Gigabyte Latest BIOS

[joeblough]

Originally Posted by joeblough

i thought i would just mention that i upgraded to F9a (mobo is GA-EX58-UD4P), and resume from S3 now seems very stable with all voltages set to auto, bclk=166Mhz and multiplier=18 (cpu = 3Ghz). thanks to gigabyte for continuing to refine the bios.

well, unfortunately even though the system wakes very reliably now even when overclocked, i've found that mprime will fail very quickly if it's started after an S3 sleep/resume cycle. yes, i am overclocking and i have most everything set to 'auto'. i know you are not supposed to do this. however, the mprime failure happens even if the system is not overclocked (BCLK=133MHz, cpu=2.66MHz) this is a UD4P with bios F9a. unfortunately i'm not running windows so i cant look at all the clock/voltage parameters after wake, but i assume that the bios must be messing up a clock frequency or a voltage after resuming from S3.

edit: i take it back. it just failed after a cold boot, no sleep. this sucks - maybe something is wrong with my cpu.

edit again: nope, it was the dram - fails memtest86 near the 11GB mark. bumped the voltage a little bit and all seems well. interesting that i never had this problem before bios F9a.

[Anpheus]

Originally Posted by joeblough

well, unfortunately even though the system wakes very reliably now even when overclocked, i've found that mprime will fail very quickly if it's started after an S3 sleep/resume cycle. yes, i am overclocking and i have most everything set to 'auto'. i know you are not supposed to do this. however, the mprime failure happens even if the system is not overclocked (BCLK=133MHz, cpu=2.66MHz) this is a UD4P with bios F9a. unfortunately i'm not running windows so i cant look at all the clock/voltage parameters after wake, but i assume that the bios must be messing up a clock frequency or a voltage after resuming from S3.

edit: i take it back. it just failed after a cold boot, no sleep. this sucks - maybe something is wrong with my cpu.

F9a on my UD4P was unstable for me and did not reliably wake. While our symptoms sound different, I would suggest going back to F8a if you can and seeing if your reliability improves.

[Lsdmeasap]

Originally Posted by joeblough

well, unfortunately even though the system wakes very reliably now even when overclocked, i've found that mprime will fail very quickly if it's started after an S3 sleep/resume cycle. yes, i am overclocking and i have most everything set to 'auto'. i know you are not supposed to do this. however, the mprime failure happens even if the system is not overclocked (BCLK=133MHz, cpu=2.66MHz) this is a UD4P with bios F9a. unfortunately i'm not running windows so i cant look at all the clock/voltage parameters after wake, but i assume that the bios must be messing up a clock frequency or a voltage after resuming from S3.

edit: i take it back. it just failed after a cold boot, no sleep. this sucks - maybe something is wrong with my cpu.

Well since you are using Auto when overclocking that is likely the whole issue you are having with resume. They system must be stable to resume properly and Auto often just will not cut it, or gives things to much voltage which can also lead to instability.

But yes, I use UD4P myself so I can assure you that turbo is disabled after a resume which would leave you with a unstable 20 multi and likely also incorrect voltages for the clocks in question. It seems to also change CAS Latency to 4, which is this is a true report to software of this change would also lead to MUCH instability as well all know DDR3 cannot do CAS4. So even if the reported 4 is correct or not, it could be lower then it should be set to and just reporting wrong, but either way I think this also is causing issues.

I have sent in a report about these issues so they are aware of it and I am sure new beta's will soon try to address these issues.

I doubt anything is wrong with your CPU, you just need to setup the BIOS properly. If you need help with that please make a thread and we can help. Even at stock speeds Auto may not properly set things up depending on your ram and ram speed, ect.

I also agree with Anpheus, if you had stability on F8a then I would go back to it as F9a offers no changes/fixes that I have observed

[joeblough]

Originally Posted by Lsdmeasap

Well since you are using Auto when overclocking that is likely the whole issue you are having with resume. They system must be stable to resume properly and Auto often just will not cut it, or gives things to much voltage which can also lead to instability.

