PCMark04 fix for Conroe: post if interested

A program i`ve asked to be written appears to have solved to about 90% the problem. Multi-thread test 3 suffers, but apart from that...works perfectly. I`m not allowed to distribute the software, but if you want it to be available, please post and say so

Kenny


This software cannot be seen as cheating. It affects core affinity- for multi-thread test 3 the CPU acts as a single-core. Thats not an unfair advantage...if anything its a disadvantage...but I get my score.

Release of the software is at the discretion of the person who coded it.

EDIT: Feb 2008. Have attached the software.Attachment 72580

Software coded by Ed Wilk. Would be good manners to credit him in all PCM04 scores using this software. Im optional. I just had the idea

It would be great to get this. We've been discussing at hwbot about removing the bm, but there should be no problem if this 'fix' comes available.

I want to know more about this for personal use. Subscribed. :)

I, too, would like to be able to run PC Mark 04.

yup yup this will be a good test.. :) I woould also like to try it.. :)

Would be very useful !

We waiting long time for this , i want to try too ;)

I want it always

http://www.geocities.com/edgemeal_software/

"Set affinity by time"

Thanks for showing interest, but still poor that only 8 guys posted here.

Ed Wilk coded it, if ya feel like thanking him in ORB results...up to yourselves

Kenny

Couldn't get it to work and get a score. How should we play with that tool ?

Multithreaded test 3 HAS to run as a single core to work.

I set:
CPU 0+1 (1 minute)
CPU 0 (1 minute)
CPU 0+1

I start the bench 15 seconds after the first CPU 0+1 instruction starts. Make sure its set to run on PCM04 in the box at the top ;)

Thanks for the fix. :toast:

Works for me :banana: :

http://service.futuremark.com/compare?pcm04=3278429

http://img148.imageshack.us/img148/2636/pcmrk04qc8.jpg

Thanks K404 and Ed Wilk! :toast:

I need easier instructions please...When I run it I get this error. How do I fix it?

Looks like Grammar Test is still erroring.

Quote:

---

Originally Posted by K404

Multithreaded test 3 HAS to run as a single core to work.

I set:
CPU 0+1 (1 minute)
CPU 0 (1 minute)
CPU 0+1

I start the bench 15 seconds after the first CPU 0+1 instruction starts. Make sure its set to run on PCM04 in the box at the top

---

Try 16 seconds :) That really is all ya need to make it work :)

I don't understand...Where exactly do I input this???
I set:
CPU 0+1 (1 minute)
CPU 0 (1 minute)
CPU 0+1

PM`d ;)

Will this work fo any dual core, currently I'm still playing with opty's and have been unable to run pcmark 04

the software will work with any dual-core, yes. It *should* solve any dual-core issues with PCM04, but i`ve only tried it on a C2D

The grammar test doesn't seem to pass with C2D even with a single core with high enough speed.

Hi! If you`re trying the affinity software at a high clockspeed, you might need to change the start point of the bench. Try 17 or 18 seconds after the first CPU 0+1 execution :)

If you`ve set affinity in task manager to completely single-core for the bench and it errors out again, I cant think of a solution, sorry :(

Yah, won't pass with a single core as i said. But the tweak is good up to 3.7GHz or sth.

:confused: :confused: How about a screenshot for the settings...I am an idiot & still don't get it..

I'm pretty sure I get how it's done, but with the cpu @ 4.5 Ghz it just will not work, grammartest (test 3 part 2) fails anyway.

Did a lot of testing/trying with one of my X2 3800+ rigs.
No way that the program did set affinity as it is supposed to do.
Added start-times for affinity set for PCMark04 (ready for bench) as described (3 "entries" in left pane).
Started bench.
No update in history (right) pane.. and AMD Powermonitor cleraly showed that no "switch" to single core occured at scheduled time.
DOuble-checked with setting affinity manually with Task Manager, which made test pass (low score of course) with AMD Powermonitir showing that only one core was used.

Either I have misunderstood the way the program should work/be used.. or there is a bug.