We
began at 100Mhz (200MHz DDR) for testing, and went up until
she wouldn't boot. To make sure that the processor wasn't holding
us back we made sure and under clocked the chip by manipulating
the multiplier such that it never ran higher than 800 or 900Mhz,
which it was designed to do. We also ran with no PCI cards,
as they could also tend to cause problems at higher bus speeds.
While running with an AGP card can produce the same negative
effect, we simply had to keep watch and try our best to not
let that be our limiting factor.
To
keep things simple we also ran each stick of DDR with a CAS
setting of 2.5. In addition, we had the RAM set to normal mode
on both boards. We used SiSoft's Sandra memory benchmark, version
2001se, as it measures the sustained memory bandwidth. Numbers
are reported in MBs (megabytes) per second, which means that
higher is better. Please take note that the first number reported
is Integer ALU & second is Float FPU.
Memory
Speed (MHz)
|
Mushkin
(ALi)
|
Mushkin
(AMD 760)
|
Crucial
(ALi)
|
Crucial
(AMD 760)
|
200MHz |
Fail |
492
625 |
390
533 |
497
622 |
220MHz |
Fail |
499
643 |
421
549 |
499
640 |
240MHz |
Fail |
511
700 |
456
634 |
512
701 |
266MHz |
Fail |
532
739 |
525
694 |
537
733 |
280MHz |
Fail |
610
761 |
543
734 |
610
760 |
290MHz |
Fail |
Fail |
582
756 |
669
783 |
300MHz |
Fail |
Fail |
Fail |
Fail |
Also,
please note that both mainboards also support separate voltage
settings for the memory. We simply ran at the highest settings
(2.7V) to maximize any overclocking potential.
The
Verdict
Clearly,
the best choice would be the Crucial
DDR. Unlike the Mushkin RAM which had some major problems with
the ALi based chipset. Not only that the despite Mushkin advertising
their RAM as overclockable the crucial was able to overclock
higher, and they don't even support overclocking the way Mushkin
does. Despite all that, Crucial's RAM is also about $10 cheaper
than the Mushkin.
by
Ryan Wissman
<<
Previous
|