What is the highest safe temp to crunch on with Bloomfield? [Archive]

Sparky

10-26-2011, 12:54 PM

I'm trying to push my clocks on my i7 finally (yeah yeah!). It is cooled by a Scythe Mugen 2 with a Yate Loon medium fan. So I used a template just to give me a starting point. Here is what I have right now:

200x20 (4.0 GHz)
Vcore 1.35
QPI 1.35
CPU PLL 1.80
PCIe 1.54
QPI PLL 1.14
IOH Core 1.20
OCH I/O 1.50

I'm not going to pretend to know what all those mean :ROTF:

Anyway, I fired up intelburntest and let it run 5 passes, which it did just fine and tells me I'm on the right track. However, I saw a max temp of 90C! :eek: Popped the side panel and it didn't make a difference, so my airflow through the case is better than I thought it was.

I know Bloomfield can take a lot more heat than I'm used to, but that just looks like way too hot for 24/7 use.

I'm thinking maybe slapping another fan on the heatsink would help, but my main limitation is probably going to be air cooling in general. But I need a target to aim for, so what is considered a safe temperature to run this at?

*edit* maybe I have a dud mount, the sink doesn't feel that hot when I touch it :confused:

D_A

10-26-2011, 01:12 PM

Bloomfield is a mental hospital near here. I think they keep the heat turned pretty low to calm the inmat ... errrrr patients.

Jaco

10-26-2011, 01:41 PM

80°C is ok for the Bloomfield. Crunching always gives a bit lower temps than stress testing so you'll be fine i guess (maybe add a second fan)

Sparky

10-26-2011, 01:44 PM

I'm packing up the PC (it is my uber powered office PC I do all my email and paperwork on :ROTF:) and taking it home with me this evening where I have all sorts of extra fans and stuff to play with. That and I can let it stress test while I'm working on other things.

By the way, how many runs of intelburntest should I do before I consider it worthy of running WCG? I know WCG is the ultimate test, but I don't want to jump into it too quickly and burn a bunch of units.

Maybe I should play it safe and just do this to it :ROTF:

http://www.xtremesystems.org/forums/attachment.php?attachmentid=111078&stc=1&d=1294888280

I probably couldn't hear it over the server room next to me anyway!

OldChap

10-26-2011, 01:54 PM

a Yate Loon medium fan??? aw come on you can do better than that... put a big ol' delta on it! Seriously, if you have one you can get pretty good @ 7v if it is too loud for you...... medium huh I got fasts on the MB sinks

crunching temps of 80-85 is good

Sparky

10-26-2011, 02:21 PM

Hey that's all I had to slap on there at the time :rofl:

Also, the fan brackets limit me to 25mm thick fans. But I'm thinking of finding a way around that :idea:

*edit* testing some fan control on an NMB 4715KL-04W-B39 P57 fan. 12V, 0.72A, from what I can tell it is somewhere in the vicinity of 120 CFM at full blast :D

Sparky

10-26-2011, 04:37 PM

Alright something seems wrong. I just got one of those fans set up on a controller but still pushing a lot more air than the yate medium was, set it as the exhaust fan of the case with some custom *ahem* cardboard ducting (:D) so it draws all its air through the heatsink (think of how Dell used to do their systems), so now there is a lot more air getting pulled through the heatsink and immediately expelled from the case. Result? Same temps or even slightly higher! :mad: And the air coming out of the back from the fan isn't near as warm as you would think it should be if the CPU temps really are 92 C :confused: Heck, those old Dells with the P4s exhausted hotter air when the CPUs were a good 20+ C cooler :shrug:

Has to be a terrible heatsink mount. Has to be. That, or the heatpipes on this Mugen 2 are dead :wth:

*edit* cranked it up to full blast and it still was at 90C :shakes:

Otis11

10-26-2011, 06:14 PM

Not that I recommend this, but for a frame of reference I ran my 920 at 90C for 13 months before the mobo finally died on me... the chip continued to work though. (Got rid of it after that so idk how long it would have lasted like that)

It was actually running the same cooler you have - wasn't the best chip either, took more volts than I was happy with, hence the temperature.

If you think the cooler is dead and want another Mugen II shoot me a PM. :up:

Otis11

10-26-2011, 06:18 PM

shoota

10-26-2011, 06:26 PM

Another reference: I run my 950 at ~3.65ghz with ~1.22v under an H50 at ~65-69C. Ambient temp is around 73F

Sparky

10-26-2011, 06:31 PM

I remounted the cooler, same temps. Heatpipes are warm, but for the chip being at 90C I can hold my hand on the heatpipes (and the base of the cooler as well as I can reach it) indefinitely. If I press down on the cooler I can knock 4C off the temps fairly quickly - I'm thinking the mounting design leaves a lot to be desired.

