HOWTO: Overclock C2D Quads and C2D Duals - A Guide v1.0

So I am about half-way through my testing of the new Xeon E3110 SLAPM rig that I just put together, and something that I find bizarre is happening.

I am running prime stable at 3.75GHz at stock voltage with bus speed 500MHz and ram at 1000 (5-5-5-18), however, when I increase the multiplier and reduce the bus speed... it craps out at 3.75 (417*9), and does not respond favorably to increasing the voltage. It will still crash orthos in about 10 seconds, no matter the boost in voltage.

I've tried increasing the MCH voltage, but it doesn't seem to affect a change.

So, what settings should I test next? I don't want to needlessly increase the Vcore to push more speed out of this chip if there's another setting I am overlooking. This machine as already kicked around at 4+GHz (prime unstable) and with a little attention could do much better.


some self-constraints:
1) I would like to stay at 500MHz bus speed max due to ram.
2) I don't want Vcore more than 1.30

Other info:

Gigabyte P35-DS3L F8b bios
 
@blink - never seen that before... usually it's the other way around (craps out at higher fsb). I have read about different straps in a board affecting stability. In other words, by pushing the fsb to 500, the board is now in a higher strap and is stable; by dropping back to 417, you might be in bad zone of the lower strap. Maybe others will have some thoughts.
 
Just got a new mobo for my E6420, the abit IP35-E. It was only $60 after a rebate. I just cranked it up to 3.2ghz with a 400mhz fsb and it ran OCCT all night long with no issues. My vcore is set to 1.41V. Max temp was about 66C with my old zalman 7000, only 25C idle. I'm just amazed at how fast it is. I won't go higher until I get a new heatsink or watercool it.
 
Binkt said:
So I am about half-way through my testing of the new Xeon E3110 SLAPM rig that I just put together, and something that I find bizarre is happening.

I am running prime stable at 3.75GHz at stock voltage with bus speed 500MHz and ram at 1000 (5-5-5-18), however, when I increase the multiplier and reduce the bus speed... it craps out at 3.75 (417*9), and does not respond favorably to increasing the voltage. It will still crash orthos in about 10 seconds, no matter the boost in voltage.

I've tried increasing the MCH voltage, but it doesn't seem to affect a change.

So, what settings should I test next? I don't want to needlessly increase the Vcore to push more speed out of this chip if there's another setting I am overlooking. This machine as already kicked around at 4+GHz (prime unstable) and with a little attention could do much better.


some self-constraints:
1) I would like to stay at 500MHz bus speed max due to ram.
2) I don't want Vcore more than 1.30

Other info:

Gigabyte P35-DS3L F8b bios
Click to expand...

I'm in the same boat sort of.. I run a 500 FSB on my SLACR Q6600.. Stable as a table..

35ghztemps.jpg



Not so stable.. fails prime within a few minutes..

512FSB-1.jpg
 
500 MHz @ 1.37 V for a Q6600 is pretty insane.. are you able to run 9x388 @ the same vcore and if so, if it p95 stable?
 
nice bus speed Officermartinez,

Will the system run the same at 8x 438MHz as it does at 7x 500?

did you have to overvolt the bus and/or MCH for that 500MHz?

That appears to be the issue that I am facing. However, I am putting some better cooling on the northbridge this weekend, so we'll see what kind of performance I'll get after that.
 
graysky said:
500 MHz @ 1.37 V for a Q6600 is pretty insane.. are you able to run 9x388 @ the same vcore and if so, if it p95 stable?
Click to expand...

Binkt said:
nice bus speed Officermartinez,

Will the system run the same at 8x 438MHz as it does at 7x 500?

did you have to overvolt the bus and/or MCH for that 500MHz?

That appears to be the issue that I am facing. However, I am putting some better cooling on the northbridge this weekend, so we'll see what kind of performance I'll get after that.
Click to expand...

