Test Results: Kingmax DDR500 and Kingmax DDR466

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.

Kingmax DDR500 Hardcore Series - 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
2-2-2-5
2.5V
329.2 INT 2892
FLT 2930
INT 4526
FLT 4524
129
433DDR
866FSB
2-2-3-6
2.5V
351.2 INT 3014
FLT 3071
INT 4853
FLT 4847
120
533DDR
1066FSB
2.5-4-4-7
2.75V
411.9 INT 3345
FLT 3362
INT 5842
FLT 5819
102
466DDR
933FSB
2.5-2-3-7
2.5V
376.1 INT 3257
FLT 3277
INT 5164
FLT 5164
112
500DDR
1000FSB
2.5-3-3-7
2.5V
396.4 INT 3341
FLT 3389
INT 5527
FLT 5528
106
528DDR
1056FSB
3-3-4-8
2.85V
416.3 INT 3477
FLT 3524
INT 5860
FLT 5866
99

It is amazing that the Kingmax BGA modules managed to run form DDR400 to DDR500 at the default 2.5V. This is the first DDR500 memory that we have tested actually to run at default voltage over such a wide range of memory speeds. The absolute ceiling for the Kingmax DDR500 modules was DDR528, and we were able to achieve this speed at voltages from 2.65V to 2.85V. Increasing the voltage at DDR528 just allowed tighter timings. At 2.65V, we could run stable 3-4-4-8 timings; at 2.75V, 3-3-4-8; and at 2.85V, the 3-3-3-8 timings as reported in the chart. While we were able to complete our benchmark tests at 2.5V at most Memory speeds, top stability was achieved at a 2.55V or 2.65V setting on the Asus P4C800-E motherboard.

Kingmax DDR466 Hardcore Series - 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
2-2-3-5
2.5V
323.8 INT 2783
FLT 2833
INT 4459
FLT 4447
130
433DDR
866FSB
2.5-2-3-6
2.5V
349.0 INT 3002
FLT 3019
INT 4840
FLT 4837
121
466DDR
933FSB
2.5-3-3-7
2.5V
372.4 INT 3179
FLT 3201
INT 5154
FLT 5146
114
500DDR
1000FSB
3-4-4-8
2.85V
388.8 INT 3446
FLT 3518
INT 5429
FLT 5398
108

We really expected Kingmax DDR466 Hardcore Series to perform about the same as their DDR500 at lower speeds, but in fact, the DDR500 is also a better performer at the DDR400 end of the range as well. That is not to say DDR466 is not a good performer, since you would never call 2-2-3-5 at 2.5V a bad DDR400 performance. It is not as extraordinary as the DDR500 modules matching the fastest timings that we have ever seen at DDR400. DDR466 performance was very competitive with the best modules that we have tested across its operating range, with the memory topping out at DDR500. We could not achieve 1 MHz higher than DDR500, but performance was stable with higher voltage at that memory speed.

Performance Test Configuration Performance Comparisons
Comments Locked

14 Comments

View All Comments

  • bldkc - Friday, March 12, 2004 - link

    By the way could you guys please start highlighting the item being tested in the graphs with a different colored bar or even different color text for the name (or both)? It makes it very difficult to find the product you are testing if I have to read every name on the graph just to find one of them.
  • bldkc - Friday, March 12, 2004 - link

    Wow, don't buy the DDR466! On page 6 it took 1114 seconds to complete! Okay, it's a typo, but still.
    #10-If you wear a thinner coat outside you won't stay as warm as if you wear a thick one. The same thing with chips.
  • Wesley Fink - Thursday, March 11, 2004 - link

    #11 - Corrected.

    #10 - BGA chips are both much smaller and thinner than TSOP chips. The electrical connections are also much shorter, generating less heat to start with than TSOP chips. I have seen data supporting 80 to 85% heat dissipation with BGA chips.
  • Shalmanese - Thursday, March 11, 2004 - link

    I sure hope Kingmax didn't provide you with 512K chips of RAM ;). (top of pg3)
  • ViRGE - Wednesday, March 10, 2004 - link

    What is it exactly about BGA chips that make them run cooler than TSOP chips? There's the size difference of course, but that doesn't account for the temp difference, does it?
  • Jeff7181 - Wednesday, March 10, 2004 - link

    It's good to see this from Kingmax... I've been wondering for quite some time now why video cards have had DDR500 memory for a few years, and it hasn't made it's way into system RAM. Now we have RAM on video cards capable of DDR1000... why can't we get similar results with system RAM?
  • Inferno - Wednesday, March 10, 2004 - link

    Something everyone may want to note, if you decrease voltage on the Kingmax sometimes that yeilds better O/Cs then raising it. I have owned alot of Kingmax TinyBGA and it usually responds negativly to more voltage. They also do benifit from heat spreaders when pushed hard and kept at default voltage.
  • Pumpkinierre - Wednesday, March 10, 2004 - link

    Good article but still more of the same. It seems like 18month old Winbond BH5 and now these BGA chips are the fastest at DDR400-433. However, are there any hints of a DDR466-500 at low latencies (CAS2) out there? I mean graphics cards have got 256Mb of <3ns DDR (dont know the latencies) so why arent the memory manufacturers using that?

    Xtreme DDR have got some PC3700+ at 2-3-3-6 using picked 5ns Samsung chips (http://www.xtremeddr.com/products/x_pc3700+.shtml)... They quote 2T command rate and some i875 mobo compatibility, which is counter to the Mushkin website advice that intel dual bank chipsets force 1T timings, and memory rated above 1T could be unstable. Perhaps you might get a coupla sticks of that and put them through your test procedure?
  • KristopherKubicki - Wednesday, March 10, 2004 - link

    Evan,

    We have achieved ludicris speed, overshot the winnebago, and gone to plaid.

    Kristopher
  • KristopherKubicki - Wednesday, March 10, 2004 - link

    BGA is the new standard on DDR2. I welcome the change.

    Kristopher

Log in

Don't have an account? Sign up now