Cas 3/1T, maximum overclock testingIn addition to our cas 3 /1T testing at a fixed 400 front side bus, we decided to find out how high our memory kits would scale at the same settings. For the record, we didn't expect the memory modules to scale very far, as the cas 3/1T setting is a very stressful set of timings indeed. We should be rewarded with some very nice bandwidth and latency results however.
To test the different modules, we just increased the front side bus, and let the memory run in sync with the FSB. We increased the FSB 5 Mhz at a time and stressed the memory for at least 20 minutes with the application "Orthos", a derivative of prime 95 for dual core cpu testing. We went on to increase our front side bus until Orthos failed within 20 minutes, and recorded that speed in our following graph.
Below you can see the different screenshots for the Orthos runs at the cas 3/1T maximum memory speed (click to open).
Corsair PC8888C4F, Corsair PC6400C3F, Team Xtreem PC6400C3
G.Skill PC8500C5, OCZ PC9200 Flex, OCZ PC8500 Reaper, OCZ PC6400 FlexThe above screenshots are combined in the following graph :
As you can see, a nice victory for the Team Xtreem kit !
Cas 4/1T, maximum overclock testingAfter the cas 3 /1T results, it's time to move to cas 4, again starting at 400 front side bus and moving up. The reason we tested this setting is the fact that synced operation still gives the best overall results, and we should be able to come close to the 500FSB mark with the very best memory kits available. For the testing, the same method was used as the cas 3/1T setting.
Below you can see the different screenshots again for the Orthos runs at the cas 4/1T maximum memory speed (click to open).
Corsair PC8888C4F, Corsair PC6400C3F, Team Xtreem PC6400C3
G.Skill PC8500C5, OCZ PC9200 Flex, OCZ PC8500 Reaper, OCZ PC6400 FlexThe above results conclude into the following chart :
This time our OCZ PC9200 kit comes out on top, with a very nice lead over the Reaper kit, and the PC6400 Team Xtreem kit.
Cas 4/2T, maximum overclock testingOur screenshots of the previous table result in the following table :
As you can see, the Dominator PC8888 rules this chart with a mind blowing 615 Mhz (1230 Mhz DDR), and this at warranted voltages. Incredible! Our second Dominator kit is also far from bad, as is our Team Xtreem kit.
Finally, all the results into an easy to use table for your reference:
Cas latency comparison: the best setting for 680i ?After seeing the above graph, one question comes to mind: which setting should I use for the 680i board ? As we mentioned before, a synced setting will probably give you the best results, and in my case everything above 400 FSB, but slower than 500 FSB, and synced, provided tremendous results. As always you'll have to make a choice between tighter latencies or higher memory speeds of course, and for that I'd like to provide you with three final screenshots: our best results in bandwidth for our three different test settings :
Bandwidth cas 3/1T, cas 4/1T and cas 4/2TSurprised? As you can see: the best results are clearly the synced ones, with a CR of 1T, but the difference between cas 3 and cas 4 are very small indeed. It's clear now that a cas 4 setting cannot really benefit from the much higher front side bus that is reached, so you should go with the setting that gives your CPU the best possible speed with a given multiplier.
About time for our value rating of the different memory kits >