E6600 Overclocking Thread

[Some1ne]

For anyone interested, here is documentation of my experiences overclocking the E6600 chip. Full system specs:

Asus P5B Wifi/Deluxe mainboard, BIOS 507, included chipset fan installed
Intel Core 2 Duo E6600 Retail (B2 revision, stepping 6)
2 GB (2x 1 GB) G.Skill DDR2-800 CL4 RAM
Enermax Liberty 500W PSU
ATI X1900XT 512 MB graphics card
2x WD Raptor 150 GB's in RAID-0
4x Seagate 7200.10 320 GB's in RAID-5
NEC 3550a burner
Creative X-Fi Platinum
Zalman CPS9500 AT w/ Arctic Silver 5
4x 120 mm case fan
Antec P180b case

...all fans with variable speeds are set to high (with the P180 the fan noise is still barely audible, it's a nice low-pitched hum), speedstep/C1E is disabled, memory ratio is set to 1:1.

At stock, with EIST enabled, the idle temp is 38 degrees, and the full load (both cores) temp goes up to about 46.5 degrees.

As of right now, the chip is running at stock vcore @ 3.0 GHz. 3.2 GHz is next. RAM voltage is 2.15V, though there was nothing that forced me to make that setting, I just did it pre-emptively since right now I just want to be stressing the CPU and not the RAM...that comes later.

Anyways, time to see if it will do 3.2 GHz at stock vcore.

Edit: one other thing, at stock, the E6600 was about 50% faster than my previous X2 3800+ @ 2.4 GHz...meaning that in superpi, for example, if the X2 took ~30 seconds to complete a run, the E6600 at stock settings would do it in ~20. I'm hoping that overclocking will get me to a 100% increase (i.e. ~15 seconds where the other one took ~30).

 

[Some1ne]

Booted successfully at 3.2 GHz w/ stock voltage, however superpi gave an error. Increasing voltage to 1.4125V.

 

[Some1ne]

Not quite sure what's up, but according to the Asus utility as well as CPU-Z, my vcore settings are not actually being applied. This is backed up by the fact that no matter how much I raise the voltage, neither temperatures nor stability change. Will have to investigate.

 

[Some1ne]

Okay, after much frustration with trying a ton of different vcores without any change in stability, I have worked out that the instability was actually being caused by the RAM. I feel like I should probably be upset, because the RAM is rated for DDR2-800 @ CL4, and here I am running it at a 1:1 ratio at DDR2-712, and I have to give it 2.30V in order to keep it stable at CL4. The rated voltage is 2.0V-2.1V. I feel kind of disappointed given the glowing reviews I saw of these modules when I bought them.

On the plus side, my E6600 is now doing 3.2 GHz stable on stock voltage. Hooray!

 

[Some1ne]

Some additional testing implies that I was being limited by the mainboard. I fixed my CPU multiplier at 6.0, picked the slowest available RAM settings, and bumped the FSB slightly to 360 MHz, and was able to produce fairly consistent errors, that would persist despite the slow CPU and RAM settings, and regardless of what voltage combinations I gave them. It turns out the magic incantation here is "FSB Termination Voltage". By raising this, the error went away permanently, so potentially my RAM is redeemed, and the mainboard is called into question a bit. I'm not entirely sure what this setting affects, but it is now at 1.40V. Wonder if that's safe.

At least now it seems I can resume the overclocking.

Edit: well, overclocking can resume tomorrow...for tonight I'm just going to leave it at a stable 3.0 GHz w/ stock vcore. My RAM is vindicated, as it is now running DDR2-1000 @CL4-4-4-15 on 2.4V (I could probably lower the voltage and/or tighten the timings, the current settings were lifted from an account of someone else overclocking the same RAM). Seems something with the board is the limiting factor (and what's interesting is that my cutoff point of ~360 MHz is very close to the 372 MHz cutoff point encountered in the anandtech review of this board), as the CPU and the RAM certainly seem to have no problem scaling gracefully. Perhaps it's the 507 BIOS...I'll have to check tomorrow.

 

[FallenHero]

I've been having some of the same problems you have in overclocking at the moment...perhaps I will install the chipset cooler and try again. Please, keep updating. I have found it very useful so far.

