Overclock.net banner
1 - 9 of 9 Posts

·
Registered
Joined
·
22 Posts
Discussion Starter · #1 ·
Let's preface this with I choose a rather easy over clock using a Q9550. nailed the PCI speed to 100, and passed prime95 and memtest's.

So as far as I was concerned the overclock was working fine.

All heck broke loose though, when I attempted to run drive benchmarks - BSOD came. Various BSOD's depending on overclock tweaking - 7F, 8E, etc. the BSOD's seemed to change depending on voltage settings, but I'm rather clueless to what would influence a PCI x8 card (btw, playing games with a HD 4870 still worked fine .. just not drive testing)

now the raid card is the PERC 5/i - so it's an external x8 PCI-E card, and I suspect that the overclock is having something to do with it.

it's definately the overclock, because running at 3.2Ghz with the CPU speed, RAM timings / speed, PCI bus speed and CPU voltage tweaked and everything else basically on auto - benchmarking the drives works fine. repeatedly.

as soon as I try to get it up to 3.4Ghz or 3.6Ghz .. just invoking the ATTO bench32 benchmarks cause an immediate BSOD - and I mean immediate - within seconds .. so I think it's either timing or something else that I'm not thinking of.

system notes:

Motherboard P5E3 deluxe X38
processor Q9550 C1
Power Supply 850W nexus
memory 2x1GB Corsair XMS DDR3-1333C9 (yes I know .. get better).
video card: Powercolor Radeon HD 4870 512MB DDR5

any ideas would be greatly appreciated - I don't want an aggressive OC, just something around 3.4 to 3.6 which should be attainable I would think.
 

·
Registered
Joined
·
603 Posts
How about your Northbridge voltage?
Have you tried increasing it a bit?

Quote:

Originally Posted by slngsht View Post
Have you tried bumping the southbridge voltage a notch? Since the southbridge handles hard drives, that could be the culprit.
Since it has the drives managed by a dedicated PCI card, the southbridge is unlikely to be causing the BSODs, IMO.
 

·
Registered
Joined
·
1,570 Posts
Quote:

Originally Posted by mcastaneda68 View Post
How about your Northbridge voltage?
Have you tried increasing it a bit?
Yep most blue screen can bet fixed with more volts to the northbridge,i would leave southbride alone.Do one change at a time
 

·
Registered
Joined
·
22 Posts
Discussion Starter · #5 ·
okay, had to pop out and overclock the sucker again .. mental note ... write down settings.. don't be stupid ..

SB on the motherboard only has two voltage settings 1.05 and 1.20 .. ick.

starting settings:

FSB strap - auto
FSB Freq - 400
PCI-E Freq 100
DRAM Freq DDR3-1332
DRAM Timing 9-9-9-24-2N

CPU Voltage: 1.28125
CPU PLL Voltage: auto
FSB Termination Voltage auto
DRAM Voltage: 1.60
NB Voltage: 1.25
SB voltage: 1.05 (only other setting 1.2)
LL calibration Enabled
PCIE Spread Spectrum disabled
CPU Spread Spectrum disabled

BSOD on bench32 instant: 0x0000007E

upped NB to 1.31 .. it just reboots, no BSOD when running bench32 - prime95 seems to be running fine with these settings.. how far can / should I continue to up the NB voltage?

tested all the way up to 1.39 volts. 1.39 has simply hung, so finally a different result.

Now i should mention the weird part of this - this is my boot drive(s) that I'm running the benchmark on .. so they obviously work in normal instances, just not running the #(*& benchmark
 

·
Registered
Joined
·
22 Posts
Discussion Starter · #7 ·
Quote:

Originally Posted by mcastaneda68 View Post
If I understand correctly, with NB Voltage at 1.31, you have got no more BSOD's, correct?

If so, leave it at 1.31, no need to increase it more.

The fact that the problem appears only when stress testing is normal. You are stressing and not only using your NB
BSOD's or just random resets .. didn't seem to matter what NB voltage I set it for up to 1.41.

tried all the FSB strap settings individually to no avail, and also the NB GTL reference. again, same scenario.

even though, yes, it's only on benching .. this is an IO intensive workstation, so it leaves that question if during heavy IO .. all heck is going to break loose when I least want it to.

amusingly, even setting it to 3.2Ghz, I can overclock the PCI-E to 110 without any issues (even running the ____ benchmark) .. so perhaps that is a better option.
 

·
Registered
Joined
·
22 Posts
Discussion Starter · #9 ·
Quote:

Originally Posted by microman View Post
NB voltage is WAY low, 1.25 is fine for stock but any OC is gonna require more than that. try 1.45v
I bumped it up to 1.41 before stopping for the evening with no real difference.

from 1.25 to 1.41 systematically it was fine .. prime95, etc .. even booting from the PCI attached raid controller - it was simply barfing on the benchmarks - which run fine at 3.2 and auto settings.

I'm sure it's something to do with the PCI bus - but even playing crysis at 3.4 .. the PCI worked fine .. so obviously PCI writing can still occur at high speeds .. just not the raid controller.
 
1 - 9 of 9 Posts
Top