Under The Hood
More and more often, we are seeing video card manufacturers go from gluing the heatsink to the chipset, to using thermal paste, and fastening the heatsink externally. It's obviously better to have thermal paste between the heatsink and chip than it is to have some cheap glue, but using external pins also makes it easier to install other heatsinks, and replace them more frequently.
The sad part, though, is most manufacturers simply can't be bothered to apply the heatsink paste thoroughly... As soon as I saw this, I wiped it all off, cleaned it up with some rubbing alcohol, put some Arctic Silver paste on it, and slapped the heatsink back on. Man, this helps a lot! We also tested overclocking with ThermalTake's new Blue ORB chipset/video card cooler..
We also put a Tennmax cooler on there, for our future shootout, but Tennmax tells us that their coolers don't quite agree with Arctic Silver Paste, and we were seeing temperatures rise way above stock. They are sending us a new unit, just in case ours is faulty (I've had it for quite a while), and some thermal paste they prefer to be used with their coolers. Look for a full Tennmax vs. Blue ORB shootout very soon!
I Have To At Least MENTION The Specs
I know you hate it when hardware reviewers use up an entire page (or TWO!) just to show you the specs you can easily check out yourself. I hate that too. I will list a couple things though, and move on, just in case you aren't entirely familiar with the GeForce 2 MX.
The GF2MX is actually based on the same core as NVidia's current flagship product, the GeForce 2 GTS (here's a review of one of those). The difference though, is that the GF2MX only supports 2 rendering pipelines (the GF2GTS has 4). Why not just stick with the original GeForce then? It also supported 4 rendering pipelines as well, right?
That's true, but the one of the big improvements the GF2 core had over the original GF was the ability to process 2 textures per pipeline in a single clock, where the GF could only process 1.
Long story short and simple: The GeForce 2 MX, because of its enhanced chipset features and clock speed, has a higher fill rate than the original GeForce card. Because it uses only SDR ram though, it will still be hampered by the low memory bandwidth, like the original SDR GeForce. This means that the GeForce 2 MX is a replacement for the GeForce chipset. In fact, I don't think you will be able to find an SDR GeForce card in stores for very long, if at all. Nvidia now has basically the entire market covered: Low end, high performance graphics - GeForce 2 MX. Middle end, high performance graphics - GeForce DDR. High end, high performance graphics - GeForce 2 GTS. And the ultimate for hardcore gamers: GeForce 2 GTS Ultra (coming soon). 3dfx who?
Back to the specs: here are some of the features of the GeForce 2 MX chipset:
Curious exactly how the GF2MX compares to the GF2GTS and even the original GeForce on paper? Here's a small table for you. These stats are all official from Nvidia.
There are the numbers for ya. I highlighted the Fill rate numbers, because those are probably the most important overall. This is of course forgetting about memory bandwidth, which is very important, particularly in 32 bit applications (and who doesn't run 32 bit applications).
And Now - The Card Itself
Now it's time to throw some real numbers at you - BENCHMARKS! First off, here are the system specs, and our benchmarking process. I'll make this real simple for you (again, no page wasted on showing you OUR specs!):
AMD Duron 900 (100 Mhz Bus)
Copyright © 2000, 2001, 2002 hardCOREware. All rights reserved.
All trademarks used are properties of their respective owners.
Use IE5 or 6 to see this page properly. 1024x768 resolution is a minimum.
32 bit colour is recommended. Our Privacy Statement