Good point.. I DID take off the factory thermal interface material (T.I.M.) on the Northbrige and I replaced it with some AS Ceramique. Whether it helped or not, I don't know. I never ran this board with the factory T.I.M. I plan on removing my AS5 on the CPU and trying some MX-2 paste. It would only be a negligible difference I'm sure, but every degree of improvement helps in a Small Form Factor Build / mATX case..

I will replace the T.I.M. on the cpu sometime today and try those other speeds.. I will probably do a couple of hours on each run to ensure decent stability if that's cool. I don't have a lot of time in this upcoming week to do very detailed testing. My setup has already proven itself to me on the 500+ FSB, so I would think anything lower "should" be OK (unless I hit a FSB hole.. LOL). I really think my two front 120mm fans are doing a GREAT job of forcing outside cooler air across my NB heatsink. It probably has a greater effect of keeping that NB cool vs. a little 40mm or 60mm fan.
 
graysky said:
500 MHz @ 1.37 V for a Q6600 is pretty insane.. are you able to run 9x388 @ the same vcore and if so, if it p95 stable?
Click to expand...

These are my IDLE temps at these given speeds. Each speed was ONLY "quick tested" for 15 minutes at full load.. I JUST finished applying AC MX-2 paste about an hour ago.. So, my temps *might* get a few degree's better over the next 3 or 4 days..

3510ghz390FSBIDLETEMPS.jpg


36ghz450FSBIDLETEMPS.jpg
 
officermartinez said:
These are my IDLE temps at these given speeds. Each speed was ONLY "quick tested" for 15 minutes at full load.. I JUST finished applying AC MX-2 paste about an hour ago.. So, my temps *might* get a few degree's better over the next 3 or 4 days.
Click to expand...

Sorry dude, but idle temps don't mean much, you gotta show the load temps after a good 30 min session of p95.
 
graysky said:
Sorry dude, but idle temps don't mean much, you gotta show the load temps after a good 30 min session of p95.
Click to expand...

That wasn't the purpose of my post. "Sorry dude", but I completely understand how to achieve full load temps. I was simply showing how versitile the Q6600 can be on different FSB combinations. I threw Core Temp in there to give a relative idea, while I was at certain incrimental CPU / FSB combinations, what temps I encountered.
 
Just wanted to say thanks for putting this guide together, it's great!
It inspired me to pull the trigger on a Q6600 and an Abit IP35 PRO yesterday.

I'll post my overclocking adventures here in a couple weeks.

cheers!
 
I consistently fail almost immediately in Prime95 (Rounding errors) on my B3 Q6600 @ 3.0GHz (333x9) but everything runs fine, in fact I've never had a more stable system. The vcore is set at 1.3 and vdroop is 1.248 at idle and 1.2 under Prime95 stress. I am running the RAM in a multiplier, perhaps if I try 400x8 at a 1:1 ratio it will be more stable?

Edit: Just tried to do 400x8 and it won't boot. Even increased voltage to CPU and RAM just a bit and still won't boot.
 
@Dark Shade - Try 1:1 @ 9x333 but I'd first recommend you use memtest (described in guide) to verify you're memory is running right. You shouldn't be failing p95 at all :) You might need to push up your motherboard voltages a tad. Are they on auto?
 
I'll memtest tonight. My voltages are set manually as I was told auto settings are bad for overclocking in general. I have +0.1v to the RAM at the moment and 1.3v to the cpu
 
I memtested using Memtest85+ and ran 4-5 passes with no errors.

I have my voltages set like so:
DDR2 - +0.2v
Northbridge - +0.1v
CPU - 1.325v

I'm running 400x8 (3.2GHz) at 1:1 RAM ratio.

I'm beginning to think this motherboard or the cpu is the problem. The only thing is, I care little for synthetic stress testers, as long as my computer runs my programs and games without issue I'm happy.

Edit: No longer stable at 400x8 (crashes 3DMark06 during 2nd test) so I am back to 333x9
 
Well, if you're convinced your vcore for the CPU isn't the problem, I'd recommend upping the motherboard vcores until you can reach p95 stability. Then begin to back off until you're unstable. It's a long process since there are so many variables. Common vcores to bump up assuming your RAM is running @ specs are the NB vcore and the IHC vcore. FSB termination is another.

