32m Low Clock Challenge

Printable View

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

03-04-2007, 08:37 PM
dinos22

OMG Vince you are scaring me man :slobber: :eek: :eek: :eek: :eek: :eek: :eek: :eek: :eek:

GJ :toast:
03-04-2007, 08:47 PM
coop

Indeed, congratulations on your time. The speed at which you did it makes it even better.
03-04-2007, 09:30 PM
Stapler

Congrats KP, that's an amazing run. You could crush the competition with Roro's clocks in 32m.
03-04-2007, 09:33 PM
dinos22

hey Vince are you sure Maxmem=600 won't give you even a more efficient score buddy...............do that and give it another crack while you're at it...........i want to see 8m 5Xs time :D for the 8m class here :banana: :banana: :banana: :banana: :banana: heheheheheh

another thing you can try depending on how good your board is .....cool that vMCH and give it some DICE and 2.0V and try for 4:5 RATIO and lower RAM clocks as the latencies are tighter.................i think i found that 500MHz on RAM at 5-4-4-8 timings was equivalent in speed to 600MHz on memory in 2:3 ratio with 5-5-4-x timings
03-04-2007, 11:10 PM
Zeus

Wow Vince, that's another job well done!! :eek2::toast: :toast:

Do you think you can top that score at higher cpu speed or are you running out of memory speed?
03-04-2007, 11:31 PM
before

Very nice Vince! Congrats :toast:

I wish I can bench these clocks one day to challenge you buddy ;) What was the temp during the bench?

Quote:

Originally Posted by dinos22

hey Vince are you sure Maxmem=600 won't give you even a more efficient score buddy...............do that and give it another crack while you're at it...........i want to see 8m 5Xs time :D for the 8m class here :banana: :banana: :banana: :banana: :banana: heheheheheh

another thing you can try depending on how good your board is .....cool that vMCH and give it some DICE and 2.0V and try for 4:5 RATIO and lower RAM clocks as the latencies are tighter.................i think i found that 500MHz on RAM at 5-4-4-8 timings was equivalent in speed to 600MHz on memory in 2:3 ratio with 5-5-4-x timings

I've stopped using maxmem myself... Doesn't really help, a good copy-waza is the key to be focused on. :p:
03-05-2007, 12:16 AM
hipro5

Quote:

Originally Posted by k|ngp|n

Guys I :banana::banana::banana::banana:ing did it! I think new overall WR in superpi 32m at only 5261mhz :D. I am posting this here because I want to share it with my low clock challenge homies :toast: . Without the help of some of you guys, I could have never have done this...really. Give credit where credit is due. I appreciate it very, very much. Before, dinos, gautum, and others I pm'd. I busted my ass for almost two weeks on this score and it paid off like hard work always does.
http://www.overclockingpin.com/32mwr.jpg

Ha, ha, ha.......Did you use the "OPB cheat" - oh sorry OPB tweak I wanted to say - Vince?......:D
03-05-2007, 12:49 AM
eva2000

Quote:

Originally Posted by k|ngp|n

Guys I :banana::banana::banana::banana:ing did it! I think new overall WR in superpi 32m at only 5261mhz :D. I am posting this here because I want to share it with my low clock challenge homies :toast: . Without the help of some of you guys, I could have never have done this...really. Give credit where credit is due. I appreciate it very, very much. Before, dinos, gautum, and others I pm'd. I busted my ass for almost two weeks on this score and it paid off like hard work always does.
http://www.overclockingpin.com/32mwr.jpg

how much vdimm was used may i ask ? :)
03-05-2007, 01:02 AM
alpha0ne

Quote:

Originally Posted by hipro5

Ha, ha, ha.......Did you use the "OPB cheat" - oh sorry OPB tweak I wanted to say - Vince?......:D

HEHEHEHHEHEHE and about 20 other names inserted there too :banana: :slapass:
03-05-2007, 04:54 AM
dinos22

Quote:

Originally Posted by before

Very nice Vince! Congrats :toast:

I wish I can bench these clocks one day to challenge you buddy ;) What was the temp during the bench?

I've stopped using maxmem myself... Doesn't really help, a good copy-waza is the key to be focused on. :p:

i haven't yet tested maxmem with all the other tweaks inluding the luna+silver+wazza+highclockinstall....just sold off my gear and looking out for something else to play with, so for the moment only E6300ES to play with lol but at least i can give it a crack in the 1066 strap and lower categories even though i'll pay a sizable penalty for B1

i think Vince should look into 4:5 as well though
03-05-2007, 05:21 AM
hipro5

Quote:

Originally Posted by dinos22

i think Vince should look into 4:5 as well though

Agree here......Less internal latency....;)
03-05-2007, 05:53 AM
before

Unfortunately, it's hard to make 4:5 working... guess it becomes a matter of chipset.
03-05-2007, 07:04 AM
k|ngp|n

Quote:

Originally Posted by before

Very nice Vince! Congrats :toast:
I wish I can bench these clocks one day to challenge you buddy ;) What was the temp during the bench?
I've stopped using maxmem myself... Doesn't really help, a good copy-waza is the key to be focused on. :p:

I was running about -120c to -125c on the cpu, still unlapped. Can do 5374mhz 1m at little bit colder like -125c to -127c/1.87v. I think with more than 1.92v, this could end up being a 5400mhz+ 3d cpu. I will go back to 3d soon, so we'll see.
I tried some diff maxmem sizes with no success on core2, doesnt seem to help like it did with fx57/maxmem104.

