miahallen Posted December 26, 2009 Share Posted December 26, 2009 (edited) Have you guys ever witnessed a bug in wPrime? I only ask because I was working out my eff in wPrime 32M when I noticed that in order to match Hicookie's result on the i5 750, I would need 100MHz more than what he ran at. So I compared the other top 10 results, and his is the only one way out of line: The #s on the graph represent the difference in (hundredths of a second) between the actual time posted, and the theoretical time based on the average eff. For reference see columns D & F. average eff = 30484.59 divided by Hicookie's time of 5.74s = 5311MHz...110MHz higher than what he was running. I'm not posting this to question Hicookie's score (the theoretical score wouldn't change the ranking anyway), but to question the accuracy of wPrime as a reliable bench. I have also seen in my own testing where errors occurred during wPrime calculation, and yet it still completed the bench as if nothing was wrong. Those scores were questionable in my mind (and they were never my best)...but I wonder if we need a fixed version of wPrime that keeps better track of bugs. edit - in the calculation above, I omitted Hicookie's result from the calculation of avg eff to avoid throwing off the avg with a score that was possibly inaccurate. Edited December 26, 2009 by miahallen Quote Link to comment Share on other sites More sharing options...
miahallen Posted December 26, 2009 Author Share Posted December 26, 2009 Doubt it...his SPI SS was captured just fine at 5.4GHz Quote Link to comment Share on other sites More sharing options...
knopflerbruce Posted December 26, 2009 Share Posted December 26, 2009 This benchmark hasn't been as thoroughly tested as superpi32m for example, so there may be some clever tweaks that alot of people don't know about. Quote Link to comment Share on other sites More sharing options...
miahallen Posted December 26, 2009 Author Share Posted December 26, 2009 Very true kb...hmmmm Quote Link to comment Share on other sites More sharing options...
Oj0 Posted December 27, 2009 Share Posted December 27, 2009 Knobflerbruce and I know a tweak which adds 10% efficiency, perhaps hicookie knows it too Quote Link to comment Share on other sites More sharing options...
knopflerbruce Posted December 27, 2009 Share Posted December 27, 2009 Knobflerbruce and I know a tweak which adds 10% efficiency, perhaps hicookie knows it too If it works on i7, it doesn't on K10. And i'm not even sure the right option is avaliable either. Quote Link to comment Share on other sites More sharing options...
Gautam Posted December 27, 2009 Share Posted December 27, 2009 Did you try 7 Miah? On i7 it's inferior, but perhaps somehow with 4 threads it's in the lead like it is with Core 2. I notice everyone but hicookie is on XP in the top 5. Quote Link to comment Share on other sites More sharing options...
miahallen Posted December 27, 2009 Author Share Posted December 27, 2009 Hum...I'll play with it Quote Link to comment Share on other sites More sharing options...
Oj0 Posted December 27, 2009 Share Posted December 27, 2009 If it works on i7, it doesn't on K10. And i'm not even sure the right option is avaliable either. I'm just saying, perhaps he has found the Intel equivalent? And yes, that option is available in most BIOSes, I'll drop you a message at the forums in a mo. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.