For the purpose of this test I will be fitting the OCZ Flex XLC into the following spec PC and connecting it up to the currently installed water-cooling system.

Once the Flex XLC is installed I will run several bandwidth tests and some game benchmarks at stock speed and timings. Then I will rerun those test and see what is the best speed I can achieve at what timings and do some comparisons against the stock settings. 

Fitting the OCZ Flex XLC was pretty straightforward, they slotted into the motherboard no problem, as you would expect.

Next was the task of plumbing it into my existing water-cooling system which was a different size bore so I had to utilise some adapters (kindly supplied by WaterCoolingUK who always come to our rescue with water cooling parts) to enable me to water cool them.

I originally wanted to go in and out of the bottom with a connecting loop at the top but the pressure of the pipe on the memory forcing them apart dictated I had to fit the pipes as above. Once this was done I leak tested, sorted out a couple of leaks on the adaptors, and carried on with the testing.

First I set the Flex XLC to its rated stock speed of 800 MHz with timings of 3-4-4-15 and the recommended 2.4v.

Once I had run all the tests with the memory at stock speed and timings I set about finding out just how fast I could get it to go at the stock timings of 3-4-4-15 at the rated 2.4v.

The best I got was 900 MHz, which it coped with with ease. I couldn’t get much higher at CAS3 so next I tried CAS4 at 1000MHz, which worked fine.

Unfortunately I couldn’t seem to persuade the memory to go any higher even at CAS5, and as my motherboard can’t supply much more voltage, its max is 2.45volts, I can’t even try upping the voltage.

Test results as follows.

Everest Memory bandwidth Read / Write and Copy.

The Memory Copy and Memory Read at CAS3 and 900mhz seem a bit odd but I ran them several times and those readings are typical of the results I got every time. I guess it maybe due to the divider I used (the P5B Deluxe seems to prefer NOT to be run on a 1 to 1 divider), as the 1000 MHz CAS4 readings were a big jump over stock. However the Memory Write speeds of the 900 MHz CAS3 were higher than the CAS4 1000 results, quite strange really and not what I was expecting at all.

Sisoft Sandra results for memory as follows.

The Sisoft Sandra results are more as you would expect and the increase in performance between 900 MHz CAS3 and 1000mhz CAS4 is quite small.

Next up is PC Mark05 a good overall test of the pc.

Here we see the strange apparent performance dip at CAS3 900MHz, again possibly due to not running the memory on a divider.

For the game tests I chose F.E.A.R. this is still a demanding game for all but the highest end of equipment. I used the in game stress test and with all settings set to maximum and a resolution of 1280x1024 I achieved the following results.

Strangely enough the results from the in game performance test show no gain in performance from increasing the memory speed, in fact I got the highest frame rate at stock settings. I am not really sure why I can only guess this is possibly down to the graphics card being bottlenecked by the rest of the system.

After doing the testing the sweet spot appears to be 1000mhz CAS4 a bit higher e.g. 1066 and the performance actually drops, although this may also be due to the fact the system became unstable and was unable to complete some tests.

Next Page - Conclusion