Scared to run Prime 95

V

Volume

2[H]4U
Joined
Jan 17, 2004
Messages
4,010
Do I have anything to worry about, running Prime95?

I have a P4 2.6c @ 3.2ghz
I am using a 6000rpm max. self-adjusting fan (see sig below). Since this is 100% cpu usage, am I in risk of burning out the cpu? Also, my voltage is from default 1.52v, set to 1.55v.
 
man, you overclocked the muther and your scared to use it? go ahead, yer fine. watch your temps for 15 mins and make sure they stay below 50ish.
 
Prime away, It will not hurt a thing. It will either work, or it will crash, or give errors. I will not hurt your CPU though. Even if it gets too hot, your p4 has thermal throttling that will keep it in check.
 
That's the thing: I have two temperature readings. One, a Hardcano hooked up to the bottom of my copper heatsink. Reads 35 idle. The other, an onboard sensor (IC7-MAX3) that reads 50 idle. Now, I know that the MAX3 has higher than usual readings, so I don't know my exact temperature. So, do I run the risk of frying anything? I'm not sure which temperature to watch. I have the BIOS set on shutdown at 70 (because 70 on this board is less than 70 in reality). I'm not sure what to do... :confused:
 
Originally posted by Dude
Prime away, It will not hurt a thing. It will either work, or it will crash, or give errors. I will not hurt your CPU though. Even if it gets too hot, your p4 has thermal throttling that will keep it in check.
Click to expand...

Alright, I'll go for it tomorrow morning. It's a bit late in the house to be running a 6000rpm fan. :D

So, if it has this thermal throttling, what's this I hear of people killing their P4's?

By the way, regarding voltages, how much is TOO much on a P4? I have mine at 1.55 (though CPU-Z reports 1.488-1.504. Also, are there any indicators on when I should up or down the volatge?
 
oh. kinda sucks if you dont know what temp yer core is at. but it cant be that high if you can get it to 3.2GHz can it?

but if you can ever figure how to get a real reading, the diode/core temp is the only one you need to worry about. at idle, and of all the temps reported by your board sensors, it will probably be the lowest one.
 
Originally posted by dobbz
oh. kinda sucks if you dont know what temp yer core is at. but it cant be that high if you can get it to 3.2GHz can it?

but if you can ever figure how to get a real reading, the diode/core temp is the only one you need to worry about. at idle, and of all the temps reported by your board sensors, it will probably be the lowest one.
Click to expand...

It's giving me a 49-50 idle on the diode core. I'm using motherboard monitor. So, I don't know any other way to test my diode temp, and still, that seems high.
 
Originally posted by Volume
Alright, I'll go for it tomorrow morning. It's a bit late in the house to be running a 6000rpm fan. :D

So, if it has this thermal throttling, what's this I hear of people killing their P4's?

By the way, regarding voltages, how much is TOO much on a P4? I have mine at 1.55 (though CPU-Z reports 1.488-1.504. Also, are there any indicators on when I should up or down the volatge?
Click to expand...

I'm pretty much an AMD guy so obvioulsy I'm no expert on intel stuff. My understanding of the thermal throttling is that once the core reaches a certain temp (sorry don't know what temp), the CPU slows itself down to reduce heat. It does it REALLY fast too. It think that most people fry their P4s with too much voltage (couldn't tell what too much is) which is a completely different monster than temps alone. You may get more expert opinions from someone over in the "strictly intel" forum.

Good luck!
 
Well, thanks a lot for the help, guys. I feel confident enough to run the benchmark now. I'll let you know how it goes tomorrow morning.
 
why are people scared to go over 50Celcius?

Some months ago i was cleaning my Athlon 1.4 heatsink and took the time to change the paste with the ocz silver paste(2celcius colder than any Artic silver btw) and when finished i start it up, quicly get in the bios, system monitoring, temp wa 50C, hum wtf, i kinda looked again maybe i was looking at Farenheight then i quicly unplugged the power supply, the last number i saw on the bios was 81C.

I though i burnt it, but after i saw that my heatsink was simply not touching the chip due to a hold down clamp reversed, i fixed it, started it up, everything was fine, temp was at 37C idle.
I do think i was like 2 or 3 second near a burnt cpu, but i dont think theres a problem before 70C under load, use motherboard monitor or any program that can shut down the pc if the temps go beyond 70C you never know... cpu fan can stop, heatsink filled with dust, ...

