I recently upgraded to version 5.2.13 from version 4.19. The CPU bencmark figures have changed significantly: reduction in Dhrystone result of approx 36%.

PC is P4-3.6GHz, Multithreading using WinXP-SP2. 2Gbyte RAM, disks are 150M-SATA. Hardware configuration unchanged since I began running CPDN last year.

Under V4.19 typical result was:
\"2006-05-01 02:09:49 [---] Benchmark results:
Number of CPUs: 2
1533 double precision MIPS (Whetstone) per CPU
2840 integer MIPS (Dhrystone) per CPU\"

Under 5.2.13 typical result is:

\"2006-05-06 02:59:15 [---] Benchmark results:
Number of CPUs: 2
1531 double precision MIPS (Whetstone) per CPU
1819 integer MIPS (Dhrystone) per CPU\"

I retested my PC using \"Passmark Performance test\" which reported very similar results to when PC was new (before XP-SP2 \"upgrade\") so I suspect that the PC is performing the same as before, but the Dhrystone results reported are different for some reason.
Is this a problem with 5.2.13? (Re-Installed as \"single user\" - no different to installing as \"server\").

Simon

---

The 5.* versions use a different algorithm to work out benchmarks than earlier versions. And they\'re still useless for this project, and still not used.
I went from version 4.05 to 4.25, so I don\'t know what they would have been like with 4.19; however, with 5.2.8, my integer has jumped a fair bit:

Benchmark results: (4.25)
1428 double precision MIPS (Whetstone) per CPU
1230 integer MIPS (Dhrystone) per CPU

Benchmark results: (5.2.8)
1719 double precision MIPS (Whetstone) per CPU
3538 integer MIPS (Dhrystone) per CPU

The 5.* versions use a different algorithm to work out benchmarks than earlier versions. And they\'re still useless for this project, and still not used.
I went from version 4.05 to 4.25, so I don\'t know what they would have been like with 4.19; however, with 5.2.8, my integer has jumped a fair bit:

Benchmark results: (4.25)
1428 double precision MIPS (Whetstone) per CPU
1230 integer MIPS (Dhrystone) per CPU

Benchmark results: (5.2.8)
1719 double precision MIPS (Whetstone) per CPU
3538 integer MIPS (Dhrystone) per CPU

Also the benchmarks in the windows version of 4.19 were badly bugged, the compiler \"optimized\" most of the code out since it did not have any output.

---

BOINC WIKI

BOINCing since 2002/12/8

Thanks Les and John for the quick replies.

I guess that my machine is indeed running OK and as you suggest the BOINC benchmarks are a bit misleading.

Simon

---