Good thing is it has done 30 passes on IBT without issue so far. Weird thing is the system doesn't want to cold boot too well. Keeps resetting several times before switching to stock, where I go in the BIOS, hit save, and it is fine :shrug:

More tinkering to do.

Otis, I appreciate the offer, but maybe my sink is OK and I just need to address a mount design flaw. The screws bottom out before the cooler even feels all that tight on the chip.

shoota

10-26-2011, 07:30 PM

I had the same issue with my V8. These cpus (and probably all cpus) need a good deal of mounting pressure methinks

Otis11

10-26-2011, 08:12 PM

I had the same thing... I forgot how I fixed it though... I think I moved the screws in toward the cooler, it was harder to mount but it got better pressure (although I might be thinking of the wrong cooler)

Sparky

10-26-2011, 08:12 PM

Well I swapped screws on the Mugen 2 with some I had lying around that have threads the whole way and I cranked that thing down.

Temps still hit 92C :shakes:

OK I'm putting it back to stock again (shush) so I can use it tomorrow at work. I'll have to mess with it later.

*edit* got curious, took the heatsink off, and did the razorblade trick on the bottom of the heatsink. Nice and flat. Maybe the nickel coating is being an issue with heat pickup? Maybe. Put the razor blade across the CPU and Bingo. There's probably the biggest problem right there - CPU is quite concave. Figures! That would explain the horrible temperatures and why the heatsink is so cool in comparison to reported temps. Probably the only way to fix this is to lap the CPU (which I have never done before).

I'd have much rathered the heatsink be busted than the CPU have a concave IHS. Really Intel, can't we figure out how to consistently make a flat IHS? :rolleyes:

Conquistador SW

10-27-2011, 01:40 AM

If it is really bad, lapping will probably help some.

Also, I saw you were using 1.35v and was wondering if you really finalized the overclock?

I ask because I have seen some interesting things with the bloomfields I got.
Both of my i7's do not need relatively very low voltages (W3520 4Ghz 1.19v and 930 needed 1.29 for 4.2Ghz).
However, these chips were very hot and even at those voltages it was hard to keep them under 75C using watercooling.
So that is why I was thinking maybe you also got such a chip and are just giving it a bit too much voltage.

s0lid

10-27-2011, 02:32 AM

My w3520 is running at 64C at 4.1GHz in wcg load.
Watercooled with monster rad (9x120mm phobya).

Vinas

10-27-2011, 05:27 AM

Probably the only way to fix this is to lap the CPU (which I have never done before).With good coolers you will not feel hot air. They dissipate the heat very quickly, if you're feeling heat then something is wrong. You can test by touching the base of the cooler, if it is noticeably warmer than the heat pipes then that could indicate a problem. But TBH your best best is to let off the voltage and let that chip run a little cooler. Hell I have a $1000 water setup and I only run my i7 860 at 4.2GHz when folding... They heat up! :p:

Sparky

10-27-2011, 05:42 AM

The base isn't very warm either and the heatpipes are only somewhat warm.

At stock 3.2GHz at stock volts, intelburntest gets to 62-65C. 30 degrees difference. Wow.

The volts were set that high because that is what the template had on it. That's the only reason lol. I'll mess around and drop the volts. It was stable through intelburntest 60 passes. Except it would not cold boot at those settings I had - it would just do a reset several times, power off, back on, couple more resets, off, on, then POST at stock. Enter BIOS and save/exit, it would reboot, do a couple resets, and POST at the 4GHz :confused: Any idea what it is telling me when that happens? Something set too high, or too low, or...? Just seems odd that once booted it seems to be stable, even at those insane temperatures.

shoota

10-27-2011, 07:16 AM

seems like a finicky board to me. which board is it again?

Sparky

10-27-2011, 07:29 AM

Gigabyte X58A-UD3R. It doesn't do any of that weird stuff at stock speeds.

Tom128

10-27-2011, 10:50 AM

I have the same question except about Sandy's. What is the max temp I should run on SB? I have my new 2600k and it's working fine, however I didn't do my research on cooling and realize not a single cooler in my collection will fit on it. So I am using the stock cooler for the moment. When playing Battlefield 3 the temps get towards 70c, I would imagine with WCG it would be even higher.

I've been an AMD guy for a long time and 70c is 15c past where I would ever let a chip run. I know Intel can/does run much higher, but I am hesitant to even put it on WCG until I either get a new cooler or find an adapter to let my water cooling run it. I might try undervolting it at stock speeds as well until I get the cooling sorted out.

Sparky

10-27-2011, 10:55 AM

Just a side note, AMD chips can do 60C, mine was there for quite some time when my water cooling needed work and I stuck a less than stellar air cooler on it, and I didn't want to lower my overclock :D

Conquistador SW

10-27-2011, 01:39 PM

I understood sandy bridge cannot take the high temperatures like bloomfield could and I personally would like to keep it at 70C max if it were my crunching rig.
But as sparky pointed out, I also found that hardware these days can take a beating.
I abuse most of my hardware and I have never ever had to RMA a component.
For example my Q9450 around 75-80C for a few years and still clocks like the day I got it.

