OCing a GeForce4 Ti4600

[gravitywell]

I've read a few posts about core and memory clocks...but when I installed coolbits, I found that my core was at 300 and mem at 650 already. Someone had already told me to put them at 300/600. Should I push em further? What should I look for as in artifacts and errors?

[ntpphong]

i dont get your question quite well since the ti4600 runs 300/650 at stock, u'd downgrade it by putting 300/600, maybe that someone meant the ti 4400 which runs at a lower speed. and you can always try pushing your card 5mhz by 5mhz to find out your limit. It also depends on the card you have. PPl who are lucky got up to 320/770, others only 310/670.

[gravitywell]

that's what I was asking...sort of. I posted my score elsewhere, and someone said try 300/600. But when I checked my settings they were 300/650. I wasn't sure what was going on.

when increasing, what do I look for?

[repo man11]

If it starts locking up, or if it shows little white specks in the display, back off. The specks are known as artifacts, and mean the memory is overheating. Be careful, you can easily go too far and fry your card. Slow and easy, and if you want to go farther, check into better than stock cooling.

[gravitywell]

ok, I kicked it up...

Running the first four tests (games)

300/650 10918
305/660 10813
305/660 10894
310/660 10908
315/665 11109
320/670 11162
*above 320 core the games hang in places, and skips*
320/675 11162
320/680 11226
320/685 11123
320/690 11265
320/695 11283
320/700 11392

and with that, I'll stop for tonight :)
feel free to post comments, good or bad, advice is also greatly welcomed.

*******
Questions...

What is the best cooling for my card?

[gravitywell]

Also, I benchmarked with all resources free, nothing else running at:

1024x768, 32 bit color
no antialiasing
double frame buffer
compressed texture format
23bit z-buffer depth
d3d pure hardware t&l

(default settings)

[gravitywell]

Looking for some advice. Should I go back to stock and really test the settings out? Full 3dMark cycles repeating? Or just continue running only the first 4 tests once?