The X1650XT is built around the new RV560 core which is manufactured using an 80nm process and is clocked at 575MHz. This is coupled with 256MB of GDDR3 memory running at 675MHz (1,350MHz effectively) and connected over a 128-bit bus. Buried away within the core are eight vertex shaders, eight pixel pipelines (with 24 pixel shaders), and eight raster operation pipelines (ROPs). This is quite a step up from ATIs previous mid-range offering, the X1650 Pro (based on the RV530 core), which could only boast five vertex shaders, four pixel pipes (12 pixel shaders) and four ROPs. It’s also, at least on paper, significantly betters than the nVidia G73 core (on which the 7600 GT is based), which can only boast five vertex shaders, four pixel pipes (12 shaders), and eight ROPs. However, the Fatal1ty 7600 GT comes clocked at 650MHz with memory running at 800MHz (1600MHz effectively) so there’s a good chance it may outperform the X1650XT.
To test the card I ran it through our usual set of benchmarks using our reference Intel 975XBX “Bad Axe” motherboard, with an Intel X6800 Core 2 Duo, coupled with 2GB of Corsair CMX1024-6400C4 running at 800MHz with latency settings of 4-4-4-12.
I ran Call of Duty 2, Counter Strike: Source, Quake 4, Prey, Battlefield 2 and 3DMark06. Except for 3DMark06, these all ran using our in-house pre-recorded timedemos in the most intense sections of each game we could find. Each setting is run three times and the average is taken, for reproducible and accurate results. I ran each game test at 1,280 x 1,024, 1,600 x 1,200 and 1,920 x 1,200 each at 0x FSAA with trilinear filtering, 2x FSAA with 4x AF and 4x FSAA with 8x AF.
I’ve recalled results from the Fatal1ty 7600 GT review to compare to, as well our reference 7600 GT results. Some results are missing due to different resolutions being used when we tested the cards but there are still enough results to get a good picture of performance.
It is quite clear that the X1650 XT is a good match for the reference 7600 GT but it consistently falls behind the over-clocked version. The reference 7600 GT consistently starts out faster but drops away much more rapidly than the X1650XT, but even as you crank the AI the overclocked Fatal1ty always stays ahead. However, the Fatal1ty board does cost £20 more than either the X1650 XT or the standard clocked 7600s so you would expect it to have a performance advantage.
Visual quality is a factor and in this area the X1650 XT wins. The 7600 GT has always been plagued by poor texture filtering and has been trumped by the fact the X1000 series can perform AA with HDR lighting. However, at this level of speed it’s possibly a moot point. The X1650 XT isn’t really fast enough to run HDR and AA together, unless you drop resolutions heavily, which isn’t what you want to do if you have an LCD display.
Also of note is the inclusion of dual dual-link, HDCP compliant, DVI-I ports on the X1650 XT that mean you can run protected HD content on two massive 30inch monitors all off the one card. Most 7600s only feature at most one dual-link DVI-I link port and HDCP compliance is also variable.
In terms of raw performance the X1650 XT is a very worthy competitor to the 7600 GT and given its better texture filtering, its ability to process both HDR and AA simultaneously, and it’s better connectivity options, I’d say it’s the better choice of the two.
However, unless you’re totally desperate I’d hold out for a couple of months until mid-range DX10 hardware is released. At the very least this will cause the price of these cards to drop.
Unlike other sites, we thoroughly test every product we review. We use industry standard tests in order to compare features properly. We’ll always tell you what we find. We never, ever accept money to review a product. Tell us what you think - send your emails to the Editor.