Hey before, if you have problems with your next container you try, you should get one of my EVO's man :D. I can walk away from the pc after a pour to get something to eat from the fridge, hit the bathroom and come back to about 2-3degress less temperature at nearly 5300mhz/1.87v mid 32m.
Benching one unit in superpi with core2 is tooo easy.
03-05-2007, 07:13 AM
before

ah you make me jealous bro! That's an impressive pot for sure. :D

Will get my new solid copper container this evening; a kind of MM rev.2 replica; but much longer. Hope it'll rock...
03-05-2007, 07:20 AM
k|ngp|n

Quote:

Originally Posted by before

ah you make me jealous bro! That's an impressive pot for sure. :D

Will get my new solid copper container this evening; a kind of MM rev.2 replica; but much longer. Hope it'll rock...

Who made it? I am sure it will be good.
Big difference between running 3d and 2d as for load on the container of course. Single threaded sp just doesnt push the chips that hard. As soon as you go over 1.9v, the heat output goes up big time though. Trick is to find that zone with the chip where the heat ouput/voltage is perfect. That's when I like benching ln2 :D
Wait till you run some 06' with quadcore :eek:, that is the toughest test of a unit IMO.
03-05-2007, 07:26 AM
before

It was made by the University of Grenoble (France) Department of Physics... a chance, my mother is an Honorary Professor of Biochemistry at the university... so it helps;

I'm waiting for QX6700 + P5N32 + 8800GTX lent by a magazine... Will be the opportunity to test the container into strong conditions :D
03-05-2007, 08:04 AM
s7e9h3n

Quote:

Originally Posted by k|ngp|n

I can walk away from the pc after a pour to get something to eat from the fridge, hit the bathroom and come back to about 2-3degress less temperature at nearly 5300mhz/1.87v mid 32m.

#1 or #2? :p: BTW, nice work on that 32m :toast:
Maybe you - or anybody else for that matter - can help me figure out what the hell is going on with this Intel rig I've got here. I thought it'd be fun to try to go sub-13 min @ 3600, but it turns out I spent most of the night trying to go sub-14 min @ 3600 . I applied almost every tweak I know of - keep in mind that I only know AMD tweaks - and didn't even break into the 13's. WTF IS WRONG????? This is on an EVGA 680i and the memory's @ 2.5V. For the record, these stix are a pair of gems as they'll boot @ 4-4-3-4, DDR2 1250 at 2.5Vmem....too bad they don't run worth a crap @ that speed though. Time to do the vmem mod....;)

http://img187.imageshack.us/img187/9...332m2k3eb5.jpg

http://img187.imageshack.us/img187/6...032m2k3lc2.jpg
03-05-2007, 08:07 AM
JMKS

Quote:

Originally Posted by k|ngp|n

Guys I :banana::banana::banana::banana:ing did it! I think new overall WR in superpi 32m at only 5261mhz :D.

:woot: :clap:
Now I know why I couldn't beat such scores clock by clock ;).

Could You post the subtimings used, please?

What about 1M 2:3?
And what is limiting now? :D

And last question: i975X with 4:5 - how high it will go?

@s7e9h3n: Command Rate 1T is very important for NV680i (and the chipset isn't really good in Pi...). And also not all memory ratios are good. If You can run 3:2 ratio linked 1T (mem @ 1066 1T of course, don't know if possible) it will be much better.

And yes, 4:5 is extremaly difficult on i965 - Commando with 2,025V @ NB (aircooled, but REALLY low temperature, ambient <0°C and HR-05 + some fans) tops @ ~425 FSB (with optimized subtimings and 1066 strap of course).
03-05-2007, 08:14 AM
eva2000

nah JMKS - 1T won't help unless you can get memory to do 520+ mhz 4-4-4-4 1T with tight sub timings. Most high end Micron D9 are doing around 430-450mhz 4-4-x-x 1T with 2.4-2.6+ vdimm only

Doesn't help for 3dmark either 1T if you have ram that does 533mhz 4-3-3-4 2T tight sub timings either > http://i4memory.com/showthread.php?t=3773

Quote:

