BSOD on CPU VTT Voltage change

C

Carter54

Limp Gawd
Joined
Feb 4, 2007
Messages
168
ALL,

Working to further OC the Comp in my sig and I seem to have hit a wall. here is what I am at.

VCORE 1.5
NB core 1.45
SB core 1.50
1.2 HT 1.45
CPU VTT 1.25
DIMM 2.175 (board overvolts a little to 2.2

I am currently able to get a stable OC at 1550 FSB using 1:1 ratio in Sync mode and a 9X multiplier. I was trying to push this to 1600 fsb and am running into all kinds of problems. anytime I adjust CPU VTT above 1.25 I get a system lock up or BSOD. I even set all clocks back to stock and this happens any Ideas on why? Any suggestion on getting my system to 1600 without being able to adjust cpu vtt anymore?

Thanks in advance carter.
 
CPU VTT above 1.25 I get a system lock up or BSOD
Click to expand...

It attempts to adjust the voltage levels of the data pulses on the FSB by raising the voltage fed to the Northbridge that is used to create the pulse level reference voltage. As you have found, typically it hurts more than helps.

My calculations indicate you are running at 3487.5 MHz. It is likely thats it.

Have you determined what the max FSB capability of the board is by running a 6x cpu mulitplier, loose memory timings and seeing how high the FSB can go ?

try with a 8x mulitplier (you will need a 425MHz FSB to acheive the same speed, start at 400 and work up 5MHz at a time, testing as you go) This will allow you to run a 1:1 memory muliplier linked which is typically more stable from a system standpoint. This may get you around an FSB wall concerning Northbridge timings but I am not an expert on Nvidia chipsets, could not hurt tho. Somewhere around 425 you will need to probally loosen memory timings as now you will be OCing your ram as well as the cpu.

Temps at current settings ?

Do you really need the 1.5 Vcore to run at your current speed ?
 
Bill here is my Sequence that I have tried.

Ran FSB up to 1400 with a 10X multiplier which is 3.5 on the cpu and all benchmarks were Ok including dual Prime 95 with 1.5 VCORE thats why its that high.

I dropped the multiplier to 8 and could get 1600 FSB no problem however I had RAM problems going over 400 at all on my ram with 4-4-4-12 timings and voltage is set to 2.15( my board seems to overvolt a little). I increase the ram voltage only to 2.175 and my board started to become unstable so I tried 2.2 and windows would make it more than 15 minutes so back down to 2.15 and increase the multiplier back up.

I then set the multiplier to 9 and started climbing using steps of 10 from 1400 to 1560 is where I am at right now with a 9X multiplier cpu 3510 and my board is not stable at 1.48 vcore but is stable at 1.5 vcore using Prime95 as a stability test.

Temps are 42/39 idle and 58 under load with 1.5 vcore the reason is ambient temp im my PC room is about 20C in the winter months.

This board seems to like alot of voltage to get stable overclocks. I have been reading several reviews of this board and they all had fairly high voltages to get a decent OC.

Anandtech review:
http://www.anandtech.com/mb/showdoc.aspx?i=2958

Hard OCP review:
http://www.hardocp.com/article.html?art=MTI3NiwsLGhlbnRodXNpYXN0

Today I will loosen my memory timings to 5-5-5-15 and see if I can get the FSB up to 425 area with an 8X multiplier. I did buy good RAM so it should be able to OC a little.

Thanks for your feedback on this Carter.
 
Results so far:

VCORE set to 1.48 (left it as is from the 9X multiplier)
All other voltages same as in original post.
RAM timings are 5-5-5-15 2.15 VDIMM
Multiplier is 8X

1600FSB booted no problem ran Prime59 15 min
1610FSB booted no problem ran Prime59 15 min
1620FSB booted no problem ran Prime59 15 min
1630FSB booted no problem PC locked up in 2 min
1635FSB booted no problem PC locked up in 1 min
1640FSB would not boot

I read in the anandtech review that there were FSB holes between 402 and 416 so I tried a few different FSB settings.

The VDIMM was reported out at 2.18 using everest ultimate edition and PC Probe so I bumped up the VDIMM to 2.175 and started over a 1630 FSB.

With Vdimm at 2.175 I am running:
1630FSB booted no problem ran Prime95 PC locked up after about 10 minutes

Its looking like my RAM might be the limiter do I need to loosen the timings more? The memory is rated to 5-5-5-15 @500mhz and 2.2V so it should be able to handle it.

Carter
 
Its looking like my RAM might be the limiter do I need to loosen the timings more? The memory is rated to 5-5-5-15 @500mhz and 2.2V so it should be able to handle it.
Click to expand...

Disclaimer, I am not an expert on Nvidia boards. One thing that comes to mind is the "SLI ready" memory option, if enabled will read (if they exist the EPP high performance profiles stored on the memory stick. Probally should be disabled for now.

I agree, it should not be your memory, reduce your cpu mulitplier to 6x, this will run the cpu at its slowest possible setting at high FSB. Download and make a memtest86+ cd and boot into that doing your same FSB value testing, Since we have taken the CPU out of the picture by running it very slow, doing this testing should tell you where your memory tops out. If you can get 425+ (1700) and a full pass of memtest with no errors it would tend to support that you have just gone as far as you can with that CPU as it has topped out. At the least you prove it is not the memory and can then start to look harder at what can be done to other thing.
 
If your worried about memory, then just lock it at it's rated speed. Then you'll know.
 
Its the memory! I cant get a full pass of Memtest86 without errors even starting at 1600 with a 6X multiplier. I took pictures of my bios settings so maybe there is a setting that I have wrong but it looks like I have a bad Memory module.

Extreme%20Tweaker%20Main%20Screen.JPG


FSB%20and%20memory%20Config.JPG


Memory%20Timings.JPG


Over%20voltage.JPG


Overclocking.JPG


Spread%20spectrum.JPG


System%20Clocks.JPG


Temp%20Monitor.JPG


Voltage%20Monitor.JPG


I will be calling crucial on Monday but any help between now and then would be helpful.

Thanks Carter
 
After my memory failed memtest I set the PC back to POR specs everything on auto and multiplier back to stock. I ran a full pass of memtest and it passes the memory was running at 800mhz 4-4-4-12 timings and voltage was on auto reading was 1.87v using pc probe and everest.

I then set the bios back to manual and unlinked the FSB and memory and started trying to figure out the theoreticle limit of my cpu. I started at 1440 fsb with a 10X multi and it booted at 3.6 ghz but I could not get stability up to 1.52 volts on Vcore I left everything else on auto. I stopped there and lowered the fsb to 1400 with a 10X multiplier at 1.5vcore and It seems to be prime95 stable it has been running for 2 hours now. I am going to run memtest overnight to see if it generates any errors. As of right now I have a stable Overclock at 3.5ghz on my cpu using 1400 fsb with unlinked memory ratios is 8:7.

Carter
 
I would be thinking about the thermals on the Northbridge now as it gets just as OCed as the CPU and it runs the memory sub-system that seems to be your weak link.
 
You must log in or register to reply here.
Back
Top