32m SPi Low Clock Challenge

[dinos22]

no big different between windows XP SP3 and windows 2k3 TW~
I have tested ago~
they were different only under 0.4s~

that's what i thought

[Kean]

XP SP3 is slower AFAIR then SP2.

[dinos22]

XP SP3 is slower AFAIR then SP2.

VERY VERY slightly

[before]

Hiwa are you using the Taiwanese Edition of Windows Server 2003 there ? Rumors say that it gives you a considerable boost in SuperPi 32M ... Gotta try it myself though first ...

Well buddy, actually the Taiwanese version is just the same one but using Taiwanese language. Nothing special; I mean for you and me, but great for Taiwanese natives.

who said that

Guess I know...

[George_o/c]

I know Xavier, I confirmed it myself. Nothing special about it. I know who said it, but I am not going to talk out loud. The one that said it, should stand up and say: "I did" ...

[hiwa]

Well buddy, actually the Taiwanese version is just the same one but using Taiwanese language. Nothing special; I mean for you and me, but great for Taiwanese natives.



Guess I know...

Hi
in TW,spanish Datacenter in not include in cd and so it's just 500Mb
Datacenter have more than 7Mb reg files which not good for Spi
that's all i think

[Shahryar_NEO]

Hi
in TW,spanish Datacenter in not include in cd and so it's just 500Mb
Datacenter have more than 7Mb reg files which not good for Spi
that's all i think

do you know how SuperPI works ?

if we know how it works we could run it better .

[dinos22]

Hi
in TW,spanish Datacenter in not include in cd and so it's just 500Mb
Datacenter have more than 7Mb reg files which not good for Spi
that's all i think

wouldnt you normally get rid of it via nlite anyways

my XP is trimmed to 150MB or something like that

[before]

Hi
in TW,spanish Datacenter in not include in cd and so it's just 500Mb
Datacenter have more than 7Mb reg files which not good for Spi
that's all i think

Well IMHO actually it doesn't change anything for Spi functionning mate. Spi computation process doesn't care about reg... In fact, windows size doesn't make pi faster or slower.

Here's a glance (monitoring the process) at Spi functionning

- Prior to start the computation: opening the app. (Here're the reg calls... I mean for both software and hardware aspects, reg isn't important after this point. Everything looks set.)
Example of events

4:58:35.3562421 PM	super_pi_mod.exe	4384	RegQueryValue	HKCU\Software\Microsoft\CTF\Disable Thread Input Manager	SUCCESS	Type: REG_DWORD, Length: 4, Data: 1

- Prior to start the computation: selecting the type of computation.
Example of events

5:01:12.7960465 PM	super_pi_mod.exe	4384	CloseFile	C:\WINDOWS.0\Fonts\vgasys.fon	SUCCESS	
5:01:21.4881158 PM	super_pi_mod.exe	4384	QueryOpen	C:\WINDOWS.0\system32\MSIMTF.dll	SUCCESS	CreationTime: 4/14/2008 2:00:00 PM, LastAccessTime: 12/19/2008 3:09:03 PM, LastWriteTime: 4/14/2008 2:00:00 PM, ChangeTime: 4/9/2009 12:00:16 PM, AllocationSize: 159,744, EndOfFile: 159,232, FileAttributes: A

- During the computation (basically dealing with spi self-created files)
Example of events

5:04:25.9159925 PM	super_pi_mod.exe	4384	ReadFile	E:\iv.dat	SUCCESS	Offset: 0, Length: 65,536
5:04:25.9160239 PM	super_pi_mod.exe	4384	ReadFile	E:\iv.dat	SUCCESS	Offset: 0, Length: 65,536
5:04:25.9163341 PM	super_pi_mod.exe	4384	ReadFile	E:\ib.dat	SUCCESS	Offset: 65,536, Length: 65,536
5:04:25.9163711 PM	super_pi_mod.exe	4384	WriteFile	E:\ib.dat	SUCCESS	Offset: 65,536, Length: 65,536
5:04:25.9174882 PM	super_pi_mod.exe	4384	WriteFile	E:\ib.dat	SUCCESS	Offset: 0, Length: 65,536
5:04:25.9175216 PM	super_pi_mod.exe	4384	ReadFile	E:\iw.dat	SUCCESS	Offset: 0, Length: 65,536
5:04:25.9206581 PM	super_pi_mod.exe	4384	ReadFile	E:\ib.dat	SUCCESS	Offset: 0, Length: 65,536
5:04:25.9276712 PM	super_pi_mod.exe	4384	ReadFile	E:\iw.dat	SUCCESS	Offset: 0, Length: 65,536