Red Maw

10-27-2011, 01:52 PM

Gigabyte X58A-UD3R. It doesn't do any of that weird stuff at stock speeds.

I have that MB for my i7 system and have had cold boot (usually only fails once though) issues once or twice. Since the only time I cold boot is when there's an extended power outage I can't tell if it's a rare or frequent failure though :shrug:

Conquistador SW

10-27-2011, 02:04 PM

Haha, you had me wondering if I ever had cold boot issues because it sounded somewhat familiar, but I cannot remember that I ever really cold booted it after the first time I powered up the system.....

Sparky

10-27-2011, 02:05 PM

Maybe I just need to update the BIOS. But I won't be able to get around to playing with it again for a while, work is going to keep me busy (end of month stuff) so I guess I can't get it clocked up for SC week :(

Otis11

10-27-2011, 06:27 PM

mreuter80

10-28-2011, 05:53 AM

Thanks Sparky for this thread. It made me check the temps of my rig.
Well, I can safely say that the i7 take a heck of temp before crashing. My machine ran with 100C - not by choice, but by accident - or laziness or however you want to call it.
The machine seemed rock solid, but I almost got a heart attack when I checked the temps last night (and I checked only because of this thread). Needless to say that seconds later the computer was open, power-washed, rinsed, waxed and polished and it runs now between 75 - 80C (@3.8 GHz) which I think is a good temp.

I don't believe that a BIOS update will help. Actually 4 GHz is already quite an achievement.
Have you tried with only one fan (I only use one)?
Have you tried new thermal paste (or did you use the stuff you found in the drawer from 5 years ago)?

Sparky

10-28-2011, 08:47 AM

I changed fan configs. one, two, push, pull, push-pull, direct exhaust, high flow, low flow...

I used MX-3 that I bought about 2-3 months ago.

I thought 4 GHz was "normal" for an i7? I almost went for 4.2 but I started with 4 just to see what happened. Seems stable (even at 90C) just hates cold boots. Like I said, I'll have to tinker with it when I have more time. I just don't know what to do about the concave IHS. You'd think they could figure out how to keep them flat :shakes:

Conquistador SW

10-28-2011, 01:46 PM

4Ghz is indeed "normal" for a i7. I would be disapointed if I got one that didn't make that.

Unfortunately, I have no idea what the cold boot isue could be. :shrug:

I wouldn't worry too much about the concave IHS too much. Of course you experience could differ, but I had both a E6600 and Q9450 which were quite concave and both lapped them. It only helped me 2 C max, so it didn't make that huge of a difference. I think you gain more by mounting your heatsink better as you discovered that pressing it down already gained you 4C. If you ever consider getting another heatsink I can say I really like the Megahalem with a gentle typhoon. Easy, quiet and very good mount. I used it for various crunching rigs; currently my dad is using the W3520 in my sig at 4Ghz in some low airflow very silent antec case under a desk with lots of dust and it never breaks 70C using the Megahalem.

Sparky

10-28-2011, 02:48 PM

The thing that got me on the mounting is I cranked it down with some other screws but that didn't make much difference. When doing that, all the pressure is pulling downward on the 4 corners... not very helpful. It needs pressure in the center more than anything else.

shoota

10-28-2011, 04:05 PM

that's my issue too. i need more pressure in the middle, not on the corners. How does the megahalems accomplish a good mount?

D_A

10-28-2011, 08:26 PM

Use the original screws but see if you can slip a nickel or something in the center between the bracket and the cooler base. It would localise the pressure in the center rather than at the corners.

I'm on some good pain killers though, so I could be talking shyte.

Sparky

10-28-2011, 08:32 PM

The problem with that is this:

http://media.bestofmicro.com/J/K/262496/original/scythe_mugen2_1366.jpg

The brackets screw on to the bottom of the cooler on each side. They don't cross through the middle (there is a small sink in the middle). Not really any way to improve center mounting pressure, which is a shame because this cooler would benefit greatly from that I would think. It's huge with lots of heatpipes in a pretty good configuration. I wonder if that center heatsink could come off somehow and I could use a different bracket on it that crosses the middle to bolt it down with...

D_A

10-28-2011, 08:43 PM

I remember reading about some guys who would lap down a silver dollar to use as a heat spreader a while back. If you could add a little silver shim a little smaller than the IHS it would concentrate the pressure more towards the center. Maybe. I'm making this up here. Old P4 heat sinks had what I believe to be exactly that, along with the crappy black compound either side of it. That would of course put two thermal interfaces in the system instead of one ... but like I said, I'm making this up as I go.