Step6 faster then Step5

Printable View

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

08-24-2006, 03:32 PM
Absolute_0

Fredyama's results show step6 faster, albeit by a hair.
08-24-2006, 03:33 PM
dinos22

Quote:

Originally Posted by Gautam

Look again at this chart by fredyama (yes I know its been linked to twice already)

http://green.ap.teacup.com/fredyama/img/1154773142.gif

If he doesn't know how to optimize Pi, I don't know who does. Like Mike said, tiny difference, but its there.

Also Mike's word is trustworthy-I don't think there's a single Conroe stepping that hasn't passed through his hands. He's mentioned that this same phenoma occured with Yonahs. No surprise its happening again.

well i don't know if you guys bothered to do some math here but i can tell you that if we take these times and blow them out to say 32M finish times of around 14minutes the difference is scores should be around 3 seconds and NOT 2 minutes M.Beier was alluding to :nono:

so comparing two processors (I assume Fredyama-san only used 1xB1 and 1xB2 for this test) is just not enough to make a conclusion on which one is faster but from results on those two we are seeing a very small difference
08-24-2006, 03:35 PM
G H Z

Quote:

Originally Posted by Absolute_0

Fredyama's results show step6 faster, albeit by a hair.

Haha good point LMAO, I need my eyes checked.
08-24-2006, 04:39 PM
Gautam

Yeah, M.Beier's results are obviously way off. (Sorry)

2 mins is impossible, a couple of secs though, yes that is possible and probable. I'm only going off fredyama's results and Mike's word.

Those comparisons done on two completely different systems are completely invalid, I won't deny that for a second.
08-24-2006, 08:29 PM
death metal

Quote:

Originally Posted by freecableguy

... B1 = B2 (except for a small twist that can't be discussed yet).

sorry for the OT, it's just that B1 and B2 reminds me of that kiddie show ;)....

anyway, threadstarter, I don't think I need to request THG or AMDtech. You posted your findings here, and as can be seen from different reply, there is huge discrepancy of result between the two. Again, if anything, thelostrican and that comparison from fred sure doesn't show such.

again, like what I replied to dinos, I may have made my reply too wordy because you missed the point (kidding or not kidding) on my earlier post because you said you can not guarantee about the 3.6GHz when I was meaning (and since it's the topic title) that it's about the results. in any case, it's easier to just use similar parts, except for CPU and compare the resuts i.e. just replace the cpu on the motherboard with another one and do the comparison.
08-25-2006, 05:42 AM
M.Beier

Updated with the run you asked for Eva2000 :)

Im gonna do some changes now, going for 13min, but please, do not use that for compare =)

EDIT:
G H Z
You need too sleep more often ;) (location, I bet its not near river Tyne neither :p: )
08-25-2006, 06:48 AM
eva2000

thanks but can you do the 32m runs with all systray apps closed as well and minimal apps running as i see your available memory in pi app at 1.6GB ish.

I just did a quick run myself and it is 0.563 seconds faster than your 1st run of 14min 06.672s

http://fileshosts.com/intel/Asus/P5W...in06s109ms.png
08-25-2006, 06:53 AM
M.Beier

Sure, but atm Im tryin' too hit those 13 mins, punishin' the mem :D

4-3-2-4 32M 500mhz running atm.. - 2.4v, gonna go for 3-3-2-4 and then get on with the other things before makin' a rerun, however, the things in systray is not of great importance as its dualcore :)

Have you used default windows settings Eva? Or performance, cause that still affects result. :)
08-25-2006, 07:00 AM
eva2000

what do you mean default windows settings ? i have system properties advance set to best appearance mode with my default windows theme since it's the only way to get pcmark05 to run
08-25-2006, 07:03 AM
IluvIntel

Quote:

Originally Posted by eva2000

what do you mean default windows settings ? i have system properties advance set to best appearance mode with my default windows theme since it's the only way to get pcmark05 to run

Whats PCMark05 got to do with this discussion on SuperPi runs?
08-25-2006, 07:06 AM
eva2000

nothing just that i'm doing pcmark05 runs right now, and in the middle of them decided to do a quick pi run at these compared settings of 9x400fsb 4-4-4-8 :)
08-25-2006, 07:31 AM
M.Beier

Quote:

Originally Posted by eva2000

what do you mean default windows settings ? i have system properties advance set to best appearance mode with my default windows theme since it's the only way to get pcmark05 to run

same :)
Just currious if you did in the run as well, since it makes a difference :)

