Gigabyte P35 DS3L not holding overclock overnight

[CDC Mail Guy]

Well, I flashed F7 BIOS and did the removing battery thing and nothing worked. F'it, I'll run stock 'cause it's not like it's slow or anything. I still get awesome FPS in Crysis, and that's what I was really going for with the C2D.

 

[Idontcare]

Originally posted by: SerpentRoyal
Make a USB boot disc. Info on the web. HP makes a popular tool to do this. It's free.

I used this link for creating a bootable USB thumbdrive: Bootable USB drive guide

It includes the link for downloading the HP software you referred to.

Using this I have (just today) flashed my D3SL bios to rev F7 and am now doing some memtest+ loops.

USB bootdrive FTW!!!

 

[Billbo]

Why do you need to make it bootable? Use QFlash and so long as the thumbdrive is formatted as FAT/FAT32 it will read it. Just make sure you access QFlash outside the bios (Press End on boot). It wouldn't allow me to save the current bios correctly when pressing F8 in there.

 

[CDC Mail Guy]

I flashed new BIOS, and it took...the motherboard just won't remember my overclock. It's still doing the double turn on and off and on thing after I save the settings, then it reverts back to stock. I might just RMA this board anyway, as another problem I have is that sometimes I won't get a video signal when I turn it on. I have to shut the PC down, disconnect the cable from the video card, plug it back in, then turn on the computer again. This only happens "occasionally", but it's still a pain in the...

 

[SerpentRoyal]

Did you run memtest86 to check RAM? Use test #5 for 50 loops.

 

[Billbo]

SerpentRoyal, I don't think this is a memory problem but I could be wrong.

Mail Guy, can you please post what you are doing to OC.
After clearing the cmos once again yesterday, I got my E2160 up to 270x9. But I have to put more Vcore on than when I first bought this board a couple of weeks ago. Back then it was running at 3gig at 1.32V or so. Now I need at least 1.3625V just to reach 2.4gig. With that value set in MIT, the Health Status screen shows 1.348V and cpu-z shows around 1.32V.

This may not be the same for you Mail Guy, but 90% of the times an OC doesn't stick and the board does the double post, something changes in the bios or somewhere. You may think it's still ok and you can just try an overclock again. But for me issues arise preventing any more successful OC's until I clear the bios. The main symptom is it will continually refuse to remember settings, just like you're getting. Another is the no video signal - you're getting that too.

So what I did yesterday is start with a high but safe Vcore, get a successful OC, then slowly reduce Vcore until orthos gives errors. Doing this the double post hasn't happened yet. And this morning - for the first time in a week - the board remembered my settings overnight.

But why it needs more voltage than when new is a puzzle. Could this all be caused by regulators going bad on some of these boards? The only other thing I can think of is a graphics card compatibility issue.

These problems are just too common for Gigabyte to ignore, IMO.

 

[CDC Mail Guy]

Originally posted by: BillboMail Guy, can you please post what you are doing to OC.

I would be more than happy to

I followed mrfatboy's way from the following link

http://forums.anandtech.com/me..._key=y&keyword1=e6750+

Robust Graphic Booster Auto
CPU Clock Ratio 8X
CPU Host Clock Control Enabled
CPU Host Frequencey (mhz) 450
PCI Express Frequency (mhz) 100
C.I.A.2 Disabled
!! Performance Enhanced with Turbo
System Memory Multiplier (SPD) [2.00]
Memory Frequency (Mhz) 800 900
High Speed Dram DLL Settings [option 1]
DRAM Timing Selectable (SPD) [manual]
CAS Latency Time 5 [4]
DRAM RAS# to CAS# 5 [4]
DRAM RAS# Precharge 5 [4]
Precharge delay(tRAS) 18 [12]
ACT to ACT Delay (tRRD) 3 [Auto]
Rank Write to READ Delay 3 [Auto]
Write to Precharge Delay 6 [Auto]
Refresh to ACT Delay 42 [0]
Read to Precharge Delay 3 [Auto]
tRD 7 [Auto]
tRD Phase Adjustment 0 [Auto]

System Voltage Control [Manual]
Ddr2/ddr3 OverVoltage Control [+.3V]
PCI-E OverVoltage Control [Normal]
FSB OverVoltage Control [Normal]
(G)MCH OverVoltage Control [+0.1V]
CPU Voltage Control [1.45000]
Normal CPU Vcore 1.35000V

Following the above example WORKED the first time I did it, with NO double posts and I WAS running at 3.6GHz. I played for a while, turned off my computer and went to bed. Upon awakening in the morning, it double posted...I like WTF (?) and then Windows loaded and I was off and running...but only at STOCK speed. This was like a couple weeks ago.


