32m SPi Low Clock Challenge

[Zeus]

Completely wrenched out my sys to squeeze that tiny bit, man that took some effort.

Could not touch mrcape's scores but manged to shave 10MHz off:



Mind you, this is just my 24/7 windows install, i had to set AI clock twister from light to moderate to get there.

Untill i find new tweaks that will help to improve my scores i'm done with sub 10 minutes. (i think)

[cheapseats]

nice one Zeus

hehe .....same time and checksum for you and mrcape

just curious, but did you run any lower vNB than 1.7x ?

i've been running CT moderate + tRD8 with 1.63-1.65V. I suppose some extra for GTR/S ICs + tight timings, but seemed a big difference for vNB.



and thanks mrcape for the answer 'all' . just hope i can find GTR/GTS sometime soon...

[mrcape]

FYI guys - vNB=1.6 (in bios, 1.58 everest) Vmem=2.07-2.12V depending on the run. I tried up to 2.2vdimm, 1.67VNB with the Ballistix to see if it could run tighter, but that didn't help.


Nice one Zeus! I bet a clean install would make a slight difference. At this point though guys, it's really all so close. I'm going to these mems for cold benching from now on.

[Zeus]

nice one Zeus

hehe .....same time and checksum for you and mrcape

just curious, but did you run any lower vNB than 1.7x ?

i've been running CT moderate + tRD8 with 1.63-1.65V. I suppose some extra for GTR/S ICs + tight timings, but seemed a big difference for vNB.


and thanks mrcape for the answer 'all' . just hope i can find GTR/GTS sometime soon...

Thanks buddy

Well, didn't really try low Vnb, set it to 1.76V which translated to 1.73V in windows. I don't know if i could do with less, haven't tried really as i wanted to be safe.
Temps were only 32ºC with watercooling so no worries there.
I hate it when a run goes well and looks like it might make it and then just hangs or errors somewhere halfway.

FYI guys - vNB=1.6 (in bios, 1.58 everest) Vmem=2.07-2.12V depending on the run. I tried up to 2.2vdimm, 1.67VNB with the Ballistix to see if it could run tighter, but that didn't help.


Nice one Zeus! I bet a clean install would make a slight difference. At this point though guys, it's really all so close. I'm going to these mems for cold benching from now on.

Thanks man.

My CSX Diablo's are GTR and i gave 'm 2.21V for those runs.

Yeah it's close but at some point beating the score above you seems an impossible task.

What i found out is that when the intial value is low, the finishing loop after the 24th loop is significantly lower. So whenever the intial value sucks i simply stop the calculation and start over again.

Comparing cheap's runs to mine you can see his 24th run is 9min 39 seconds but he still manageses under 10 minutes because his initial value is very low.

My initial value sucks and therefore i need to be at 9min 37 to crack 10 minutes. At 9min 38 i cannot break 10 minutes.

It's just an observation but i have no idea how to obtain a low initial value.

I'm thinking about making a harddisk dedicated for SuperPi, this over half a year old windows install is just not good enough for this stuff.

[mrcape]

I was getting crazy low initial values with the ballistix, and when the drive was fresh. Then again, as i went to lower freq, starts times were all 7.0+ and came back to beat similar scores in later loops.

[cheapseats]

Not sure if this adds much to what's been said on initial values, as its really just an example of the same thing

The main points of interest (i hope) being a reasonably direct comparison between the the two different levels of initial values, as seen in posted 9.xx runs.

the two runs (599x7.5 - links below) also show two distinct patterns for how all the 9.xx runs seem to behave at certain points in a run (initial value / Loop 1 / Loop 24 to finish).

the only (intended) difference between the my two run setups was running SuperPi from either 'C' (windows) partition, or 'D' (empty) partition.


SuperPi in 'D' (599x7.5 - 10m00.047s)

http://www.xtremesystems.org/FORUMS/...&postcount=412

typical 9.xx 'Pattern 1' : high initial value 6.9-7.0s / < 28s Loop 1 / 21-22s finish stage

dami1stm / Zeus / mrcape (GSkill) runs look like this



Superpi in 'C' (599x7.5 - 09m59.391s)

http://www.xtremesystems.org/FORUMS/...&postcount=419

typical 9.xx 'Pattern 2' : low intial value 6.4-6.5s / > 28s Loop 1 / ~ 20s finish stage

3oh6 / ToyTen / youngpro / mrcape (Ballistix) runs look like this



In my case, running SPi from a different partition to windows somehow produces Pattern 1 (high initial values etc), but running Spi from 'C' (windows) partition, produces Pattern 2 (low initial values etc). It's not just initial values that change from Pattern 1 to 2 - Loop times are also differently spaced between the runs, and are not comparable loop by loop.

my compared runs so far, show Pattern 2 (low initials) to finish slightly faster overall, and for me, faster by at least the difference between the Pattern 1 and 2 initial values (0.5-0.6s)

The difference is small as measured here, but neither qualifying 9.xx run of mine was fast enough using the 'slower' alternative. (high initial values and Pattern 1)

[CryptiK]

You guys are really good at this and your systems are very fast, really nice work

This is my first sub 11 min run ever Got a new e8400 and it's much better than my old C0. Nothing special efficiency wise but I was happy to go sub 11mins.

