ATI Radeon CrossFireX Review
ATI divides its products into segments based on price, power requirements and performance. The complete breakdown is as follows:
- Ultra Enthusiast: 5970, Eyefinity 6 and CrossFireX
- Enthusiast: 5870, 5850, 5830
- Performance: 5770, 5750
- Mainstream: 5670, 5550, 5450
To date we have benchmarked and reviewed the products in the Ultra Enthusiast and Enthusiast segments, along with the Performance and Mainstream segments. I tested the Eyefinity6 in CrossFireX during its review. Here, I am testing CFX performance of the original 5870, 5850 and 5770.
Architecture & Specs
Below is a spec block with comparing outlining the 5870 E6 through the 5770:
Card | GPUs | Transistors | Max Memory | Shaders | Clock (MHz) | TDP (Watts) | MSRP* | ||
Core | Mem | Idle | Max | ||||||
ATI Radeon HD 5870 E6 | 1 | 2.15B | 2GB | 1600 | 850 | 1200 | 27 | 188 | $479 |
ATI Radeon HD 5870 | 1 | 2.15B | 1GB | 1600 | 850 | 1200 | 27 | 188 | $399 |
ATI Radeon HD 5850 | 1 | 2.15B | 1GB | 1440 | 725 | 1000 | 27 | 151 | $324 |
ATI Radeon HD 5770 | 1 | 1.04B | 1GB | 800 | 850 | 1200 | 18 | 108 | $174 |
As these cards have been on the market some time, price is based on current average |
The original 5870 and the 5870 E6 have identical specs - outside of the RAM and power draw. The 5850 reduces the shaders and both clock speeds. The 5770 reduces shaders further, but returns the clock speeds to that of the 5870.
Notes on Benchmarking & Data
There are a number of instances where moving from a single card to CFX produces strange performance comparisons. These instances occur in the most demanding tests, where the single card either fails to run the test or barely runs the test. When the single card posts 1fps, and the CFX combination posts 5fps, this is mathematically a 500% improvement. In reality this is a 4fps improvement, and still unplayable. When the single card won't run the test at all (effectively 0fps), it works to an infinite improvement mathematically.
In the instances either of these circumstances occurs, I have left the percent improvements out of the charts and analysis. The raw improvement is still in the graphs, but will not play into our overall analysis.