I am kinda freaking out!

I am set at 1.65v and according to SpeedFan it flips between 1.65 and 1.63.

That is alot of air going through that HS Freeagent! I had a tornado on my SI-97 but at higher speeds it actually caused the CPU to heat up some, so I changed it out to the Panaflo in my sig. I can run it at full speed and it still cools, and is a tad bit quieter!

DrSpanky: You could be right that with a differnt BIOS, you do not need to run TRAS at 11. Let us know what you find out with your testing.

wow, nice chip BC|Wulf, I need about 1.775 for that speed in your sig

It sucks the big one. With my old mobo i could hit 2700mhz at something like 2.03 to 2.05 and it was stable. now if i wanna do that i have to pump even more in to allow for the droop and then the vcore still stupidly erratic. hence why im down to 2500mhz.

But never fear for there is a way round this issue... adding an extra phase on the mobo for supplying the vcore. The guy who did it set 2v in bios and 2.023 at idle and 2.022 under load. So there was still the difference from what was set in bios but the idle to load insanely low fluctuations.


ok i ran some test with changeing my tras from 11 all the way down to 5.
i used sissoft sandera memory bandwith benchmark. and from what i can see there is little to no difference in running any of them. i ran them at least twice on each one and there was like 5mbps difference from all the settings i tested, you would prob get the same jump in results running the same benchmark a few times on your current settings.

as ive seen its made hardly any difference to "memory bandwith". do u know of any other software i could use to see if theres any other performances boosts else where from these settings?

Not too sure.. but i know superpi likes timeings

when i get some time later i will do a table comparing tras settings and superti results and a rendering test. with the lower tras i shaved 11secs of a image i was rendering

got any other test u want me to try? im thinknig 3dmark 2001.. what u lot think

dr:
i tested ram settings down to 5 & from the benches, my highest was when i was set @ 11 (i think its a nf7 thing........whereas the other board's may like it lower)

cuda70383 said:

dr:
i tested ram settings down to 5 & from the benches, my highest was when i was set @ 11 (i think its a nf7 thing........whereas the other board's may like it lower)

Click to expand...

its a nforce 2 thing. i did a few test and a tras of 11 is beeting 5 in certain things so looks like im gonig back to a tras of 11

freeagent said:

wow, nice chip BC|Wulf, I need about 1.775 for that speed in your sig

Click to expand...

Thanks! It could be because I am running CPC off whereas you are running with it on.

Also is anyone running their system with the CPU Interface disabled. Mine is and if I enable it I have to pump up the voltage alot to get thing stable.

Thanks for doing that testing DrSpanky! At least we have some proof that 11 is the magic number.

BC|Wulf said:

Thanks! It could be because I am running CPC off whereas you are running with it on.

Also is anyone running their system with the CPU Interface disabled. Mine is and if I enable it I have to pump up the voltage alot to get thing stable.

Click to expand...

Its funny you say that.. i get better results, and overclocks with cpu interface disabled, if i can recall, i failed some of the same of the tests i passed when disabled, with enabled, weird eh? this is the first bios ive tried with that charachteristic.

I'm running disabled too (and CPC off - dont remember which bios I'm using). Been running for a few weeks at 1.9V, but I'm trying to bring that down a little now to see what I can get away with.

Edit: not much less. 1.8V wont boot, 1.85 is doing ok so far.
Edit 2: well, I think I'll be staying at 1.9.

i havent tryed playing with cpu interface... maybe worth trying


i ended up trying 1t's last night... what can i say as i expected a no go overclock was unstable as hell, wouldnt post with memory set at x,2,2,x timings. im sticking with 2t

Thought i would test the difference between having cpu interface: enabled/disable so hes my results

cpu interface: enabled
12*220 = 2646mhz
2.5,2,2,11 <-- memory timings

SiSoftware Sandra tests

Cpu arithmetic
10948 - 10948 - 10948
4167 - 4167 - 4167

multi-media
25003 - 25003 - 25002
26510 - 26510 - 26511

memory bandwith
3369 - 3373 - 3376
3186 - 3795 - 3197

each Sandra was ran 3 time to test and then i marked down the results.

render test
there a 4 tests with the render benchmark and there listed from lowest lmage size to largest ( left to right )
theres were run 2 times to test results
18s, 38s, 48s, 4min 43s <---- First Run
16s, 28s, 47s, 4min 39s <---- Second Run



super pi results
00.219 16k
00.515 32k
01.203 64k
02.813 128k
07.531 256k
18.219 512k
--------------------


Cpu interface: Disabled
12*220 = 2646
2.5,2,2,11


SiSoftware Sandra tests

Cpu arithmetic
10950
4167

multi-media
25006 - 25006 - 25005
26513 - 26513 - 26513

memband
3364 - 3371 - 3375
3173 - 3183 - 3186

render test

18s, 28s, 47s, 4min 34s <---- First Run
16s, 27s, 47s, 4min 35s <---- Second Run


super pi results
00.219 16k
00.500 32k
01.141 64k
02.640 128k
07.250 256k
17.594 512k

render test is a benchmark u can download here
http://www.ocforums.com/showthread.php?t=392976

hmmm...have you tried going higher on the cpc off? usually you can go a little higher???

also, surprised at how close the scores were with ON vs OFF


i'm awaiting a gig of OCZ Platinum's & geForce 6800 so i'll be going through the same thing. my Ballistix 512mb can't do Battlefield2 all the time....some maps choke it....so are fine.

1t is cpc enabled and the 2t's are cpc off.

Over all i would say having Cpu interface: Disabled works out better, got better results on super pie and render tests. When i got to the 512k test u could see more of a difference so i assume if i did longer calculation there would be an even greater improvement