 

[xoduzz786]

yes please keep us up to date! I am looking to build a conroe rig within the next 2-3 weeks and most likely i will have the same exact specs as you do

edit: i have heard the P965 based motherboards will only go to around ~400 for fsb with the 4mb core2 duos (e6600, e6700, x6800) and past 450+ with the lower models (e6300/e6400). Maybe thats why youve hit a wall at ~370 fsb?

 

[RichUK]

Are you using the G Skill PC2 6400 HZ sticks?

Also what BIOS are you running, try flashing to 502 or 507 and test again if you haven?t already.

Upping the FSB termination voltage, increases the voltage down the FSB tracers so to stabilise the increased FSB frequency between the processor and MCH. This also results in electrical noise which will destabilise the FSB if the voltage is set too high. AMD's HTT use's lower voltage and is a serial interconnect. This is why HTT is far more superior to the FSB model.

 

[Some1ne]

Are you using the G Skill PC2 6400 HZ sticks?

Yes, those are the ones...specifically, these:

http://www.newegg.com/Product/Product.asp?Item=N82E16820231065

Also what BIOS are you running, try flashing to 502 or 507 and test again if you haven?t already.

Started with 507, downgraded to 405 last night after I visited the Ausu website and all of a sudden the 507 link was gone and 405 was the only one left.

Upping the FSB termination voltage, increases the voltage down the FSB tracers so to stabilise the increased FSB frequency between the processor and MCH. This also results in electrical noise which will destabilise the FSB if the voltage is set too high.

Thanks, that's helpful...maybe I should try the 1.30 V setting? Also, I know on AMD systems they say to disable Spread Spectrum when overclocking, does that also apply to Intel systems as well?

yes please keep us up to date! I am looking to build a conroe rig within the next 2-3 weeks and most likely i will have the same exact specs as you do

Okay, here's the morning update...I left the system running at 3.0 GHz, stock vcore, DDR2-1000 overnight with dual instances of prime95 going. No errors.

Prior to turning in last night, I visited the Asus website and found that the link to the 507 BIOS had been removed. Thinking that this meant that there could be a problem with 507, I tried a downgrade to 405 (the only version that remained on their website). It did not improve my stability, and removed my ability to manually lower the CPU multiplier in the BIOS. I kept an archival copy of 507 around though, and plan to put it back on tonight.

I tested my RAM settings a bit. 2.4V is indeed necessary for CL4-4-4-15 @ DDR2-1000+. 2.3V was unstable, and trying either 2.3V or 2.4V @ CL4-4-4-12 was very unstable.

Also, ran 3dMark 05, got a bit over 12,000 @ 3.0 GHz, with stock settings on the video card. The demos don't "load" anymore, they just start.

So now that I know what my RAM will do, after work I want to put BIOS 507 back on, and use that to set my CPU multiplier to 6.0. I will then set my FSB to 400 MHz, and my RAM divider at 1:1. This will give me 2.4 GHz on the CPU, and DDR2-800 on the RAM. The only overclocked component will be the FSB. I will then see if the system is unstable despite the fact that only the FSB is being overclocked, and if there's anything I can do to make it stable at that speed. If I succeed, I'll take a straight shot at running the CPU at 3.6 GHz, and see what happens.

Edit: and also, a couple of articles I found regarding overclocking on the P5B have suggested that there is some bug in the BIOS that is limiting the maximum FSB speed thta the board will run at. Hope Asus fixes this soon.

 

[Some1ne]

Something to note while my workspace rebuilds...I ran HDTach on my RAID arrays yesterday, and both of them appear as though they are being limited to SATA 1.5 Gbps speed. Both arrays show a sustained transfer rate of ~181 MB/sec (for the raptor array, the graph indicates this sustained rate across the first 2/3 rds of the volume...the line is virtually flat, while for the RAID-5, the rate is sustained across a good 85%+ of the volume), which translates to 1.45 Gbps, remarkably close to the 1.50 Gbps allowed under the SATA-1 spec.

Is there something special I need to do to enable SATA-II/3.0 Gbps transfer rates? It seems like I've finally built something that might be able to benefit from the extra bandwidth.

