Test
software will be
SiSoft
Sandra 2004
PiFast
CDex 1.51
TMPGEnc
Unreal Tournament 2004
Quake 3: Arena
All
of the test software applications were run 3 times and the results
averaged between the 3.
SiSoft
Sandra
Although
a synthetic benchmark, it's a popular one, freely available if
you wish to make comparison benchmarks. We will be testing the
memory speeds, using the 32-bit 2004 version.
|
|
Ram
Int Buffer
|
Ram
Float
|
|
Mushkin
2-2-2-6 @ DDR400
|
2808
|
2633
|
|
Mushkin
3-4-4-8 @ DDR400
|
2740
|
2549
|
|
Ultra
PC4000 3-4-4-8 @ DDR400
|
2793
|
2580
|
Higher
is better
As
expected, when run at 3-4-4-8 @ DDR400 speeds, both the Ultra
and the Mushkin perform on par with one another. However when
you decrease the latencies on the Mushkin to its standard SPD
timings of 2-2-2-6, then the SiSoft Sandra's results show a
nice increase.
PiFast
A
good indicator of CPU/Motherboard performance is PiFast version
4.2, by Xavier Gourdon. We used a computation of 10000000 digits
of Pi, Chudnovsky method, 1024 K FFT, and no disk memory. Note
that lower scores are better, and times are in seconds.
|
|
Time
in Seconds
|
|
Mushkin
2-2-2-6 @ DDR400
|
55.86
|
|
Mushkin
3-4-4-8 @ DDR400
|
57.16
|
|
Ultra
PC4000 3-4-4-8 @ DDR400
|
57.01
|
Lower
is better
PiFast
shows similar results, with a nice decrease in the overall time
taken with low timings.
CDEX
I
took a CD (Bad Company - 10 from 9), and ripped it to one long
414MB .wav file. I then turned around and encoded that wave file
to a 320Kb/s sample rate MP3. I am sure there is a negligible
HD performance influence in this test but with the amount of memory
we are testing, it shouldn't be more than a fraction of a second.
|
|
Time
in Minutes
|
|
Mushkin
2-2-2-6 @ DDR400
|
2:09
|
|
Mushkin
3-4-4-8 @ DDR400
|
2:13
|
|
Ultra
PC4000 3-4-4-8 @ DDR400
|
2:13
|
Lower
is better
CDEX
is more CPU dependent than memory speed dependent, however the
memory speed and quantity do play their part in the overall scheme
of things. Despite this, once more the lower latencies from the
Mushkin win out. The difference may seem marginal but over a greater
time period with larger files, then that same differential will
apply.
TMPGEnc
MPEG Encoding
Video
editing is a taxing chore, and we'll be testing the Mushkin using
TMPGEnc 2.512 to encode an Animatrix short of 150MB to a DVD quality
mpg of 5000kbps. Note that lower scores are better.
|
|
Time
in Minutes
|
|
Mushkin
2-2-2-6 @ DDR400
|
12:12
|
|
Mushkin
3-4-4-8 @ DDR400
|
12:16
|
|
Ultra
PC4000 3-4-4-8 @ DDR400
|
12:16
|
Lower
is better
TMPGEnc
is another application that while the CPU plays a heavy part in
the encoding process, having lower timings can serve to decrease
the encoding time considerably, especially if you are encoding
a larger file then we have here.
Unreal
Tournament 2004 - as convoy @ 640x480, Min Details
Running
UT2004 at a low resolution and with minimum details should put
the load more on the CPU and Memory.
|
|
Frames
Per Second
|
|
Mushkin
2-2-2-6 @ DDR400
|
66.29
|
|
Mushkin
3-4-4-8 @ DDR400
|
64.25
|
|
Ultra
PC4000 3-4-4-8 @ DDR400
|
65.93
|
Higher
is better
The
as-convoy demo shows off a lot of scenery and moving imagery,
and can be quite taxing. The lower timings on the Mushkin 2-2-2-6
@ DDR400 certainly give a slight boost to the scores, and while
it may not be a huge number difference, every little bit helps
when you are tweaking.
Quake
III Arena - Demo Four @ 640x480, Min Details
Using
a default 640x480 configuration at lowest settings, we have attempted
to remove the load from the graphics card and put the emphasis
on the CPU, Memory and Subsystem's performance.
|
|
Frames
Per Second
|
|
Mushkin
2-2-2-6 @ DDR400
|
292.1
|
|
Mushkin
3-4-4-8 @ DDR400
|
286.1
|
|
Ultra
PC4000 3-4-4-8 @ DDR400
|
287.3
|
Higher
is better
Like
the UT2004 numbers, having the lower timings once again provides
a small boost.

Final
Words
This
has been for me a first look at a Mushkin
product in quite a while, well over a year and it has once again
been a positive experience. It can be kind of hard to believe
that lower timings, at least on an AMD system, can have as much
effect on application performance as they do. While the numbers
shown here may only show a slight differential, in the grand scheme
of things for those who do tweak, every little bit helps, so combined
with other system tweaks the low timings could contribute a lot.
This is especially useful for those who are not comfortable overclocking
their systems, as they may not get a boost from increasing the
FSB and Memory speeds, they can get a slight boost from the lower
than average timings.
Of
course true enthusiasts will want to overclock, and the samples
I have here have shown themselves to be capable of near PC4000
speeds which is quite a jump from PC3200. Timings had to be loosened
to reach the higher speeds but the headroom is there. If you are
a mild overclocker of P4 systems, or you run an AMD system, then
this ram would be suitable for you. If what you are after is 250MHz
and up, forget it, this ram was never designed for that and you
should be looking elsewhere. If you're not an overclocker, then
the slight boost from the lower timings at DDR400 speeds, perhaps
combined with some other system tweaks will certainly appeal.

Pros:
Low timings of 2-2-2-6, 2-2-2-5 timings were capable on these
samples @ DDR400, Fetching black on black heatspreaders and PCB's,
Headroom for overclocking to near DDR500 speeds with relaxed timings,
gold contacts
Cons:
Price - at $320
its around $40-$50 more than the majority of the competition,
however once into reseller hands that price will likely even out
to around the same.
Bottom
Line: Overclocking isn't something that everyone is comfortable
doing, but the lower timings from this Mushkin 1GB L2V2 PC3200
Dual Channel Kit will help go towards wringing every last drop
of performance from your system.
If
you have any questions or comments, feel free to use our Forums
to discuss this review.
HOME