I say that because not all mtboard bios have these security, ie cpu fan check and temp check. Like my Abit KG7.
 
Oh, I have those security options set in my BIOS. Wouldn't boot up without them. :p
 
Originally posted by embro
why are people scared to go over 50Celcius?

Some months ago i was cleaning my Athlon 1.4 heatsink and took the time to change the paste with the ocz silver paste(2celcius colder than any Artic silver btw) and when finished i start it up, quicly get in the bios, system monitoring, temp wa 50C, hum wtf, i kinda looked again maybe i was looking at Farenheight then i quicly unplugged the power supply, the last number i saw on the bios was 81C.

I though i burnt it, but after i saw that my heatsink was simply not touching the chip due to a hold down clamp reversed, i fixed it, started it up, everything was fine, temp was at 37C idle.
I do think i was like 2 or 3 second near a burnt cpu, but i dont think theres a problem before 70C under load, use motherboard monitor or any program that can shut down the pc if the temps go beyond 70C you never know... cpu fan can stop, heatsink filled with dust, ...

I say that because not all mtboard bios have these security, ie cpu fan check and temp check. Like my Abit KG7.
Click to expand...

I think AMD can handle more heat then Intel...though I dunno much about Intel =\
 
ive seen and touched an AMD Thunderbird run @ 90C. Nice to gain thicker skins when using AMD :D

(sticky fan + cheap mobo that does not shutoff on thermal danger or fan failure)
 
PRIME 95 RESULTS:

Got to Test 5 (blend test mode):
FATAL ERROR: Rounding was 0.4975585938, expected less than 0.4

Torture test ran 10 minutes - 1 error, 0 warnings.
Execution halted.


Sooo... what's this mean? *Look at first post to see my o/c.
 
Originally posted by Volume
PRIME 95 RESULTS:

Got to Test 5 (blend test mode):
FATAL ERROR: Rounding was 0.4975585938, expected less than 0.4

Torture test ran 10 minutes - 1 error, 0 warnings.
Execution halted.


Sooo... what's this mean? *Look at first post to see my o/c.
Click to expand...


means you OC'd too far
 
Bah! As long as my system doesn't mess up on everyday tasks I'll keep the o/c.
 
Originally posted by Volume
Bah! As long as my system doesn't mess up on everyday tasks I'll keep the o/c.
Click to expand...

Yep, as long as prime is not the primary use of your computer, you should be fine.
 
If your system is not stable, do not fold or any other distributed computing programs. You will not turn in valid results.

If you want to risk crashing the system, that is your deal.
 
though i burnt it, but after i saw that my heatsink was simply not touching the chip due to a hold down clamp reversed, i fixed it, started it up, everything was fine, temp was at 37C idle.
Click to expand...
i killed a 1.3 duron that way :rolleyes:

maybe 3 or 4 seconds on , had heatsink cockeyed, so only half touching chip .

no smoke , no heat , but alas , no boot no more.:mad:
 
When I first booted my watercooling setup a few weeks ago, the block wasn't completely on. BIOS temp read 75C and climbed quickly shutting down at 90C. I'm using the CPU right now so no harm was done, (P4 [email protected]). Northwoods are killed by voltage, not heat. I'd highly suggest you don't go over 1.65v. Chips were dying when people used 1.75v+ on them.
 
Originally posted by SmokeRngs
If your system is not stable, do not fold or any other distributed computing programs. You will not turn in valid results.

If you want to risk crashing the system, that is your deal.
Click to expand...

Stability is not a problem. It was a processing error, which does not bother me.
 
Ive seen lots of cpus fail prime, i lost faith in it since they had no issues with any other program.
 
Originally posted by Volume
Stability is not a problem. It was a processing error, which does not bother me.
Click to expand...

Umm... that means it's not stable. The CPU is not calculating the correct results. That is why you should not run any type of distributed computing programs. You will turn in bad results that way.
 
Originally posted by SmokeRngs
Umm... that means it's not stable. The CPU is not calculating the correct results. That is why you should not run any type of distributed computing programs. You will turn in bad results that way.
Click to expand...

Okay, and as I said, its stability is not a problem because I can do my everyday tasks just fine.
 
Now, I know that the MAX3 has higher than usual readings, so I don't know my exact temperature
Click to expand...

