Opinions on prime95

[Keel]

Recently, the upgrade bug bit me and i got a new Barton 2500+ along with an Abit NF7-S and some Buffalo pc3700 (winbond BH-5 chips). I wanted to see what the rage was about setting the fsb directly to 200mhz, reaching xp 3200+ speeds (i have an ax-7 with a 12v L1a providing cooling) however prime95 will not pass!!! I think the system is pretty stable but no matter what I do (even raised vcore up to 1.8 at one point), prime will fail in under 3 hours.

Prime did finally pass though (ran for 18hours until i stopped it) when the 2500+ was run at 10.5x200 with a 1.725 vcore. At this setting, temps and noise are fine with me (my college dorm is reputed to be a noisy one).

What're your opinions of prime95?

[DryFire]

allot of people use it as a stability test. but i'd rather just use F@H as one.

it's pretty stressful on your cpu and most likly your ram although it will not flush out your gfx card flaws.

maybe you hit the point Diminishing return?

[SometimesWarrior]

The problem with F@H is that, if I recall correctly, it won't detect errors, but will instead pass them back to the server with the results. Your system could be spitting out bad data due to system instability, but you won't know you have a problem unless the client crashes completely.

That's why programs like Prime95 and CPUBurn are useful: they do a serious stress-test, but they check the results of the test to make sure you aren't getting any errors.

Personally, I wouldn't run a distributed computing client unless I knew my machine was going to generate good data for the cause. That means it needs to pass a 24-hour test of CPUBurn or something similar.

To someone who knows more about this stuff than I do: is my concern about sending bad data to F@H justified, or am I off-base?

[Ralf Hutter]

All Prime95 (in the "tortue test" mode that we use for stability testing) is doing is making a mathematical calculation and comparing the results to the known correct answer. If there is an error in one of these calculations Prime95 shuts down and notes that an error occured. It's pretty simple to me, if you are getiing errors in simple calculations your system is not stable. That's all a computer does after all, it manipulates "0's" and "1's". If it can't do that without genrating incorrect calculations something is not right.

Here's some info taken from the Prime95 Help FAQ:

This program is a good stress test for the CPU, memory, caches, CPU cooling, and case cooling. The torture test runs continuously, comparing your computer's results to results that are known to be correct. Any mismatch and you've got a problem! Note that the torture test sometimes reads from and writes to disk but cannot be considered a stress test for hard drives.

CAN I IGNORE THE PROBLEM?

Ignoring the problem is a matter of personal preference. There are two schools of thought on this subject.

It is likely that most programs you run will not stress your computer enough to cause a wrong result or system crash. A few games stress your machine and a system crash could result. Stay away from distributed computing projects where an incorrect calculation might cause you to return wrong results. You are not helping these projects by returning bad data! In conclusion, if you are comfortable with a small risk of an occasional system crash then feel free to live a little dangerously!

The second school of thought is, "Why run a stress test if you are going to ignore the results?" These people want a guaranteed 100% rock solid machine. Passing these stability tests gives them the ability to run any program with confidence.


FREQUENTLY ASKED QUESTIONS

Q) My machine is not overclocked. If I'm getting an error, then there must be a bug in the program, right?

A) Unfortunately, no. The torture test is comparing your machines results against KNOWN CORRECT RESULTS. If your machine cannot generate correct results, you have a hardware problem.

Q) How long should I run the torture test?

A) I recommend running it for 24 hours. The program has been known to fail only after several hours of operation. In most cases though, it will fail within a few minutes on a flaky machine.

Q) Prime95 reports errors during the torture test, but other stability tests don't. Do I have a problem?

A) Yes, you've reached the point where your machine has been pushed just beyond its limits. Follow the recommendations above to make your machine 100% stable or decide to live with a machine that could have problems in rare circumstances.

Q) A forum member said, "Don't bother with prime95, it always pukes on me, and my system is stable!” What do you make of that?
Or
We had a server at work that ran for 2 MONTHS straight, without a reboot I installed Prime95 on it and ran it - a couple minutes later I get an error. You are going to tell me that the server wasn't stable?

A) These users obviously do not subscribe to the 100% rock solid school of thought. THEIR MACHINES DO HAVE HARDWARE PROBLEMS. But since they are not presently running any programs that reveal the hardware problem, the machines are quite stable. As long as these machines never run a program that uncovers the hardware problem, then the machines will continue to be stable.

