| View previous topic :: View next topic |
| Author |
Message |
Dennis
Guest
|
 Posted: Thu Apr 26, 2007 12:19 am Post subject: Fastcode Trim B&V 0.6.3 |
 |
|
Hi Everybody
As soon as I have the remaining benchmark results I can adjust the
subbenchweigths.
Then we know which functions can be removed and bench stability will look
better that with 220 functions.
Thanks to CodeGear I am able to build with D2007 in the next release. I have
used D2006 so far.
I think that the B&V is beginning to look like a 1.0.0. Bench stability is
nearly OK (Se benchstability.xls in release 0.6.2 - the tests labeled 4 are
for the latest release - and make a few extra tests). I think the validation
is quite good. I also think that the benchmark is realistic. The function
size file is complete and so is the build all system. FastMM4 validation is
also included.
Opteron bench stability is awfull but I think my PC is to blame. Could
somebody try run bench test on another AMD 64 (X2) ? I get 1.4-1.9 % max
errors on the other PC's. Opteron is at 17.5% !!!!
Best regards
Dennis Kjaer Christensen |
|
| Back to top |
|
 |
Jouni Turunen
Guest
|
| Posted: Thu Apr 26, 2007 2:03 pm Post subject: Re: Fastcode Trim B&V 0.6.3 |
|
|
Hi Dennis,
| Quote: |
Opteron bench stability is awfull but I think my PC is to blame. Could
somebody try run bench test on another AMD 64 (X2) ? I get 1.4-1.9 % max
errors on the other PC's. Opteron is at 17.5% !!!!
|
Just uploaded Northwood results. Bench test shows 2.0% on Presler,
will start bench test on X2 when I leave office. At the moment
I don't have access to any Core 2 Duo pc.
I'm still pretty much away from Fastcode stuff due to personal
reasons but I try to find time to run benchmarks. Situation will
clear and I will return some day :)
If anyone needs to contact me by email, please use gmail address.
Regards,
Jouni
--
The Fastcode Project: http://www.fastcodeproject.org/ |
|
| Back to top |
|
|
Jouni Turunen
Guest
|
| Posted: Fri Apr 27, 2007 8:11 am Post subject: Re: Fastcode Trim B&V 0.6.3 |
|
|
Hi Dennis,
| Quote: |
Opteron bench stability is awfull but I think my PC is to blame. Could
somebody try run bench test on another AMD 64 (X2) ? I get 1.4-1.9 % max
errors on the other PC's. Opteron is at 17.5% !!!!
|
X2 run overnight shows 3%, a lot better than your 17.5.
Regards,
Jouni
--
The Fastcode Project: http://www.fastcodeproject.org/ |
|
| Back to top |
|
|
Dennis
Guest
|
| Posted: Fri Apr 27, 2007 6:53 pm Post subject: Re: Fastcode Trim B&V 0.6.3 |
|
|
Hi Jouni
Thanks.
2-3% is still not good enough and I hope we can get better numbers after
removing a few functions. 220 functions is very many.
Best regards
Dennis Kjaer Christensen |
|
| Back to top |
|
|
Dennis
Guest
|
|
| Back to top |
|
|
|
FAQ
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
