Page 7 of 10
Test Setup and Overclocking
The regular Fatal1ty AN8 was a pretty good overclocker and continues to be a great performing motherboard. The Fatal1ty AN8 SLI is no different from the previous board. I proceeded to find the highest HTT allowed with dual sticks of memory installed. The first test will be conducted with the memory divider at 1:1. The second test will be with a divider, so that the RAM wouldn't pose as a bottleneck. All memory timings were left at 3-4-4-8 2T and the LDT multiplier at 1x and CPU multiplier at 6x. I made sure that nothing would bottleneck the chipset and motherboard potential.
The Fatal1ty AN8 SLI overclocked rather well with the use of a divider. 315HTT seems to be the max when it comes to 1:1 ratio, even when the memory is capable of much more.
There is one problem with the Fatal1ty AN8 SLI that many enthusiasts have been encountering. The sample I received also has this problem. It seems to vary from board to board, depending on your luck. As usual, my bad luck punched me in the face and refused to let me run 2x512MB memory at 1T command rate. This seems to ONLY happen with an AMD Athlon 64 3000+ Venice core CPU. At first I thought it was the BIOS, but I quickly found out that it was not the cause. I went through four different versions before concluding that it was my CPU. I figured that it might be my old date/week CPU since there have been enthusiasts who were able to run a 3000+ Venice without a problem. I switched my week 04 Venice to a week 17 Venice. The problem was still there. This compatibility issue is very specific to AMD Athlon 64 3000+ Venice core CPUs and using 2x512MB memory and a command rate of 1T. I tried different kits of memory, both TCCD and Winbond UTT and they all refused to run 1T. The strange thing is that running 2x256MB solved the problem. I was able to push well beyond 300HTT with 2x256MB and 1T, but who runs 2x256MB nowadays? This compatibility issue is extremely annoying, yet minor. The problem is very specific to a certain CPU and the use of 2x512MB memory. Unfortunately, the specific CPU is an extremely popular processor. Many enthusiasts purchase a 3000+ Venice for the low price and the high overclocking potential. Just make sure that you keep this problem in mind when you make your purchase decision. If you're not using a 3000+ Venice, no need to sweat bricks. You will certainly not have a problem with the board. The funny thing is that I have the regular AN8 SLI sitting here and it does not have this problem. The boards are identical in design, layout, and BIOS. There is no difference other than the Fatal1ty name and the chipset cooling. Perhaps Abit can fix this problem in the near future. It is very disappointing.
We just received word from a fellow enthusiast in Australia who reported that he encountered similar problems. The CPU he was using was an AMD Athlon 64 X2 3800+. The memory he was using was 2x512MB of Corsair's PC3200 XL TCCD. He was forced to run the command rate at 2T and loosening his timings to 2.5-4-3-7 in order to run a stock 200HTT. This further extends the problem with the motherboard. It just goes to show that the Fatal1ty AN8 SLI has problems with more than just a specific CPU. I hope Abit realizes this problem and quickly fixes it, because quite frankly, it has angered many loyal Abit fans, myself included.