[aphonos]

There has been some info regarding Prime95 on the Radified.com blog. Pasted below for your reference:

From: 18August2003

Regarding yesterday's post, I received mail from Joshua this morning, who says:

The rig you’re building excites me. I read where you plan to use Prime-95 to stress-test the system. Funny thing about these new P4 HT chips is that Prime-95 only runs the CPU at 50%. Since the program doesn't take advantage of Hyper-Threading, and the P4 HT chip effectively has 2 CPUs on the die that share a common instruction bank, older programs such as Prime-95 only use half of the CPU’s power.

If you want a good stress-test, I suggest running both Prime-95 and Motherboard Monitor’s Heat Up utility together, which is enabled via MBM's Launcher. Simply click on the little "+" sign and navigate to the Heat Up directory inside MBM5's program folder. That will run the HT CPU at 100%, and give you a worst case heat scenario .. and a good test environment.

I wrote to George, who is in charge of Prime-95, and asked if he planned to modify the code to take advantage of Hyper-Threading. His response:

In short, no. Hyper-threading is not really 2 CPUs on one die. Prime95 uses the chip's one floating point unit so much that a second Prime95 would not find many FPU cycles to use. You could run a second Prime95 in another directory, but to put a higher load on the chip try an app that runs primarily integer operations.

From 19August2003

In reference to yesterday's post, Alex sent word about the Heat Up utility that comes with his Motherboard Monitor:

Hi Rad.
A small flame appears in the system tray when the program (MBM5HU.EXE) starts. Clicking that icon will stop the program. The code is very simple:

procedure THeatupThread.Execute;
var
X: Real;
begin
Randomize;

repeat
if Terminated then Exit;
X:=Random(65000)+3;
X:=X /
ArcTan2(Sin(arctan(cos(Random(65000)+3)))*Pi,Sin(arctan(cos(Random(65000)+3)
))/Pi);
until 1=0;
end;

There are no menus or settings for it. Tis an endless calculation that goes on and on. It's not perfect in that it will run only 1 CPU on a dual-CPU machine. Feel free to publish the code. It could turn into a nice project as I am sure others could create even bigger CPU hogs.

Regards, Alex

FWIW,

[Ralf Hutter]

There has been some info regarding Prime95 on the Radified.com blog. Pasted below for your reference:

From: 18August2003

Regarding yesterday's post, I received mail from Joshua this morning, who says:

The rig you’re building excites me. I read where you plan to use Prime-95 to stress-test the system. Funny thing about these new P4 HT chips is that Prime-95 only runs the CPU at 50%. Since the program doesn't take advantage of Hyper-Threading, and the P4 HT chip effectively has 2 CPUs on the die that share a common instruction bank, older programs such as Prime-95 only use half of the CPU’s power.

If you want a good stress-test, I suggest running both Prime-95 and Motherboard Monitor’s Heat Up utility together, which is enabled via MBM's Launcher. Simply click on the little "+" sign and navigate to the Heat Up directory inside MBM5's program folder. That will run the HT CPU at 100%, and give you a worst case heat scenario .. and a good test environment.

I wrote to George, who is in charge of Prime-95, and asked if he planned to modify the code to take advantage of Hyper-Threading. His response:

In short, no. Hyper-threading is not really 2 CPUs on one die. Prime95 uses the chip's one floating point unit so much that a second Prime95 would not find many FPU cycles to use. You could run a second Prime95 in another directory, but to put a higher load on the chip try an app that runs primarily integer operations.

From 19August2003

In reference to yesterday's post, Alex sent word about the Heat Up utility that comes with his Motherboard Monitor:

Hi Rad.
A small flame appears in the system tray when the program (MBM5HU.EXE) starts. Clicking that icon will stop the program. The code is very simple:

procedure THeatupThread.Execute;
var
X: Real;
begin
Randomize;

repeat
if Terminated then Exit;
X:=Random(65000)+3;
X:=X /
ArcTan2(Sin(arctan(cos(Random(65000)+3)))*Pi,Sin(arctan(cos(Random(65000)+3)
))/Pi);
until 1=0;
end;

There are no menus or settings for it. Tis an endless calculation that goes on and on. It's not perfect in that it will run only 1 CPU on a dual-CPU machine. Feel free to publish the code. It could turn into a nice project as I am sure others could create even bigger CPU hogs.

Regards, Alex

FWIW,

Ahh, very interesting! I've been running 2 separate instances of Prime95 on my HT CPUs even though some folks are claiming the one instance actually does load the CPU 100% (even though Task manager shows only 50%). I've found that I get temps about 1-2°C higher while running two instances as opposed to running only one instance, that's why I've continued to run 2 x Prime95.

Just for grins I'll try 1 x Prime95 and 1 x Heat Up to see how it works.

Thanks for the heads up aphonos!