|
SiSoft
Sandra "STREAM" Memory Benchmark
For
this test we did not have a CPU in house that was capable of a
133MHz. Front Side Bus Speeds. Instead, we used a Pentium 3-450 that
was capable of 124 MHz. with the utmost stability. In this test, the
Toshiba PC100 module was not stable at the 124
MHz.
bus speed at a CAS Latency Setting of 2, so it was set to 3. With
this setting we experienced good stability with the Toshiba module.
However
the EMS Memory was rock solid stable at 124MHz.X4.5 for a total
558MHz. with a CAS2 setting. What you see below is the difference
that CAS2 makes with respect to system memory speed.
Click
for larger view...
|
PC100
CAS3
|
EMS
- PC133 CAS2
| As
you can easily see a CAS2 setting makes a difference and the
stability of the EMS DIMM was perfect at this speed. Since the
module is designed to operate at 133MHz. w/ 3,2,2 settings, 124 MHz.
at 2,2,2 was a walk in the park for the EMS module. The Toshiba DIMM
fell down badly even vrs. the Sandra base system. I can only assume
the base system was running at 100 MHz. FSB and was set to CAS2.
With this in mind it is easy to see how lower quality SDRAM can hurt
your overclocked performance. Due to the fact that the standard
SDRAM DIMM could only run at CAS3 with the 124MHz. FSB setting, we
were forced to throttle back memory bandwidth significantly. This
doesn't mean our P3-450 @ 558MHz. was slower overall than the base
Sandra System, a P2 @ 450. The CPU speed boost plays heavily as
well, of course. However, the system with a stable CAS2 setting at
124MHz. system bus enjoys the best of both worlds.
Wintune
Memory Performance Test
|
PC100
CAS3
|
EMS
- PC133 CAS2
| Once
again, a measurable difference between CAS2 and 3 and a measurable
difference in memory subsystem performance.
| 