The raptor array shows a seek time of ~8.5 ms, the RAID-5 shows about 13.5 ms.

 

[Some1ne]

Some things to report...I'm not entirely sure what thing it is that did it, but since coming home I have:

1. Flashed the 405 BIOS back to 507.

2. Re-enabled C1E, disabled the TM2 option, left EIST disabled.

3. Lowered by CPU multiplier from 9 to 8.

4. Set my RAM divider back to 1:1.

5. Decided to avoid FSB settings between 350 MHz and 400 MHz like the plague (everyone seems to hit problems here with this board).

6. Set all voltages to "Auto" (*including vcore*), except for my RAM, which remains ar 2.4V (the system appears to be automatically scaling up the CPU voltage for me...for example, at 3.6 GHz the monitor will report a vcore of 1.55...I might be concerned, but temps stay withing acceptable levels)

...so because of 1 or more of these things, the system now acheives a much better overclock. I have found that:

at 3.5 GHz (437x8) the system is prime95 stable.
at 3.6 GHz (450x8) the system is stable for any level of Superpi, but will not pass prime95.
at 3.7 GHz (463x8) the system will boot to desktop, but crash as soon as you try to do anything.
at 3.8 GHz (475x8) the system will crash when booting windows.
at 4.0 GHz (500x8) the system will not POST.
at 3.0 GHz (500x6) the system will not POST (indicative of instability coming from mainboard and not CPU).

When running prime95, the temps hover around ~60 degrees. I'm not overly concerned, because during actual usage I will almost never be putting this much load onto the CPU, and I definitely won't ever be doing it for me than 30 seconds at a stretch, at most (until truly CPU demanding multithreaded games come out, then maybe).

At the moment, I'm not sure if the increasing instability at speeds > 3.5 GHz is being caused by the mainboard or the CPU, because I am dangerously close to the 500 MHz cutoff point above which the board will not POST. I could test this somewhat by dropping the CPU multiplier down to 6 or 7, and I probably will at some point. For now though, I'm content with the 3.5 GHz overclock (it's close enough to my realistically hoped-for overclock of 3.6 GHz, if not quite anywhere near my super-optimistic overclock of 4.0 GHz), and will probably make an attempt to kick my RAM up to the next speed grade, and then leave it at that for awhile, and install programs, and maybe get a game on the new PC and see how it handles.

In case it makes a difference, the order in which I flashed my BIOS was:

302 -> 507 -> 405 -> 507

E6600 @ 3.5 GHz (for now), prime stable. Good enough.

 

[n7]

You gotta increase chipset voltage if you want to be able to run over ~450 most likely.

vMCH i believe it's called, though i'm not sure if that's the correct term in Asus's bios.

 

[Some1ne]

It actually has a bunch of chipset-sounding voltage settings:

ICH chipset voltage
NB Vcore
SB Vcore

...I just tried bumping all of them (plus the FSB voltage), and setting a CPU multiplier of 6.0, and the system again failed to POST at 500 MHz FSB.

Also, RAM is not stable at DDR2-1100, CL5-5-5-15 (it's close, though). Going back to DDR2-874 @ CL4-4-4-15

 

[lopri]

What are you using to monitor temps? I've become to trust CoreTemp more than anything else.

 

[Some1ne]

I've been using the included Asus monitoring software, "PC Probe II" or something like that.

Used to use SpeedFan, but it does not seem to work with the P5B.

 

[secretanchitman]

wow...almost identical to my comp...congrats on your OC!

 

[dasmokedog]

Much cheaper than the 6600:

Core 2 Duo E6300 1.86GHz 2Meg L2 Cache
------------------------------------------------------
CPU-Z:
3499.98MHz http://valid.x86-secret.com/show_oc?id=112855

SuperPI:
1M: 00m 16.593s Checksum: 5EC49D70
http://www.rigshowcase.com/img/3463TEjR/6133.jpg
8M: 03m 10.812s Checksum: A1A08EA2
http://www.rigshowcase.com/img/3463TEjR/6134.jpg

