Date: December 8th, 2003
Article by: Nathan Glentworth (Owner, Head Editor & Hardware Reviewer)
Product was donated by: Corsair Micro
<--Shop for this item

BENCHMARKING SETUP AND COMPOSITION

Intel P4 2.4C 800FSB Retail CPU Check Prices
Abit IC7-G Max II Advance Canterwood Motherboard (BIOS Revision Version 16): Check Prices
1GB corsair TwinX XMS4000pro High Performance Dual Channel Memory Check Prices
512 megs Corsair TwinX Dual Channel Series PC3200 Low Latency Memory Check Prices
512 megs OCZ Platinum Dual Channel Series PC3200 Low Latency Memory Check Prices
Vmem = 2.8volts
Vcore = 1.60volts
ATI All-in-Wonder 9800pro Check Prices
Catalyst 3.5 Videocard Drivers
Windows XP Professional SP1 Check Prices
Seagate Barracuda V SATA ATA-150 HD Check Prices
Intel Application Accelerator ver2.3
Hyperthreading and PAT Enabled


Benchmarking Programs Used:

- 3DMARK2001SE
- PCMARK2002 (MEMORY)
- SISOFT SANDRA MEMORY BANDWIDTH BENCHMARK
- GLEXCESS v1.2V BENCHMARK
- QUAKE 3 Q3CRUSH DEMO BENCHMARK


Processor Configurations Used:

- 2.4C @ 2.4Ghz (DDR400) (1:1) (2-5-3-2 Ram Timings) OCZ & Corsair
- 2.4C @ 3.0Ghz (DDR400) (5:4) (2-5-3-2 Ram Timings) OCZ & Corsair / 1:1 DDR500 Corsair XMS4000
- 2.4C @ 3.06 (1:1 DDR510 Corsair XMS4000)

Benchmarking Explained:

With the different configurations seen above, I will be testing the Corsair XMS4000 memory clock for clock against the equivalent Corsair TwinX PC3200LL Module and the OCZ PC3200 Platinum Memory I have reviewed previously.

The one main difference which will be labeled in every graph is that I test the slower Corsair and OCZ at 2-5-3-2 memory timings whereas I could only test the XMS4000 at 3-8-4-4. Even at DDR400, the XMS4000 would not boot even into the BIOS if 2-5-3-2 timings were applied. But as you will see, timings don't play a huge role in memory performance. When it comes to raw data transfer, megahertz matters and this is where the XMS4000 pulls away from everyone.

Everything else is absolutely identical right down to the software and drivers. I will not be testing this memory at the CPU's highest overclock seeing to do so, I would be using the memory divider and essentially underclocking the memory below it's rated speed. *Note: It did reach 3.45Ghz without a sweat.*

What I will show you is just how much of an overclock you can obtain through pushing up the front side bus and keep the divider at it's 1:1 stock ratio. As you will see in the next pages, I was able to push this memory to 510Mhz or 255FSB while still keeping the divider stock. This brings about a massive overall performance boost to the whole computer and contributes to a noticeable reduction in OS and program load and execution timeframes.

Let's see how it performed....