Solved (maybe): Help with OC settings and stability on GA-P35-DS4 and C2D e7500

[TylerS]

Hi folks - want to thank everyone for their help. I think I have this beast beat now. If you are interested, please have a peek at my latest post for an update.

Hello everyone:

I have been asking some piecemeal questions to try to get a bit more speed out of my rig, but in my fiddling I have found I have some stability issues and thought I would make a more compressive post to see if anyone can help me with the fine tuning.

The issues:
- random crashes while computer at idle: computer freezes or reboots -- I am not on the computer when this happens, so I cannot comment on exactly what happens at freeze
- BSOD after 3 to 4 hours of Prime95

Toubleshooting already done:
- Full Mallware/Anti-virus/Rootkit scans with MS Security Essentials and AVG Anti Virus 2013
- Minor bios tweaks do no appear to change much:
-- dropped DDR2 OverVoltage Control to +0.35V (was maybe less stable)
-- seems to run fine at "full stock/default" settings (though I haven't done any long stress runs)

System:
C2D e7500 @ 3.93
Gigabyte GA-P35-DS4 (v.2.0)
6GB PC 800 Patriot @ 859
Asus 7870 (stock)
Corsair HX620 psu
Thermaltake Water 2.0 Performer (cooler)

BIOS:
Robust Graphics Booster___________ [Auto]
CPU Clock Ratio ____________ [11 X]
Fine Clock Ratio ____________[N/A]
CPU Host Clock Control_ [Enabled]
CPU Host Frequency (MHz)__________ [358]
PCI Express Frequency (Mhz)_______ [100]
C.I.A. 2__________________________ [Disabled]
System Memory Multiplier (SPD)____ [2.40]
Memory Frequency__________________ 800 859
DRAM Timing Selectable_______ SPD __ [Auto]

Performance Enhance = [Standard]

CAS Latency Time________________ 5
Dram RAS# to CAS# Delay_________ 5
Dram RAS# Precharge Delay_______ 5
Precharge Delay (tRAS)__________ 16

ACT to ACT Delay (tRRD)_________ 3
Rank Write to READ Delay________ 3
Write to Precharge Delay_________ 6
Refresh to ACT Delay______________ 52
Read to Precharge Delay__________ 3
Static tRead Value_______________ 7
Static tRead Phase Adjust________ 0 [Auto]

System Voltage Control____ [Manual]
DDR2 OverVoltage Control__ [+0.40V]
PCI-E OverVoltage Control_ [Normal]
FSB OverVoltage Control___ [+0.15V]
(G)MCH OverVoltage Control [+0.125V]
Loadline Calibration______ [Enable]
CPU Voltage Control_______ [1.36875V]

Limit CPUID Max. to 3.....................: [Disabled]
No-Execute Memory Protect............: [Enabled]
CPU Enhanced Halt (C1E)................: [Enabled]
CPU Thermal Monitor 2(TM2) ...........: [Enabled]
CPU EIST Function.........................: [Enabled]
Virtualization Technology................: [Enabled] Enabled if you use Vmware/Virtual PC

Integrated Peripherals
Legacy USB Storage Detect___________________[Enabled]

Temp readings (averages of Core Temp, HWMonitor)
CPU: Max 56C
System: Max 60C

Questions:
1] Any idea to the likely cause of instability/crashes?
2] What settings should I consider adjusting to improve stability but at the same time maintaining a similar OC level?
3] It has been suggested that +0.40 (Base of 1.8V + 0.40 = 2.2V) is too high for DDR2 Ram (thanks Idontcare – very helpful responses) – is this the case in all instances: I don’t remember 100% but I know the Patriot sticks I have need a bit more volts than normal.
4] How are my FSB and (G)MCH (is this Northbridge?) levels? Could these be the problem?

5] And the question I don't really want to hear the answer to, am I just asking too much of this set-up?

Any other feedback would be appreciated.

Thank you.

 

[TylerS]

I thought I would pst a quick update since no one has any ideas. I am assuming that means things are okay(ish) with my settings and/or my computer is so old that no one can really help much.

The BSOD Stop Code is 0x0000007e (plus more codes which I didn't catch yet.) It seems it may be driver related, but I thought I would ask the experienced overclockers here if it is possible that it is still an issue with stability of my overclock settings?

Is there better forum section for this request?

Thanks.

 

[sm625]

"DDR2 OverVoltage Control__ [+0.40V]"

That is freakishly scary.

System Memory Multiplier (SPD)____ [2.40]

Can you change that to 2.0?

Also

CPU Voltage Control_______ [1.36875V]

Dont be afraid to punch that up by 0.050 or more. You have a cool running chip. It can take quite a bit more voltage.

 

[VirtualLarry]

You shouldn't need to boost the GMCH voltage, nor the FSB voltage, for only a dual-core and 358FSB. You also shouldn't need +0.40v on the DDR2 voltage. Unless you have some crazy overclocking memory that requires high volts. I would drop that down to +0.2v.

