|
VisionTek couldn't have picked a better time to switch teams.
Since VisionTek flipped its allegiance, NVIDIA has struggled to overcome early GeForce FX problems and endured an ugly benchmark cheating/optimization fiasco. At the same time, ATI's new line of R3x0-derived Radeons has dominated mid-range and high-end enthusiast markets. I suppose one could accuse VisionTek of jumping on the ATI bandwagon, but can you blame them?
Today we've wrangled VisionTek's latest high-end graphics card, the Xtasy 9800 Pro 256MB. As I'm sure you've already guessed, the card is based on ATI's Radeon 9800 Pro, and comes packed with 256MB of memory. Loading up graphics cards with gobs of memory has become a clever marketing trick for manufacturers looking to spice up their low-end and mid-range graphics cards, but often those cards lack the horsepower to take advantage of the extra memory. Can the Xtasy 9800 Pro's R350 graphics chip push enough 3D pixels to make use of the card's extra memory? Read on to find out.
The specs
We reviewed ATI's Radeon 9800 Pro back in March, so I won't dwell too much on the graphics chip's capabilities. 128MB and 256MB flavors of the Radeon 9800 Pro use the same R350 graphics chip clocked at 375MHz, so they're essentially identical when it comes to feature support, DirectX capabilities, and the like.
| GPU | ATI R350 |
| Core clock | 380MHz |
| Pixel pipelines | 8 |
| Peak pixel fill rate | 3040 Mpixels/s |
| Texture units/pixel pipeline | 1 |
| Textures per clock | 8 |
| Peak texel fill rate | 3040 Mtexels/s |
| Memory clock | 700MHz |
| Memory type | BGA GDDR2 SDRAM |
| Memory bus width | 256-bit |
| Peak memory bandwidth | 22.4GB/s |
| Ports | VGA, DVI, and S-Video outputs |
| Auxiliary power connector | Four-pin Molex |
Not only do 128 and 256MB Radeon 9800 Pros differ in memory size, the two card flavors also feature different memory technologies running at different clock speeds. 256MB cards have a 20MHz advantage over 128MB cards in memory clock speeds. That 20MHz lead translates into an extra 640MB/sec of memory bandwidth, which should give 256MB cards a very slight advantage in scenarios where memory bandwidth becomes a key bottleneck.
Part of the reason for different memory clock speeds for 128MB and 256MB Radeon 9800 Pro cards is the fact they use different memory types. 128MB cards use plain old DDR SDRAM, while 256MB cards use swanky new GDDR2 chips. GDDR2 is a relatively new technology that's supposed to enable higher clock speeds, but given the Radeon 9800 Pro 256MB's paltry 20MHz memory clock speed advantage, it doesn't look like GDDR2 has a significant clock speed advantage just yet.
Since the R350's memory controller was designed to work with both DDR and GDDR2 SDRAM, ATI didn't have modify its GPU chip to support the new memory type.

| In the lab: AMD's Radeon RX 460 graphics card | 47 |