Ivy Bridge Lowered Expectations Club

[moonbogg]

@ Arkaign...damn that sucks. Stock chip unstable = borkfish

Also, I bet your 2500k performs the same as my 3930K in BF3. I got more threads and you got more clock speed. I bet it just about evens out. The moment I switched my 2600K for the 3930K was the precise moment that i went full retard on this gaming rig.

regarding voltage....something wrong with the OP's chip for sure. This 3930k does 4.3 stable at 1.28. I turned off all of those stupid high power, high current, high high high garbage settings in BIOS. Everything is on regular, low power settings and the chip is solid at 4.3. I couldn't get stable at 4.5 no matter what I did. All depends on your luck of the draw.

 

[WhoBeDaPlaya]

http://www.overclock.net/t/1291703/ivy-bridge-overclocking-guide-asus-motherboards
is a fairly simple guide to overclock ib and will get you in the ball park of 4.6
-and for those using a $20.00 cooler on a $1500.00 build what can I say.

Actually, I tried a $17.02 cooler - a Zerotherm ZT-10D on sale at MC. Since this thing beats the Silver Arrow and Venomous X, I'd hardly call it a slouch
The TIM I used was IC Diamond, again pretty much the best.

 

[Arkaign]

@ Arkaign...damn that sucks. Stock chip unstable = borkfish

Also, I bet your 2500k performs the same as my 3930K in BF3. I got more threads and you got more clock speed. I bet it just about evens out. The moment I switched my 2600K for the 3930K was the precise moment that i went full retard on this gaming rig.

regarding voltage....something wrong with the OP's chip for sure. This 3930k does 4.3 stable at 1.28. I turned off all of those stupid high power, high current, high high high garbage settings in BIOS. Everything is on regular, low power settings and the chip is solid at 4.3. I couldn't get stable at 4.5 no matter what I did. All depends on your luck of the draw.

Hahaha, you could always de-lid and go for 5Ghz

 

[WhoBeDaPlaya]

Hahaha, you could always de-lid and go for 5Ghz

I think I'd rather stick with a (possibly) 5GHz SB rather than a 5.2GHz equivalent IB.

 

[masterxfob]

stable at 4.4G @ 1.26V @ 90C. 4.5G gets way too hot and BSODs. would like to deLid, but i'm a p*ssy.

 

[Meghan54]

The TIM I used was IC Diamond, again pretty much the best.

As long as you don't include Indigo Extreme. Don't get me wrong, IC Diamond is quite good, but nothing earth shattering. Have seen testing where ICD is bested by Prolimatech's PK-1 TIM and Phobya HeGrease, among others. Of course, this was with a minimum of 5 mounts of the cpu cooler, something most "testing" sites don't do.


As for temps on IVB cpus, this is what I got on a 4.4GHz OC on a 3570k, with TX-4 and a Corsair H100 cooler w/Scythe GTs, while running IBT on maximum setting, on its sixth run out of 10---took the capture here because nothing had changed over the last 4 runs.....(I was streaming ESPN at the time I tested and had 12 browser windows open with around 100 tabs open in them.)

 

[Termie]

As long as you don't include Indigo Extreme. Don't get me wrong, IC Diamond is quite good, but nothing earth shattering. Have seen testing where ICD is bested by Prolimatech's PK-1 TIM and Phobya HeGrease, among others. Of course, this was with a minimum of 5 mounts of the cpu cooler, something most "testing" sites don't do.


As for temps on IVB cpus, this is what I got on a 4.4GHz OC on a 3570k, with TX-4 and a Corsair H100 cooler w/Scythe GTs, while running IBT on maximum setting, on its sixth run out of 10---took the capture here because nothing had changed over the last 4 runs.....(I was streaming ESPN at the time I tested and had 12 browser windows open with around 100 tabs open in them.)

The key to your result is voltage. It looks like you're at stock. While everything else is nice (H100, TX-4, etc.), it's not what kept you in the low 70s.

You are one of the lucky ones. That's the thing about IVB - it's a total crapshoot.

 

[WhoBeDaPlaya]

On another note, WTF is with the anemic VRMs on some boards? Granted, I have never owned anything S1156, but even the budget EX58-UD3R doesn't flinch when driving an OCed i7 920 (130W TDP stock, God knows how much when OCed).

In contrast, the Asrock Z77 Pro4 goes crying to mama when driving a 2500K @ 4.6GHz (~120W TDP according to CoreTemp) and throttles the multiplier down to 33x periodically.

As long as you don't include Indigo Extreme. Don't get me wrong, IC Diamond is quite good, but nothing earth shattering. Have seen testing where ICD is bested by Prolimatech's PK-1 TIM and Phobya HeGrease, among others. Of course, this was with a minimum of 5 mounts of the cpu cooler, something most "testing" sites don't do.

My bad, I should have phrased that better. Meant to say that ICD is pretty darned good

 

[bl00tdi]

Depends on the chip. My first 3570k refused to be stable at 4.6 regardless of voltage and required 1.36v to be stable @ 4.5. My current unit is rock solid @ 4.8/1.35v. Luck of the draw.

 

[piesquared]

Those 22nm process characteristics sure seem inconsistant.

 

[Ferzerp]

Mine starts needing more voltage than I want to use at 4.5Ghz, so I leave it 4.4Ghz.

I have one of the release ones, w/ TIM replaced so it stays cool (max temp 66C or so).