Make sure that your memory clock is set to 1:1, or 2.0 in gigabyte speak, not 2.4.

Also, your tRAS should be higher, try 18.

You should have more vcore headroom, depending on cooling.

Edit: TBH, you should start over. Reset CMOS settings to optimized defaults, set host clock control to enable, set memory multiplier to 2.0, set memory timings to manual, set 5-5-5-18. Enable CPU voltage control. Set vcore to 1.3v.

Then boot windows, run OCCT:linpack or Prime95 small FFTs.

Test for 10-20 minutes. If no errors, boot back into BIOS, increase host clock by 10.

Test again. Repeat. If errors, then increase vcore.

If temps get too high, drop host clock by 5-10, and test again. This should be your stable OC.

Edit: Don't go above 1.38v though. That's the max for a 45nm Core2 CPU.

 

[tweakboy]

I thought I would pst a quick update since no one has any ideas. I am assuming that means things are okay(ish) with my settings and/or my computer is so old that no one can really help much.

The BSOD Stop Code is 0x0000007e (plus more codes which I didn't catch yet.) It seems it may be driver related, but I thought I would ask the experienced overclockers here if it is possible that it is still an issue with stability of my overclock settings?

Is there better forum section for this request?

Thanks.

To me that is low voltage problem. Your monster OC of 3.9Ghz needs some or a lot of juice. Also PLL voltage take it to max NB voltage take it up a notch or two. Then let us know if it crashes still. thx gl

 

[TylerS]

Thanks. To be honest, I was trying to cut corners and jump ahead on the process. I knew it was probably a bad idea. I wasn't going in completely blind as I ha done the process before, but I was trying to jump way ahead simply because I installed a new cooler.

One question: you say set ram to 1:1, but that puts ram at under 800 until I et up to 400 fsb. Aren't I losing a bunch of ram performance?

 

[TylerS]

Edit: Don't go above 1.38v though. That's the max for a 45nm Core2 CPU.

Thanks VirtualLarry and SM625 -- great advice above that has help smooth things out.

I started over from scratch and was able to get to 4.0 GHz running smoothly but not (yet) fully stable. I have run both 11x363 and 10x400 - 10x400 kicked back an error on Prime95 (Core 2 only) after 55 minutes. I didn't run 11x363 past 10 minutes – I intend to test it.

Current voltage settings:
System Voltage Control____ [Manual]
DDR2 OverVoltage Control__ [Normal]
PCI-E OverVoltage Control_ [Normal]
FSB OverVoltage Control___ [Normal]
(G)MCH OverVoltage Control [Normal]
Loadline Calibration______ [Enable]
CPU Voltage Control_______ [1.36875V]

Temps (HWMonitor): e7500 reading: CPU 58C max
Motherboard reading: System 45C max; CPU 67C max

I want to try to bump up vCore a bit more - the next available setting is 1.375v (after that I'm over 1.38). However, BIOS Health screen reads this as 1.380v. But I ran it for a couple minutes and HWMonitor shows vCore at under 1.37V.

Which do I trust?

Is it better to be cautious and just trim down FSB to the stable point at 1.36875v? (Per HWMonitor max vCore was 1.360v.)

I’m going to test this but since I am asking about vCore, would there be any reason why 11x363 would be better than 10x400? You are left with a lower FSB, so I assume it is less stressful on the CPU.

EDIT: Nevermind this question. Idontcare answered it very well in a reply to a previous post here.
Going with 11x363 drops RAM speed to 726 at 1:1; I want to run at rated 800 (or as close as I can) but to be honest I want to only because that’s the spec for the RAM and it bugs me if it isn’t at or above it. Is there any real world loss of performance by dipping to 726? Am I just being silly?

 

[VirtualLarry]

1.38v is under load. You can set the BIOS slightly higher.

However, I do have some experience with an E5200 degrading, set to 1.425v in BIOS, 1.4v under load. Was stable at 3.75, then 3.65, then only 3.5 (with a subsequent reduction in vcore setting too).

It has been said that 1.4v is "safe", but after that degradation experience, I tend to be slightly more conservative.

 

[TylerS]

1.38v is under load. You can set the BIOS slightly higher.

Just to follow up then, HWMonitor readings can be trusted for vCore?

 

[TylerS]

Another question on reading vCore: I am not getting any readings past a max vCore of 1.360 no matter how high I set the CPU Voltage Control in BIOS. I am reading it with HWMonitor, SpeedFan and CPUID -- all show the vCore never exceeds 1.360 at load even with a BIOS setting of 1.38XXX (sorry do not remember the exact, but it's just over 1.380).

What does this mean? Is there something inhibiting voltage? Are the readings incorrect?

Does the motherboard/CPU simply think it doesn't need to pull that much voltage? If it is this, is there a way to "coax" it to?

