32m Low Clock Challenge

[Johnny Bravo]

Hello, Johnny Bravo-san.
I understand what you say and your question is very reasonable.

First question,
I think MAXMEM shows the effect by combining with COPY-WAZA.
My Copy-Waza size is "MAXMEM size + 100-200MB".
Without MAXMEM, Copy-Waza size will be very large.
And there is no difference in MAXMEM tweaking itself.

Second question,
I use the tool named "MEMEAT" to adjust the physical memory more in detail after physical memory is roughly adjusted with MAXMEM.
However, the effect is a little (about 0.5-1.0 second).
I'm sorry I can't explain well any more.

DEWA DEWA ～

NewBeetle thanks for the answers your explanation is clear even with the language difference MEMEAT sounds interesting I will have to experiment with it.

[luihed]

Wow... thanks for all the tips NewBeatle.... You are very generous with regards to sharing your knowledge...... May the overclocking gods reward you with a very fast cpu.......

[eva2000]

thanks Newbeetle i dropped ~4 seconds off my best 400FSB 4:5 500mhz memory 32M time with maxmem 512MB

[dinos22]

eva are you running RAS TO CAS WRITE DELAY = 2 or is that a glitch


try 600MHz maxmem too....i had slightly better times with that over lower values

lower values was better for shorter runs for me

[eva2000]

maxmem = 600MB = 13min 23s which is ~4 seconds slower heh

yes RAS TO CAS Write delay = 2 memset tweaked no glitch it isn't RAS TO CAS Read delay (tRCD)

[dinos22]

interesting

i have to try two timings changes with my sticks and maxmem 500......that should get me to NewBeetle/KP times if my system behaves the same as yours george

[over@locker886]

I don't know but in XP we can shot 1M 5480, with 2k3 max is 5400, similar with 32M, ~100mhz less dunno why Do u have best 3600 32M with quadGT can u try it? I'll be very interested to see that.

hmm,interesting..

Today I will come back my cpu,what can run 514FSB, so I will test tonight..

[Misko]

hmm,interesting..

Today I will come back my cpu,what can run 514FSB, so I will test tonight..

I think it can go under 13min and if someone manage to make it, it would be great to see 514mhz ram under 13min run, good luck with your benching

[Misko]

Hello, Misko-san.
I tested it.
It increases for about five seconds on hdd.
And i am using "ERAM for Windows Ver 2.11".

tnx a lot NewBeetle for that information.

[Johnny Bravo]

NewBeetle (or anyone else for that matter) can you explain what autogear does that simply assigning cpu priority using task manager doesn't? It seems that the program assigns priority to running threads - though the application is in Japanese so I don't exactly know what I'm doing with it

[Viss]

http://www.xtremesystems.org/forums/...32&postcount=4

[Johnny Bravo]

http://www.xtremesystems.org/forums/...32&postcount=4

yeah saw that and another post from Fredyama, was wondering if there is no difference then why are most of the Japanese team still using it? I was seeing if it offered something more

[shimmishim]

just want to thank you guys for this thread. it has helped me find tweaks for getting the best possible pi time and also for setting goals to achieve at clockspeeds. i'm trying to see if i can break the 13minute barrier with 3.65 or less ghz on my 6600. so far i've been able to tweak down from 14m10s to 13m25s at 3.6. only 25 more seconds! lol.

@Johnny - thanks for this analysis you did.
http://www.ocxtreme.org/forumenus/showthread.php?t=332

i actually found that the luna theme made things faster in win2k3 (the most interesting tweak i've ever tried).

it actually lowered my 1M superpi time at 3.6 from 14.064 to 13.938!

[andre X_X]

so close to sub 15m

[erwinz]

my 32M run..



full screen
http://img139.imageshack.us/img139/5225/jhgjghey8.jpg

validation..
http://valid.x86-secret.com/show_oc?id=189682

[over@locker886]

32M
4*512MB quad-channel used,and copy-waza only 336MB. I tested maxmem=512MB/550MB but 600MB faster..I think I can faster with 2*1GB memory and faster test hdd(used 15GB Quantum Fireball)



1M
Yesterday, I no tested 1M with 1:1. This monday result.

[staxm]

QuadGt (vdroop,vmch modded)
e6300@1.45v,TT BigTyphon
2x1GB crucial ballistix pc6400@2.75v



ballistix need too much voltage for cas4-4-4-4@514FSB and that subtiminigs,for spi1m@530 need 2,825v....did anyone have same problem?

[Misko]

32M
4*512MB quad-channel used,and copy-waza only 336MB. I tested maxmem=512MB/550MB but 600MB faster..I think I can faster with 2*1GB memory and faster test hdd(used 15GB Quantum Fireball)

Yes faster hdd will improve Spi result, also waza must be above maxmem. You have beautiful spi 1M result, is it done on the same windows as 32M run?

[over@locker886]

Yes faster hdd will improve Spi result, also waza must be above maxmem. You have beautiful spi 1M result, is it done on the same windows as 32M run?

Thx. Same windows. What do you think? 2*1GB I will get better results than 4*512MB in Quad-GT?

[Misko]

Thx. Same windows. What do you think? 2*1GB I will get better results than 4*512MB in Quad-GT?

Yas I think it will, which ide driver did u use? default win or u apply intel drivers (and did u put intel driver for chipset, my experience say it is better to use win default drivers for chipset but with hdd I dunno know yet)

[over@locker886]

Yas I think it will, which ide driver did u use? default win or u apply intel drivers (and did u put intel driver for chipset, my experience say it is better to use win default drivers for chipset but with hdd I dunno know yet)

default win ide driver, no install drivers..

[shimmishim]

QuadGt (vdroop,vmch modded)
e6300@1.45v,TT BigTyphon
2x1GB crucial ballistix pc6400@2.75v



ballistix need too much voltage for cas4-4-4-4@514FSB and that subtiminigs,for spi1m@530 need 2,825v....did anyone have same problem?

yes. 2.825 for 530 1M seems a bit excessive especially for ballistix.

i can do 4-4-3-4 ddr2-1040 or 520mhz memtest looping #5 with only 2.2 volts

[andre X_X]

14.XX Category

3133.0......andre X_X..............14.58.890.....8x391 .....587mhz.....4-5-4-4.....2/3

[Poliniboi#40]

Cool info

[hadji_a]

for the 14.xx category