XP90-C: It can't hang at 4Ghz! Any suggestions on a diff. HS?

[Jeeka_abx]

When squashing a movie in DVD Shrink, Probe is reporting that my 3.2E@4.0Ghz temps are at 54C/129F and that is when my machine resets.... I thought that these were supposed to hang until around 60C?! Either that or the readings in Probe are not accurate.

My 3.2E sl7pn e0 can do EVERYTHING accept compensate for the new heat inroduced by more volts... Any long loads on the chip, like After Effects rendering or squashing of a DVD in DVD Shrink causes the system to crash...

I lapped the ThermalRight XP-90c and have AS5 on it, so the temps are beautiful at idle: 40C/104F....even at load they hover at 50C/122F for about 5 minutes, then the temps go up and, bam, instant restart .

Could this be a HS that can't hang anymore? These xp-90C's are really really good HS's though....

Could it be a wobbly PSU? I have my P4P800-e Deluxe Droop modded so the volts are VERY stable between load and idle....

Is there a HS/Fan combo that is better than the Thermalright xp-90c (for a socket 478 3.2E PrescHott) ????

Thanks!
jeeka.

[fvbounty]

Originally Posted by Jeeka

When squashing a movie in DVD Shrink, Probe is reporting that my 3.2E@4.0Ghz temps are at 54C/129F and that is when my machine resets.... I thought that these were supposed to hang until around 60C?! Either that or the readings in Probe are not accurate.

My 3.2E sl7pn e0 can do EVERYTHING accept compensate for the new heat inroduced by more volts... Any long loads on the chip, like After Effects rendering or squashing of a DVD in DVD Shrink causes the system to crash...

I lapped the ThermalRight XP-90c and have AS5 on it, so the temps are beautiful at idle: 40C/104F....even at load they hover at 50C/122F for about 5 minutes, then the temps go up and, bam, instant restart .

Could this be a HS that can't hang anymore? These xp-90C's are really really good HS's though....

Could it be a wobbly PSU? I have my P4P800-e Deluxe Droop modded so the volts are VERY stable between load and idle....

Is there a HS/Fan combo that is better than the Thermalright xp-90c (for a socket 478 3.2E PrescHott) ????

Thanks!
jeeka.

Try a higher output fan on your xp-90, thats a easy thing to try and won't cost much, my xp-90 at full load runs between 47 to 52C depending outside temps...off topic how do you like your new ram?

[h20-ski]

Sounds like a psu being pushed to its limits. My OCZ psu can't handle my 3.2C oc'ed along with the X800 XT PE oc'ed, it just hard locks especially in 3DMark 05

[Jeeka_abx]

My PSU is an Antec NeoPower 480W.... The items on hooked up to it are:

MB, 6600gt agp w/fan, 2 DVD Burners, Floppy, 2x160GB WD Drives, 80GB WD Drive, 120BG WD Drive, 200GB Seagate SATA Drive, 2x120mm Case Fans, 92mm Panaflo HS Fan, Small 40mm Fan on Swiftech MX-159 HS

That could not be stressing the PSU THAT Much could it?

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

I will try a higher speed fan than the 92mm one I have on there now as a no-cost test....

If I do need a more powerful PSU, which would you recommend based on what you see above?

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

Off Topic,
I am LOVING this RAM, man. I can run so many hi-memory programs like After Effects, Photoshop, Illustrator at the same time and not even begin to dig into the swap file. Games are more enjoyable, system response is great. I just wish it could hit 250Mhz @ 2.5-3-3-6 like my PC3200C2PT 1GB Kit could (Which is for sale by the way)

[h20-ski]

Actually may be getting close. My estimate is that you are running almost 400W loaded. My OCZ PSU with the 3.2C @ 3.7, X800 XTPE @ 540/570, 2 Raptors, with just those components is stable, but as soon as I plug in the cd-roms, wireless card, and extras is not longer stable with my estimated peak load at 380W. But my Server running a 550W PSU has a peak load of 620W.

[FightingChance]

I don't know about running into the top of the PSU load; wouldn't playing games generate more load then encoding? (Encoding doesn't engage the video card like a game would)

[Jeeka_abx]

Hmm. I dumped a high rpm fan on there and spun that sucker up, it made it farther into the encoding process but it crapped out about 5 minutes after where the encode crashed without a super high rpm fan strapped to the XP-90C.

The droop mod for the p4p800-e deluxe on mine sits at 1.488v when i set it to 1.45v in the BIOS and flutters between 1.472v and 1.508v under load.

The CPU almost always craps out when the temps are at 54C under load for about 30 seconds.... MBM5 states the same temps..... so I am guessin' that the sensor has to be inaccurate......

Hmm.....

Any thoughts?
Water Cooling? worth it?
jeeka.

[h20-ski]

Hmmm watercooling is it worth it? That is like is it worth buying an X1800 or an X1600. If you are content with what the X1600 provides then no, but if you think that it is worth the money then yes. I myself find that it is worth the money

[R_S_Eddy]

Try turning the fan over, so its drawing air through the heatsink and blowing it out of the side of the case. Some heatsinks respond better like this when there is high air flow and high heat output.

If you go water cooled you`ll need a fairly decent set up to get a noticeable improvement over what you allready have.

Maybe You have just hit the limit of the cpu.

[h20-ski]

Well gaming does not stress the cpu as much as encoding. And while GPU's are getting up there in power requirements, a Prescot at 4GHz still takes the cake as far as power consumption goes.

