Inno3D iChill 8600GTS XStriker3 Review - Page 3 of 6
By Steve Angelly - June 5, 2007
SYNTHETIC BENCHMARKS
AQUAMARK3
Multiple runs were made at each of the resolutions
identified in the graphs below. The results reflect the average. All runs
were made with no antialiasing (AA) and anisotropic filtering (AF) set at 4x.
The results were very consistent with less than 2% variation and,
considering more than 3 runs were made at each resolution, that is exceptional.
Also of note, temperatures did not register more than 47C on the GPU. Compared
with the other GTS card I ran through this benchmark, it runs around 10C cooler.
This is a good initial indication that the iChill cooling setup on the
8600GTS XStriker3 is there for more than just looks.
Aquamark 3 Results
The results at each resolution, with the benchmark's default settings of no
AA/4AF, indicate good performance for a GTS card. There is nothing exceptional here, but
it is a mainstream entry. Scaling was consistent, and stability was excellent.
Plus, it runs very cool at this benchmark and noise from the fan is
not noticeable. I would say this initial test produced a very good result when compared with
the previous generation DX9 cards.
3DMARK06
With 3DMark06, the score is influenced more by CPU test results than it ever was before.
Shader Models 2.0 and 3.0 are used exclusively for all vertex and pixel processing.
The engine builds the shaders for each material in HLSL (high level shader language) and
compiles in runtime by the GPU drivers. This makes for an up-to-date, stressful
benchmark for the latest graphic cards and systems. I loaded it up and tested the
8600GTS XStriker3 to see how the card would perform with this high
performance, low-budget test system.
3DMARK06 - Results
The Inno3D iChill 8600GTS XStriker3 did well, mimicking
the reference design 8600 GTS results for the most part. It was a little behind
in some test results in the benchmark. I did not post the comparison results, as the reference
design I have been using was a slightly higher overclocked version (721/1050), which gave it a
slight edge. The results above were achieved with the card at default clocks as
it came right out of the box (705/1050) running in the test system used in this
review. The GPU on this particular graphics card has about 10% more headroom so
the 3DMark06 score can exceed 6100 3DMarks. However, I found the memory on
this card not as overclocker 'friendly' and more satisfied with the default clock
setting. Again, temperatures did not enter in as the cooling solution Inno3D has
provided for the 8600GTS XStriker3 maintained the GPU temperature under 47C. I
took several independent temperature readings off the base of selected memory
chips during benchmarking and did not notice any rise in temperature that would cause concern.