Low Benchmark Result With GridRepublic

  
Stephen Schroder
ID: 40050
Posts: 4

When I was running my projects under BOINC 6.6.36 my floating point benchmark was ~1744 MIPS.  With GR 5.10.30 my FP benchmark result is ~1145 MIPS.  Any thoughts on this?

 Steve

(Monday, September 07 09, 8:55 PM)
Spark
ID: 15261
Posts: 141

Thoughts on the varyng benchmark numbers.

  • The current GridRepublic distribution is based on 5.10.30 is an older version of BOINC.  There may have been a change as how the benchmarks are conducted since version 5.10.30 of BOINC compared to the latest 6.6.36 distriubtion. 
  • If there was no change in the benchmark code, activity on a computer while the benchmarks are running can impact the final numbers. Either in the background or activity by the user.

BOINC Versus GridRepublic Information

(Tuesday, September 08 09, 1:12 AM)
Stephen Schroder
ID: 40050
Posts: 4

Thanks for the reply, Spark.  When I conduct benchmarks I make sure there are no other controllable activities on my computer.  It could be that the algorithms are different, but since benchmark algorithms are synthetic I may not see any difference in real world computation times.  Time will tell.

 Steve

(Tuesday, September 08 09, 2:53 AM)
Post   Previous   Page 1 of 1   Next