 

[utahraptor]

Well I did tweak things a little bit. It seems certain applications can prevent the 45 multi from engaging even when set if not every core is set to 45.

Mine is set:

1 Core: 45
2 Core: 45
3 Core: 44
4 Core: 43

If I have steam open and it is doing nothing and I stress test 1 thread it will never user 45. If I close steam it will.

Also I was like another poster in this thread able to maintain 44 on all cores with stability and not terrible temperatures. 45 however seems to hit a voltage wall needing huge increases that I care not go into. (I stopped testing at 1.31 or 1.32) I ended up hard setting my PLL to 1.8 as it was on auto. I did run 1.65 for awhile and it seemed stable, but I just figured I would leave it at 1.8 since the heat difference was hardly noticeable.

I also wondered if I needed to increase my RAM past 1.5 to 1.525, but it did not seem to matter as I think the RAM was already stable at 1.5 and 2,133.

I think I am happy now leaving the system at 1.29 volts and 43, 44, 45, 45.

I am able to leave the Corsair H100 on low this way which is very quiet.

 

[Global688]

Haven't had an issue with my 3770k running @ 45 multi. Temps are even fine loading all the cores. When I built mine first thing I did was up the multi to 45, installed windows and didn't have an issue. Now I just run it stock no point in what I do (more cores are nice + HT) for any more speed.

The problem is everyone expects 120$ in WC to cool their processor. I looked at the corsair kits and they seemed to be very inadequate to cool anything over stock. If you really want a cool cpu 120$ just gets you started in a good water-block. Have to spend the $$$ if you want to go big.

 

[BallaTheFeared]

I've had 3 i5-2500k's.

First wouldn't do a lick past 5051MHz, after that it was unstable. It required 1.525v for 5GHz.

Second would do nearly 5.3GHz with 1.525v but would not go as high as 5.3GHz but a few times and was very unstable when doing it.

Third one I got does 5.5GHz with 1.52v, I've never tried to go higher lol.

There is a lot of variation within the same family of chips is my point, I've seen people who got stuck around 4.4GHz with SB, but that could be due to the end-user. I've never had a chip that wouldn't do 5GHz, and I've never had a problem with heat with SB with my custom loop.

Some people are just going to be more willing to push harder than others too, or go to extremes by removing the IHS or running liquid water directly over their $2000 PC.

 

[jacktesterson]

Stuck with my 2500k as I'm not comfortable with Delidding and my few buddies who did go Ivy have not been happy with temps or O/C's

Long story short, if you have Sandy Bridge, stick with it.

 

[exar333]

Stuck with my 2500k as I'm not comfortable with Delidding and my few buddies who did go Ivy have not been happy with temps or O/C's

Long story short, if you have Sandy Bridge, stick with it.

I think this is mostly true. For the build I just completed, I really just went with the cheaper option (3570k) because with the budget cooling setup I was making, (CM 212+) getting higher than 4.5ghz on the 2500k was unlikely anyways. A IB @ 4.2-4.3 is actually going to be a little better in most conditions.

On a high-end WC setup, SB and IB are not too different. It's always a crap-shoot on getting 5 ghz and higher anyways. IB has the advantage of being faster per clock anyways, so it doesn't need to hit quite the same numbers as SB does to get the same speed in most applications. If pure Ghz makes you happy, SB is probably the best all-around option.

 

[OVerLoRDI]

My ib has been wonderful. I have it at 4.5ghz and temps are in the low 60s with prime95.

I was going to delid and put new paste in there, but my temps seem a lot better than what others have. Granted I have a custom wc loop with a decently thick 240mm rad. But I was surprised that it was that much better than a H100.

 

[WhoBeDaPlaya]

Strangely enough, for my application (scientific number crunching), I don't see any real IPC difference between Sandy and Ivy. Since I was stuck at 4.2 on the "dud" Ivy I got, Sandy wins hands down (4.6 - 1.28V, 65C under IBT load).

 

[Idontcare]

Strangely enough, for my application (scientific number crunching), I don't see any real IPC difference between Sandy and Ivy. Since I was stuck at 4.2 on the "dud" Ivy I got, Sandy wins hands down (4.6 - 1.28V, 65C under IBT load).

My experience is the same. There might be a 0.2-0.3% improvement in IPC for my apps but that is it.

 

[gramboh]

Haven't read the whole thread, but my results since launch day:

3770K
Asus P8Z77-V Pro (latest BIOS)
Noctua NH-D14 with stock fans in a Corsair 550D with stock fans + 1 PWM intake

CPU = 4.5GHz via offset using all turbo/power saving options (all 4 cores turbo to 45 x 100)
CPU voltage peaks at 1.2V in 100% LinX (AVX) or Prime95 27.7 test, peaks at 1.21V in BF3/other games (lower load), idles at 0.97V/1.6GHz. All BIOS settings are stock except for XMP profile, voltage offset +0.03, Turbo ratio 45 (1-4 cores), LLC medium (25%).

CPU cores peak at 70-79C in LinX, 61-69C in BF3 after a few hours (with video card heat exhausting into case).

So 4.5GHz is very possible on air, and I haven't really done extensive tweaking/testing as I'm happy with the speed/temps.

Edit: Using the TIM that came with my NH-D14 and I re-seated the heatsink twice during installation (fan cable issues) without reapplying. Did not delid.

 

[bl00tdi]

My 4.8ghz is on air without delidding. Mid-80's in aida stress test and stable in everything I use it for from FSX to transcoding video and that's about all I really care about regarding stability.