The Radeon X1950 Pro
Recently, ATI invited us to a press briefing where we were given more details about the new Radeon X1950 Pro graphics card.
Although the new card may seem like a slightly improved version of the Radeon X1900 GT, it's actually more than that.
The new Radeon X1950 Pro signals a new beginning for ATI, as far as their multi-card strategy goes. The days of master cards and dongles are numbered. They were just plain silly, weren't they?
Now that ATI has a new system to replace the CrossFire dongle, they have not been shy in acknowledging the problems and disadvantages of the CrossFire dongle. But let's leave CrossFire for later and take a look at the new card in detail.
Specifications
The Radeon X1950 Pro is based on the new ATI RV570 GPU, unlike its X1950 brethren - the X1950 XTX and X1950 CrossFire Edition which are based on the R580+.
While all three X1950 cards come with 8 vertex shaders, the X1950 Pro will have 25% fewer pixel shader ALUs and ROPs than its bigger brothers.
In addition, the X1950 Pro's GPU will be running at a slower clock speed of 575 MHz, giving it a fill rate of 6.9 Gigatexels/s. Its 256MB GDDR3 frame buffer will run at 690 MHz (1380 MHz DDR), giving the card a memory bandwidth of 44.16 GB/s.
That means the new Radeon X1950 Pro is really an improved version of the Radeon X1900 GT, with 15% more memory bandwidth. In fact, ATI will be phasing out the X1900 GT, essentially replacing it with the slightly faster X1950 Pro.
Armed with more memory bandwidth, the Radeon X1950 Pro was meant to challenge the NVIDIA's GeForce 7900 GS, with a target price of just US$199 or EUR 199. Let's compare the Radeon X1950 Pro against some of its compatriots and the GeForce 7900 GS.
Radeon |
Radeon |
GeForce |
Radeon |
|
Architecture |
R580 |
RV570 |
G71 |
R580 |
Manufacturing Process |
0.09 Micron |
0.08 Micron |
0.09 Micron |
0.09 Micron |
Transistor Count |
384 Million |
330 Million |
278 Million |
384 Million |
DirectX Support |
9.0c |
9.0c |
9.0c |
9.0c |
Interface |
PCIe |
PCIe |
PCIe |
PCIe |
Vertex Pipelines |
8 |
8 |
7 |
8 |
Vertex
Shader Version |
3.0 |
3.0 |
3.0 |
3.0 |
ROPs |
16 |
12 |
16 |
12 |
Pixel Shader Engines |
48 |
36 |
20 |
36 |
Pixel
Shader Version |
3.0 |
3.0 |
3.0 |
3.0 |
Core
Speed |
625 MHz |
575 MHz |
450 MHz |
575 MHz |
Fill
Rate |
10,000 MTexels/s |
6,900 MTexels/s |
9,000 MTexels/s |
6,900 MTexels/s |
Memory
Bus Width |
256-bits |
256-bits |
256-bits |
256-bits |
Memory Type |
GDDR3 |
GDDR3 |
GDDR3 |
GDDR3 |
Memory
Speed |
725 MHz |
690 MHz |
660 MHz |
600 MHz |
Memory Bandwidth |
46.40 GB/s |
44.16 GB/s |
42.20 GB/s |
38.40 GB/s |
For more specifications of these and many other cards, you can check out our Desktop Graphics Card Comparison Guide.
The Radeon X1950 Pro appears to be slower than the GeForce 7900 GS in fill rate and is roughly its match in memory bandwidth. However, we must not forget that the ATI card has almost twice as many pixel shader engines as the NVIDIA card.
The Radeon X1950 Pro can apply multiple effects on each pixel almost twice as fast as the NVIDIA card. Of course, the X1950 Pro's pixel throughput is lower than that of the GeForce 7900 GS because it has fewer ROPs.