9x333 for a q6600 isn't too bad... that's a 25 % boost overstock. 8x400 would be 33 % but your MB doesn't officially support a 400 MHz FSB (the P35 that is)... you might not be able to run 8x400 stable at all... not all hardware [CPUs and MBs] are created equally. I would look closer at your board vcores first though.
 
I kind of don't want to increase the voltage much more as I am already running 45C at idle

Maybe try another motherboard. I do have a revision 1, so maybe the 2.2 version or a completely different motherboard would work. Not that I would do that, I'm not that desperate
 
Yeah, but 45 C for your NB isn't that high for a P35 actually. 55 or 65 and I'd start to worry. Have you tried re-seating the NB HS? Might benefit from removal/cleaning and maybe a lap job.
 
Hey guys I need your overclocking expertise here. I am not sure what the best options for my setup might be and i'm having a hard time figuring it out. Give me some of your suggestions if you could as to what might be the best options.
These are my current settings. A few days ago, I had the memory clock at 533 and RAM at 1066 which my crucials are rated, but not sure if using the 2.5 divider-which I am using now...is better than using the auto SPD setting...seems faster...I hope as stable...have yet to do a prime or orthos test at this point. I have never been able to see the 1:1 ratio in Cpu-z, so if you guys could give me a tip on how to achieve that also, that would be great.

hardforum.jpg
 
Hey graysky, terrific guide . I have one question : the PCI clock synchronization Mode is not an option anywhere on the abit ip35 pro bios, and I was wondering if not having it will fry the other pci devices you mentioned. My goal was a mild oc of 3.0. I did get it last night, 333x9 and set my ddr2 800 for 1:1 (667), but not knowing what else to look for makes me feel as if I should go back to the stock specs.
 
first off I'd like to say that this is an excellent guide, nice work graysky

Now on to my request for some assistance. For reference my setup is:

asus p5n32-e SLI
Q6600 G0 - VID 1.25
2x1GB Geil (model GX22GB6400UDCA=)
XFX 8800GTS
Corsair 620HX PSU
2 Silverstone FM121-B fans (1 intake, 1 exhaust)
Koolance Exos water cooler (on the CPU, it's an old cooler but it still works well for just the CPU)
40mm fan installed on NB
2 striker PWM cooling fans on the PWM heatsinks
4x GPU ramsinks on the SB

I had been running this system with an e6420(overclocked to 3.0ghz) up until the $189 Q6600 deals started to show up last month and I finally bit and snagged one. Over the last 2 weeks I've been working on overclocking it and what I'm at so far is:

FSB: 1400
CPU: 3150 (350x9)
RAM: for testing I left it at 1:1 so its currently 700mhz

Volts:
CPU: 1.35v (1.289 full load after vdroop)
RAM: 2.1v
NB: 1.4v
SB: 1.5v
VTT: 1.25v

This is what I consider to be stable. I've run 10hour 4 thread prime95 small FFT and 10 hour 4 thread prime 95 blended. CPU temps at the 10 hour mark were 57c,57c,52c,53c. (which I was happy with). The issue is, I figure the CPU can go higher since my volts really aren't that high yet(3.6ghz being the target), but no matter what I do, I can't get the box to post at an FSB of 1600. For example, I set my multi to 7 (400x7 would leave the CPU at 2.8ghz and remove it from the overclock equation since it runs happily at 3.15ghz), and I set the ram to unlinked at 667mhz (to remove it from the equation). I bumped the NB volts up to 1.5 (which I imagine is the main voltage to worry about when raising the FSB) and still no post. Raised it to 1.55 and still no post. I've read many places that the p5n32-e sli isn't a great quad overclocker, and perhaps that's what's holding me back now (which I can accept if it is indeed the issue) but before I admit defeat(a semi defeat, as 3150 is still decent) I thought i'd solicit some input/suggestions as to what else I may try to get it to post at the 1600mhz FSB. Anyone have any suggestions?
 
