Test Results: Corsair XMS4400 & OCZ PC4400

To test overclocked stability, we used the very demanding Gun Metal 2 - Benchmark 2, which pushes systems with its DX9 routines. To be considered stable for test purposes, Gun Metal, our Quake3 benchmark, UT2003 Demo, Super PI, Halo, and Comanche 4 had to complete without incident. Any of these, and in particular Super PI and Gun Metal, will crash a less-than stable memory configuration.

Corsair XMS4400v1.1 TwinX - 2 x 512Mb Double-Bank
Speed Memory Timings & Voltage Quake3 fps Sandra UNBuffered Sandra Standard Buffered Super PI 2M places
(time in sec)
400DDR
800FSB
2-3-3-7
2.55V
318.1 INT 2663
FLT 2639
INT 4442
FLT 4445
133
500DDR
1000FSB
2.5-3-4-7
2.65V
390.2 INT 3147
FLT 3228
INT 5549
FLT 5520
108
533DDR
1066FSB
2.5-4-4-7
2.75V
411.9 INT 3345
FLT 3362
INT 5842
FLT 5819
102
550DDR
1100FSB
3-3-4-8
2.75V
422.9 INT 3415
FLT 3529
INT 6054
FLT 6008
98
560DDR
1120FSB
3-3-4-8
2.75V
436.8 INT 3606
FLT 3538
INT 6124
FLT 6101
95

Corsair XMS4400 did not perform any better or reach higher overclocks when using 2.85V for the memory. In fact, performance was often worse or we got a no-boot condition when 2.85V was selected. 2.75V allowed us to get all we could from Corsair's DDR550.

OCZ PC4400 Dual-Channel Kit - 2 x 512Mb Double-Bank
Speed Memory Timings & Voltage Quake3 fps Sandra UNBuffered Sandra Standard Buffered Super PI 2M places
(time in sec)
400DDR
800FSB
2-3-3-7
2.55V
319.3 INT 2686
FLT 2653
INT 4455
FLT 4422
133
500DDR
1000FSB
2.5-3-4-7
2.65V
387.7 INT 3146
FLT 3246
INT 5516
FLT 5470
108
533DDR
1066FSB
3-3-4-7
2.75V
412.2 INT 3416
FLT 3450
INT 5846
FLT 5888
102
550DDR
1100FSB
3-3-4-8
2.75V
421.9 INT 3446
FLT 3518
INT 6040
FLT 6003
98
571DDR
1142FSB
3-3-4-8
2.75V
438.5 INT 3752
FLT 3642
INT 6231
FLT 6237
94

OCZ PC4400 reached approximately the same overclock at 2.75V that we found with Corsair's DDR550. That is, both reached about DDR560 at 2.75V. The OCZ did not have a problem with 2.85V, however, and continued on to 571DDR. Frankly, neither of these results with DDR550 shows much headroom. 560 and 571 represent just 1.8% to 3.8% over the DDR550 specification. This is much less "reserve" than we normally see with Premium memory.

Performance Test Configuration Performance Comparisons
Comments Locked

13 Comments

View All Comments

  • retrospooty - Thursday, February 19, 2004 - link

    Nice results at 5:4 vs 1:1 ... See, I told you so, after reading one of your older Ram articles from a few months back... 5:4 2-2-2 beats 1:1 2.5-4-4-8 anyday =)

    I'm glad you tested it !
  • kamper - Thursday, February 19, 2004 - link

    holy tiny target market, batman!

    how many people require memory to run at 275fsb default?
  • Pumpkinierre - Thursday, February 19, 2004 - link

    Very good memory article once again Wesley. I dont know where this 'low latency memory doesnt make any difference' argument started but from my observations, it is false. OCZ usually test on an ABIT IC7 so I wonder why it did't make the grade on the DDR550?

    Any hints as to when DDR500@ 2-2-2-5 is going to become reality. It seems like the old BH5s are still the lowest latency mem. chips and they've been around for a year now. 1:1 at low latencies and PAT aggressive is the way to go. That's why I run my 2.6c at 2.8 even though it goes stable to 3.3 but then I got to run the mem. at 5:4 (no PAT) and performance improvement is barely noticeable.

Log in

Don't have an account? Sign up now