BTW: Mem freq and timmigns changed, 13:2x, but I'll go for those darn 12:59 :D SHouldnt be much of a problem :D
08-25-2006, 07:38 AM
dinos22

Quote:

Originally Posted by M.Beier

same :)
Just currious if you did in the run as well, since it makes a difference :)

BTW: Mem freq and timmigns changed, 13:2x, but I'll go for those darn 12:59 :D SHouldnt be much of a problem :D

so you shaved 40 seconds on tighter timings and increased RAM frequency ALONE
08-25-2006, 07:40 AM
M.Beier

Quote:

Originally Posted by dinos22

so you shaved 40 seconds on tighter timings and increased RAM frequency ALONE

Yes :toast:
08-25-2006, 07:42 AM
dinos22

Quote:

Originally Posted by M.Beier

Yes :toast:

what RAM frequency and timings exactly

find that a bit hard to believe

screens please
08-25-2006, 07:46 AM
M.Beier

Quote:

Originally Posted by dinos22

what RAM frequency and timings exactly

find that a bit hard to believe

screens please

Screen up after next run :)

500mhz 4-3-2-4.
Havnt used memset yet.. But yeah its 38-39 secs I belive, dont remember exact numbers :)

EDIT:
Here goes
http://img174.imageshack.us/img174/3541/2cx5.jpg
Didnt use cpu-z after run, since it was during the phase of looking "whats best, 400mhz 3-2-2-4, 500mhz 4-3-2-4 or 600mhz 4-4-4-4, last one failed wouldnt even boot - and dont wanna add more vdimm, 2.4v is fine, 3.0v for some benchies yes, but not 32M :D )

Still testing, looking into copy-waza and some other stuff soon :)

- another EDIT :(
Memset gave less then 10 seconds, now that sucks :( - A pretty nice burst in PiFast though (thelab.gr pifast) 19.70 secs.
08-25-2006, 07:50 AM
dinos22

kick arse RAM there man
08-25-2006, 07:50 AM
eva2000

Quote:

Originally Posted by dinos22

so you shaved 40 seconds on tighter timings and increased RAM frequency ALONE

yup very easy dinos with upclock divider and tighter timings :)
08-25-2006, 07:53 AM
Metroid

This is all about the users :lsfight:
08-25-2006, 07:56 AM
M.Beier

Quote:

Originally Posted by dinos22

kick arse RAM there man

Its a quite alright 320$ pair :D (2*1GB Crucial Ballistix PC6400)

But I belive I was lucky with them, newer uses other chips as far as I've heard.
08-25-2006, 07:59 AM
dinos22

Quote:

Originally Posted by eva2000

yup very easy dinos with upclock divider and tighter timings :)

yeah i didn't imagine he was going to use 4-3-2 timings :D

i'll squeeze the life out of my fattys on monday and see if they can do the same clocks..........would be interesting to see how far behind the allendale would be..........hmmm fingers crossed the mobo can keep up with the CPU hehehehe
08-25-2006, 08:04 AM
eva2000

Quote:

Originally Posted by M.Beier

Its a quite alright 320$ pair :D (2*1GB Crucial Ballistix PC6400)

But I belive I was lucky with them, newer uses other chips as far as I've heard.

that's another factor in pi times.. memory modules used can have a minor or slight effect on pi times even if the same timings are used :)
08-25-2006, 08:07 AM
M.Beier

Hmm, I slightly disagree, its just a matter of getting all timmings identical, sadly you cant set all timmings :\
A64 // BH5 proved that pretty much I guess :D

Could you do a run with 400 1:1 with memset window open as well? Just so I can try the exact same settings afterwards :)
But seems the difference is 30sec to me, using same mobo and so on, you have a P5W right next too ya, dont ya? - Read that in another thread, something about a cheaty 1.85v @ vmch.. (man thats like getting a free vmod with it, hehe)
08-25-2006, 12:40 PM
TeamUSOffice

post for OPB:toast:

The cpu: b2 s6, L627a664
board: asus p5b dx bios dated 057
all air cooled

http://img166.imageshack.us/img166/9...803packjw3.jpg

no tweak:eek:

http://img423.imageshack.us/img423/1525/36gb2uu8.jpg
08-25-2006, 12:58 PM
M.Beier

Hello TeamUSOffice & Kev'

Another mobo as well, P5W DH

BTW: That tRAS & precharge is a bit lower.. - Also, those settings, they're quite aggressive, isnt standart mode in bios, is it?

HMm, is that memset 2.1? - I still use 2.0, time too update :D

Sweet result on 965P, seems like its Gigabyte only that scores poor in 32M ?

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