Community
Inventor Forum
Welcome to Autodesk’s Inventor Forums. Share your knowledge, ask questions, and explore popular Inventor topics.
cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

Analysis of Graphics Tweaks / Reg Edits / Drivers

12 REPLIES 12
Reply
Message 1 of 13
Anonymous
290 Views, 12 Replies

Analysis of Graphics Tweaks / Reg Edits / Drivers

Using Andy Shelter's excellent graphics benchmark, I ran it lots of times
with different nVidia drivers, OpenGl tweaks, suggested registry hacks, 1
and 2 monitors, and with/without the statistics posting onscreen hack.

In all but one trial, I rebooted the machine between trials.

Short of a Chi-Squared statistical analysis, my instincts tell me that there
is no significant difference in the results with ANY combination of
settings.

See Excel spreadsheet posted in CF.

-Jack
12 REPLIES 12
Message 2 of 13
Anonymous
in reply to: Anonymous

Very interesting.
I said it in Andy's thread and I'll say it here. I don't trust these
numbers.


"Jack Brentner" wrote in message
news:1013E178544D2FC0DA171760A7AFCC88@in.WebX.maYIadrTaRb...
> Using Andy Shelter's excellent graphics benchmark, I ran it lots of times
> with different nVidia drivers, OpenGl tweaks, suggested registry hacks, 1
> and 2 monitors, and with/without the statistics posting onscreen hack.
>
> In all but one trial, I rebooted the machine between trials.
>
> Short of a Chi-Squared statistical analysis, my instincts tell me that
there
> is no significant difference in the results with ANY combination of
> settings.
>
> See Excel spreadsheet posted in CF.
>
> -Jack
>
>
Message 3 of 13
Anonymous
in reply to: Anonymous

One can postulate any number of new hypotheses.

To name a few:
H1) The benchmark results are suspect.
H2) The test equipment does not respond as expected to the chosen inputs.
H3) Significant result differences are driven by other unidentified inputs
or combinations thereof.

Verification of
H1: Run similar trials with another benchmark tool.
H2: Perform similar trials on other equipment with the same benchmark tool.
H3: Choose other inputs or parameters and combinations thereof and re-run
(hardware, op system, other opengl tweaks, etc.).

One can draw no other conclusion from my analysis other than that "the
benchmark provided no significant different results with the inputs chosen
on that particular machine". At the least, my results must be confirmed
independently before any other conclusions can be drawn.

Thanks Many,
Jack


"Ron Crain" wrote in message
news:EF089B81EB15AD063E42823F17687CA6@in.WebX.maYIadrTaRb...
> Very interesting.
> I said it in Andy's thread and I'll say it here. I don't trust these
> numbers.
>
>
> "Jack Brentner" wrote in message
> news:1013E178544D2FC0DA171760A7AFCC88@in.WebX.maYIadrTaRb...
> > Using Andy Shelter's excellent graphics benchmark, I ran it lots of
times
> > with different nVidia drivers, OpenGl tweaks, suggested registry hacks,
1
> > and 2 monitors, and with/without the statistics posting onscreen hack.
> >
> > In all but one trial, I rebooted the machine between trials.
> >
> > Short of a Chi-Squared statistical analysis, my instincts tell me that
> there
> > is no significant difference in the results with ANY combination of
> > settings.
> >
> > See Excel spreadsheet posted in CF.
> >
> > -Jack
> >
> >
>
>
Message 4 of 13
Anonymous
in reply to: Anonymous

Jack,
I'm excited about your extensive tests.
After viewing your system specs I come to this conclusion:
Your system seems to have a bottleneck with your P3-650 because your results
are so bad. On other systems the 750XGL gets about 30FPS in shaded mode. On
your system only 10 FPS.
You can not see any different results because your graphics board is waiting
on the CPU in EVERY case.
The results on other systems are very different because of
- different hardware (CPU, GPU, memory)
- different Inventor graphics settings (I recommend to use IV default
settings to assure we compare apples with apples)
- different OS, different displaydriver and displaydriver settings (this is
probably the smallest factor)

