First up, I tested the PC3200 in asynchronous mode at DDR400 speeds on my Albatron microATX nForce 2 board, naturally in dual channel mode. All timings were by SPD and I booted into Memtest86, a very handy tool that allows you to test memory in a non-windows environment without any risk of data corruption. Unfortunately, even with timings set to 'by SPD', the memory made 5 errors per pass at the same point during test 5. Setting the timings to a more relaxed speed or reducing the memory speed by 10MHz reduced the errors to zero, but nonetheless the Twinmos has got off to a shaky start. To give it a fair test I checked the voltages of the memory in the BIOS and actual voltage was at around 2.44V.
If you remember from the previous pages, Twinmos rate this memory up to DDR400 at 2.6V. Since the Albatron has no V-DIMM adjustments, I had to rebuild my Asus A7N8X based system to continue AMD based testing. At 2.6V on the Asus (2.62V actual) I was able to get the memory to run happily all day in Memtest without errors at the suggested SPD timings. Setting CAS to 2 and timings to 6-3-3 also had no effect on the stability at this voltage - no problems. Seems this memory does exactly what it says on the tin on the nforce 2 platform!
Moving over to the Canterwood based Abit IC7-G will allow us some more flexibility to test the memory at various speeds and timings, as well as verify that the memory is compatible with this rather picky platform. The Twinmos booted up without issue in dual channel mode on the canterwood and once memory voltage was set to 2.6V, once again the RAM showed itself to be fully stable at its advertised settings.
Overclocking
I am happy that the Twinmos PCB is of a sufficient quality to handle elevated voltages, so have no qualms about hiking the V-DIMM up to 2.8V to test the overclockability of this memory. The maximum speed I was able to obtain at the SPD timings was 413MHz stable, not bad at all for memory of this price point. At 2.8V, with timings set to CAS 2, 5-2-2 i was able to hit 390MHz, with the optimum compromise of timings and clock speed being 406MHz at CAS 2, 6-2-3. At these settings Sandra reported some excellent results, 4615MB per second is not to be sniffed at.