Sandra:
Dhrystone: 32268 MIPS -->AMD FX2-62(20225 MIPS)
Whetstone: 22206 MFLOPS -->AMD FX2-62(17084 MFLOPS)
http://www.rigshowcase.com/img/3463TEjR/6129.jpg

Integer x8 iSSE4: 193522 it/s -->AMD FX2-62(52653 it/s)
Floating-Point x4 iSSE2: 104339 it/s -->AMD FX2-62(56980 it/s)
http://www.rigshowcase.com/img/3463TEjR/6130.jpg

RAM Bandwidth Int Buff'd iSSE2: 7982 MB/s
RAM Bandwidth Float Buff'd iSSE2: 7961 MB/s
http://www.rigshowcase.com/img/3463TEjR/6132.jpg

Check out the Cache test!!
http://www.rigshowcase.com/img/3463TEjR/6131.jpg

BIOS Settings:
http://www.rigshowcase.com/img/3463TEjR/6135.jpg
http://www.rigshowcase.com/img/3463TEjR/6136.jpg
http://www.rigshowcase.com/img/3463TEjR/6137.jpg

Gutz:
http://www.rigshowcase.com/img/3463TEjR/6138.jpg
http://www.rigshowcase.com/img/3463TEjR/6139.jpg
http://www.rigshowcase.com/img/3463TEjR/6140.jpg
http://www.rigshowcase.com/img/3463TEjR/6141.jpg
http://www.rigshowcase.com/img/3463TEjR/6142.jpg
http://www.rigshowcase.com/img/3463TEjR/6143.jpg

 

[Pabster]

Originally posted by: dasmokedog
Much cheaper than the 6600:

Did you even bother to read the thread title?

 

[dasmokedog]

Originally posted by: Pabster

Originally posted by: dasmokedog
Much cheaper than the 6600:

Did you even bother to read the thread title?

No time to read, sorry.

 

[Some1ne]

Much cheaper than the 6600:

Core 2 Duo E6300 1.86GHz 2Meg L2 Cache

True, though that's really just sort of a judgement call regarding whether or not you think the extra cache is worth it. Obviously when you look at the prices of the chips alone the extra cache doesn't justify close to double the price, but in the face of a $2K+ build, it's basically adding less to the cost of the build than the sales-tax was that I had to pay, and in that context, I think it's worth it. For example, my superpi 1M time is 14.750, a little more than 10% faster than yours, despite the fact that my RAM is currently clocked a bit slower. For close to twice as much money on the CPU alone, that's probably not worth it, but for a ~7% surcharge on the entire system in order to get better performance by a factor of 10% to 15% under certain circumstances (especially if any of those circumstances involve gaming...which remains to be seen), then it can be justifiable.

But anyways, that's beside the point...the point was, good news, and very bad news.

The good: Played Q4 at the max resolution my monitor will allow, Ultra quality, all settings maxed, and it's smooth as butter...very nice.

The bad: The system has been hit by a W32.Licum infection, which is kind of baffling because as far as I know I set everything up on it just like the one before it, and the previous system had no problems with W32.Licum (or any other virus, for that matter). It's pretty rampant, and even if I clean out all the infected files, if I let the system idle for a bit, the infection returns...overnight it spread to basically every .exe in /windows/system32, so I think that means my install is pretty much screwed at this point (and right after I got everything configured the way I like it). Sucks. Does anyone know if there's a removal tool for this thing (or why my new PC keeps getting re-infected, while the one sitting next to it with the same OS and programs installed is fine)?

 

[secretanchitman]

Originally posted by: Some1ne

Much cheaper than the 6600:

Core 2 Duo E6300 1.86GHz 2Meg L2 Cache

True, though that's really just sort of a judgement call regarding whether or not you think the extra cache is worth it. Obviously when you look at the prices of the chips alone the extra cache doesn't justify close to double the price, but in the face of a $2K+ build, it's basically adding less to the cost of the build than the sales-tax was that I had to pay, and in that context, I think it's worth it. For example, my superpi 1M time is 14.750, a little more than 10% faster than yours, despite the fact that my RAM is currently clocked a bit slower. For close to twice as much money on the CPU alone, that's probably not worth it, but for a ~7% surcharge on the entire system in order to get better performance by a factor of 10% to 15% under certain circumstances (especially if any of those circumstances involve gaming...which remains to be seen), then it can be justifiable.