Andy



"Jack Brentner" wrote in message
news:5199353292B5E6F4BEE877981CD1D6F6@in.WebX.maYIadrTaRb...
> One can postulate any number of new hypotheses.
>
> To name a few:
> H1) The benchmark results are suspect.
> H2) The test equipment does not respond as expected to the chosen inputs.
> H3) Significant result differences are driven by other unidentified inputs
> or combinations thereof.
>
> Verification of
> H1: Run similar trials with another benchmark tool.
> H2: Perform similar trials on other equipment with the same benchmark
tool.
> H3: Choose other inputs or parameters and combinations thereof and re-run
> (hardware, op system, other opengl tweaks, etc.).
>
> One can draw no other conclusion from my analysis other than that "the
> benchmark provided no significant different results with the inputs chosen
> on that particular machine". At the least, my results must be confirmed
> independently before any other conclusions can be drawn.
>
> Thanks Many,
> Jack
>
>
> "Ron Crain" wrote in message
> news:EF089B81EB15AD063E42823F17687CA6@in.WebX.maYIadrTaRb...
> > Very interesting.
> > I said it in Andy's thread and I'll say it here. I don't trust these
> > numbers.
> >
> >
> > "Jack Brentner" wrote in message
> > news:1013E178544D2FC0DA171760A7AFCC88@in.WebX.maYIadrTaRb...
> > > Using Andy Shelter's excellent graphics benchmark, I ran it lots of
> times
> > > with different nVidia drivers, OpenGl tweaks, suggested registry
hacks,
> 1
> > > and 2 monitors, and with/without the statistics posting onscreen hack.
> > >
> > > In all but one trial, I rebooted the machine between trials.
> > >
> > > Short of a Chi-Squared statistical analysis, my instincts tell me that
> > there
> > > is no significant difference in the results with ANY combination of
> > > settings.
> > >
> > > See Excel spreadsheet posted in CF.
> > >
> > > -Jack
> > >
> > >
> >
> >
>
>
Message 5 of 13
Anonymous
in reply to: Anonymous

>>> Your system seems to have a bottleneck with your P3-650 because your
results
are so bad >>

I 100% agree with this. On my home comp my P3 - 1Ghz running @ 1250Mhz
166FSB is a bottleneck for my Geforce4 TI4200 128Mb, I have come to this
conclusion after running a large number benchmarks (3DMark2001).
See my system specs here http://service.madonion.com/compare?2k1=4595876
Message 6 of 13
Anonymous
in reply to: Anonymous

Yes I suspect as such. I'm not even sure the MB supports beyond AGP 1X. I
purchased the 750 XGL at the time there was a promotion from PNY and
solidjerks for $399 - couldn't pass it up. The Voodoo5 card was pukin' and
diein' on the XP operating system, and 3dfx underground drivers turned out
to be vaporware.

Would like to see a similar analysis for the 750 XGL in a worthy system,
like oh say a dual 2.0 Xeon 😉

Again, the results indicate that a low-end machine (read P3) will not
respond to differing input.

