avg. time (sec): 29108 - 25444 - 25546
No. of results: 4 - 5 - 1
Now I have 5 results by app 4.17. Avg. time is 25303 sec (min. 24771 - max. 26067 sec) that seems 4.17 is about 0,5% faster on My Athlon64 2800+ (socket 754) but by 4.16 was time of crunching between 24478 and 25920 sec.
On my second computer (cpu Intel(R) Pentium(R) M processor 1.60GHz) I have 44 WUs h1_0335.0 (credit 19,55) crunched by 4.17 avg. time 3448 sec. min. time 3436 sec and max. time 3455 sec. On this comp I have only 1 Wu crunched by 4.01 and this is WU h1_0728.5 (credit 174,29) time 38975 sec. So by 4.01 I got 16,1 credits per hour and by 4.17 I got 20,4 credit per hour
Well... I still think v4.17 is a bit slower on this machine, but I haden't enough data under 4.16 to make an accurate comparison.
Question at 0.001 credit ... Do you guess what my dilemma is ? :-)
"Vanitas, vanitatum et omnium vanitas]"
Ecclesiast
"Entia non sunt multiplicandam praeter necessitatem" (OKHAM)
589494 is an AMD Athlon 1200+ processor running Slackware 10.2 with kernel of 2.4.31. It has had no issues with validation, but when going from 4.16 to 4.17, it has slowed somewhat. So, on older AMD Athlons, it seems to be less efficient than 4.16. 4.16, however, was significantly faster than 4.01. We appreciate your hard work on making our machines more efficient in their contributions.
Hello all, my A64 3700 just started crunching WUs with the 4.17 beta app and here are some early results for small WUs:
4.01 avg. for last 5 results - 2362.6 s
4.17 avg. for first 5 results - 2007.2 s (all validated)
Approximate speed increase : 15%
In the stderr_txt ouput of all results crunched with 4.17 I am seeing something I have never seen before, have a look
Quote:
2006-08-02 22:36:48.0649 [normal]: Start of BOINC application 'einstein_S5R1_4.17_i686-pc-linux-gnu'.
2006-08-02 22:36:48.0814 [normal]: Started search at lalDebugLevel = 0
2006-08-02 22:36:48.3104 [normal]: Checkpoint-file 'Fstat.out.ckp' not found.
2006-08-02 22:36:48.3104 [normal]: No usable checkpoint found, starting from beginning.
Detected CPU type 1
small x
small x
small x
small x
small x
2006-08-02 23:10:13.8260 [normal]: Search finished successfully.
What does 'small x' mean? The number of times it appears differs for each WU?
Great work Bernd, keep it up.
There are 10^11 stars in the galaxy. That used to be a huge number. But it's only a hundred billion. It's less than the national deficit! We used to call them astronomical numbers. Now we should call them economical numbers. - Richard Feynman
My AMD 2700+ is 300-400
)
My AMD 2700+ is 300-400 seconds slower running 4.17 v. 4.16.
Norm
RE: Comparing 4.01 - 4.16 -
)
Now I have 5 results by app 4.17. Avg. time is 25303 sec (min. 24771 - max. 26067 sec) that seems 4.17 is about 0,5% faster on My Athlon64 2800+ (socket 754) but by 4.16 was time of crunching between 24478 and 25920 sec.
On my second computer (cpu
)
On my second computer (cpu Intel(R) Pentium(R) M processor 1.60GHz) I have 44 WUs h1_0335.0 (credit 19,55) crunched by 4.17 avg. time 3448 sec. min. time 3436 sec and max. time 3455 sec. On this comp I have only 1 Wu crunched by 4.01 and this is WU h1_0728.5 (credit 174,29) time 38975 sec. So by 4.01 I got 16,1 credits per hour and by 4.17 I got 20,4 credit per hour
On my AMD Athlon 64 3000+
)
On my AMD Athlon 64 3000+ machine the time wnt from about 2950 with 4.01 to about 2550 with 4.17.
JLP's Blog | Get Firefox
I'm seeing about a 14%
)
I'm seeing about a 14% speedup with 4.17 compared to the stock app on my coppermine celeron 800. Times for short wus are around 7800 sec/WU.
Regards,
M
I went back to 4.16 on my
)
I went back to 4.16 on my Pentium II (no SSE) and things look better. Memory usage is much higher than on 4.17 and I think this is a good sign.
Tullio
Good afternoon
)
Good afternoon crunchers
AuthenticAMD AMD Athlon64 3500+ Linux 2.6.13-15.8-default (Suse 10.1)
Computation times ( in seconds )
v4.01 average: 55562,95 v4.01 std dev: 836.75 [sample of 15 results]
v4.16 average: 44060,07 v4.16 std dev: 56.75 [sample of 3 results]
v4.17 average: 44878,15 v4.17 std dev: 156.48 [sample of 4 results]
Compare v4.16/v4.01 = 20.70% faster
Compare v4.17/v4.01 = 19.23% faster
Compare v4.17/v4.16 = -1.86% slower
Well... I still think v4.17 is a bit slower on this machine, but I haden't enough data under 4.16 to make an accurate comparison.
Question at 0.001 credit ... Do you guess what my dilemma is ? :-)
"Vanitas, vanitatum et omnium vanitas]"
Ecclesiast
"Entia non sunt multiplicandam praeter necessitatem"
(OKHAM)
589494 is an AMD Athlon 1200+
)
589494 is an AMD Athlon 1200+ processor running Slackware 10.2 with kernel of 2.4.31. It has had no issues with validation, but when going from 4.16 to 4.17, it has slowed somewhat. So, on older AMD Athlons, it seems to be less efficient than 4.16. 4.16, however, was significantly faster than 4.01. We appreciate your hard work on making our machines more efficient in their contributions.
Good afternoon,679091
)
Good afternoon,
679091 AuthenticAMD mobile AMD Athlon XP-M 2400+ Linux 2.6.8-24.24-default (Suse 9.2)
Computation times ( in seconds )
v4.01 average: 31881.45 v4.01 std dev: 775.56 [sample of 15 results]
v4.16 average: 27903.87 v4.16 std dev: 301.39 [sample of 5 results]
v4.17 average: 27759.45 v4.17 std dev: 343.96 [sample of 8 results]
Compare v4.16/v4.01 = 12.48% faster
Compare v4.17/v4.01 = 12.93% faster
Compare v4.17/v4.16 = +0.52% faster
"Entia non sunt multiplicandam praeter necessitatem"
(OKHAM)
Hello all, my A64 3700 just
)
Hello all, my A64 3700 just started crunching WUs with the 4.17 beta app and here are some early results for small WUs:
4.01 avg. for last 5 results - 2362.6 s
4.17 avg. for first 5 results - 2007.2 s (all validated)
Approximate speed increase : 15%
In the stderr_txt ouput of all results crunched with 4.17 I am seeing something I have never seen before, have a look
What does 'small x' mean? The number of times it appears differs for each WU?
Great work Bernd, keep it up.
There are 10^11 stars in the galaxy. That used to be a huge number. But it's only a hundred billion. It's less than the national deficit! We used to call them astronomical numbers. Now we should call them economical numbers. - Richard Feynman