It is about twice slower than optimized S-39L for windows on similar machine.
(Athlon XP 2800+ (linux) to Sempron 2800+ (Windows))
It's not clear to me that you can compare CPU times. From the average time it takes for the two computers to finish similar loads it would seem that the speed is not that much different. Granted, I don't know what else your two systems are doing. But from experience, when I had my windows and linux running the standard clients, linux would finish in less wall time comparable units, and both are workstations. I would even argue that the linux box was doing more things (encoding/watching movies, compiling kernels) than the windows machine (mostly typing/compiling)
It is a bit off-topic, but I thought you would like to hear that you can run your beloved Linux E@H bits also under Windows with coLinux. Even the graphics do work, albeit deadly slow. My first (and only) result is at http://einsteinathome.org/host/604036
Performancewise it is reported to be suprisingly quick, if the CPU time reported is true then it is not much above my Akosf results if this Intel p4 1.8MHz machine. Admittedly my personal feeling is that it is slower. What really frightened me is that the performance reported in the boinc-client is at about 50%: Linux 436.26/1027.49 million ops/sec vs Windows 916/1781.32 million ops/sec, floating/integer, respectively.
Whatever, www.colinux.org is good fun, the X server is of cygwin.com.
I think the reason if this problem the low-precision sin/cos lookup table.
I will suggest to Brend the usage of my interpolation algorithm.
It gives about ~2300000 times more accurate sin/cos values an the speed is about the same.
Hey, since a fortnight has passed, I was wondering if there are any news on that front? Bernd, have you managed to integrate akosf's suggestion or are there more obstacles to it?
Cheers
:
your thoughts - the ways :: the knowledge - your space
:
I think the reason if this problem the low-precision sin/cos lookup table.
Hey, since a fortnight has passed, I was wondering if there are any news on that front? Bernd, have you managed to integrate akosf's suggestion or are there more obstacles to it?
To change the precision needs very big circumspection. Usually the validator gives 0 credit to the more accurate algorithms, because the results of the two others are same (but less accurate).
albert_4.55_i686-pc-linux-gnu
)
albert_4.55_i686-pc-linux-gnu
I have only 5 invalid results out of 90 results
25279882
25327326
25070117
24837662
24774141
It is about twice slower than optimized S-39L for windows on similar machine.
(Athlon XP 2800+ (linux) to Sempron 2800+ (Windows))
Just had my first validate
)
Just had my first validate error using 4.55. The previous errors were cleared up by dropping the graphics capability.
Found 3 invalid results out
)
Found 3 invalid results out of approx. 50.
25656010
25765717
25765723
All were checked against standard apps. from different platforms (Linux/Windows or Linux/Mac).
Michael
Team Linux Users Everywhere
OK, of 21 completed with
)
OK, of 21 completed with Albert v4.55 I've had one invalid result. Seventeen passed and three are pending.
Nothing travels faster than the speed of light with the possible exception of bad news, which obeys its own special laws.
Douglas Adams (1952 - 2001)
RE: It is about twice
)
It's not clear to me that you can compare CPU times. From the average time it takes for the two computers to finish similar loads it would seem that the speed is not that much different. Granted, I don't know what else your two systems are doing. But from experience, when I had my windows and linux running the standard clients, linux would finish in less wall time comparable units, and both are workstations. I would even argue that the linux box was doing more things (encoding/watching movies, compiling kernels) than the windows machine (mostly typing/compiling)
RE: albert_4.55_i686-pc-lin
)
Is that a K7 or K8 based Sempron?
I have some invalid results
)
I have some invalid results too
http://einsteinathome.org/workunit/7248285
http://einsteinathome.org/workunit/7164144
http://einsteinathome.org/workunit/7061952
http://einsteinathome.org/workunit/6985064
http://einsteinathome.org/workunit/6980760
http://einsteinathome.org/workunit/6887120
An important point is, that all invalid results are procced by old Athlon 64 processors with Clawhammer Core (130nm).
My newer Dual Core Toledo and Winchester have done only valid results.
It is a bit off-topic, but I
)
It is a bit off-topic, but I thought you would like to hear that you can run your beloved Linux E@H bits also under Windows with coLinux. Even the graphics do work, albeit deadly slow. My first (and only) result is at
http://einsteinathome.org/host/604036
Performancewise it is reported to be suprisingly quick, if the CPU time reported is true then it is not much above my Akosf results if this Intel p4 1.8MHz machine. Admittedly my personal feeling is that it is slower. What really frightened me is that the performance reported in the boinc-client is at about 50%: Linux 436.26/1027.49 million ops/sec vs Windows 916/1781.32 million ops/sec, floating/integer, respectively.
Whatever, www.colinux.org is good fun, the X server is of cygwin.com.
Cheers,
Steffen
RE: I think the reason if
)
Hey, since a fortnight has passed, I was wondering if there are any news on that front? Bernd, have you managed to integrate akosf's suggestion or are there more obstacles to it?
Cheers
:
your thoughts - the ways :: the knowledge - your space
:
RE: RE: I think the
)
To change the precision needs very big circumspection. Usually the validator gives 0 credit to the more accurate algorithms, because the results of the two others are same (but less accurate).