Test Results: Corsair XMS4404v1.1

To be considered stable for test purposes, Quake3 benchmark, UT2003 Demo, Super PI, Aquamark 3, and Comanche 4 had to complete without incident. Any of these, and in particular Super PI, will crash a less-than stable memory configuration.

AMD Test Results


Corsair XMS4404v1.1 (DDR550) - 2x512Mb Double-Bank
CPU Ratio at 2.4GHz Memory Speed Memory Timings
& Voltage
Quake3
fps
Sandra UNBuffered Sandra Standard
Buffered
Super PI 2M places
(time in sec)
Wolfenstein - Radar - Enemy Territory fps
12x200 400 DDR 2-3-3-10
2.5V 1T
501.0 INT 2641
FLT 2764
INT 6051
FLT 601
82 108.2
11x218 436 DDR 2-3-3-10
2.6V 1T
510.1 INT 2680
FLT 2855
INT 6439
FLT 6372
81 109.7
10x240 480 DDR 2-3-3-10
2.75V 1T
522.7 INT 2903
FLT 3095
INT 6684
FLT 6609
80 112.0
9x267 533 DDR 2.5-3-3-10
2.75V 1T
533.3 INT 3045
FLT 3266
INT 6972
FLT 6885
78 113.3
8x305
(2.44GHz)
Highest 1T Mem Speed
610 DDR
2.5-3-4-10
2.8V 1T
569.2 INT 3248
FLT 3467
INT 7531
FLT 7441
77 116.6
8x318
(2.54GHz)
Highest 2T Mem Speed
636 DDR
3-4-4-10
2.85V 2T
568.7 INT 3088
FLT 3149
INT 7011
FLT 7549
76 122.6
9x295
(2.75GHz)
Highest Performance
590 DDR
2.5-4-3-10
2.8v 1T
572.8 INT 3346
FLT 3538
INT 7712
FLT 7549
72 122.9

Corsair handily turns in the highest memory performance that we have yet seen on the Athlon 64. The highest 2T value of DDR636, the Highest 1T value of 610, and the Top Performance of 9x295 at 1T are all records on the AMD platform. It is clear that Corsair aimed for best performance at the top with this DDR550 memory, since performance at DDR400 is not up to the same standards achieved at the top. Corsair says that modules will typically do 2-2-2 timings at DDR400, but the best that we could do at any voltage is 2-3-3 at DDR400. Corsair selected Samsung TCCD chips for absolute best performance at the top and they clearly succeeded in reaching this goal. The compromise was a bit less performance at DDR400 - at least with the modules we tested.

Intel Test Results


Corsair XMS4404v1.1 (DDR550) - 2x512Mb Double-Bank
Speed Memory Timings
& Voltage
Quake3
fps
Sandra UNBuffered Sandra Standard
Buffered
Super PI 2M places
(time in sec)
RCW-ET
fps
400DDR
800FSB
2-3-3-5
2.5V
331.0 INT 2720
FLT 2728
INT 4457
FLT 4453
130 70.1
433DDR
866FSB
2-3-3-5
2.55V
358.1 INT 2910
FLT 2827
INT 4784
FLT 4802
120 77.4
466DDR
933FSB
2-3-3-5
2.65V
384.5 INT 3107
FLT 3187
INT 5162
FLT 5189
112 83.0
500DDR
1000FSB
2-3-3-5
2.75V
409.1 INT 3348
FLT 3396
INT 5567
FLT 5561
104 89.6
533DDR
1066FSB
2.5-3-3-5
2.75V
430.3 INT 3580
FLT 3623
INT 5947
FLT 5851
98 94.9
572DDR
1144FSB
2.5-4-4-6
2.85V
450.1 INT 3763
FLT 3799
INT 6348
FLT 6273
93 97.8

The pattern is similar on the Intel memory platform, although the domination by Corsair at the top of the performance charts is not nearly so clear. Corsair lags the best TCCD memory in performance at DDR400, but by DDR466, the superior high-end performance of Corsair DDR550 has established itself. As speed further increases, the advantage of the Corsair DDR550 widens.

