I decided to play a bit this afternoon. I wanted to see the throughput differences with various memory configurations. Before you post and flame me... I AM NOT TRYING TO SELL YOU ON A SPECIFIC CONFIGURATION. I'm just posting the averaged results to my tests. Each configuration was tested 10 times and the score was averaged. The tests were done in Single User mode, CLI only. No Networking, GUI, services... nothing. Config 1: 2x 512MB, Apple OEM Geekbench Score: 363 Config 2: 4x 512MB, Apple OEM, All DIMMs on Riser 1 Score: 359 Config 3: 4x 512MB, Apple OEM, 2x Riser 1, 2x Riser 2 Score: 418 Config 4: 2x 1GB, Dell OEM Geekbench Score: 359 Config 5: 4x 1GB, Dell OEM, All DIMMs on Riser 1 Score: 357 Config 6: 4x 1GB, Dell OEM, 2x Riser 1, 2x Riser 2 Score: 421 Config 7: 2x 512MB, Apple OEM, 2x 1GB Dell OEM, 2x Riser 1, 2x Riser 2 Geekbench Score: 347 Config 8: 2x 512MB, Apple OEM, 2x 1GB Dell OEM, All DIMMs on Riser 1 Score: 358 Config 9: 4x 512MB, Apple OEM, 4x 1GB, Dell OEM, 2x2 Riser 1, 2x2 Riser 2 Score: 411 Config 10: 4x 1GB Dell OEM (2x R1, 2X R2), 2x 512MB Apple OEM (2x R1) Score: 361 My best result came from 4x 1GB DIMMs, configured in DIMM1/2 on each riser. The console in rEFIt confirms this is Quad Channel. The worst result was 2x 1GB on Riser 1 and 2x 512MB on Riser 2. The console confirmed that this is dual channel. The results were not as dramatic as I expected. I did expect closer to 2x the throughput between Dual and Quad channel mode. There is an improvement of about 18%, and this will show up in memory intense apps. I think the increase would be more if the benchmark used larger blocks of data.