You might need additional voltage to the MB components such as NB, IHC or FSB term. I'm too lazy to google you memory... are they DDR2-800?
 
heh, always something that I forget

yeah, the memory is ddr2-800

on my e6420 I ran it at 860mhz 4-4-4-12 with no issue
 
Well, I updated my bios from the 1203 "recommended" to the latest 1404 last night and can now hit the 1600fsb (3.6ghz). I was able to get it up to 5 hours small FFT before it died but my volts are getting high (1.45 I think) and my temps were getting very warm (67c,67c,63c,63c) so I think I'm going to settle in around 3.4ghz (unless I can sort out why core 0 and 1 are 4-5c higher than core 2 and 3 (but that's a different issue)

It should be a fun weekend of tuning though!
 
snapper said:
heh, always something that I forget

yeah, the memory is ddr2-800

on my e6420 I ran it at 860mhz 4-4-4-12 with no issue
Click to expand...

Awesome.. I have (4) of my computers in my house running the exact same settings on each of my E6420's! The E6420 is often, the "forgot about" processor.. But it hauls, none-the-less!

snapper said:
Well, I updated my bios from the 1203 "recommended" to the latest 1404 last night and can now hit the 1600fsb (3.6ghz). I was able to get it up to 5 hours small FFT before it died but my volts are getting high (1.45 I think) and my temps were getting very warm (67c,67c,63c,63c) so I think I'm going to settle in around 3.4ghz (unless I can sort out why core 0 and 1 are 4-5c higher than core 2 and 3 (but that's a different issue)

It should be a fun weekend of tuning though!
Click to expand...

I have two (about to be 3) quadcore setups. Up to about a month ago, my QX6700 was always 4c - 5c diifference (on a phase change setup). My current Q6600 in my HTPC setup is also different.. Under full load, their is a 5c difference.

30ghz500FSBFULLLOADTEMPS.jpg
 
officermartinez said:
Awesome.. I have (4) of my computers in my house running the exact same settings on each of my E6420's! The E6420 is often, the "forgot about" processor.. But it hauls, none-the-less!



I have two (about to be 3) quadcore setups. Up to about a month ago, my QX6700 was always 4c - 5c diifference (on a phase change setup). My current Q6600 in my HTPC setup is also different.. Under full load, their is a 5c difference.
Click to expand...

Ahh, given what you've told me I'm not going to worry about trying to get those 2 cores cooler. Seems that must be normal.

And yeah, those 6420 were so cheap when they came out and really performed quite well. I have it sitting here staring at me, so I may need to pick up a new motherboard for it :D
 
Hi, just finished my build and I want to try overclocking my Q6600 to 3.2ghz. I'm using G.Skill 2x2GB DDR2 1000 RAM.

If I want to leave my multiplier at x9 for the CPU I would have to up my FSB to 355mhz. My question is how can I set the memory frequency to 355mhz in order to run 1:1? Your guide said only the following options were available:

1. FSB : DRAM
2. 1:1 = 333 MHz : 667 MHz
3. 4:5 = 333 MHz : 833 MHz
4. 2:3 = 333 MHz : 1,000 MHz
5. 3:5 = 333 MHz : 1,111 MHz
6. 1:2 = 333 MHz : 1,333 MHz

I think I'm being a noob, but is there a way to more precisely change the memory frequency so I can get 1:1? Thanks.
 
gatomakato said:
Hi, just finished my build and I want to try overclocking my Q6600 to 3.2ghz. I'm using G.Skill 2x2GB DDR2 1000 RAM.

