Why it kept saying I got a PIII? - Page 2

Prime95 · 2002-08-31, 01:39

Everything looks fine. Those look like acceptable P4 timings.

Is there any chance the server difficulties the last week have caused our problem? Some users report mprime reacts badly when the server is unavailable - losing plenty of CPU cycles.

Try running mprime -m and set iterations between outputs at 100 or 1000. Then Test/Continue and see if iteration timings are OK (22 milliseconds). If so, does the the wall clock match up to your observed
iteration timings?

Does FreeBsd support a power-saving sleep mode? If so, disable it. That too can reduce throughput drastically.

garo · 2002-08-31, 02:08

My mprime running 22.7 crashed earlier this week because of server troubles. So definitely the server caused problems. Another option is to turn on verbose and see what the iteration timings are. Say

mprime -d > log.txt

That way you should find out how much time it's really taking.

ebx · 2002-08-31, 13:58

Once it starts crunching the second exponent, things are better. Here is some indicators:

M8312093 completed P-1 at Aug 31 00:35:59 2002. At Aug 31 06:36:47 2002, it reported progress 987109/8312093. That is more than 1/10 of the work in 6 hours. Whole thing should be done in about 50 hours.

Estimate is also 2 days. It said 5 days for the last exponent.

We will see.

We can enter my above bench mark as a data point for P4A 2.2G running DDR133. Sorry for the sticky mouse.

garo · 2002-08-31, 19:40

Again, don't trust the completion estimate too much. It can be wrong for a variety of reasons. Loos at the actual iteration times by using the -d switch.

ebx · 2002-09-03, 19:08

The estimates were not so bad most of times. It said 5 days for my last double check and it did take 5 days. This time, it said 2 and took 2. It is a P4, verified! :D Now I can get it some serious work.

Thanks!

2002-08-31, 01:39	#12
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 17·487 Posts	Everything looks fine. Those look like acceptable P4 timings. Is there any chance the server difficulties the last week have caused our problem? Some users report mprime reacts badly when the server is unavailable - losing plenty of CPU cycles. Try running mprime -m and set iterations between outputs at 100 or 1000. Then Test/Continue and see if iteration timings are OK (22 milliseconds). If so, does the the wall clock match up to your observed iteration timings? Does FreeBsd support a power-saving sleep mode? If so, disable it. That too can reduce throughput drastically.

2002-08-31, 02:08	#13
garo Aug 2002 Termonfeckin, IE 2⁴·173 Posts	My mprime running 22.7 crashed earlier this week because of server troubles. So definitely the server caused problems. Another option is to turn on verbose and see what the iteration timings are. Say mprime -d > log.txt That way you should find out how much time it's really taking.

2002-08-31, 13:58	#14
ebx Aug 2002 101 Posts	Once it starts crunching the second exponent, things are better. Here is some indicators: M8312093 completed P-1 at Aug 31 00:35:59 2002. At Aug 31 06:36:47 2002, it reported progress 987109/8312093. That is more than 1/10 of the work in 6 hours. Whole thing should be done in about 50 hours. Estimate is also 2 days. It said 5 days for the last exponent. We will see. We can enter my above bench mark as a data point for P4A 2.2G running DDR133. Sorry for the sticky mouse.

2002-08-31, 19:40	#15
garo Aug 2002 Termonfeckin, IE 2⁴×173 Posts	Again, don't trust the completion estimate too much. It can be wrong for a variety of reasons. Loos at the actual iteration times by using the -d switch.

2002-09-03, 19:08	#16
ebx Aug 2002 101 Posts	The estimates were not so bad most of times. It said 5 days for my last double check and it did take 5 days. This time, it said 2 and took 2. It is a P4, verified! :D Now I can get it some serious work. Thanks!