 

[VirtualLarry]

Another question on reading vCore: I am not getting any readings past a max vCore of 1.360 no matter how high I set the CPU Voltage Control in BIOS. I am reading it with HWMonitor, SpeedFan and CPUID -- all show the vCore never exceeds 1.360 at load even with a BIOS setting of 1.38XXX (sorry do not remember the exact, but it's just over 1.380).

What does this mean? Is there something inhibiting voltage? Are the readings incorrect?

Does the motherboard/CPU simply think it doesn't need to pull that much voltage? If it is this, is there a way to "coax" it to?

This can be an issue with "vdroop". Giving it more juice, doesn't help, because your mobo VRMs are at their limit.

 

[TylerS]

This can be an issue with "vdroop". Giving it more juice, doesn't help, because your mobo VRMs are at their limit.

So there is no point really in setting vCore much higher than 1.360, or whatever setting maxes it to that?

 

[VirtualLarry]

Possibly not. Look, your components are older, if they have been run overclocked all this time, then they may have degraded, and pushing them to even higher overclocks may be impossible for them, in their current state.

Just get what you can out of them, fully stable, and start saving up for your next upgrade.

 

[tweakboy]

Dude, just run stock clocks on everything BIOS defaults no GPU OCing then tell us if your getting the BSOD,,, gl Also try installing latest WHQL drivers from nvidia or ati.. gl

 

[TylerS]

your components are older, if they have been run overclocked all this time, then they may have degraded, and pushing them to even higher overclocks may be impossible for them, in their current state.

Thanks for this. I was wondering if this was an issue all along, but no one really mentioned it so I continued to push.

Also - it seems I may have a bad set of RAM (I still have some testing to be sure). I have pulled the 1x2 GB set and now I am having a much better time of it.

I'm currently running 3.85GHz through the ringer and it's coming out clean so far. 20 passed clean on LinX, 1000% coverage HCI Memtest (well to be honest 950ish% as I wanted to stop it and start another before heading to work), running Prime95 Large now. I'll update my final settings once I have hit them in case anyone is curious or can get any help from it.

I may be able to squeeze another 0.05GHz out of this, but I doubt I will try. I shoudl also be okay with 4GB of RAM instead of 6 -- I'll miss the extra 2GB at points, but if it has bought me another year I'll be okay with it.

I just want to say thank you to everyone who has helped out here! It really does make a big difference. Basically you have let me wait until Haswell comes along and saved me $450 to $550 dollars right now.

Cheers!

 

[TylerS]

Hi folks,

I just wanted to post an update, mostly incase someone else is searching the forum for answers that this string can help with.

I have the computer running stable and very smooth right now. I’m still not ready to declare it “healthy” but I think it’s very close: LinX – 20 passes no errors; Prime95 Large – 12 hours no errors; CPU temps never exceed 65C. I still need to run HCIMemtest.

Key changes to BIOS settings:
CPU Clock Ratio ____________ [10 X]
CPU Host Frequency (MHz)__________ [385]
PCI Express Frequency (Mhz)_______ [100]
C.I.A. 2__________________________ [Disabled]
System Memory Multiplier (SPD)____ [2.00]
Memory Frequency__________________770
CAS Latency Time________________ 4
Dram RAS# to CAS# Delay_________ 4
Dram RAS# Precharge Delay_______ 4
Precharge Delay (tRAS)__________ 12
DDR2 OverVoltage Control__ [+0.30V]
PCI-E OverVoltage Control_ [Normal]
FSB OverVoltage Control___ [Normal]
(G)MCH OverVoltage Control [Normal]
Loadline Calibration______ [Enable]
CPU Voltage Control_______ [1.32XXXV]

I am now running only 2 sticks (2x2GB) RAM.

Fresh install of Windows 8 (definitely not loving it yet).

I can chalk up a lot of problems mentioned in this string and others to an unstable Windows 7 environment. Something definitely went wonky there and it’s beyond me to fix it. It won’t even operate smoothly, crash-free at stock settings on my hardware. I'm dual booting Win7 and 8 and Win7 is still very f***ed.

I probably WON’T push the clock any higher, I think 3.85 will do what I need with little difference to be had from going to say 3.95.

I WILL put my other ram sticks (2x1 GB) back in at some point to see if it is still stable with them.

 

[TylerS]

After some solid testing and time clocked, I would say I have this thing beat. My final OC at 3.85 GHz seems to be very stable and meets idontcare's "gold standard" for stability.

I am back up to my full 6 GB of RAM.

I tried again at 3.95 GHz without any luck. Too bad, because there is a noticable degredation of Skyrim performance going from 3.95 to 3.85. You wouldn't think it, but it is there.

It's frustrating that a lot of my struggle to get here was due to an unstable OS, which I suspect was broken registry items but may have been a virus/mallware/etc.; however this exercise probably helped me discover that sooner.

Thanks again everyone for all your help!