less credit with 4.25 boinc manager?
7 Mar 2005 7:55:36 UTC
Topic 188334
(moderation:
HI
I have just changed from boinc version 4.19 to 4.25. For the first two results since the "upgrade" the average credit claimed for my computer's WU's has dropped from 85/88 to 67/68. (boo-hoo, stamp-stamp, tantrum) I ran the benchmark test and discovered the measured interger speed has fallen from roughly 4800 to 3430. I can't remember what the old floating point speed was but i'm pretty sure it has also dropped. The actual seconds taken to finish a WU shown on the results list is still the same.
Windows xp home fully updated, computer id 17944
you can lead a horse to water but a pencil must be led
less credit with 4.25 boinc manager?
)
HI
Just to confirm my third WU is also claiming lower credit than before the "upgrade", but it still took the same time to complete as before.
you can lead a horse to water but a pencil must be led
With Boinc v4.25 the
)
With Boinc v4.25 the benchmark results for Windows have been reviewed. As a result of this, the numbers of benchmark are lower compared to the previous version of Boinc.
This is what I got:
with Boinc v4.19
--- - 2005-01-26 18:35:17 - Version Change Detected (4.13 -> 4.19); running CPU benchmarks
--- - 2005-01-26 18:36:28 - 1282 double precision MIPS (Whetstone) per CPU
--- - 2005-01-26 18:36:28 - 1935 integer MIPS (Dhrystone) per CPU
with Boinc v4.25
5/03/2005 13:26:15 - Version Change Detected (4.19 -> 4.25); running CPU benchmarks
5/03/2005 13:27:17 - 1277 double precision MIPS (Whetstone) per CPU
5/03/2005 13:27:17 - 1092 integer MIPS (Dhrystone) per CPU
Due to the fact credits are calculated taking into acoount the CPU time and the benchmark result it is normal that the claimed credit is lower with v4.25 against v4.19 as the credit is a multiplication of CPU time and benchmark numbers.
Greetings from Belgium
Thierry
Hi T Van, Thanks
)
Hi T Van,
Thanks for the information
So the moral for the anally obsessed results freaks out there is don't upgrade. Bet quite a few revert back to 4.19 when they discover this.
I was worried something had gone wrong, resulting in a slower cpu and was ready to delete the program and start anew.
you can lead a horse to water but a pencil must be led
> So the moral for the
)
> So the moral for the anally obsessed results freaks out there is don't
> upgrade. Bet quite a few revert back to 4.19 when they discover this.
> I was worried something had gone wrong, resulting in a slower cpu and was
> ready to delete the program and start anew.
Well, you shouldn't be worry about that. Have a look to the FAQ's concerning the granted credit against claimed credit.
I switched over to v4.25, having more capabilities compared to the v4.19. Looks running fine since 2 days without noticable problems.
Greetings from Belgium
Thierry
Hi T Van I understand how
)
Hi T Van
I understand how the granted credit system works. My point was, because 4.19 will still run the old benchmark test ; when two or three machines running 4.19 report in the credit granted will be about 85 ; when two or three machines running 4.25 report in the credit granted will be about 67 (using my machine as a nominal template). The glory hunters among us will see this as a disincentive to upgrading BOINC. I still think some who have upgraded will revert back to 4.19 after realising this situation.
My main concern was that something had gone wrong with the instal of 4.25 causing my machine to run slower, thus the lower benchmark figure . Your info about the altered benchmark standard cleared up any worry on that score, plus of course the fact that it still took much the same time to complete a WU as before.
you can lead a horse to water but a pencil must be led
> I understand how the
)
> I understand how the granted credit system works. My point was, because 4.19
> will still run the old benchmark test ; when two or three machines running
> 4.19 report in the credit granted will be about 85 ; when two or three
> machines running 4.25 report in the credit granted will be about 67 (using my
> machine as a nominal template). The glory hunters among us will see this as a
> disincentive to upgrading BOINC. I still think some who have upgraded will
> revert back to 4.19 after realising this situation.
IMHO, sooner or later everybody will have to upgrade to 4.25 once it is not a development version anymore. So at that moment there will be no discrepancy anymore.
Greetings from Belgium
Thierry