Error Rates - Page 3

Daffy · 2002-08-31, 22:53

It seems there is a strong correlation between errors and heat. I know Prime95 tries to get as much done as possible as fast as possible => more heat. I really feel it with my Athlon 1400 and I don't want to use A/C. During the day, it's hot and I need to stop the program for a couple of hours to avoid destroying the PC too early.

Why not checking the temperature as well and decide to switch to some other type of work - less stressing - or stop the program when the CPU gets hot for a couple of minutes ? Would that help ?

Any way to send a command from an external tool to put prime95 into sleep mode automatically ?

I am using motherboard monitor and it's possible to start an application when the PC gets too hot. Would be nice if I could tell Prime95 to stop for 5 minutes and re-start.

Pi Rho · 2002-09-03, 17:10

I plowed through "Lucas_V.txt" and "BAD" to get some more results that may be of use. "Good" represents the double and triple checked results in Lucas_V.txt and "Bad" counts the incorrect results found in the range of exponents.

Exponents 0.0-1.3MM Good: 93568 Bad: 252 => 0.27%
Exponents 1.3-2.6MM Good: 71315 Bad: 1354 => 1.86%
Exponents 2.6-3.9MM Good: 69715 Bad: 1815 => 2.54%
Exponents 3.9-5.2MM Good: 66099 Bad: 2425 => 3.54%

Overall Good: 300697 Bad: 5846 => 1.91%

It might be interesting to see the counts over the different FFT ranges as well. I haven't been that ambitious yet.

As an additional check, I counted 131511 exponents from the range of 809-5,200,000. Can someone verify this?

asdf · 2002-09-04, 01:27

Quote:

As an additional check, I counted 131511 exponents from the range of 809-5,200,000. Can someone verify this?

i got 361,268 (you do mean prime numbers between 809 and 5.2 million right?)

Pi Rho · 2002-09-04, 11:43

I was just counting exponents found in Lucas_V.txt that have been double checked properly. I imagine that some of the exponents in that range were factored instead of being LL'd.

PageFault · 2002-09-06, 02:27

Doing that would be the worst possible thing you could do to the box. Thermal stress is not good. Computers, as with all other electronics, last the longest if they stay at a constant temperature.

Running these projects 24/7 increased the life of my hardware.

Quote:

Originally Posted by Daffy

During the day, it's hot and I need to stop the program for a couple of hours to avoid destroying the PC too early.

Why not checking the temperature as well and decide to switch to some other type of work - less stressing - or stop the program when the CPU gets hot for a couple of minutes ? Would that help ?

Daffy · 2002-09-06, 04:37

What temperature are you running your PC ?

My box stays at a constant 41 degrees C and the CPU goes from 49 degrees C to 55 degrees C. Up to 57 degrees when it really gets hot. At that point, I stop Prime95. It's too hot. And I have 7 fans in this PC (3 for the case, 1 for the power supply, 1 for the graphics processor, 1 for the CPU and 1 for the chip set) !

As you can see, there is no much difference between low and high temperature and I doubt this stresses the CPU very much. However, if your CPU runs very hot too long, you will kill it faster (and other things as well).

It's not like shutting down the PC. This can be stressful. Shutting down the PC when very hot is certainly not good. I lost my hard drive last saturday after shutting it down, just after posting my message and deciding not to do any LL work for that day because of the heat...

I am not sure if this failure was linked to the heat but it was surely a very bad surprise! :(

garo · 2002-09-06, 04:51

I think 57 is a bit hot. Are you sure your CPU fan is a good one? You may need to change that.

outlnder · 2002-09-06, 07:37

If you are running a TBird, 57C is about average without massive cooling. It is a little hot for an XP but still not above what the processor was made to take. If it is a Pentium of any family, it is a bit hot.

I don't worry about my CPU lifetime because I will probably replace it before it dies, under any circumstances.

I think great cooling is only necessary if you overclock or intend to never upgrade.

Daffy · 2002-09-06, 13:53

It's a thunderbird (Atlhon 1400). I am considering upgrading to an XP 1900+ eventually which is the max my KG7-R can support apparently. But still, it's hot inside. I wonder if I should find a way to open the box. I scared though. The box is on the floor and my feet are very close...

The 57 degrees C were reached when the temperature was reaching 30 degrees C outside. It's not a cooler 53 degrees C. I don't know what kind of heat sink I am using. The motherboard and the CPU / heat sink came together already assembled when I ordered them.

By the way, the failure was a RAID-0 stripe failure. The RAID controller declared a broken stripe and hops! All data lost... I had a 4 month old copy on an other drive which allowed me to restore things pretty fast. But I lost my assignment: 90 % complete!

svempasnake · 2002-09-06, 14:34

Quote:

Originally Posted by Pi Rho

As an additional check, I counted 131511 exponents from the range of 809-5,200,000. Can someone verify this?