But yes, I use UD4P myself so I can assure you that turbo is disabled after a resume which would leave you with a unstable 20 multi and likely also incorrect voltages for the clocks in question. It seems to also change CAS Latency to 4, which is this is a true report to software of this change would also lead to MUCH instability as well all know DDR3 cannot do CAS4. So even if the reported 4 is correct or not, it could be lower then it should be set to and just reporting wrong, but either way I think this also is causing issues.

I have sent in a report about these issues so they are aware of it and I am sure new beta's will soon try to address these issues.

I doubt anything is wrong with your CPU, you just need to setup the BIOS properly. If you need help with that please make a thread and we can help. Even at stock speeds Auto may not properly set things up depending on your ram and ram speed, ect.

I also agree with Anpheus, if you had stability on F8a then I would go back to it as F9a offers no changes/fixes that I have observed

well i just edited my post - it turns out it was the dram, and was not related to sleep at all. it eventually failed mprime even after a cold boot.

suspecting the memory, i left memtest86 running overnight and found that between 11 and 12GB there were failures. the memory is rated at 9-9-9-24 and 1333MHz, and that's how i have them set in the bios - no overclock for the ram, just the CPU. if i do not tweak the bios settings at all, when bclk=133Mhz and cpu=2.6Ghz, the bios actually pics 8-8-8-19 (IIRC) for the memory, so in fact i had to manually set it to 9-9-9-24 and lower the memory multiplier when setting up the CPU overclock. anyway i increased the Vdimm to 1.580V and memtest86 is happy now, at least for 1.5 iterations.

i actually did not have stability on F8a. the system would die about 1/2 the time resuming from sleep. in fact F9a has been very stable for me resuming from sleep. like 100% stable. i have not repeated my mprime testing yet because i want to let memtest86 run for a few more hours to really prove that the memory is now stable.

sorry for the bad science and the noise. i'm really confused though, since F9a has been very stable from an emprical point of view. like i said i cant see all the multipliers and voltages because i'm not running windows. i actually have turbo mode turned off right now, maybe that's why i don't see any problems?

[Octopuss]

I was just thinking, why is Stasio posting the AHCI BIOS version when it's the same old one?
And how's the status on P35 boards anyway? :)

[rge]

I am using the modded bios 8b with the new AHCI ROM posted by noangel in post 943. Can someone request that GB put this new AHCI ROM in next beta version of 8 for extreme.

The new AHCI ROM in modded 8b bios cuts 8 seconds off boot time for the extreme. The annoying 8 second delay of dots where the hard drive search occurs on unmodded bios 8b with OLD AHCI ROM is no longer tolerabe, now that I have been spoiled with the better ahci rom......just would like all future bioses to have the new ahci rom...if someone would pass along the request I would appreciate it.

[tijgert]

Well... I'd like to have that modded bios then! Care to share?

[GHz]

NoAngel's post:
http://forums.tweaktown.com/298926-post947.html

I flashed it to my UD5 and it's working well so far :)

[tijgert]

Yeah sorry about that. You actually posted the message number instead of just hint at it... but you responded before I could correct my oversight. Your extra link only makes me appreciate you more :)

[casperse]

Hi All

I am going crazy trying to get S3 working on my Gigabyte Board

I am using Windows 7 last build and when I try S3 the PC goes to sleep but when I try to power it on again it boots from scratch and I get a "Windows was closed down to prevent...bal bal bla" etc...

And in the Event log I can se the "HALL error message"...
The platform firmware has corrupted memory across the previous system power transition. Please check for updated firmware for your system."

My setup is this:

MB: GA-EX58-UD5 (Running last official Gigabyte Bios)
RAM: Corsair XMS3 DDR3 1600MHz 6GB CL7-TR3X6G1600C7G
CPU: Intel Core i7 I7-920 8 MB

When reading this Thread people are saying that in order to get S3 working you need to:

HPET=off
enable EIST

I have tried all this and also disabling SATA (running IDE Legacy).
And the other way running ACHI/RAID with Intel Matrix Manager...
Nothing works so I am writing here because it must be a BIOS PROBLEM?

Please Help / Advice....
Regards
//
Casperse