We have two Visiontek Xtasy Ti4600s here, and I decided to use the one where we've had less success overclocking. It happens to be coincidence as well, that this Ti4600 runs about 2C higher than the other.
Depending on how the manufacturer decided to attach the heatsink, will determine how easy it is to take off. I can tell you that Visiontek uses some kind of adhesive, and it's a real pain to get the heatsink off the card. I wanted to do the installation quickly, so the trick with putting the card in a ziplock bag, and freezing it overnight (this will make the adhesive brittle, thus weaker) wasn't in the plans. So instead, I played Quake 3 for about two hours, and uninstalled the card (who says testing isn't fun?). After removing the stock push pins, a little twist and turn, and off came the heatsink.
Click to enlarge
I should mention that once you've done the above, you probably just voided your warranty.
You should thoroughly clean the GPU. I prefer to use my girlfriend's nail polish remover (yes, it's really hers) to remove thermal material. Afterwards, I wipe the remover off with some rubbing alcohol, and the GPU is clean. After that, you can use the included packet of thermal paste, or if you have a higher quality paste (like it'll make any difference), to apply on the GPU. I shouldn't have to tell you that you only need a thin amount. I chose not to lap the GeForce4 Copper Cooler, since the finish was already excellent.
Click to enlarge
Installing the ramsinks is a snap. If you know how to use double-sided tape, you'll have the ramsinks on in no time. Just a note: There are no instructions on installation included with the GeForce4 Copper Cooler. Now, installation is straightforward, but there are some arguments about which ramsinks you should use on the front and back. The majority of reviews I've seen slap the ramsinks on the front of the card. To me, this doesn't make a whole lot of sense, as the first fin will negate airflow over the other fins. Also, because the back of the video card doesn't have active cooling, I thought it would be better to put the larger heatsinks in the back of the card. For the record, our overclocking results with the ram was the same under both scenarios (big sinks in front, or big sinks in back), but the back ram did run 1C cooler with the larger heatsinks.
We're going to look at a few areas where such a kit may prove useful, if at all. To be honest, I'll be expecting lower temperatures across the board, which as we know, is always a good thing, but more importantly, we will be examining if overclocking will be more successful with the GeForce4 Copper Cooler. We will also be examining the temperatures while overclocked.
The Test Setup will include the following:
Visiontek Xtasy Ti4600
Lian-Li PC65U, with only stock intake and exhaust fans on
Thermaltake Hardcano 2 to gauge the temperatures
Ambient system temperature will remain about 32C during testing
To stress the card:
5 x Quake 3 Timedemos, followed by 4 x 3D Mark 2001 Benchmarks
One temperature probe will be attached on the back of the card, directly behind the GPU, during the tests. I considered attaching the probe to the edge of the GPU, but since the center of the GPU is where the majority of the heat will be generated, I felt behind the GPU would net more accurate results. Keep in mind that this isn't a true measure of the core temperature, but by placing the probe between the GPU and heatsink will almost certainly result in inaccurate temperatures.
One temperature probe will be placed on the back of the card, on the edge of the memory. We did it this way because I feel that the memory as a whole will reflect the same temperatures in the middle and on the edge.
When the computer isn't doing anything, both the stock cooler and the GeForce4 Copper Cooler perform very closely top one another. When the going gets tough though, the GeForce4 Copper Cooler pulls ahead (or behind if you want to look at the numbers that way), and records readings as much as 7C lower.
I think the graphs speak for themselves. Obviously, the BGA ram does benefit with some form of cooling, and runs 3C cooler with the ramsinks. Note that the temperatures for the ram are for the back ram modules, where there was no active cooling. The front ram modules were about another 2C cooler due to the air moving from the GPU cooler.
Previous Page - Introduction
Next Page - Overclocking and Final Words