4374.2......CryptiK..............10.59.172......9 x 486.....607.5mhz.....5-5-5-15....4/5...P45

[mrcape]

Hey Cheap, it's still a mystery to me as I've run all from a separate partition and have had 6.4 - 7.1 starts. Also, my best PP run in this range started with 7.062.

Just some more data to make you go hmmm. The only way I could gauge whether to kill a run was to compare it to runs that looked similar to my own in the same range from other users. If mine was ahead in more than half of the loops after ten loops I'd let it go. If it looked dead even, or it was alternating (I'd win a loop, then they), I'd go by whichever was winning the alternate loop by more.

[mrcape]

11.xx - 11.59.625s - 3555.6 - 6x592.6

e8600
Thermalright Ultra-120
Asus Rampage Extreme (0401)
G.Skill HZ 1x1GB PC3-12800 CL7
Evga 9800gx2, no driver
Seagate 7200.12 SATA2 (3GB partition for OS / 4.5GB partition for SPi/page)
Corsair HX1000
WinXP (Nlited)
Silver Theme
CDT/maxmem 600/reg tweaks

[mrcape]

Is this thread dead or what lol.

[Benji Tshi]

No time to clock these days...I'll be back soon, probably with E8600

[cheapseats]

mrcape, really impressive 11.xx run . Beats mine (not posted) by more than your 9.xx runs do.

some problems here.... cross-linked files and bad sectors after using Memset save option with CL6 timings. Forgot to delete save before setting 950mhz/CL7 in the BIOS...

my new XP install and image of the old one, are both slower so far by 1-3 seconds in 9.xx

[mrcape]

Did you quick format? Got any other drives to try?

I just set up a few drives for LN tomorrow, got to remember to do 8.xx on the way down.

[cheapseats]

mrcape, sorry for the late reply i have a couple of other disks, but I could only get low initials with the 320GB samsung - and no low initials at all now. The slow new XP install is on a new samsung disk, and the OS image i also mentioned is really a full partition copied over from another disk, so the existing system partition is deleted first. Not sure if it's good way to do it - just easy. The OS + partition i used for my posted runs was copied from the same original.

[mrcape]

I'm going to wipe my raptors with dban and reformat. One was goofing around today when i was beating on it. I had a nice 8.xx but I deleted it by mistake! I need a better e8600, 822a435 would be nice.

Anyway here's a 7.xx -

[jmax_oc]

Very well tweaked run mrcape

[mrcape]

Thanks jmax_oc. Now I need either a cherry e86 or an e87 to add the freq to the mix.

[Zeus]

No more updates? Where's 3oh6 gone?

Soon i will make some entries in the amd class i hope.

[Benji Tshi]

I bet he's trying to improve his way of tweak on i7

[ToyTen]

3482.9......ToyTen.............9.59.812....21x165. 8 .....995.1mhz.....8-7-6-18....2/12..X58



1,55vQPI/2,1vDDR cpu under water(at 22°C)
I didn't tweaked subtimings in BIOS yet too used having felix's awsome log...(who said chronicle-addicted to facility ? )

XP sp3 nlightened of my own,
615 maxmem on a 15Go partition WD(74Go 16Mo), still gotta try some other OS I got from Hellas just to lasy George I don't forget, besure of it

And here is a 8min T_Z:

3900.5......ToyTen.............8.59.641....21x185. 7 .....928.7mhz.....7-6-5-17....2/10..X58

[rge]

Core i7 940 on water with help of open window to 3C outside temps. Bios set at 7,6,6,18 software reads 22?

[bluto]

I'm going to wipe my raptors with dban and reformat. One was goofing around today when i was beating on it. I had a nice 8.xx but I deleted it by mistake! I need a better e8600, 822a435 would be nice.

Anyway here's a 7.xx -

Very nice MrCape. I have similar set-up minus your tweaking skills so maybe you can help me. Sub 8min 32M has been so frustrating and cost me liters of ln2. Everytime I pass 24 loops around 7m58sec and the window comes up saying calculation done my pc freezes. Done this to me about 5 times. What do you think I should look at adjusting first. I was at 6.15ghz 2.02v/5:8 cas7

[NewBeetle]

7.xx & 8.xx Category

M/B: ASUS P6T Deluxe (BIOS:0703)
M/B Bed: V-TEC ARC Bed Rev.3.3
CPU: Core i7 Extreme 965 (3836A767)
CPU cooling: Alphacool NexXxoS XP Silver Socket 775
Chipset Cooling: stock + FAN
Mem: A-DATA DDR3-2133X 1GBx3
VGA: MSI NX7900GTO 512MB PCI-EX
HDD: Western Digital WD360ADFD (AHCI)
PSU: SilverStone OP1200
OS: Windows XP SP3
Remark: 2 Core set by BIOS

7.xx


8.xx

[Benji Tshi]

Really nice NB-san
I have seen your thread about AData and i was praying for you to don't try LCC 32M But now, i'm defeated...I'll try to go under 3894 this afternoon

[George_o/c]

Jody, I was just going through the second post, re-calling some really fast runs I've seen there, and noticed that in 10.xx category, in hipro's run (3923MHz 10.59.xxx) you accidentally wrote that he was using an X58 motherboard He's using a pretty normal Rampage Extreme X48