Overclocked X6 1075 stress test

Hi guys.

So I'm new to overclocking, this is the first cpu I've tried to overclock. Right Now I've got it clocked up to 3.82 Gz and it seems to be relatively stable. However, when I run prime95 on the small FFT test and it shoots up to 75 C in under about a minute, climbing steadily. That kind of temp spike made me nervous so I stopped the test. Is it normal for that kind of a temp spike when doing a stress test? It idles at about 30 to 35 and when under load from a game it ranges from 40 to 50 depending on the load, maybe it gets up to 55.

Thanks for any help.

Specs:

CPU Voltage: 1.432; stock ~ 1.32
CPU - NB Voltage: 1.38; stock ~1.3
NB Voltage - 1.2; stock ~ 1.1

No Aftermarket Cooler
NZXT Phantom case with no additional fans

X6 1075 @ 3.82
8GB 1663 Patriot RAM
Radeon HD 5850 1GB
MSI 870a Fuzion Motherboard (have heard this board is finecky when OCing)
650 Watt Thermaltake Power Supply
1 TB HDD

Hello and welcome to the forum

Prime95 will get your temps as high as they are ever likely to go and beyond, i think what your looking at there are the socket temps, and 75c is high, its to high.

You were right to be nervous and right to have stopped the test, you should not let your socket temps go over 62 or 65 at most, its not actually the socket temps that matter, your core temps are more important but if the socket gets to hot it could damage your motherboard, at this stage my advice is don't prime stress it with those settings.

whats more the AMD stock cooler is close to useless for overclocking.

We need to know more about your system before we can advise you more.

You will also need some software like HWMonitor which will give you, and us a clear understanding of whats going on (including your core temps) and CPUID is useful for looking at what the settings are actually operating at.

For now install the software, set your CPU back to stock, Run HWMonitor, run Prime again (with stock CPU settings and clock) and keep your eye on those temps, if they reach over 64c stop the test. take a screen cap of HWMonitor while the CPU is warm and stressed, then upload the screen cap in this post.

What frakk said.

AMD states the maximum safe temp for most of their processors is around 70 C. You really do need a good aftermarket cooler to overclock that CPU. But beware that if your case is less than 8" wide it probably won't provide the clearance to the side panel necessary to accommodate a 120 mm fan-based tower style cooler which are the good ones. If that is the case you might need to look at a water cooling kit or get another case.

trents said:

What frakk said.

AMD states the maximum safe temp for most of their processors is around 70 C. You really do need a good aftermarket cooler to overclock that CPU. But beware that if your case is less than 8" wide it probably won't provide the clearance to the side panel necessary to accommodate a 120 mm fan-based tower style cooler which are the good ones. If that is the case you might need to look at a water cooling kit or get another case.

Click to expand...

And on that note it looks like your case is wide enough for something like a CM Hyper 212+

If you want to do some 'proper' Overclocking and have a bit of spare cash (its not much) you will need it, or something like it.

Just as the year comes to an end, I have seen two posts back to back with some of the clearest information to be found. Good luck to you two guys in the New Year and keep up that awesome spot on work. RGone...ster.

Thanks for the quick reply.

Quick update:

So later last night I found out 3.82 was not stable (bsod after about 40 minutes of gaming) so I took it back down to 3.74 which I'm almost certain is stable as that what's it's been set at for the last week or so with no problems. This speed also seems to require much lower voltages. I'll examine the bios and get exact numbers later today. Also, I forgot to mention I was monitoring temps during the stress test with CoreTemp, so unless the program's a liar those temps I were reporting were the core temps, however I've never even heard of socket temps and don't know what they were during the test. I'll download those programs and re-run later today and update.

Thanks Again!

Okay, "Seig" did not tell us what cpu temp monitoring application he was using and now there are MANY. Actually probably too many. Too many for any of us to have a real baseline of use to determine if the application is a known liar or does a pretty good job.

CoreTemp is probably pretty accurate and does not bother to show 4 cores split out on a four core AMD processor since there is only "one" core temp signal exitting the cpu anyway. That is why HWMonitor shows the core temps the same for however many cores are in and AMD porcessor.

So from my looking around, CoreTemp probably is okay for reading the temp of AMD cpu since the temp it will show is in fact "just" the core temp and not a bunch of stuffz.

Now that said, I still prefer HWmonitor since it has a table readout of voltages, that I do use as well as the core temp even if it shows multiple temps on AMD even though there is only one temp from the cpu itself and not 2, 3, 4, or 6 core temps as is displayed onscreen.

"Socket" temps are measured from the CPU socket area of the motherboard. "Core" temps are measured directly from the processor die itself and are therefore more useful in analyzing what's happeing temp wise with the processor, i.e., its a more direct measurement. Most of the time, "CPU" temp refers to socket temp. The exception that is CoreTemp where "CPU" temp is actually core temp. The temp reading you get in bios is a CPU socket temp and most all hardware monitoring programs that give temp readouts will take the reading from bios.

Like RGone, I'm partial to HWMonitor because it gives more info than does CoreTemp.

By the way, you should run Prime95 blend for at least 2 hours to determine if you overclock values are stable. Prime95 is the defacto stability tester for the overclocking community.

Those look like idle temps. Run Prime95 blend for 20 minutes with HWMonitor open during the test and report back with another pic. Yes, you have the concept for CPU/socket temps vs. core temps down correctly. Core temps are particularly useful when unlocking cores because when that is done the core temp reading is disabled. If you know the differential between core temps and CPU temps before you unlock cores then you can estimate the core temp from the CPU temp after unlocking.

Ruh Roh, I feel a *waft* coming on. Ole "Frakk" was only showing "Seig" what to look for in HWMonitor. He weren't trying to toilet us nuttin bouten his rig. Hehehe. Crazy RGone...ster outen hear.

Was just illustrating whats what in HWMonitor.....

And because you asked.... this is @ 3.5Ghz in Win diagnostic mode. its not that good i know, my 3 year old cooler was not all that when i bought it

I can't wait for the CM 412S to hurry up and get here....

Oh, and there's no 'wafting' during this tress test, the crack-head i hired ran off with my case side panel....

Frakk said:

Oh, and there's no 'wafting' during this tress test, the crack-head i hired ran off with my case side panel....

Click to expand...

He, he, ho, ho! (Belly laugh). Frakk, you crack me up!

We need a guy like you around here. Thanks for coming on board.

^ +1 ^

Thank you, its good to be here.