After looking at both Intel and AMD performance results, it is obvious that Corsair has binned Samsung TCCD for this memory with the clear goal of performance at the top. Corsair rating the memory at DDR550 instead of the DDR400 used in most other TCCD memory is clearly justified with this emphasis on the top-end of memory performance. Corsair had to compromise DDR400 performance a bit - at least on the modules that we tested - to achieve the incredible top-end results that we see in these benchmarks.

Intel Performance Test Configuration Performance Comparisons
Comments Locked

23 Comments

View All Comments

  • DonCornelius - Saturday, January 22, 2005 - link

    Can anyone tell me why we can't get 2-2-2 timings on 1GB size DIMMs? The only DIMMs I see with this timings are 512MB and 256MB. Is this marketing or a limit on the technology?
  • Live - Thursday, January 6, 2005 - link

    If what PrinceGaz is sayimng about memory on the AMD platform is true I think it would warrant some clarification from Anandtech. If money is an object what gives best bang for the buck. Cheap memory and faster CPU or the other way around?
  • PrinceGaz - Wednesday, January 5, 2005 - link

    from #20- "#19 - The $82 PQI Turbo stuff at newegg is 2.5-3-3 timing RAM. The cheapest you can get a 2-2-2 512MB stick of RAM at newegg is the Patriot for $107."

    Given that we've already got a 2.5-3-3 timings with the PQI, and it was the module that at most was 3% slower on memory-bandwidth bound applications with the Athlon 64, I think that answers my question about why budget memory has not been covered.

    You may as well save still more money and get brand-name value-products for an AMD box, unless you are going for a high-end overclocking system with an FX-55 where every component is the best in it's class. Even if overclocking you aren't going to suffer because there is no such thing as an asynchronous memory frequency with an Athlon 64 (there is no Northbridge between the CPU and memory) so just set the budget memory to "DDR333" and you'll be fine for overclocking up to about DDR500.

    Actually when you combine the S939 Athlon 64's lack of dependence on memory bandwidth with it's onboard memory controller that ensures any memory speed is equally efficient; when building a mid-range Athlon 64 box you may as well just get cheap brand-name DDR400 and run it at what ever speed it is happiest with after overclocking your CPU. Which makes all these high-end memory review articles pointless for all except extreme overclockers.
  • eetnoyer - Wednesday, January 5, 2005 - link

    #19 - The $82 PQI Turbo stuff at newegg is 2.5-3-3 timing RAM. The cheapest you can get a 2-2-2 512MB stick of RAM at newegg is the Patriot for $107.
  • kmmatney - Tuesday, January 4, 2005 - link

    With NewEgg having PQI turbo 3200 at $82, I thinks that's the best deal, probably worth the extra $10 or so over value RAM. In this review it performed almost as well as the top of the line stuff.
  • Googer - Tuesday, January 4, 2005 - link

    Correction:

    the standard JDEC complient ram does not need to be included in the overclocking tests.
  • Googer - Tuesday, January 4, 2005 - link

    All memory should be tested agains JDEC Standard Ram
    using JDEC standard Timings For DDR400. Standard ISSUE Crucial (not ballistix) should also be included as a base compairson for all DDR400 Tests.

    When it comes to overclocking the JDEC complient ram
    does not need to be tested becuase that was never the intent of its design.
  • Fricardo - Tuesday, January 4, 2005 - link

    #'s 6,12,14,15

    Here here. I couldn't care less about timings...it's not worth the cash to get ever so slightly more performance. I'd just like some decent RAM that'll let me overclock an A64, nothing fancy.
  • miketheidiot - Tuesday, January 4, 2005 - link

    Cheap memory review! Enough of this expensive junk.
  • JustAnAverageGuy - Tuesday, January 4, 2005 - link

    #13

    I think he was thinking more along the lines of say Kingston\Corsair ValueRAM which runs for around $65-70 for a 512MB stick.

    Zebo did something similar (see CPU & OC forums), but I've been waiting for the AnandTech review.

Log in

Don't have an account? Sign up now