Why are, for example, Power Macs getting more points or credits for processing at around the same rates as a say AMD MP 1800+?
Example:
Power Mac: 7786 sec, 136.89 credits
AMD MP1800+: 8892 sec, 25.77 credits
Why?? Shouldn't they get the same?
Copyright © 2024 Einstein@Home. All rights reserved.
Unequal credits between systems for processing at the same rates
)
Their respective boinc clients probably benchmarked the cpu's differently. This tends to happen when mixing different Operating Systems.
such things just should not be writ so please destroy this if you wish to live 'tis better in ignorance to dwell than to go screaming into the abyss worse than hell
RE: RE: Why are, for
)
I'm not talking benchmarks, but getting equal credit for processing at the same amount of time.
Shouldn't the two examples have around the same amount of credits?
RE: Shouldn't the two
)
Bionc Wiki - Computation of Credits
It's all about the benchmarks. 7786 sec.=136.89 credits, that is some major tweak and peak.
It will be the high claim and
)
It will be the high claim and not used for the granted credit. So it is not really a problem.
:)
98SE XP2500+ @ 2.1 GHz Boinc v5.8.8
RE: RE: Shouldn't the
)
apparently some major tweak and peak does influence the result consensus if the workunit is small....
Example of granted credit on a possibly tweak system/optimised client
pay attention to the second result ID....
RE: RE: RE: Shouldn't
)
With the second ID, while the granted credit is the same for all three, you are doing the processing in less time, so you do gain something between 3 to 5 times.
RE: With the second ID,
)
So am I to understand that Mac users are looking at the possibility that they will be sidelined in terms of processing times and credit claims should Einstein@home introduces clients with prodigious Akosf's optimisations for the x86 platform..., just like what you've lamented when comparing the currently unoptimised x86-client benchmarks against the Altivec-optimised PPC-clients'...?
RE: So am I to understand
)
Sounds like much the same thing, as the AltiVec-optimized code has given G5 users a large advantage for the past 6-8 months, though it didn't do anything for the non-G5 Macs, which are the majority of Macs crunching here.
I'm sure that someone will do much the same for Macs - they are, after all, a brilliant and proud lot. :-)
microcraft
"The arc of history is long, but it bends toward justice" - MLK
RE: Sounds like much the
)
Someone named Bruce infact. Once they upgrade thier servers to handle the increased load he's promised new official versions of the client based on akos's original improvements. Some of the others might be in it as well since they've been corresponding back and forth on algorithms.
RE: Sounds like much the
)
The G4 PPCs have AltiVec as well, so I don't see how that would make the difference in comparison to G5s. All I know is that my 400-MHz G4s seem to get through WUs in a similar time to Intel & AMD hosts with several times the clock speed—or, as is more often the case, two or three times as long as other systems with seven or eight times the MHz rating.