[dinos22]

oh guys it is not unusual to have half a second difference or ever slightly more between runs because of a couple of loops called bonus loops dont consistently throw out expected results

so someone getting a second quicker doesnt necessarily mean its got anything to do with the OS

I know who claimed that he shaved off 10 seconds and i could understand him saying that because he publicly says he's all brute force and not much into tweaking so if he did get some great tips and everything setup for him he could have seen good gains i guess

but OS to OS its laughable to suggest there are significant differences like that

now i'm starting to sound like Cpt Obvious hahah

[hiwa]

wouldnt you normally get rid of it via nlite anyways

my XP is trimmed to 150MB or something like that

my TW lite with all reg tweaks on is fastest about 2 seconds of my TW orginal 500mb @ 4000 MHz without waza
yes sure if you know all that files you can lite it with nlite

Thanks before for info

[cheapseats]

2303.9......cheapseats.........17.59.609.....6x384 .....768mhz.....6-5-4-12....1/2...X48

my CPU clock doesn't drop to 2302.x, so this doesn't beat Hyperhorn's run...

anyway, i needed tighter timings just to break 18 mins

XP sp2 / LSC on / core 1 affinity / realtime / explorer closed / 512-512 page file /
1 disk / 2 partition / maxmem 680 / SpiTweaker v1.0 CDT (680MB file) D>C (SPi C)

[cheapseats]

2063.9......cheapseats.........19.56.844.....6x344 .....688mhz.....5-4-4-12....1/2...X48

Low Clock Challenge !

XP sp2 / LSC on / core 1 affinity / realtime / explorer closed / 512-512 page file /
1 disk / 2 partition / maxmem 680 / SpiTweaker v1.0 CDT (680MB file) D>C (SPi C)

[cheapseats]

3969.0......cheapseats.........10.59.719.....7x567 .....907mhz.....7-5-5-15....5/8...X48

tRCD 5

XP sp2 / LSC on / core 1 affinity / realtime / explorer closed / 512-512 page file /
1 disk / 2 partition / maxmem 680 / SpiTweaker v1.0 CDT (680MB file) D>C (SPi C)

[dinos22]

my TW lite with all reg tweaks on is fastest about 2 seconds of my TW orginal 500mb @ 4000 MHz without waza
yes sure if you know all that files you can lite it with nlite

Thanks before for info

wow interesting
thats a bit difference

guess i'll have to get it and test it out

[alex_tpc]

4.376g sub 8m

first post but been lurking for ages. new entry for 7.xx category @ 4.388g. so near to Dsjjang's 4.376g sub 8m, yet so far ...

i7 920 D0 + Corsair Dominator GTs on R2E
vanilla Windows XP with most services off
waza 1.9GB C: --> D:, D:--->C:, C:--> D:on .bat file run from command prompt window

4388.9......alex_tpc............7.59.890....21x209 ....1045mhz.....8-8-7-20....2/10..X58

[George_o/c]

to XtremeSystems !

Nice run there dude ! Do you use Dominator GT CL8 ? I'm surprised they can't do CL7 @ this frequency, or you just didn't have enough time to test it ?

Hmmm 1896MB CDT-IV + 589MB Maxmem should work better for you - I mean in comparison with normal 1896MB CW and that weird Maxmem you're currently using, but generally you should search for a CW file size that's lower than that one

[alex_tpc]

to XtremeSystems !

Nice run there dude ! Do you use Dominator GT CL8 ? I'm surprised they can't do CL7 @ this frequency, or you just didn't have enough time to test it ?

thanks. actually, am using 2 sticks 2000cl7 and 1 stick 1866cl7 (which is harder to keep up with the rest). reason being, there's weird stuff happening with this R2E and D0 + 2000 cl7 combo . bios and windows won't see the third stick even if cpu-z does. if i run with the 3rd 2000 stick, results are lower.

full set of 1866 work just fine with the D0. full set of 2000 work just fine on my old 965 (though the 3rd stick "disappears" sometimes with certain settings). read somewhere that this could be due to vqpi and vcore interaction but haven't pinned it down.

am also "babying" the mem oc as this r2e kills mems instantly with certain combinations of vqpi and vcore even within intel spec. when i get a new board (classified w/o nf200 I think; haven't decided), might try for more (and see if the problems with the 2000cl7 + D0 combo disappears).