If I want to leave my multiplier at x9 for the CPU I would have to up my FSB to 355mhz. My question is how can I set the memory frequency to 355mhz in order to run 1:1? Your guide said only the following options were available:

1. FSB : DRAM
2. 1:1 = 333 MHz : 667 MHz
3. 4:5 = 333 MHz : 833 MHz
4. 2:3 = 333 MHz : 1,000 MHz
5. 3:5 = 333 MHz : 1,111 MHz
6. 1:2 = 333 MHz : 1,333 MHz

I think I'm being a noob, but is there a way to more precisely change the memory frequency so I can get 1:1? Thanks.
Click to expand...

Well, you'll have to change those 333's to 355's

1:1 = 355 : 710
4:5 = 355 : 888
2:3 = 355 : 1,065
5:8 = 355 : 1,136

You can see that the 2:3 divider would be pushing the limits of your memory... try it and see if you can stabilize it. If I were you, I might try 4:5 or 1:1. Sorry, I just noticed your past was way back on the 25th of April... since some time has passed, what have you done?
 
OK, just got finished upgrading the guide to version 1.6!

• Added a detailed section to help you find the minimum stable CPU and MB vcore settings. Check it out (near the bottom of the guide entitled, “Stress Testing and Minimizing Your Vcores”)!
• Updated the CPU table
• Provided a less than $5 method you can use to shave off some NB load temps (in the thermal management section at the end of the guide).
 
haha nice. I was actually gonna do add a fan to my P35-T2RS NB just like that
 
zenius said:
haha nice. I was actually gonna do add a fan to my P35-T2RS NB just like that
Click to expand...

Dude, it absolutely rocks. I just did about 9 hours of x264 encodes and that NB temp never went above 44 °C. It's idle now with the room temp being 21.5 °C (71 °F) and the inside of the case being 28.6 °C (83.5 °F) and the NB reads 41 °C. The temps I typed over the pic in the guide were running p95.

Post a pic of your job and report the results when it's finished.
 
Ok heres my problem,

FATAL ERROR: Rounding was 1.407374884e+014, expected less than 0.4
Hardware failure detected, consult stress.txt file.

I run prime between 1 and 5 minutes I get fatal error
Running Large FFT

Setup is:
Cosmos-s replaced all fans with 1600 scythe's, q6600 w/ zalman 9700, corsair tx750w,
ocz reaper hpc 2gbx2, 8800 gts 512, maximus formula, g-5 mouse

cpu voltage is 1.216
9x 333--> 3Ghz
ram voltage is 2.20 volts
5-5-5-15
temp on the failed core is 36C
all other cores read 36-44C

my hunch is the voltage but i was hoping to get some help from you guys,
any tips would be arrpreciated,

-Rob
 
@rob - did you read the guide recently? I updated it with a section for just this problem. You might need to up another vcore on the board - instabilities aren't always CPU vcore related.
 
Version 1.6.1 is up. Totally re-wrote the section on memory which now includes a discussion on both DDR2/DDR3 and formulas you can use to calculate max supported FSB of a given module based on it's DDRX-Y and PCX-Y designations. Also re-ordered the first part of the guide.
 
what about the HT link? does adjusting them down from 5x make any difference in stability? I have a 389x9 fsb for 1556fsb
 
graysky.
I bought the DFI LT P35 T2R board for my second Q6600 sys along with 4 X 2 GB sticks of A-DATA 800/6400 ram and a PCP&C 750 quad psu. I was hoping to surpass the easily achieved Abit IP35-Pro Q6600 I've had running at 3.3 Ghz but the DFI board fuzzed out at 3.3 on my second phase of OC'ing.
BTW...the heatsink is a measly Asus Triton 70 due to the jumongous northbridge heatsink on the DFI board.
Any suggestions on places to start getting it up?
TIA!
 
Anybody willing to help me with some suggestions on incremental BIOS adjustments.
I'm really lost with all the VDD, VTT and GTL ratio stuff I've read and reread in the AT article on OC'ing the board but the fact that I have different ram and am totally lost trying to understand what he's saying...can someone translate in simple terms how to get my components up to about 3.3, 3.4 or more...please?
 
You must log in or register to reply here.
Back
Top