Overclocking i5 750 - may have a winner

[The Sauce]

Update:
Max stable o/c - 4400 MHz @ 1.425v. 60% o/c.
Final 24/7 o/c - 4200 MHz @ 1.325v.

Prod Code: BX8060515750
MM#: 903877
FPO/Batch#: L934B414
S-spec: SLBLC

Original thread follows:

----------------

Just spanked this thing together, went straight for the gold.

Initial overclocking attempt: 4.1 GHz (21x multi, 195 FSB, vCORE 1.35)
Temps 60C-65C Prime Large FFTs


In the reviews I have seen so far I havent seen people hitting this with less than 1.45 - 1.5 vID, let alone prime stable. I finally may have scored a decent overclocker. That never happens to me. Will update as things progress.



The system:

i5 750
Megahalems HS
ASUS P7P55D-E
Corsair 620HX
Corsair XMS3 1600 2x2gb
HAF 922
Video: waiting for 5830 (using old card for testing)
WD Raptor 150gb

 

[aclim]

damn very nice. It passes 24/7 stable?

 

[MrK6]

Most of the reviews out there were done at launch and I think reviewers then still hadn't fully appreciated the differences between Bloomfield and Lynnfield. For instance, you'll see reviewers shoot 1.35V+ VTT through the chips, where Intel recommends a max of 1.21V VTT. That said, with a slight change in technique and appreciate for the platform, 4GHz+ is easy. I think you have a good chip there, or at least better than mine, a I need 1.375V to get 4.1GHz stable (might be cooling limited for me though). Nice work!

 

[The Sauce]

damn very nice. It passes 24/7 stable?

I got 8 hours, then manually halted the test. Too eager to see the limit to wait 24 hours between tests. I think 8 hours is adequate anyway. Once I hit a wall I'll do some 24h testing.

 

[The Sauce]

MRK6, what is VTT? Are you talking about vCORE/vID? Because it seems most people are safely going up to 1.4 - 1.5 from what I am seeing. I picked 1.35 because I thought it would be a nice, safe starting point.

 

[MrK6]

MRK6, what is VTT? Are you talking about vCORE/vID? Because it seems most people are safely going up to 1.4 - 1.5 from what I am seeing. I picked 1.35 because I thought it would be a nice, safe starting point.

VTT is different from Vcore. There's a couple of different voltages to look at when overclocking the 750.

Vcore - core voltage, Intel specifies a maximum "safe" value of 1.55V, but you'll probably run into cooling issues first.
VID - processor "stock" voltage
VTT/QPI voltage - Intel specifies a "safe" max of 1.21V, stabilizes higher BCLK/RAM speeds
PLL voltage - I believe the "safe" maximum here is 1.98V, it could be lower though.

The only CPU-related voltages I've had to change are Vcore and VTT. Everything else on my system is at stock.

 

[aclim]

im jealous of you guys

 

[The Sauce]

OK, primed for 2 hours @ 2.1GHz/1.35v.

Bumped to 2.2 GHz (21x200 1.35v). Priming now. Temps still low to mid 60's. Looking good. Maybe have a good batch/stepping here? I'll post it once I have a ceiling.

 

[The Sauce]

Up to 4.3 (204 x 21) 1.35v. Priming. Temps now mid to upper 60s.

 

[ScorcherDarkly]

I've got an i5 750 I'm going to overclock once I finish modding my case. I think I've about decided to just do a stock voltage OC as high as I can get it and leave the turbo modes on. Tom's Hardware did a test on the different OCs of a 750 and that came out as one of the more efficient ones.

 

[The Sauce]

That sounds interesting. I may do that as well and compare the two. For now I want to see what my max 4-core o/c is.

I started getting errors at 4.3. Had to bump volts to 1.375. Temps now mid 70s. Added a second CPU fan. Priming...

Wondering...is CPU burn-in still a relevant issue? We used to talk about that all the time but I never see it come up anymore.

 

[imported_RJR]

I don't know if you have changed this or not, but on another thread here you stated your vtt was set to 1.4v. The max as stated by MrK6 is 1.21v. When I first got my MB UD4P the Bios stated max at 1.31v and an updated Bios has it down to 1.21v now.

Also the "safe" vcore is 1.4v and the absolute max is 1.55v (MAY cause degradation above 1.4v) as per Intel. FYI.

 

[The Sauce]

I don't know if you have changed this or not, but on another thread here you stated your vtt was set to 1.4v. The max as stated by MrK6 is 1.21v. When I first got my MB UD4P the Bios stated max at 1.31v and an updated Bios has it down to 1.21v now.