Serpent, I downloaded and burned ISO of Memtest86 a minute ago, and I will run it the next time I start this damn thing.


Ooops...Can I say damn here?

 

[Billbo]

And you're turning off all the power saving crap I take it? What voltage does the pc health screen show for vcore compared to what you've set in MIT?

Did you read through my previous post? I'd suggest before running memtest to clear the bios just to make sure those problems I mentioned aren't there.

 

[Idontcare]

Guys - new DS3L owner here - I tried reading thru this thread to see if what I am about to post has already been tried or attempted...but this thread is a crazy mess (no offense!)

Had my board <24hrs. Flashed to F7. Cleared CMOS, all that jazz. All was good.

Started overclocking, noticed after a few intentional reboots (initiated thru windows XP) that the board "lost" my overclock settings...seemed a lot like what you guys are talking about.

BUT it was easily fixed, for me at least. When I change my BIOS settings (say FSB to 333MHz, etc) I exit the submenu's to get back to the main BIOS screen (push escape key) and then I press F11 to save my bios.

Pick any save slot, type a name and hit enter. Pressing enter pops you back to the main menu again. Then I press F10 to save and exit.

Viola, no lost OC settings anymore. Everytime I tweak a BIOS setting I make sure and save the BIOS with F11 (over-write an existing one if needed) and then F10 to save and exit.

Hope that helps if my problem was indeed the same as yours.

 

[Billbo]

Yeah that's another one of the many "fixes" I've tried with no luck. Another is to set the Halt on Errors to 'none'. Doesn't work here.

Mail Guy, have you at least attempted an overclock but leaving System Voltage Control on Auto for a couple of days? That seems to make my system stable but it runs too hot. If that works maybe you need a little more volts on the MCH/FSB/Ram.

What I noticed with my system today is even though the ram is supposedly rated at 1.8V for 800MHz, when Voltage Control is set to Auto the board is supplying nearly 2.1V to it. But I'm not OC'ing the ram so I don't know why the extra is needed. But this could be part of the instability probs on my board.

 

[CDC Mail Guy]

I did it, I did it!!! Thanks Idontcare for the tip of saving new profile, I think that's what did it. Also Billbo, I changed Voltage Control to Auto, but my temps are okay so far.

Here is a picture of everything I got goin' on http://star.walagata.com/w/gonzo69/1249842241.JPG

I got some weird 5-6-6-17 thing going on with the RAM though that I'll have to work with in the near future. But for now, and after several restarts from Windows, I am up and running @ 3.6 with great temps on air. I'm sure I forgot to add something, but I'll add to this "crazy mess" later as I think of it. ( Idontcare )

 

[Billbo]

Got my fingers crossed for you Mail Guy. Hope it sticks.
BTW, what are your load temps? No way I could get 26 deg idle in a 27 deg room

 

[SerpentRoyal]

Savings the profile works for some users.

I would reflash the BIOS and reset CMOS. Bump FSB up by 5MHz and see if this setting will stick. This tiny overclock should not result in BIOS reset if the board is okay.

 

[CDC Mail Guy]

Originally posted by: SerpentRoyal
Bump FSB up by 5MHz and see if this setting will stick. This tiny overclock should not result in BIOS reset if the board is okay.

So bump up CPU Host frequency to like 455 to do this?

 

[SerpentRoyal]

Yup. Abit boards increase the CPU's FSB by 3 to 4MHz from default. If the Gigabyte board cannot maintain this small overclock, then the board is defective.

 

[CDC Mail Guy]

Okay, I increased the CPU Host frequency to 455, and it's been holding for a few hours now. FSB is up to 1820 and 3.64GHz, and temps holding the same as at 450. Thanks for the help everyone.

