max core/mem settings?

[mikill]

in this forums experience, what is the highest reported core/memory settings attained for a ti4600 128, msi brand w/def. settings at 297/650. also, i have setup a fan for it to blow directly on the heatsink on the back of the card. with this setup i have, it never gets hot. both heatsinks, gpu/mem are cool to the touch. bear in mind though, before rigging the fan up it used to get pretty hot, hence me doing the fan trick. i dont intend to leave it at the o/c'd settings, but i'm curious to see how far this will clock up from the defaults of course while monitoring the temps.

thanks in advance for any info

 

[LeeJend]

Poor assumption. Just because the surface of the heatsink feels cool to the touch doesn't mean the GPU die temperature is. It could still be dangerously hot. See if there is a GPU temp monitor for your card before getting too adventurous.

 

[champ2005]

Only the Geforce FX line have cpu temp monitors.
Im sure you can get to around 315/675 at least, just remember go up by around 3-4mhz core / 5-6mhz memory at a time, and then just go down when it gets unstable.

 

[mikill]

thanks for info. i so far have gotten it to 310/700. left 3dmark looping last night, got up this morning and still looping just fine.

LeeJend, wish i had a good way to read die temps. maybe there is some other way? if so any info would be appreciated.

 

[LeeJend]

Wish I had a good way to read die temps too. I'd be friggin rich, lol.

I hate trusting the sensors, even the CPU internal diodes, because the Bios's apply scaling factors to them and there have been major screw ups. I forget who it was but one MB maker hosed it once.

If you can find a cheap digital thermometer with a really small sensor head:
Place the sensor right over the center of the die on top of the heatsink. Then cover that sensor head with a small dab of putty to insulate it from the cooling air flow and to hold the sensor in place. We do this at work. The putty we use is affectionately called "Monkey Dung" (smell and appearance). Keep the amount of putty small or it will reduce heatsink effectiveness.

Warning, many of the "Radio Shack" digital thermometers top out at 120 F. Useless for overclcoking work (or for monitoring my attic temp during a TX summer).

You won't get a good idea of die temp this way but you can see when the temperature starts to go up and how fast it goes up. Small uniform increases in temp as you go up are safe. If the temp starts to rise more rapidly, BACK the speed OFF, QUICKLY!

The attached picture is what I use. I cut the big rubber cover off the sensor and insulated the exposed wires. It fits between most heatsink fins.

Best I can do. After a friend and I murdered a couple of video cards we decided to limit our overclocking to the CPU.

 

[mikill]

thanks for reply

your mentioning temps, i was just wondering, what are 'safe' temps for the gpu and if possible for the memory chips? plus do you think the mem heatsink is pasted? if so, any easy safe way to remove it, so i can monitor the temps on them as well?

btw, the pic's somewhat fuzzy. what kind/model is yours? thanks again

 

[LeeJend]

Don't know what the safe temps are for the GPU's. Memory chips don't seem to fry they just make mistakes and crash you or create artifacts on the screen. When they cool down they always seem to recover.

I've heard some real horror stories in here about people pulling the GPU or RAM off their video cards when trying to get the heatsinks off to put on a better cooling solution. If it's just a strip heatsink on the RAM then it is most likely pasted. If the ram cooler is part of the overall GPU cooler those are usually not pasted.

Note old heatsink paste can become as hard as glue so be careful if the card is old.