Bug.... Incorrect computation of Benchmark and credit given

Shaitan
Shaitan
Joined: 15 Mar 05
Posts: 14
Credit: 67595
RAC: 0
Topic 188648

I first started looking at this in another thread.

Another cruncher reported this:
AAdjuster's A64 cpu at 2650 MHz
Benchmark results:
Number of CPUs: 1
2561 double MIPS (Whetstone) per CPU
6991 integer MIPS (Dhrystone) per CPU
Finished CPU benchmarks

AAdjuster's WU's completed
2111885 555363 19 Mar 2005 23:13:51 UTC 20 Mar 2005 8:48:34 UTC Over Success Done 16,812.55 92.59 pending

2105486 521654 19 Mar 2005 18:33:36 UTC 20 Mar 2005 4:08:22 UTC Over Success Done 16,921.39 93.19 93.19

2099553 552594 19 Mar 2005 12:37:39 UTC 19 Mar 2005 23:13:51 UTC Over Success Done 16,853.52 92.82 pending

My Results FX53 2639 Mhz

Benchmark results:
Number of CPUs: 1
2538 double precision MIPS (Whetstone) per CPU
4681 integer MIPS (Dhrystone) per CPU
Finished CPU benchmarks

Recent results at said speed:

2207550 81562 21 Mar 2005 1:32:50 UTC 21 Mar 2005 10:58:16 UTC Over Success Done ...16,869.13... 70.56 pending

2187045 572354 20 Mar 2005 20:50:56 UTC 21 Mar 2005 6:15:30 UTC Over Success Done ..16,876.17... 70.59 pending

2186978 572338 20 Mar 2005 10:29:02 UTC 20 Mar 2005 23:16:56 UTC Over Success Done ...16,975.91... 71.00 pending

You can see by the benchmarks scores the benchmark for the client is flawed....
The AAdjuster's Dhrystone benchmarks are out of wack.

The results are clearly at odds with each other.
Another interesting thing is that AAduster's credits are inflated even though the times are nearly identical..

My claimed credts all range in the 70's area where all of his/hers are in the 90's
very strange indeed.

Saenger
Saenger
Joined: 15 Feb 05
Posts: 403
Credit: 33009522
RAC: 0

Bug.... Incorrect computation of Benchmark and credit given

> My claimed credts all range in the 70's area where all of his/hers are in the
> 90's
> very strange indeed.

Am I right, that the other om´ne crunches with the outdated client 4.19 on a windoze machine?
And you are using 4.25 and/or another OS?

This is a known (and disposed) bug from the earlier clients. Windows used somehow shortcuts on the benchmarks, and get therefore better Dhrystones, and claimed therefore more credit.

Grüße vom Sänger

Shaitan
Shaitan
Joined: 15 Mar 05
Posts: 14
Credit: 67595
RAC: 0

I'm using the 4.25 client on

Message 8953 in response to message 8952

I'm using the 4.25 client on WInXP as I just joined a few days ago and it was the only one available as far as I could see.

If it's a known bug, why are they still accepting WU's from the 4.19 client?

Saenger
Saenger
Joined: 15 Feb 05
Posts: 403
Credit: 33009522
RAC: 0

> I'm using the 4.25 client

Message 8954 in response to message 8953

> I'm using the 4.25 client on WInXP as I just joined a few days ago and it was
> the only one available as far as I could see.
> If it's a known bug, why are they still accepting WU's from the 4.19 client?

Because Boinc is a multi project project ;)

Seti recommends 4.25, and 4.19 is not available any longer @Boinc, unless you know the deep link.
But as 4.19 still delivers the correct scientific results, and 4.2x is still dev-stuff for other-than-windoze, some projects still have it in their DL-stock, perhaps not to confuse the n00bs.
It's just a inconvenience from the scientific POV, so it's a bit down on the to-do-list.

Grüße vom Sänger

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.