Practically yes :) : In the lucas_v.txt copy I've got, downloaded it about 10 days ago, "DISTINCT EXPONENT" here counts to 131563. /S-O

2002-09-03, 17:10	#24
Pi Rho Sep 2002 2²×11 Posts	Some more error numbers I plowed through "Lucas_V.txt" and "BAD" to get some more results that may be of use. "Good" represents the double and triple checked results in Lucas_V.txt and "Bad" counts the incorrect results found in the range of exponents. Exponents 0.0-1.3MM Good: 93568 Bad: 252 => 0.27% Exponents 1.3-2.6MM Good: 71315 Bad: 1354 => 1.86% Exponents 2.6-3.9MM Good: 69715 Bad: 1815 => 2.54% Exponents 3.9-5.2MM Good: 66099 Bad: 2425 => 3.54% Overall Good: 300697 Bad: 5846 => 1.91% It might be interesting to see the counts over the different FFT ranges as well. I haven't been that ambitious yet. As an additional check, I counted 131511 exponents from the range of 809-5,200,000. Can someone verify this?

2002-09-04, 11:43	#26
Pi Rho Sep 2002 54₈ Posts	Counting exponents I was just counting exponents found in Lucas_V.txt that have been double checked properly. I imagine that some of the exponents in that range were factored instead of being LL'd.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Electrical Service Rates	storm5510	Hardware	70	2019-10-14 21:14
error rates and P-1 test	drakkar67	Prime Sierpinski Project	9	2008-05-26 14:29
error rates	drakkar67	Prime Sierpinski Project	12	2006-04-21 17:26
LL Test Rates and GIMPS Promotion	Primenut	Lounge	14	2003-06-09 09:32
Error rates revealed	Prime95	Math	1	2002-09-01 00:10

2002-08-31, 22:53	#23
Daffy Aug 2002 31 Posts	It seems there is a strong correlation between errors and heat. I know Prime95 tries to get as much done as possible as fast as possible => more heat. I really feel it with my Athlon 1400 and I don't want to use A/C. During the day, it's hot and I need to stop the program for a couple of hours to avoid destroying the PC too early. Why not checking the temperature as well and decide to switch to some other type of work - less stressing - or stop the program when the CPU gets hot for a couple of minutes ? Would that help ? Any way to send a command from an external tool to put prime95 into sleep mode automatically ? I am using motherboard monitor and it's possible to start an application when the PC gets too hot. Would be nice if I could tell Prime95 to stop for 5 minutes and re-start.

2002-09-06, 04:37	#28
Daffy Aug 2002 31 Posts	What temperature are you running your PC ? My box stays at a constant 41 degrees C and the CPU goes from 49 degrees C to 55 degrees C. Up to 57 degrees when it really gets hot. At that point, I stop Prime95. It's too hot. And I have 7 fans in this PC (3 for the case, 1 for the power supply, 1 for the graphics processor, 1 for the CPU and 1 for the chip set) ! As you can see, there is no much difference between low and high temperature and I doubt this stresses the CPU very much. However, if your CPU runs very hot too long, you will kill it faster (and other things as well). It's not like shutting down the PC. This can be stressful. Shutting down the PC when very hot is certainly not good. I lost my hard drive last saturday after shutting it down, just after posting my message and deciding not to do any LL work for that day because of the heat... I am not sure if this failure was linked to the heat but it was surely a very bad surprise! :(

2002-09-06, 04:51	#29
garo Aug 2002 Termonfeckin, IE 2²·691 Posts	I think 57 is a bit hot. Are you sure your CPU fan is a good one? You may need to change that.

2002-09-06, 07:37	#30
outlnder Aug 2002 2·3·53 Posts	If you are running a TBird, 57C is about average without massive cooling. It is a little hot for an XP but still not above what the processor was made to take. If it is a Pentium of any family, it is a bit hot. I don't worry about my CPU lifetime because I will probably replace it before it dies, under any circumstances. I think great cooling is only necessary if you overclock or intend to never upgrade.

2002-09-06, 13:53	#31
Daffy Aug 2002 31 Posts	It's a thunderbird (Atlhon 1400). I am considering upgrading to an XP 1900+ eventually which is the max my KG7-R can support apparently. But still, it's hot inside. I wonder if I should find a way to open the box. I scared though. The box is on the floor and my feet are very close... The 57 degrees C were reached when the temperature was reaching 30 degrees C outside. It's not a cooler 53 degrees C. I don't know what kind of heat sink I am using. The motherboard and the CPU / heat sink came together already assembled when I ordered them. By the way, the failure was a RAID-0 stripe failure. The RAID controller declared a broken stripe and hops! All data lost... I had a 4 month old copy on an other drive which allowed me to restore things pretty fast. But I lost my assignment: 90 % complete!