Pifast Benchmark

Printable View

Show 100 post(s) from this thread on one page

05-01-2003, 09:54 AM
reglisse

with the 4.2 version :)

http://membres.lycos.fr/reglissement.../pifast10m.jpg
05-01-2003, 10:07 AM
JCviggen

Quote:

Originally posted by reglisse
you uses Version 4.1 of PIFAST but we use the last Version 4.2

Quote:

**PIfast challange uses Version 4.1 of PIFAST to maintain a standard please use per download link and submit 'Total computation time'.

4.2 is shorter, and so its also a bit easier. So for the bench, only 4.1 counts. doesnt matter that a faster version exists, the fastest 4.1 time will be the fastest in 4.2 anyway, only its a little harder and requires more stability. In other words, a better challenge than 4.2
05-01-2003, 10:26 AM
reglisse

sorry but i'm french and my english is very bad ... :rolleyes:

i'm ok with you said, but for on my site we decide to take the 4.2 version for the bench and 50Millions of digits

the 4.1 version more difficult for the P4 ;)
05-01-2003, 10:59 AM
PMM

Me and M@tt decided to keep the 4.1 version due to the fact
that results and validation would tie up with the results that
were once part of Kain's Pifast Challange before it wen't Off
line.

Kains Pifast always used the 4.1 version as hosted on Hexus
together with the Hexus batch file for setting the parameters
to do the calculation.

When David at Hexus offered us the hosting at Hexus.net we
set everything up as per before so those that competed in Kain's
Pifast challenge were able to re-submit there last scores as per
Kain's and carry on as normal which I knew were valid from cached pages of Kains Site.

It was dicussed if we were to use 4.2 but everyone prefered sticking with 4.1.

So 4.1. it shall be :)
05-01-2003, 01:23 PM
CodeRed

Quote:

Originally posted by M@tt
:toast: CodeRed I will have to try a little harder tomorrow

Try 7-2-2-2, it worked for me.

I aint finished yet, the board is still stock and my prom still runs R134a.
05-01-2003, 01:56 PM
hipro5

Quote:

Originally posted by PMM
Me and M@tt decided to keep the 4.1 version due to the fact
that results and validation would tie up with the results that
were once part of Kain's Pifast Challange before it wen't Off
line.

Kains Pifast always used the 4.1 version as hosted on Hexus
together with the Hexus batch file for setting the parameters
to do the calculation.

When David at Hexus offered us the hosting at Hexus.net we
set everything up as per before so those that competed in Kain's
Pifast challenge were able to re-submit there last scores as per
Kain's and carry on as normal which I knew were valid from cached pages of Kains Site.

It was dicussed if we were to use 4.2 but everyone prefered sticking with 4.1.

So 4.1. it shall be :)

One question.....Is it possible that Version 4.1 is "more optimised" for an AMD CPU???.....Perhaps the V.4.2 is more accurate for BOTH Intel and AMD....Just a question..... :)
05-01-2003, 02:20 PM
JCviggen

actually I think 4.1 is more raw power code, whereas 4.2 "assists" the P4 in avoiding its weak points.... or you could say that its just optimized code which enables the P4 to make most of the optimisations
05-01-2003, 05:38 PM
xgman

Duration : 59.61 secondson 4.1.

so what does that mean anyway.
05-01-2003, 09:36 PM
reglisse

Quote:

Originally posted by hipro5
One question.....Is it possible that Version 4.1 is "more optimised" for an AMD CPU???.....Perhaps the V.4.2 is more accurate for BOTH Intel and AMD....Just a question..... :)

the V. 4.2 is "more optimised" for 512ko of the p4 ... it's all
05-01-2003, 10:23 PM
cpulloverclock

Quote:

Originally posted by JCviggen
I did some more tweaking today, and I took down the 50s mark :)

http://users.pandora.be/JCviggen/PI.JPG

Good job, nice oc
is it with ur 1700+?
05-02-2003, 12:29 AM
M@tt

Quote:

Originally posted by CodeRed
Try 7-2-2-2, it worked for me.

I aint finished yet, the board is still stock and my prom still runs R134a.

