Which optimized client (I assume you mean application) is that, and where can one get it? I haven't noticed any being offered for Macs here in the six months since I joined.
1) athlon 64 3500+, Windows XP, long h-WUs: 19,3 C/h
2) athlon 64 3500+, Linux, short h-WUs: 26,7 C/h
3) Intel Celeron M, Linux, short h-WUs: 15,6 C/h
4) Athlon Xp 2200+, Linux, short l-WUs: 21,7 C/h
it wonders me much, why i get more credits with 4) than with 1).
and i just can't compare, it could be:
a) the faster Linux Client
b) generaly more Credits for short WUs
c) generaly more Credits for l-WUs
or all three together... no clue :( But in all benchmarks the 3500+ scores about twice as good as the 2200+......
edit: in general amd seems to be better for einstein
I thought credits are server generated and not platform and operating system dependent. Very very strange!!
yes, the credits for one WU are fixed. I just calculated how much Credits i get in one hour ( Credits * 3600 / Crunchtime ).
And i still wonder why the 2200+, linux, short l-WUs gets more than the 3500+, windows, long h-WUs :(
Ill now let the 3500+ only crunch under Linux und wait till i get long h-WUs there too, then i can see how much the linux client makes about this....
Which optimized client (I assume you mean application) is that, and where can one get it? I haven't noticed any being offered for Macs here in the six months since I joined.
My older box will probably be project hopping in the near future since the credit differential is roughly double the difference in proccessor speeds. I knew it was hurting performance wise but didn't realize it was that bad until now.
Before everybody goes mad on these figures, I am not seeing a flat line graph for the different units being crunched. e.g.
I get approx 10cr for 45min units = 13.3/hr
20cr for 1 hr units = 20/hr
and 178cr for 10hr units =17.8/hr
This is on a Pent M 1.86GHz (SSE2)
Other computers in this house, not all on same account, also show the non-linearity.
I have two windows computers, one is an AMD Athlon XP 1700+ (computer 417623) and the other is an Intel P4 2GHz (computer 217888).
The AMD definitively is SSE capable and the P4 (P4 Northwood) also is SSE capable.
Here are my results:
AMD: 51,365 sec for 179.15 credits (12.5 cr/h) see this WU
P4: 70,200 sec for 30.62 credits (1.6 cr/h)
see this WU (not yet uploaded due to network constraints, but my quorum partner claimed 30.62 credits.
Do you have any idea what is going on here?
[edit]just noticed that this quorum computer 430747 is claiming far to less credits![/edit]
[edit2]...remains the fact that Athlon XP1700+ is 30% faster than Pentium 4 2Ghz![/edit2]
Yes, your quorum partner is using BOINC 4.43 which is not flops aware so you can completely disregard that claim. There is no way you will get that low as the server will grant you both the true value based on the actual work content. You are using 5.2.13 which is flops aware. I would expect that your claim will be much higher and you wont have a problem.
Also, your computer #217888 is not an Intel P4 at all. It is listed as an AMD 1700+. You don't seem to have any P4 2.0G Northwoods in your complete list of 7 machines. It's still working on S4 stuff and hasn't even started on the S5 yet. BTW, when it does start you are probably going to be in deadline trouble as you have left your cache too large and your BOINC client doesn't know (yet) how long these new results are going to take. It will only find that out when it finishes the first one and then there will be a sudden large re-evaluation upwards of how long all the others are going to take. You are almost certain to have work that you cannot complete within the deadline. You should reduce your cache to less than 1.0 days and update ASAP.
RE: My data: […]
)
Which optimized client (I assume you mean application) is that, and where can one get it? I haven't noticed any being offered for Macs here in the six months since I joined.
RE: hm... with einstein i
)
I thought credits are server generated and not platform and operating system dependent.
Very very strange!!
Athlon
Stay tuned and keep crunching
RE: I thought credits are
)
Credits are of course server generated.
BUT credits/hour is platform/CPU dependant...
Udo
yes, the credits for one WU
)
yes, the credits for one WU are fixed. I just calculated how much Credits i get in one hour ( Credits * 3600 / Crunchtime ).
And i still wonder why the 2200+, linux, short l-WUs gets more than the 3500+, windows, long h-WUs :(
Ill now let the 3500+ only crunch under Linux und wait till i get long h-WUs there too, then i can see how much the linux client makes about this....
RE: RE: My data: […]
)
Odysseus,
you can get the optimized client 5.4.9 here: http://members.dslextreme.com/~readerforum/forum_team/boincbeta.html
Be aware that this does not have any influence on the credit anymore, it just delivers higher benchmark results since it is Altivec-Enhanced.
Intel Pentium P4 (Northwood -
)
Intel Pentium P4 (Northwood - 512 kB L2 Cache) 3GHz (HT on);
Win XP SP2; long WUs:
2x 9,7 Credits/hour = 19,4 Cr/h
A64x2 @ 2.6gig:
)
A64x2 @ 2.6gig: 41.5credits/hour
Athlon (nonSSE) @ 1.48gig: 6.2credits/hour
My older box will probably be project hopping in the near future since the credit differential is roughly double the difference in proccessor speeds. I knew it was hurting performance wise but didn't realize it was that bad until now.
Before everybody goes mad on
)
Before everybody goes mad on these figures, I am not seeing a flat line graph for the different units being crunched. e.g.
I get approx 10cr for 45min units = 13.3/hr
20cr for 1 hr units = 20/hr
and 178cr for 10hr units =17.8/hr
This is on a Pent M 1.86GHz (SSE2)
Other computers in this house, not all on same account, also show the non-linearity.
Andy
there must be something
)
there must be something horribly wrong...
I have two windows computers, one is an AMD Athlon XP 1700+ (computer 417623) and the other is an Intel P4 2GHz (computer 217888).
The AMD definitively is SSE capable and the P4 (P4 Northwood) also is SSE capable.
Here are my results:
AMD: 51,365 sec for 179.15 credits (12.5 cr/h) see this WU
P4: 70,200 sec for 30.62 credits (1.6 cr/h)
see this WU (not yet uploaded due to network constraints, but my quorum partner claimed 30.62 credits.
Do you have any idea what is going on here?
[edit]just noticed that this quorum computer 430747 is claiming far to less credits![/edit]
[edit2]...remains the fact that Athlon XP1700+ is 30% faster than Pentium 4 2Ghz![/edit2]
Udo
RE: Do you have any idea
)
Yes, your quorum partner is using BOINC 4.43 which is not flops aware so you can completely disregard that claim. There is no way you will get that low as the server will grant you both the true value based on the actual work content. You are using 5.2.13 which is flops aware. I would expect that your claim will be much higher and you wont have a problem.
Also, your computer #217888 is not an Intel P4 at all. It is listed as an AMD 1700+. You don't seem to have any P4 2.0G Northwoods in your complete list of 7 machines. It's still working on S4 stuff and hasn't even started on the S5 yet. BTW, when it does start you are probably going to be in deadline trouble as you have left your cache too large and your BOINC client doesn't know (yet) how long these new results are going to take. It will only find that out when it finishes the first one and then there will be a sudden large re-evaluation upwards of how long all the others are going to take. You are almost certain to have work that you cannot complete within the deadline. You should reduce your cache to less than 1.0 days and update ASAP.
Cheers,
Gary.