To
begin with, I will boot to the OS with the stock speed
and timings to test the performance of the modules
and their performance without tweaking the system
at all. Checking for the lowest timings that the modules
could handle, I found that the modules could take
3/2/2/8 timings at a considerably high speed. Then
I tested how high the modules could run with 3/3/3/9
and 4/4/4/12 timings. Finally, I tested the modules
while running the specified 5/5/5/15 timings, in order
to check their maximum clocking capability. Voltage
was always set at 2.2V inside the BIOS, which is the
stock rating of these modules, hence 100% safe as
well. Trying a somewhat higher voltage which the particular
motherboard can support (2.3V) made only a tiny difference,
so I decided to skip it.
Note 1 : Stability
is tested by running Memtest for at least 30 minutes
at the test settings to ensure that there are no errors.
Note 2 : The
comparison between the tested module speeds and timings
is not recommended, as the motherboard FSB and the
processor speeds differ. They all take place in order
to show the average performance and capability of
the modules, not direct comparison between the settings.
Memory
at 360MHz (720MHz DDR), 3/2/2/8 timings (240MHz FSB,
3600MHz processor speed)
What I did
like about these modules best of all, was their capability
to handle the lowest possible DDR-2 timings, 3/2/2/8,
at a so high speed without any burning in at all or
any kind of effort. The low timings provide excellent
bandwidth efficiency and very strong unbuffered tests
despite the low (if you can call 720MHz low) speed
of the modules. These modules would be one of my first
recommendations for users that do not have CPUs which
can run at high FSB speeds but want to boost their
performance via low timings.
Memory
at 390MHz (780MHz DDR), 3/3/3/9 timings (234MHz FSB,
3510MHz processor speed)
By loosening
the timings up to 3/3/3/9, the modules did not gain
much speed, edging at 390MHz (780MHz DDR). However,
their performance increased considerably, especially
considering that the FSB and CPU speed are both lower
this time round. The timings are not much looser,
so the extra MHz do give a performance boost.
Memory
at 475MHz (950MHz DDR), 4/4/4/12 timings (237MHz FSB,
3555MHz processor speed)
By loosening
the timings up to 4/4/4/12, the modules gained quite
some speed, up to 475MHz (950MHz DDR), which was very
close to their rated speed for 5/5/5/15 timings. Although
the modules did not gain much raw speed, given the
looser timings, the Everest tests prove that their
high speed will do more than keep up in many applications.
The Sandra benchmarks did get a performance boost,
albeit a slight one.
Memory
at 520MHz (1040MHz DDR), 5/5/5/15 timings (260MHz
FSB, 3900MHz processor speed)
At the stock
timings of 5/5/5/15, the modules ran up to 520MHz
(1040MHz DDR), which is not far from their specifications
but I suspect that it was the chipset which was failing
on me and not the modules. The modules ran flawlessly
at 520MHz for hours without any problems, but at 525MHz
I could not get the PC to boot. Probably the Asus
motherboard doesn’t like driving these particular
modules any higher. The performance increases are
considerable here across all tests, despite the very
loose timings, however take notice that the FSB and
CPU speed are considerably higher than in the previous
tests also.
|
