begleysm

Hello, I'm new to the forums but I just built a brand new computer with the parts listed below.


I had it up and running without any OCing or modding and got a score of ~10300 on 3DMark06 (4 tests over 3 reinstalls of windows all gave similar results). I noticed that, on POST, my RAM was registering as DDR2 800 and I decided that, since Crucial guaranteed that it would run at 1066 5-5-5-15, I would up the speed and cut down the timing.

Now, I am pretty computer savvy but this is my 1st foray into OverClocking of any kind. I went into the BIOS, got into the RAM section, changed the multiplier from 4.00 to 5.33 (changing the RAM clock from 800 to 1066 MHz). Then I changed the timings from AUTO to MANUAL and knocked em down to 5-5-5-15. Restarted my comp and it basically failed miserably. I got a few bad sounding beeps, my computer restarted and it, appears, that the BIOS was kind enough to reset the RAM settings for me. So I went back in to the BIOS upped the multiplier back to 5.33 and left the timings on AUTO. This time the computer booted into Windows but it ran INCREDIBLY slowly (15+ seconds to register a mouse click or keyboard press). Also, my harddrive was making an audible click every 5 seconds or so.

I reformatted my HDD and reinstalled Windows and it booted up ok. Except now I get a score of ~8200 (2 tests) on 3DMark06 and my motherboard complains of either a MAIN BIOS CHECKSUM ERROR or BACKUP BIOS CHECKSUM ERROR on upwards of 75% of the times I start the computer. On that 25% when it doesn't happen it boots into Windows fine. I have tried flashing the BIOS (it is now version F4e) and I have tried taking out the CMOS battery to reset the BIOS. I've also let memtest86+ run for about 4 hours (about 4 passes) and got no errors.

My best guess is that this BIOS CHECKSUM thing is more of a coincidence (then as a result of my changing the RAM timings) and that my MOBO is shot. My plan right now is to call Gigabyte on Monday and then, most likely, RMA the thing on Tuesday after Gigabyte isn't able to help me.

I would appreciate any suggestions on the follow...
- how to go about diagnosing or fixing this BIOS CHECKSUM problem.
- why did my RAM timing changes fail so miserably?
- are there any RAM settings saved on the RAM itself or is it sort of "passive".
- any opinions on whether my meager overclocking attempts could have damaged something?
- any other reason I would see a ~20% performance hit on 3DMark06 besides my mobo is fecked or I ruined my RAM somehow?

Thanks,
Sean

jameskelsey

Did you short the pos to neg when you took the battery out or use the CMOS jumper to clear the CMOS?

The Duke

Checksum can happen if you have changed the BOOT order in any way.

I had that 5 second click when I killed a Seagate 7200.10, sometimes it gets seen in the BIOS some times it don't... now it just don't! So, if you have an issue with the HDD, it will screw with the Boot Order and cause a Checksum error!
I'd run a diagnostic on the HDD.

__________________

ericeod

When you flashed the bios, did you make sure to restore bios defaults first? I've seen a lot of problems when the bios is flashed with the OCed settings.

The board also defaults to 1.8v to the ram. So you have to make sure to give the ram the prpper voltage. The Specs from crucial say that the ram should be set to 2.2v, so you will have to add +0.4v to RAM Overvoltage Control.

__________________
eVGA 680i Overclocking guide .......... ASUS P5E/Rampage Formula Thread.......... 4.4GHz CPU-Z

noobdown

thats what i was thinking aswell

when you change the ram setting did you change the voltage at all?

__________________

ATi 4830 Club

smokinbonz

From what i have heard 4 sticks of ram stress the chipset more than 2, Not that its your problem ,not sure if it does limit people on their OC's . Have you set the voltage to your ram? It should be 2.2V I believe that is the voltage for the rated speed.


I think after you get everything squared away, next time make sure the voltage is set correctly for the ram and set the speed but leave the timings on auto, Boot up check out the timing in cpuz and then try to tighten from there

begleysm

Thanks for all the replies. I did not short the pos to neg posts when I took the battery out. Last night I reset the CMOS by shorting the CLR_CMOS jumper and after 4 or so restarts I have not seen the BIOS CHECKSUM error again. Also, I did not up the voltage when I modified the RAM timings and freq multiplier.

I have, however, had the computer just hang before the POST twice. I still am suspect of this motherboard.

My biggest concern right now is my score drop for 10300 to 8200 implying a 20% decrease in general gaming/graphics performance. I think everything is as it was so I don't know what would be causing this. My first guess is the motherboard and my second is the RAM.

And honestly, I'm a bit scared to try to OC anything since that first attempt appears to have left me with lasting damage of some kind... although, as I said the RAM made it through 4 passes of memtest86+ without an error.

Could my attempt at OCing the RAM (without increasing the voltage) damage the RAM in such a way that it slows down, but doesn't have any data errors?

Thank,
Sean

jameskelsey

A bad OC can cause data corruption in the OS so maybe do a repair install of windows if everything is at default in the BIOS.And never flash a BIOS with OC settings,always set to default first.

begleysm

I figured out the performance hit problem thanks to jam over at Computing.Net. There is a bug present in the L3 cache of the AMD Phenom 9500 and 9600 CPUs.

http://techreport.com/discussions.x/13724

This bug can occasionally cause a system to hang due to a race condition. AMD release a "patch" that cured this race condition but at the cost of a 10-20% performance hit on the CPU. The patch is applied through a BIOS update. AWARD implemented this patch in version F3 of their BIOS. I had patched to F4e. Once I learned about this suspect patch it was as simple as going into the BIOS and disabling the AMD TLB erratum patch.

As for the BIOS CHECKSUM ERROR, it appears that clearing the CMOS in the right way (using the CLR_CMOS jumper) fixed it. I had tried clearing the CMOS by removing the battery, but this was not sufficient. Thanks to MRCOMPUTER over at ********central.com for this suggestion.

If anyone has any similar problems and would like more details on my experience, feel free to email me. s.begley AT vanderbilt.edu

Thanks,
Sean

jameskelsey

Glad you got it figured out.You can clear the CMOS without the jumper if you remove the battery but you then need to short the negative and positive battery contacts.