Also the "safe" vcore is 1.4v and the absolute max is 1.55v (MAY cause degradation above 1.4v) as per Intel. FYI.

Yah I looked into that. I saw several places where it was recommended that Vtt should be safe up to 1.4. I think even ASUS specs it as safe at that level.

 

[The Sauce]

I've got an i5 750 I'm going to overclock once I finish modding my case. I think I've about decided to just do a stock voltage OC as high as I can get it and leave the turbo modes on. Tom's Hardware did a test on the different OCs of a 750 and that came out as one of the more efficient ones.

Just read your article. I wasn't really happy with their study. First of all they do not test with varying cores operating to see what the real-world maximum stable Turbo boost speed might be. What they do is find the 4-core max speed and then just assume that that will be the safe maximum for 1-2 cores, and then set their base clock accordingly. Hence, they do not test turbo over 3.2 GHz. That is completely bogus logic. You should be able to go much higher than the 4-core max (in this case 4.2) with just 1-2 cores running. This article was based on assumptions rather than real testing.

The question is, how do you test just 1-2 cores to see what the max is?

 

[MrK6]

Yah I looked into that. I saw several places where it was recommended that Vtt should be safe up to 1.4. I think even ASUS specs it as safe at that level.

1.4V VTT max was for Bloomfields (i7 920, etc.). Like I said, a lot of early reviews assumed the specs were the same, and they aren't. I'd be willing to bet that's why we saw those fried socket 1156 reports early on, but not anymore. People woke up, saw and respected the differences, and problems solved.

 

[The Sauce]

Followed your advice and dropped Vtt back to 1.2. Shouldn't really matter much anyway since I am not overclocking the FSB that much.

Anyway, running into stability problems here. Think I may have hit that point of diminishing returns. Now running 4350 (21x207) @ 1.4v. Interestingly, temps not increasing that much. Still mid 70s. I think I was happier at 4200 @ 1.35v and temps in the mid 60s, but it's all about finding the ceiling, so I will push onward. Don't think that I'll take the Vcore over 1.4v, though, so this may be it.

 

[The Sauce]

Ok. 4350 (21x207) @ 1.4v looks relatively stable. Temps now mid to upper 70s.

 

[MrK6]

Followed your advice and dropped Vtt back to 1.2. Shouldn't really matter much anyway since I am not overclocking the FSB that much.

Exactly. And you know, there's probably some guy (or more) out there who's been running 1.4VTT since he got the chip in September and will run it for another two years with new issues. However, my attitude is Intel designed the chip and they know their stuff, so better safe than sorry and heed their recommendations. I'd hate to see you lose such a good chip.

Anyway, running into stability problems here. Think I may have hit that point of diminishing returns. Now running 4350 (21x207) @ 1.4v. Interestingly, temps not increasing that much. Still mid 70s. I think I was happier at 4200 @ 1.35v and temps in the mid 60s, but it's all about finding the ceiling, so I will push onward. Don't think that I'll take the Vcore over 1.4v, though, so this may be it.

FWIW, if I Prime my chip with Large FFT's, I'll hit 80-82C. I don't think I've cracked 65C yet while gaming, so look at this as your worse case scenario. As long as you're keeping load temps <85-90C, you're fine. Technically these chips are rated up to 100C, I believe.

Ok. 4350 (21x207) @ 1.4v looks relatively stable. Temps now mid to upper 70s.

Now you're getting into the impressive range. Very nice overclock .

 

[The Sauce]

4400 MHz @ 1.4v! Priming... no good.

4400 @ 1.425v - priming...looking good...

...still good. Max temps hitting 81C. I think this is where I stop and let it burn in and hopefully get things pretty stable like that.

I don't think 1.425 is too high to run 24/7, huh? My idle temps are 30s.

Well this has far and away been the easiest, and best overclock I have ever done. And I've been doing it for over 20 years - Intel 80286 was my first. Very sweet. Still very fun. I think the Celeron 300a had about as good an overclock ratio, but nothing close since.

 

[imported_RJR]

And I've been doing it for over 20 years - Intel 80286 was my first.

Geeze, I didn't start until the 386. Glad to see there are some old farts like us still around.

On another forum I frequent, a young man decided to OC his i7 920 with an EVGA board and didn't do much research on Bios settings. He didn't notice the "+" in the space he was entering his vtt voltage and just presumed what he set was the voltage. Needless to say he got it up to 2.1v, and now he's RMAing the board. You may want to keep it under 2.0v, just to be on the safe side, hahaha.