Originally Posted by s7e9h3n

#1 or #2? :p: BTW, nice work on that 32m :toast:
Maybe you - or anybody else for that matter - can help me figure out what the hell is going on with this Intel rig I've got here. I thought it'd be fun to try to go sub-13 min @ 3600, but it turns out I spent most of the night trying to go sub-14 min @ 3600 . I applied almost every tweak I know of - keep in mind that I only know AMD tweaks - and didn't even break into the 13's. WTF IS WRONG????? This is on an EVGA 680i and the memory's @ 2.5V. For the record, these stix are a pair of gems as they'll boot @ 4-4-3-4, DDR2 1250 at 2.5Vmem....too bad they don't run worth a crap @ that speed though. Time to do the vmem mod....;)

you need 600+ mhz on memory on either 965/975 or 680i platform to even come close and probably need B2 stepping core2 for 10-20 seconds clock for clock faster times than B1

http://fileshosts.com/intel/Asus/680...3m13s859ms.png

try 600mhz 4-3-3-4 to see
03-05-2007, 08:23 AM
s7e9h3n

Quote:

Originally Posted by JMKS

s7e9h3n: Command Rate 1T is very important for NV680i (and the chipset isn't really good in Pi...). And also not all memory ratios are good. If You can run 3:2 ratio linked 1T (mem @ 1066 1T of course, don't know if possible) it will be much better.

Lol....1066@1t? Sheesh....that's not gonna be too easy. I'll try 32m @ as hgih as 1t will take me and post back...thanks...

@ Eva: Exactly how much over 600+? ;) My sticks are capable of it. Just need a tad bit more Vmem....Maybe it's time I try running 4x1gb ddr2 1200 @ 4-3-3-4 :hehe:
03-05-2007, 08:25 AM
k|ngp|n

Quote:

Originally Posted by s7e9h3n

#1 or #2? :p: BTW, nice work on that 32m :toast:
Maybe you - or anybody else for that matter - can help me figure out what the hell is going on with this Intel rig I've got here. I thought it'd be fun to try to go sub-13 min @ 3600, but it turns out I spent most of the night trying to go sub-14 min @ 3600 . I applied almost every tweak I know of - keep in mind that I only know AMD tweaks - and didn't even break into the 13's. WTF IS WRONG????? This is on an EVGA 680i and the memory's @ 2.5V. For the record, these stix are a pair of gems as they'll boot @ 4-4-3-4, DDR2 1250 at 2.5Vmem....too bad they don't run worth a crap @ that speed though. Time to do the vmem mod....;)

heh #3 ;)
and def what eva said Steve. First thing I learned when I started benching 32m was to ditch the b1 stepping. Pretty significant difference in time between the two steppings. Something to do with cache utilization?? Not sure.
03-05-2007, 08:33 AM
eva2000

well my ram is capable of 640mhz 5-4-3-4 2T at 2.5v but had to loosen sub timings too much that it was slower than my 600mhz 5-3-3-3 very tight sub timings at 9x400FSB.. so i'd say you'd need 622-640mhz 4-3-3-4 with same tight sub timings as at 600mhz 4-3-3-4 to get their on 680i platform.

I just don't have the ram to do that and I limit myself to under 2.5-2.6v vdimm as I like my ram and don't want to kill them hehe
03-05-2007, 08:53 AM
s7e9h3n

Quote:

Originally Posted by k|ngp|n

heh #3 ;)
and def what eva said Steve. First thing I learned when I started benching 32m was to ditch the b1 stepping. Pretty significant difference in time between the two steppings. Something to do with cache utilization?? Not sure.

It's gotta have something to do with the cache as the 1m times seem at least close to par....I'm not in any position to ditch this cpu as it's not mine :p: But maybe it'll be good for a few 1m passes @ -100C on the auto :D Oh well, I guess for 32m, I'll just have to wait for a few quads from AMD ;)

@ EvaL OK...I'll try tighter...but for some 1m runs instead. I guess we've established that this cpu is a POS in 32m :p:
03-05-2007, 08:57 AM
i found nemo

Quote:

Originally Posted by Stapler

^If I remember his thread right, he tested Super Pi on two seperate installs, with the same settings except one install was done at 2000mhz and another at 2800mhz. The 2800mhz was faster in both 1m and 32m. OPB has also been using this tweak for quite a while, and installs windows at over 5300mhz IIRC.

??? bull i install windows o/c'ed all the time and i get 32.7xx in 1m ... so ur saying if i install windows on stock it will be even slower?
03-05-2007, 09:05 AM
eva2000

only a prob if you are in for 3.6ghz challenge.. if you can clock B1 higher than B2 to a pt then overall 32M might be faster than B2 but with -100C monster of yours I'm sure you'll be able to max out any cpu LOL

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

All times are GMT -8. The time now is 12:14 AM.

XtremeSystems