 

[CDC Mail Guy]

http://star.walagata.com/w/gonzo69/1249842241.JPG

I guess I could just edit my post above this, but I don't know if it will bump or not

Anyway. CPUZ or CPUID is all ____ed up. Still showing multiplier as 6, so it makes me wonder if anything else is wrong (like the 1820.2 FSB it says I have). Obviously the RAM speeds are wrong. Anything else out there that will show my correct information?

 

[Billbo]

Sounds like you have EIST enabled. It will drop the multi when not under load.

 

[CDC Mail Guy]

Originally posted by: Billbo
Sounds like you have EIST enabled. It will drop the multi when not under load.

Okay I disabled it. Still shows multiplier at 6. Everything running great, so not a big deal.

 

[Idontcare]

I wanted to post about one more "quirk" of this DS3L board that I got snarled up in...again if this has already been discussed in this thread then my apologies for wasting your time reading it again here.

If you change BIOS settings, say change FSB from 300MHz to 333MHz, etc and for whatever reason you reboot the computer (press reset button, or even just tell windows to do a restart instead of a shutdown) before you do at least one hard shutdown then the BIOS assumes the prior BIOS settings were unstable and it lapses back to stock settings.

I found this out because I OC'ed my Q6600 from 333 to 367FSB...booted into windows, all was stable. Ran small FFT prime95 for a couple hours, all was stable. Then I did a windows update, which once concluded it rebooted (restarted) my computer rather than doing a complete shutdown.

Upon the reboot the bios reset my overclock because it did not see a shutdown/restart sequence since my most recent BIOS changes.

So to confirm this I reset my overclocks, booted into windows, then told windows to shutdown the computer. Once shutdown I wait 5-10s before pushing the power button. Viola all is good, after this I can have the computer restart (rather than full shutdown) and it keeps my OC settings without fail.

Have been messing around with the BIOS all day and this "trick" works like a charm.

 

[Billbo]

Not sure if that always works Idontcare. I noticed similar things but it hasn't worked every time for me.
My problems still persist. Was about to go and buy another PSU today in case that was the cause until I found even more posts of people having similar issues, swapping PSU's, CPU's and everything else but the problems persist.

Gigabyte is playing the Tech Support ping-pong game with me so I'm a little peeved.
I'm just going to wait until I can get to the city next week and grab another DS3L board because I like it's features. I couldn't possibly be unlucky a second time. If that doesn't fix the problem then at least I have an excuse to build another comp with the spare board Otherwise, I better be able to RMA it.

 

[Idontcare]

Yeah one thing I noticed was the CPU temp (as near as I could figure) made a difference at the time of the power-on.

If I reboot immediately after running a small FFT test (temps will be still in the high 30's for about 30s) then the mobo double-posts and loses its OC settings.

It is not consistent, but since I became paranoid about it I began letting the idle temps drop to <35C before I reboot (or even shutdown and reboot) and I haven't had a double-post since I began doing this.

Could be completely a coincidence though. These boards have personality to say the least!

 

[mingsoup]

Same EXACT problem after cold cold boots (overnight.) Currently I tried taking out my battery/PSU, reflash to F7, take out my battery/PSU again, then loaded up all my settings, did a save into profile and then a save and quit. I'll report what happens.

I also turned off half on errors. I will also shutdown and power on before I do the shutdown for the night, to try incorporate all ideas from this thread.

 

[Billbo]

I think it's just luck that any of these "fixes" work. Some people have reported their problems going away by themselves after a few weeks/months. Others have had perfectly good boards for several months before the problems show up. The annoying part for me is not knowing when. For the first week I was so happy to have a board that utilized S3 standby mode properly. Now I can't trust it in case it doesn't wake correctly, as often it will come back with no video signal or it reboots or locks up.

 

[VirtualLarry]

Originally posted by: SerpentRoyal
Yup. Abit boards increase the CPU's FSB by 3 to 4MHz from default. If the Gigabyte board cannot maintain this small overclock, then the board is defective.

A dirty trick used to win benchmarks, when reviewers run performance benchmarks on motherboards. That's a negative in my book. If you set it to a certain FSB, it should run that FSB, period.