Actually, the IC7 has a more accurate temperature readout than the hardcano. The mobo reads the on-die thermal diode and the hardcano reads the outside of the package. The die is always 10 to 20C hotter than the heatspreader.

I changed motherboards from the Gigabyte GA-8KNXP to the IC7-MAX3 and they both use the on-die diode and they both turn in the same numbers.
 
There could be a number of things limiting your overclock from being stable, most likely the vcore...

Make sure your ram has enough voltage and is stable at whatever clocks / timing you have set..

Make sure your NB has enough voltage too, although i dont think you can adjust that on the IC7..

Dont be afraid of 50c or higher, p4's throttle which will save you no matter what. you could remove the heatsink and it wont fry.. unless it was a super high overclock like 3.8ghz+ or more and the instant lack of cooling may harm it but that's a different story,

You should be able to do 3.2 pretty easily off that cpu though..

Also to you non-stable prime users, how can you call your pc "stable" if it obviously fails as a result of unstability in prime95.
 
Originally posted by syztem
...Also to you non-stable prime users, how can you call your pc "stable" if it obviously fails as a result of unstability in prime95.
Click to expand...

How long do you run prime before you call it "stable"?
 
I would recommend an hour at least after everything has warmed up and to run two instances to test hyperthreading.
 
Originally posted by syztem
...Also to you non-stable prime users, how can you call your pc "stable" if it obviously fails as a result of unstability in prime95.
Click to expand...

Originally posted by smahoney
I would recommend an hour at least after everything has warmed up and to run two instances to test hyperthreading.
Click to expand...

So how do you know it's not going to give errors after two hours?
 
Originally posted by Volume
PRIME 95 RESULTS:

Got to Test 5 (blend test mode):
FATAL ERROR: Rounding was 0.4975585938, expected less than 0.4

Torture test ran 10 minutes - 1 error, 0 warnings.
Execution halted.


Sooo... what's this mean? *Look at first post to see my o/c.
Click to expand...

Failing in 10 minutes is very unstable to leave it . For example I did a winXP FIX (after screwing my system with overclocking it currupted winXP), and did not reload my chipset drivers, My prime95 would fail very quickly like yours. But I had other problems also like Applications that I would normally run for hours like Overnet would crash out on me after a few hours, my system was very unstable! And you might not realize. I bet running a demanding game like quake3 or wolfenstein will surly crash on you after a while.

So what I did was reformatt and install chipset drivers properly and my system was smooth as butter, Games NEVER crash on me anymore, Overnet ran without ever crashing once!. But I actually ran prime95 again and it would fail in 2 hours this time, It did this a couple times and now I can run prime95 without failing, don't konw what changed...

You have to figure out why prime95 is crashing so quickly, run prime95 at stock speeds, run memtest, Becuase you have a problem.
 
I just upped the vcore a bit (according to BIOS, it's a little over 1.55, but CPU-Z reports 1.52). I'll try Prime95 again later on.

I have another question:
I had to loosen my RAM timings quite a bit to run the 3.2 overclock. The RAM is running at 198mhz (200 default). Would increasing on the voltage on the RAM let me run it at default or even better timings? Default timings for it are 2-3-3-7 and I had to set them at 3-4-4-8. What's the limit on increasing voltages for RAM?
 
Originally posted by Dude
How long do you run prime before you call it "stable"?
Click to expand...

24 hours is the usual amount. by then its had many chances to run ALL the tests at least a few times, so if anything was going to break, itd have done it by then. for memory I dont use prime95, i use memtest86 at the settings I want to run the ram at. prime95 would work just as well but i dont like being on the desktop and not being able to use it (prime95 testing memory takes up all of it so you cant really do anything)
 
Originally posted by kronchev
for memory I dont use prime95, i use memtest86 at the settings I want to run the ram at. prime95 would work just as well but i dont like being on the desktop and not being able to use it (prime95 testing memory takes up all of it so you cant really do anything)
Click to expand...

thats only if you're running Prime at high priority. by default its at low, which lets me run IE and stuff smoothly. of course that steals time from Prime, but thats fine if its not really hogging the CPU for too long.
 
oh, some will say they run all kinds of benching/burning programs alongside Prime to be sure theyre system is stable, as if multiple programs sharing 100% processor/memory load is more stressful than just one, but that sounds like BS to me. i mean, is it harder to lift a barbell with four 25lb weights than one with two 50lb weights? no.
 
You must log in or register to reply here.
Back
Top