
 

Attachments

Last edited:
  • Like
Reactions: Damian Höster

RolloZ170

Well-Known Member
Apr 24, 2016
6,520
2,017
113
Kizune, Bert, anyone else... if you also have Servermemory running overclocked. Does it really show in HWinfo? Maybe you other than me, really succeeded in overclocking, or is it similar as here?
josef used RDIMM with XMP profile
BIOS 0215 and HWinfo 7.63 5205 <- check this version ???
edit: Kizune's QYFS reported 3161mhz and used W790 Ace BIOS 0403 w. HWinfo 7.62 5200
 

DHamov

Active Member
Jan 12, 2024
112
27
28
what is the ASUS BIOS reporting ?
CPU-Z ?
Asus Bios Reports 6200 Or whatever i set it. in the first screen, not the screen where it is set. So first i believed the bios and not HWinfo. In windows Taskmanager (performance and then memory), and AIDA64 report what i set so say 6200. But cpuz and HWinfo say 2400 / 4800.
In Linux demicode says the properties of ram and as installed speed what is set in Bios (6200).

josef used RDIMM with XMP profile
I know, i am not doubting his results. I was doubting my results.
And i am also curious if others who overclock server memory, have a successful overclock or not? If the main boost is DR vs SR, this would speak for c741 serverboard.

Still Need to check hwinfo version, but since cpuz reports the same i think HWinfo is right.
 

Kizune

Member
Dec 2, 2022
65
49
18
aida64_mem.pngaida64_mobo.pnghwinfo64.png

The stuff hwinfo64 reports is weird. I mean - for some reason it counts memory reference clock as 50 instead of 100. So, yeah, super-strange.
 
  • Like
Reactions: DHamov

RolloZ170

Well-Known Member
Apr 24, 2016
6,520
2,017
113
The stuff hwinfo64 reports is weird. I mean - for some reason it counts memory reference clock as 50 instead of 100. So, yeah, super-strange.
this may be true because ASUS W790 is a OC board. mem.ref clock <> BCLK
 

DHamov

Active Member
Jan 12, 2024
112
27
28
The stuff hwinfo64 reports is weird. I mean - for some reason it counts memory reference clock as 50 instead of 100. So, yeah, super-strange.
Well, i could of course live with the factor of 2, and some BCLK factor.
The bigger difference is that in your screenshots HWinfo actually reports something else than 2400. So i think your overclock worked, and mine does not. Whatever i tried so far, HWinfo and Cpuz always report 2400/4800. With this Sage i could not even get it to change to 2600/5200 or 2800/5600 even though the memory is 5600 Micron.
Did you use manual timings and voltage settings? Or some other magical setting? Or do you have any idea what i might be doing wrong?
 

DHamov

Active Member
Jan 12, 2024
112
27
28
well, 7000Mt CL40 should not work with 1.1V - you know how to use the CMOS Reset Button ?
or set to 3600..
Sorry but i do not really understand what you mean. I am not expecting 7000MT to work.
I only used 6200 vs 4800 as an example because the bios says it runs both without complaining. And i wanted to compare the speeds, with enough difference in speed setting that it would be measurable w.r.t. the measurement error/noise. There seems not enough difference, hence i conclude that my overclock is not working. Now Kizune provided a screenshot of his HWinfo, and for him it seemed to have worked. So probably i am doing something wrong.
 

Kizune

Member
Dec 2, 2022
65
49
18
Well, i could of course live with the factor of 2, and some BCLK factor.
The bigger difference is that in your screenshots HWinfo actually reports something else than 2400. So i think your overclock worked, and mine does not. Whatever i tried so far, HWinfo and Cpuz always report 2400/4800. With this Sage i could not even get it to change to 2600/5200 or 2800/5600 even though the memory is 5600 Micron.
Did you use manual timings and voltage settings? Or some other magical setting? Or do you have any idea what i might be doing wrong?
I’m not doing anything special, just set the memory speed in bios. Not playing with the voltages or timings. Maybe you need to try different versions of bios. I remember on some versions the memory overclock was not working in exactly the same way - system was ignoring the settings I made and down clocked memory back to the 4800.
 

RolloZ170

Well-Known Member
Apr 24, 2016
6,520
2,017
113
I only used 6200 vs 4800 as an example because the bios says it runs both without complaining. And i wanted to compare the speeds, with enough difference in speed setting that it would be measurable w.r.t. the measurement error/noise. There seems not enough difference, hence i conclude that my overclock is not working
ASUS BIOS also can return autom. to defaults if you set some NOT working OC settings.
 

DHamov

Active Member
Jan 12, 2024
112
27
28
ASUS BIOS also can return autom. to defaults if you set some NOT working OC settings.
I understand, but there seem two different modes of doing that.
1) One mode is that you set it in the bios xxxxMT, Bios reboots and tries tried, does not succeed, reboots again, and you see some default settings in the bios, and apps like HWinfo, and all other apps. like 4800MT, or some times even 4000MT.
2) Now another way is, after setting it in the bios, it reboots, tries, then continues to start up the computer shows the set value in bios first screen xxxxMT, windows task manager, Aida64, linux demicode. But not in HWinfo cpuz. And is actually not running the settings.

Both things happen. Another thing i will try when i have time is do the same with QYFS.
 

maskhidden

New Member
Aug 21, 2020
2
0
1
Seattle
I just got the 8461V + SAGE SE combo. What seems unclear to me and others, from me reviewing the post is a clear answer on what is the highest rated overclockable RAM you can get?
 

RolloZ170

Well-Known Member
Apr 24, 2016
6,520
2,017
113
I just got the 8461V + SAGE SE combo. What seems unclear to me and others, from me reviewing the post is a clear answer on what is the highest rated overclockable RAM you can get?
5600 easy - 6000 should work (XMP) but remember this is still silicon lottery.
note there is a relation memory clock vs IDLE package power.
i personaly don't overclock, i got a cheap CPU and hope for a long life.
 

Bert

Well-Known Member
Mar 31, 2018
950
476
63
45
In same memory frequency and similar CPU specifications, E0 has higher memory bandwidth than D0
How is this possible? Isn't memory bandwidth determined by memory freq and channel count?

Also can someone share the differences between E0 and D0 chips? I understand E0 has "fewer bugs" but what are those bugs and any other change?

Also how can I get a list of all E0 chips, I am considering to switch to E0 now that we know that they work on Asus W790 MBs (not on ASRock).