Will do, Thanks CodeRed

I also have a NF7-S V2.0 to try again and the Prommie is having a service today ;) I am still using the 8RGA+ at the moment.
05-02-2003, 01:36 AM
JCviggen

Quote:

Originally posted by cpulloverclock

is it with ur 1700+?

yup :)
the CPU can run this higher, but the FSB wont let me... doing NF7 cap mod today I think
05-02-2003, 01:50 AM
hipro5

Quote:

Originally posted by JCviggen
actually I think 4.1 is more raw power code, whereas 4.2 "assists" the P4 in avoiding its weak points.... or you could say that its just optimized code which enables the P4 to make most of the optimisations

Is it because some AMD guys want that???.....just another question.......If an Intel guy benches the V.4.2 and an AMD guy as well , the "deference" between those two is it going to be LEES(in seconds) than the V.4.1...?????....I think as the software is getting better we SHOULD go on with it..!!!....It's like you benching 3DMark2000 and NOT the 3DMark2001 or 2003.....Just wondering.....

EDIT : I was just at Pi-fast's page and I DIDN'T read ANYTHING about "optimized code which enables the P4 to make most of the optimisations" in V.4.2.....It JUST says as the V.4.1 version about a bit faster and repaired some bugs(same as they did at V.4.1).....I think that something "fishy" around here..... :stick:
05-02-2003, 02:55 AM
CodeRed

Ok hipro I'll bite (just this once), so whats your fastest v4.2 time
05-02-2003, 03:04 AM
reglisse

yes but with pifast v 4.2, 10M is too short, 50M is good no short no long ... look my site for the best time
05-02-2003, 03:31 AM
hipro5

Quote:

Originally posted by CodeRed
Ok hipro I'll bite (just this once), so whats your fastest v4.2 time

I don't have V.4.2 "fixed" bench to test it......If someone has it please give me a link to download it.....Thanks in advance..... :)
05-02-2003, 03:52 AM
reglisse

http://numbers.computation.free.fr/C.../download.html
05-02-2003, 03:55 AM
hipro5

Quote:

Originally posted by reglisse
http://numbers.computation.free.fr/C.../download.html

Thanks but.....I didn't ment that....I mean a "fixed" V.4.2 as the V.4.1 that you don't have to "put" numbers and the other things it asks you to run it........Just "click" and run..... ;)
05-02-2003, 04:01 AM
reglisse

for 10M or 50M of digits ?
05-02-2003, 04:06 AM
reglisse

:) sorry i don't understand all

for 50M of digit here :

http://membres.lycos.fr/reglissementhe/tele.php3

for 10M of digits

edit reglage.txt and change

4096 --> 1024

5000000 --> 1000000
05-02-2003, 05:14 AM
sysfailur

Very nice benches people! :) I just RMA'ed my a7n8x. We'll see what I can do with a rev 2.0 *rawr*
05-02-2003, 06:23 AM
hipro5

Quote:

Originally posted by reglisse
:) sorry i don't understand all

for 50M of digit here :

http://membres.lycos.fr/reglissementhe/tele.php3

for 10M of digits

edit reglage.txt and change

4096 --> 1024

5000000 --> 1000000

Thanks A LOT...... :toast:
05-02-2003, 06:25 AM
hipro5

.........NOW......Can someone with an AMD system bench this V.4.2 at about 2700MHz Please.......Thank you...... :)
05-02-2003, 08:01 AM
JCviggen

1 Attachment(s)

Quote:

Originally posted by hipro5
.........NOW......Can someone with an AMD system bench this V.4.2 at about 2700MHz Please.......Thank you...... :)

you ask, we bench :)

not exactly 2700, but only 10 MHz off
05-02-2003, 09:08 AM
hipro5

Quote:

Originally posted by JCviggen
you ask, we bench :)

not exactly 2700, but only 10 MHz off

Thanks JC....... ;) :toast:

Now.......I think that IF we ALL try the V.4.2 one , the Intel "times" are going to be FASTER than the AMD's ones.......Bet anyone???? :D

I think THAT'S why you ALL AMD guys "insist" for V.4.1 which "could" have a "bug" for Intel CPU's..... :stick:

Show 100 post(s) from this thread on one page