P55-UD3R incorrect CAS latency setting

[Csybe]

Hi,

I have a P55-UD3R (BIOS ver. F5) and a Geil Value Kit 2x2GB 1600MHz 8-8-8-28 1.6V.

My problem is that the CAS latency is not what I set up in the BIOS, and I'm unable to run it at its rated timings.

There is always a difference between the CL set in the BIOS and the real value according to CPUZ and Everest. Here is what I have tested so far (voltage 1.6V):

6-8-8-28 in BIOS -> 10-8-8-28 according to CPUZ and Everest
7-8-8-28 -> 11-8-8-28
8-8-8-28 -> 11-8-8-28
9-8-8-28 -> 12-8-8-28

With SPD or XMP settings, it won't POST - rapidly repeating short beeps then restart at default settings notifying me that the system failed to boot last time. Also, I thought all I'll have to do is just set 5-8-8-28 in BIOS to reach the desired 8-8-8-28 timings, but it won't post with these settings either.

I had the similar issue with BIOS F4 and F5b... any ideas?

Thanks!

[Csybe]

Interesting, it seems to change as I play with my BCLK and QPI multiplier as well

[FELIX]

...could you try with cpu-tweaker.
Thank.

[massman]

Cas latency cannot be read correctly by CPU-Z, it's a known bug, probably because the CPU and IMC are now on seperate dies. Because there's a longer latency between IMC and CPU (through QPI link), the reported Cas Latency is higher.

What you set in the bios is applied correctly, though.

[Csybe]

...could you try with cpu-tweaker.
Thank.

Thanks for your reply. CAS latency and Command Rate is not active for me, I cannot change them.

[Csybe]

Cas latency cannot be read correctly by CPU-Z, it's a known bug, probably because the CPU and IMC are now on seperate dies. Because there's a longer latency between IMC and CPU (through QPI link), the reported Cas Latency is higher.

What you set in the bios is applied correctly, though.

Well if that's true that's the best what could happen. Do you know any software yet that is capable of reading timings correctly from Clarkdale IMC? Thanks for your help!

Edit: assuming that it reports CAS latency only 1 higher (9 instead of 8) when I use higher QPI ratios, You are right

[stasio]

Well if that's true that's the best what could happen. Do you know any software yet that is capable of reading timings correctly from Clarkdale IMC? Thanks for your help!

Edit: assuming that it reports CAS latency only 1 higher (9 instead of 8) when I use higher QPI ratios, You are right

Which CPU you using?
Did CPU-Tweaker show correct?Btw,CAS latency you cann't change but Command Rate yes.
Also you can try:
HWM BlackBox 2.0
HWiNFO32 3.37-540
SIW 2009.10.22
MaxxMEM²v.1.74

[Csybe]

Which CPU you using?
Did CPU-Tweaker show correct?Btw,CAS latency you cann't change but Command Rate yes.
Also you can try:
HWM BlackBox 2.0
HWiNFO32 3.37-540
SIW 2009.10.22
MaxxMEM²v.1.74

It shows 10-8-8-28-1T (8-8-8-28-1T is set in BIOS), CL and Command Rate cannot be changed, their repsective dropdown menus are inactive. I'm using an i5 650, trying to stabilize at 4.5GHz

Also, thanks for the links, I have tried hwinfo and maxxmem - one reported 9-8-8-28, and other one reported 10-8-8-28.

[FELIX]

Thanks for your reply. CAS latency and Command Rate is not active for me, I cannot change them.

...I know that, I just want to know if Cas# is read correctly with cpu-tweaker.

[massman]

Well if that's true that's the best what could happen. Do you know any software yet that is capable of reading timings correctly from Clarkdale IMC? Thanks for your help!

Edit: assuming that it reports CAS latency only 1 higher (9 instead of 8) when I use higher QPI ratios, You are right

I don't know any software that reads it out correctly at this time, but again: whatever you set in the bios is the CL you apply.

Try to use even higher QPI ratios. If you go 'too high' the CL will even drop below what you've set in the bios. See screenshot:

[Csybe]

massman: thanks, I'm can confirm that. Unfortunately, at the moment I'm a bit struggling to get my 650 stabilized in my P55-UD3R and 2x2GB Geil 1600MHz 8-8-8-18 Kit... The board seem to run fine at 200MHz BCLK, but the CPU maxes out at 4.2GHz which I don't believe... I also have problems with higher QPI multipliers, I'm at only 2450MHz (CPU-Z). :S

Btw, when pushing too far, under Prime95 torture test, in most cases my rig just stops responding or I get a BSOD saying "a clock interrupt was not received on a secondary processor" - is that normal?

[pangingIII]

about the CAS latency is not what you set up in the BIOS

i don't have your board but i use P55A-UD3P and i5-660 and i think this solution below might work for you.

-Load Optimized Defaults in bios, save/exit bios setup and reboot, and then reset bios to your liking.
-or, Set the bios the way you want it, shutdown and restart computer and next time you are in Windows the CPU-tweaker should correctly report you bios's memory timings
if you change memory timings, memory clock speed and/or QPI link multi again try redo the step.

[Csybe]

pangingIII: thanks, unfortunately, it does not work for me - after startup, CPU-Z reports 11-8-8-28, CPU Tweaker 6-8-8-28. Anyway, it's not a big problem if the correct timings are applied - I'm sure developers will release newer revisions of their utils to work around this issue.

[stasio]

Csybe,
can you try this CPU-Z version:

CPU-Z 1.53.2

-fix bug version
-released 05.02.2010

[pangingIII]

i am not sure that this has to do with either the board's BIOS + CPU's IMC or windows software or all of them.
anyway, i decide to check out some bios's memory tRD settings i just found that when i lower the tRD value enough the windows software reports correctly the memory timings and as the same that i set in bios.



next i would check system stability with the tRD value settings when i can.

[FELIX]

I think "Static tRD" is the same timing that "Read and merging-write window" in cpu-tweaker. (in this case, I'll rename it in the finale version)
It's possible that when you change the "Read and merging write" timing with cpu-tweaker, Cas# change in cpu-z,
cause these timings are exactly at the same register address in P45 chipset.
I use an other address in cpu-tweaker, but I'm not sure that it's right...

[Csybe]

Do we know what the "Performance Enhance" does with memory settings on Gigabyte P55 boards?

stasio: not fixed, still reports cL of 11

[massman]

I think "Static tRD" is the same timing that "Read and merging-write window" in cpu-tweaker. (in this case, I'll rename it in the finale version)

I think you're right!

[FELIX]

I changed reading Cas# in this version: CPU-Tweaker14b9
I'm not sure it's the best way, but it seem to be better than the old reading...