But anyways, that's beside the point...the point was, good news, and very bad news.

The good: Played Q4 at the max resolution my monitor will allow, Ultra quality, all settings maxed, and it's smooth as butter...very nice.

The bad: The system has been hit by a W32.Licum infection, which is kind of baffling because as far as I know I set everything up on it just like the one before it, and the previous system had no problems with W32.Licum (or any other virus, for that matter). It's pretty rampant, and even if I clean out all the infected files, if I let the system idle for a bit, the infection returns...overnight it spread to basically every .exe in /windows/system32, so I think that means my install is pretty much screwed at this point (and right after I got everything configured the way I like it). Sucks. Does anyone know if there's a removal tool for this thing (or why my new PC keeps getting re-infected, while the one sitting next to it with the same OS and programs installed is fine)?

restart in safe mode, run a virus scan, then it should be gone.

 

[dasmokedog]

Originally posted by: Some1ne

Much cheaper than the 6600:

Core 2 Duo E6300 1.86GHz 2Meg L2 Cache

True, though that's really just sort of a judgement call regarding whether or not you think the extra cache is worth it. Obviously when you look at the prices of the chips alone the extra cache doesn't justify close to double the price, but in the face of a $2K+ build, it's basically adding less to the cost of the build than the sales-tax was that I had to pay, and in that context, I think it's worth it. For example, my superpi 1M time is 14.750, a little more than 10% faster than yours, despite the fact that my RAM is currently clocked a bit slower. For close to twice as much money on the CPU alone, that's probably not worth it, but for a ~7% surcharge on the entire system in order to get better performance by a factor of 10% to 15% under certain circumstances (especially if any of those circumstances involve gaming...which remains to be seen), then it can be justifiable.

But anyways, that's beside the point...the point was, good news, and very bad news.

The good: Played Q4 at the max resolution my monitor will allow, Ultra quality, all settings maxed, and it's smooth as butter...very nice.

The bad: The system has been hit by a W32.Licum infection, which is kind of baffling because as far as I know I set everything up on it just like the one before it, and the previous system had no problems with W32.Licum (or any other virus, for that matter). It's pretty rampant, and even if I clean out all the infected files, if I let the system idle for a bit, the infection returns...overnight it spread to basically every .exe in /windows/system32, so I think that means my install is pretty much screwed at this point (and right after I got everything configured the way I like it). Sucks. Does anyone know if there's a removal tool for this thing (or why my new PC keeps getting re-infected, while the one sitting next to it with the same OS and programs installed is fine)?

Oh man that sucks. I am sorry to hear of your misfortune. Does McAfee or Symantec have a tool? Good luck with the removal. I do hope you can remove it vs a complete reinstall.

 

[EdgeTech]

Make sure and disable System Restore before doing the virus scan and thanks for the updates. They are very useful.

 

[Some1ne]

Well, telling the virus scanner to quarantine parts of my OS pretty much ruined the install, and a rescue install didn't fix it, so I just reformatted the partition and started over. This time first thing I did was install antivirus software (and this time I left the "filesystem realtime protection" enabled...I usually turn that off). No viruses on the new install yet, so hopefully that will stay resolved this time. Also installed CoreTemp, and I like it because under load it consistently reports temps between 3 and 5 degrees cooler than the Asus utility. Wonder which one is more accurate though.

Also, saw a post on a different forum that said that if I set a PCI-E clock of 110 MHz, and a FSB voltage of 1.4V or 1.45V it should improve my FSB stability a bit (a lot of people are reporting that the board will stop POST'ing between 450 and 500 MHz, regardless of CPU speed, which is consistent with what I've found with mine). I'll try that once I'm done patching up my system again.

 

[Gar818]

Hey, i got this board, so far im in love with it, also wich cpu options do people leave enabled, and wich disabled, because i got an increase in some benchmarks with some of them enabled, for example, CPU Spread spectrum, do u guys enable this or disable it?

c1e support
max cpuid
vanderpool
cputm
peci

and also what is EIST i see people disabling this feature.....i dont see a option for this in my bios. thanks.