Hi all,
I've recently done a pretty much full system upgrade. As such, I've been
doing a lot of benchmarking to test the performance in various different
states.
One of my benchmarking tests was to run the 3DSetup program included with
F1 Challenge '99-'02. It does a test and reports triangle performance.
So I'd pick a resolution, run it and record the results. Then I would
enable FSAA and anisotropic filtering at various levels, re-run it and
record results. But I noticed that a bit later when I went to a
resolution I had already tested, I got much lower figures. Then I
eventually tried turning the settings off and on, and I was getting the
same triangle performance with full 6X fsaa, 16X quality aniso, than
without!
Well, it turns out that the changes I was making to the fsaa and aniso
settings were not always taking effect -- even though I clicked "apply"
and when prompted by the applet whether I wanted to apply the settings
"now" or "later", I chose "now".
Rebooting the system seemed to get the changes to stick.
Quite an annoying problem though, that destroyed the validity of my
tests.
So anyways, for those that say enabling full fsaa and aniso incurs no
performance hit, be certain that the settings are indeed being activated!
Ped Xing <[email hidden]> wrote in
news:[email hidden]:
To get sensible benchmarks, you should really be rebooting between changes
to the config so that you're starting as clean as possible each time. You
don't know what stuff might have memory leaks etc that would affect later
runs. This would presumably ensure that the ATI driver changes kick in too.
--
* -
* - Nicotine, valium, vicadin, marijuana, ecstasy, and alcohol.
* - C-c-c-c-c-cocaine ...
* -
The symptoms you describe are what happen in a multi-monitor setup (is that
your situation?). Ths bug has existed in the Catalyst drivers after v.3.2
D.
"Dale Green" <[email hidden]> wrote in
news:[email hidden]:
> The symptoms you describe are what happen in a multi-monitor setup (is
> that your situation?). Ths bug has existed in the Catalyst drivers
> after v.3.2
No, single monitor. The card has 2 ports though.
One Punch Mickey <[email hidden]> wrote in
news:[email hidden]:
> To get sensible benchmarks, you should really be rebooting between
> changes to the config so that you're starting as clean as possible
> each time. You don't know what stuff might have memory leaks etc that
Now I know it, but what a pain!
"Ped Xing" <[email hidden]> wrote in message
news:[email hidden]...
I've seen the same in other CAT's, like disabling v-sync to run a benchmark
and it's still enabled.
