Test System Setup
Processor(s): Intel Pentium 4 2.4GHz "C"
Motherboard(s): Abit IC7 MAX3 (Supplied by Altech Computers)
Video card(s): PowerColor Radeon 9800 Pro (Supplied by Altech Computers)
Hard Disk(s): Maxtor SATA 80GB 7200RPM (Supplied by Altech Computers)
Operating System Used: Windows XP Professional SP1
Drivers Used: Catalyst 3.5
Software Used: SiSoft Sandra Max 3, PC Mark 2002, 3DMark 2001 SE, 3DMark 2003, Unreal Tournament 2003, Jedi Knight 2 , Commanche 4 and Quake 3
Due to the massive amount of memory being tested, we choose to cut down on the amount of tests when compared to our last Dual Channel roundup. We have chosen to include SiSoft Sandra Max 3 as it really does give us the maximum theoretical speeds of memory and it gives people a good idea on how these numbers affect real world gaming. PCMark 2002 is another benchmark that is purely numbers, it gives us an idea on what the maximum performance of the memory is but when it comes to real world gaming we don't always see these numbers translate correctly.
We couldn't run a set of benchmarks without the famous 3D Mark 2001 and 3D Mark 2003. While some people will take 3D Mark 2003 results more seriously then others, we couldn't help but include them for people to get a better grasp of the speeds and to use for their own comparisons.
We included a list of gaming benchmarks including UT2003, Jedi Knight 2, etc. We have run all gaming benchmarks in two resolutions - 640 X 480 and 1280 X 1024. 640 X 480 gets rid of the video card bottle neck and concentrates on getting a higher frame rate with faster processor speeds and quicker memory. While this is all good and well a lot of people don't play at this resolution so we have opted to run the benchmarks at 1280 X 1024 as well.
We have chosen to run all the memory at a list of different speeds. All memory was ran at DDR400 using the timings 2.5-4-4-7 and 2.7V. We then ran all the memory at default speed with SPD - in the event that the memory was PC3200 or DDR400, we left their default test as the original tests we first spoke of. Finally we moved into maximum overclocking raising the memory voltage between 2.9V to 3.2V and we attempted to achieve the maximum overclock of each brand of memory. While at this voltage extra cooling came into the equation. We wouldn't recommend running voltages over 2.9V for extended periods of time without added cooling. A maximum overclock was not counted unless it could complete all tests multiple times and the relaxed timings 3-4-4-8 were used all across the board to get the maximum overclock.
When it comes to the graphs there are a few different ways the memory can win. First up we will be looking for the fastest memory at DDR400 speeds. Since we aren't all overclockers, some people will want to simply run their memory at processor spec. Another is the fastest memory when it comes to PC3500/PC3700 speeds as well as the advantage these two different speeds make. Finally we have the fastest overall memory which we would expect will be one of the PC4000 modules. All these different numbers will give people a better idea of what is best for them. Finally we will look at overclocking which will give the more power hungry user an idea of what they can expect and if it's really necessary to spend big on PC4000 memory.
Hardware Run Down
There have been a few hardware changes since our last round up. The most important has been the upgrade to the Abit IC7-MAX3, this board has quickly become known across the hardware community as one of the best, if not the best Canterwood motherboard on the market. One of the best features on this motherboard when it comes to testing memory is the maximum voltage of 3.2V. Most motherboards on the market don't go above 2.8V and this can limit your overall overclock of the memory, with 3.2V we are able to get the most out of the memory which gives our readers a better off aspect of what the memory is capable of.
We have moved from the P4 2.8C that we had last time over to a 2.4C. The 2.4C is known as a better overclocker and gives us the chance to try and remove another bottleneck that we may have with maximum FSB. We have continued using the Radeon 9800 Pro as it is still one of the fastest video cards on the market, besides the new 9800 XT which we didn't have access to during the time of testing.
Without any further ado, let's see how the memory modules go at DDR 400 and beyond!