[Jeeka_abx]

Crap,
For some reason, the 3.2E @ 4.0Ghz won't hold stable after the droop mod anymore.... Thing is, I dumped it down to 245FSB (3920Mhz) and dropped the vCore down one notch to 1.4375v (that's BIOS, it came to 1.488v in Probe)....and it held up under encoding into 57C/134F no problem, and finished the encode. Reason it got hotter with less volts is due to the fact that I closed the window and door and baked the room up basically......

If you remember, it was crapping out at 54C/129F @ 4.0Ghz, 1.45v Bios (up to1.508v in Probe)........

So this probably tells me that I have pretty much max'd out my OC and WON'T be able to stay at 4Ghz at all times....Sound logical? It can't be the heat then right???


I would like to see what you guys think. I can do all other work except any long loads on the CPU (encoding/rendering, etc...)

Help! I am sooooooo close! What else could I do? Put sinks on the fets? the XP-90c/fan blows down on some of the stuff it overhangs, near the back panel stuff.....

What else could I do that's cheap and widely known to have benefits.....

Thanks a ton!
jeeka.

[h20-ski]

What are your MOSFET temps? For the most part aircooled cpu's don't have a hard time keeping MOSFETs in operating temperatures. Have you tried to run strictly CPU intensive programs ie P95 small FFT's? Because that is odd that the cpu won't encode at 4GHz for more than 5 minutes but will do so at 245FSB. My experience is that if it continues to crap out after a certain short period of time, it will crap out eventually at a slightly slower speed, but longer intervals.

[FightingChance]

Maybe electron drift has started to set in and the maximum OC you can get will continue to degrade.

[yamawho]

I have the same mobo and CPU cooled with zalman 7700cu.

I had it over 3.8 Ghz but running OCCT torture test, it would reboot after about 5 mins. I lower the OC to 3.75 Ghz and was able to run OCCT over two hours with temps in the mid 50's.

Last week it rebooted playing doom 3 but I didn't change anything yet. This is the time of year when outside temps are mild and I haven't started using the A/C yet so it warms up in the basement when the 3 gaming rigs are running. When the A/C is turned on, it gets very cold in the basement.

I was ponderng changing the zalman and getting a XP120. It's been awhile since I looked into getting a HSF ... which is top dog these days for air cooling ?

[.stev]

The XP90 series is a top of the line cooler. How's the rest of your setup cooling in the case? And what are you internal case temps? Having the XP-90 cooling with warm air isn't fun.

Also, note, the XP90 and other heat pipe solutions work extremely well when they are face up vs. sitting sideways in a tower case. Not too long ago, I spent time understanding how heat-pipes REALLY work. This info comes from a site with both users and manufactures postings.

I saved the part that talks about heat-pipes working on their sides ...

The biggest assumption here would be that a heatpipe (even when using capillary action) will indeed perform the same in any possible orientation. No offense, but isn't it a little presumptuous for us to assume that each heatpipe would handle the same amount of wattage (15W)? I am absolutely positive that the orientation of the heatpipe will always matter, my friends - The manufacturer even knows this enough to include a diagram of the best-implemented orientation. The bulk of the inner material needs to coagulate nearest to the point of heat application for the design to be efficient - at all. This is exactly where the rubber hits the road, and equations won't apply unless they're in the correct context

The heat-transferring substance on the inside won't do much good if the applied outside heat is at the top of the heatpipe. Kind of like trying to heat up something from the top-down. Its just not as effective as putting the heat underneath it.
In essence, we could divvy up heatpipe efficiency based upon individual orientation - so in a vertically aligned scenario:

Top heatpipe: 30W
Middle heatpipe: 10W
Lower heatpip: 5W

While the combined output is still 45W, the contributing amounts of each heatpipe play a large factor in that role.

Now, what if the middle and lower heatpipe were both better aligned, like the top one? Then the heatpipe design would ROCK.

The only problem is, different motherboards have different alignments. So unless you knew exactly which way yours was aligned...

The other possible issue here is whether or not the fan is actually blowing onto the HS (as is supposed), rather than sucking air from the HS...

Being that the area of resistance would then be shifted, we would incur a further drop in back-pressure in the latter example (due to the inherent resistance of the HS). Velocity may or may not be increased in order to compensate, depending upon the exact design of the fan, HS, and even the ambient airflow measured within the case. There are many, many other outside variables involved here.

Therefore we can only trust these equations only to a very limited extent.

I don't think we should use them very widely in scope, as some other sites do.
This has proven self-evident time after time, in that some HSF's work better with the fan blowing outward, rather than blowing inward - by as much as 3C.

So what I'm saying is, the end really does justify the means here - While I absolutely and irrefutably do agree that any Thermaltake all-copper HSF will kick the pants off of any Swiftech HSF, its because we're looking at the bottom line results - not because of any formulaic equation.

People want to make heatpipes out to be hundreds (or even thousands) times better than the thermal conductivity of copper. However, unless they're implemented in the proper fashion, they're just really more copper (when push comes to shove) - with a slight twist (that being the heatpipe design itself).

The true effectiveness of the heatpipe will involve the measurement and application of the the 'thickness' of the copper tubing itself into the equation, in addition to the thermal conductivity of the inner material (with and without wicking/capillary action).

Again, if all the heatpipes could be oriented in the proper fashion, then we'd probably see a significant difference, but until a manufacturer realizes this, then we're stuck with heatpipe HSF's that only work best in a horizontal position.

You can see the simple heat-pipe required orientation from Thermalright here (just change the word DOT to a . ):

wwwDOTsilentpcreviewDOTcom/files/images/sp94-97/orientation.jpg

I do find it interesting that MANY review site will test heat-pipes in an open system sitting horizontal on a test stand. This gives the heat-pipe the most advantage.