|
 |
|
Test System |
|
Please take a look
at our system specs for reference |
|
|
Iwill
P4R533-N Motherboard
512MB of PC1066
RDRAM
ASUS P4T-E i850 Motherboard
512MB PC1066 RDRAM
Common Hardware:
Intel Pentium 4 2.4GHz
NVIDIA GeForce4 Ti 4600
(Drivers - v.28.32)
IBM 30GB ATA-100
7200RPM Hard Drive
Sound Blaster Live!
Value |
A Few
Words About The Benchmarks:
In setting up our test machines, we install Windows XP
on a formatted, FAT32 hard drive. After installing
the relevant drivers, we disabled system restore, all of
the graphical enhancements in Windows XP, and the
Automatic Update feature. The desktop on each test
bed is set to 1024x768, 16-bit color and a 75Hz refresh
rate.The Iwill
board in this piece was setup with performance defaults
enabled and the Asus board was set to performance
defaults with the "RDRAM Turbo" option enabled. |
We'll
continue testing with a brief look at some synthetic
benchmarks from SiSoftware's Sandra tests.
 |
|
SiSoft Sandra, eTesting Labs Winstones and
Comanche 4 |
|
Iwill versus Asus |
|
|
CPU
2.4GHz
|
Multimedia 2.4GHz
|
|
Memory PC1066
 |
ATA100 Drive Perf.
 |
All of the
Sandra numbers you see above are right where they should be
for this type of setup. In fact, the scores are right
on top of the reference numbers provided by the Sandra
software in these graphs. Notice however, that the CPU
clock is detected at slightly less than 2.4GHz. Iwill
chose to setup the board with modest and stable PLL timings,
unlike what we have seen from manufacturers like Asus and
Abit. Specifically, the Asus board that we used for
reference testing in this article, overclocks the CPU ever
so slightly even at default speed, inching the PLL up a MHz
or two in the process. Please take this into account
when viewing the remainder of the benchmarks.
These Winstone
scores, for all intents and purposes, are pretty much
identical. The Business Winstone side of things
favored the slightly more robust timings of the Asus board.
We should also note that, although both boards were setup
for PC1066 memory and 533 FSB at standard BIOS default
settings, the Asus board has an "RDRAM Turbo Mode" function
available in its BIOS. As such, with it enabled, it
does seem to edge past the Iwill P4R533-N. Regardless,
the differential is negligible.
We've started to
incorporate this benchmark into our tests for CPUs and
motherboards, as well as graphics cards. It is a great
DX8.1 test but also is hugely CPU and system bandwidth
dependant. So it's a good indicator of overall
performance as well. Here the picture is pretty much
the same, with both PC1066 configurations coming in neck and
neck. We've included a PC800 score on the Asus board,
for reference. Believe it or not, 3 fps in this test
is fairly significant.
 |
|
MadOnion 3DMark 2001SE Software Rendering |
|
3DMark Scores that
stress the CPU |
|
In this test,
we've set up MadOnion's 3DMark for Software Rendering for
T&L. This will place more of an emphasis on CPU
performance, since the host processor is required to perform
transform and lighting techniques algorithmically, in
software, versus through our test system's GeForce4 card.
Here again we
illustrate the performance edge PC1066 RDRAM speeds lend to
the Pentium 4 versus PC800. The added overall system
bandwidth increases approximately 9%. Also, once again
the slightly more aggressive timings of the Asus board give
it the edge versus the P4R533-N but the difference in
minescule.
PCMark 2002, Quake 3 and The Rating |