Playing with SPI 1M and OS.

[Epox4life]

E6600 stock, SPI Tweaking... Memory Settings... 2:3, 400mhz @ 5-5-5-15

The overall time is not important. Look at the difference between them.
Nothing special just interesting to see. I'll probably add more results later.

2.4ghz, numproc1, maxmem600.
PF = 49.5mb = explorer open

SPI 1M = 21.500sec
SPI 1M = 21.484sec
SPI 1M = 21.485sec

PF = 39mb = explorer open

SPI 1M = 21.484sec
SPI 1M = 21.469sec
SPI 1M = 21.453sec

----------------------------

2.4ghz, numproc1, maxmem256.
PF = 36mb = explorer open

SPI 1M = 21.484sec
SPI 1M = 21.469sec
SPI 1M = 21.453sec

----------------------------

2.4ghz, maxmem256, 2 cores.
PF = 36mb = explorer open

SPI 1M = 21.453sec
SPI 1M = 21.453sec
SPI 1M = 21.469sec

2.4ghz, no maxmem, 2 cores.
PF = 36mb = explorer open

SPI 1M = 21.500sec
SPI 1M = 21.516sec
SPI 1M = 21.485sec

Update* Fresh OS + extra tweaks and i've got it down to 21.313sec.

SPI 8M compare, 2 cores to 1. Without copy-waza.

2.4ghz, maxmem600, 2 cores.
SPI 8M = 4.14.172min

2.4ghz, maxmem600, numproc1.
SPI 8M = 4.14.328min

[Onepagebook]

nice job.
You can also try:

explorer.exe keep, call taskmanager and set its priority to lowest. and minimumizeto task barafter done for copy-waza if you do copy-waza.

** actually 256mb is a bit too many, I was thinking about 205~208mb
OPB

[Gautam]

Good to see you here Kev. ;)

[Onepagebook]

Why not;)
I think this is gonna help a bit;
qutoe from my own reply @my WR thread at XS

table:

http://img261.imageshack.us/img261/7157/superpi1mha5.jpg

"Well, before I did that bench, I really made a table to think about any possibilties that either I can win over them by the delta between initial and 1st iteration or the 19th iteration and pi value output results. So, according to this, if you wanna win by getting faster results from these two points I set. you have two ways. I can give the hint:

** before that. Gautam ask me from 2 days ago. He wonder why I use tras as 1 instead of 4. Certainly I know 1 won't be fastern than 4 or 8; but this will affect the hint I would like to give.

Master GoriLLakoS is about correct.
the initial result can be faster only base on:
1. Good setting of permanent paging files (Set that file @ other HD not partition)
2.Good result depends on the writing speed on both I/O chip from Motherboard and HD block size especially on Raid (0 for the most) HD system.
3. Good result will be gained from the write to read , read to write even write to write along with precharge timing of command related settings. well what are those settings? You definitely can figure those out especially from Memset 3.0.


Hope this help little bit.


OPB

[Epox4life]

Thanks, i'll try these things and post more results.

Nice to see you post at OCF.

[greenmaji]

** before that. Gautam ask me from 2 days ago. He wonder why I use tras as 1 instead of 4. Certainly I know 1 won't be fastern than 4 or 8; but this will affect the hint I would like to give.
OPB

:welcome: back OPB :D

Ok, the part of your post that I highlighted and colored there, at first I thought you were refering to the lower part of your post, but I think you might have started a thought and didn't finish it :shrug:

Any way you could elaborate?

[El<(')>Maxi]

Nice thread Epox, you have some interesting numbers in comparisons with setup already. I'm gonna jump in on this next week sometime but you'll probably have everything figured out by then :D

Hi OPB :D

[Epox4life]

Something weird going on here...

SPI 1M time of 21.375sec... hmm. < 1 stick of g.skill 1gb.

*Update
Put second stick back in there was a time of 21.328sec.

[Epox4life]

Can someone with a 975X board run SPI 1M at the same speed? I'm interested in the difference in times.
"2.4ghz C2D(266fsbx9), Memory Settings... 2:3, 400mhz @ 5-5-5-15" <---- these settings.