Hmmm 1896MB CDT-IV + 589MB Maxmem should work better for you - I mean in comparison with normal 1896MB CW and that weird Maxmem you're currently using, but generally you should search for a CW file size that's lower than that one

thanks for the tip. I tried lower maxmem before on core 2's and they gave me lower scores. haven't tried with i7 so maybe i'll try again.

btw, i think i gained (lost?) 5 seconds on this waza. how much more is possible?

[3oh6]

update done

this run won't be the fastest sub 7min by a long shot, but it is the first on the board. let's go guys, sub 5100MHz should be easy for this time slot.

i7 965 (3836A287) 25 x 205.5 (boot at 205.5*23) // MMouse Rev3 w/LN2
EVGA X58 3X SLI Classified (S21H) // bone stock
Corsair Dominator-GT 3x2GB PC3-1866 CL7
EVGA 295GTX (no drivers)
Seagate 7200.9 80GB SATA2 (3GB partition for OS / 4GB partition for SPi/page)
Corsair HX1000W
WinXP Pro SP3 (custom nLite'nd job)

5137.0MHz SPi 32M = 6m 59.734s
click for full size including memset...

[hiwa]

nice one here is sub seven


965 i7 3837AXXX Cooled By Colosso
P6T WS Pro Bios:0302 StockCooling
G.Skill PerfectStorm 16000 CL7 7-7-7-20 rest Auto Active Stock Aircooling
2K3 TW 32m Edition (3 services[RPC, EventLog, SHD],maxmem :600,copywaza : 1625MB,Priority = High,Cpu:0,9 proccess )

Vcore:1.62
Vdimm:1.68
Vqpi: 1.55

5100MHz 6.58.953



will try sub 7 with classfied soon need work better tonight i'll try sub 9 with classi

[cheapseats]

3257.9......cheapseats.........12.59.750.....6x543 .....905mhz.....7-5-5-15....3/5...X48

tRCD is my new favorite memory timing..

XP sp2 / LSC on / core 1 affinity / realtime / explorer closed / 512-512 page file /
1 disk / 2 partition / maxmem 680 / SpiTweaker v1.0 CDT (680MB file) D>C (SPi C)

update done

Thank you

a couple of things i noticed:

I made a mistake with my 10.xx details - divider is 5:8 not 1:2

and dark blue is much too flattering for my 19.xx run (45nm)

[cheapseats]

4431.5......cheapseats.........09.59.906...7.5x591 .....945mhz.....7-6-5-15....5/8...X48

(Edit 08/05/09: 3oh6 please ignore this run in next update )

slight change to my pre-run procedure - now waiting (a lot) less time between waza and starting Spi,
but it's only the same as others have recommended.. 35s

XP sp2 / LSC on / core 1 affinity / realtime / explorer closed / 512-512 page file /
1 disk / 2 partition / maxmem 680 / SpiTweaker v1.0 CDT (680MB file) D>C (SPi C)

[Zucker2k]

This any good for this clock? No tweaks, regular OS.

[alex_tpc]

a bit more push (no more "babying"):

4364.1......alex_tpc............7.59.828....21x207 ....1039mhz.....7-7-6-19....2/10..X58

i7 920 D0 + Corsair Dominator GTs on R2E
vanilla Windows XP with most services off
waza 1.9GB C: --> D:, D:--->C:, C:--> D:on .bat file run from command prompt window
maxmem = 620 (lower gives lesser score for me)

[alex_tpc]

8,xx and 9.xx entries

i7 920 D0 + Corsair Dominator GTs on R2E
vanilla Windows XP SP3 with most services off
waza 1.9GB C: --> D:, D:--->C:, C:--> D:on .bat file run from desktop shortcut

8.xx
3854.1......alex_tpc............8.59.031....19x202 ....1014mhz.....7-7-6-18....2/10..X58



9.xx
3434.1......alex_tpc............9.59.406....17x202 ....1010mhz.....7-7-6-18....2/10..X58