Thanks Many,
Jack

"Andy Shelter" wrote in message
news:923BEBFEDA08B0AFE2D358AFAA819AEC@in.WebX.maYIadrTaRb...
> Jack,
> I'm excited about your extensive tests.
> After viewing your system specs I come to this conclusion:
> Your system seems to have a bottleneck with your P3-650 because your
results
> are so bad. On other systems the 750XGL gets about 30FPS in shaded mode.
On
> your system only 10 FPS.
> You can not see any different results because your graphics board is
waiting
> on the CPU in EVERY case.
> The results on other systems are very different because of
> - different hardware (CPU, GPU, memory)
> - different Inventor graphics settings (I recommend to use IV default
> settings to assure we compare apples with apples)
> - different OS, different displaydriver and displaydriver settings (this
is
> probably the smallest factor)
>
> Andy
>
>
>
> "Jack Brentner" wrote in message
> news:5199353292B5E6F4BEE877981CD1D6F6@in.WebX.maYIadrTaRb...
> > One can postulate any number of new hypotheses.
> >
> > To name a few:
> > H1) The benchmark results are suspect.
> > H2) The test equipment does not respond as expected to the chosen
inputs.
> > H3) Significant result differences are driven by other unidentified
inputs
> > or combinations thereof.
> >
> > Verification of
> > H1: Run similar trials with another benchmark tool.
> > H2: Perform similar trials on other equipment with the same benchmark
> tool.
> > H3: Choose other inputs or parameters and combinations thereof and
re-run
> > (hardware, op system, other opengl tweaks, etc.).
> >
> > One can draw no other conclusion from my analysis other than that "the
> > benchmark provided no significant different results with the inputs
chosen
> > on that particular machine". At the least, my results must be confirmed
> > independently before any other conclusions can be drawn.
> >
> > Thanks Many,
> > Jack
> >
> >
> > "Ron Crain" wrote in message
> > news:EF089B81EB15AD063E42823F17687CA6@in.WebX.maYIadrTaRb...
> > > Very interesting.
> > > I said it in Andy's thread and I'll say it here. I don't trust these
> > > numbers.
> > >
> > >
> > > "Jack Brentner" wrote in message
> > > news:1013E178544D2FC0DA171760A7AFCC88@in.WebX.maYIadrTaRb...
> > > > Using Andy Shelter's excellent graphics benchmark, I ran it lots of
> > times
> > > > with different nVidia drivers, OpenGl tweaks, suggested registry
> hacks,
> > 1
> > > > and 2 monitors, and with/without the statistics posting onscreen
hack.
> > > >
> > > > In all but one trial, I rebooted the machine between trials.
> > > >
> > > > Short of a Chi-Squared statistical analysis, my instincts tell me
that
> > > there
> > > > is no significant difference in the results with ANY combination of
> > > > settings.
> > > >
> > > > See Excel spreadsheet posted in CF.
> > > >
> > > > -Jack
> > > >
> > > >
> > >
> > >
> >
> >
>
>
Message 7 of 13
Anonymous
in reply to: Anonymous

For what it's worth...I have a 700XGL and these are my results (P4 2.0GHz,
1/2 gig ram)

Shaded = 22.1 FPS
Hidden = 8.1 FPS
Wireframe = 16.6 FPS
Multiple Windows = 23.3 FPS
Average = 17.5 FPS

better than a lot of the 750s

--
Sean Dotson, PE
http://www.sdotson.com
Check the Inventor FAQ for most common questions
-----------------------------------------
"Jack Brentner" wrote in message
news:64327BFACA760CBA23645078A68DFB35@in.WebX.maYIadrTaRb...
> Yes I suspect as such. I'm not even sure the MB supports beyond AGP 1X. I
> purchased the 750 XGL at the time there was a promotion from PNY and
> solidjerks for $399 - couldn't pass it up. The Voodoo5 card was pukin' and
> diein' on the XP operating system, and 3dfx underground drivers turned out
> to be vaporware.
>
> Would like to see a similar analysis for the 750 XGL in a worthy system,
> like oh say a dual 2.0 Xeon 😉
>
> Again, the results indicate that a low-end machine (read P3) will not
> respond to differing input.
>
> Thanks Many,
> Jack
>
> "Andy Shelter" wrote in message
> news:923BEBFEDA08B0AFE2D358AFAA819AEC@in.WebX.maYIadrTaRb...
> > Jack,
> > I'm excited about your extensive tests.
> > After viewing your system specs I come to this conclusion:
> > Your system seems to have a bottleneck with your P3-650 because your
> results
> > are so bad. On other systems the 750XGL gets about 30FPS in shaded mode.
> On
> > your system only 10 FPS.
> > You can not see any different results because your graphics board is
> waiting
> > on the CPU in EVERY case.
> > The results on other systems are very different because of
> > - different hardware (CPU, GPU, memory)
> > - different Inventor graphics settings (I recommend to use IV default
> > settings to assure we compare apples with apples)
> > - different OS, different displaydriver and displaydriver settings (this
> is
> > probably the smallest factor)
> >
> > Andy
> >
> >
> >
> > "Jack Brentner" wrote in message
> > news:5199353292B5E6F4BEE877981CD1D6F6@in.WebX.maYIadrTaRb...
> > > One can postulate any number of new hypotheses.
> > >
> > > To name a few:
> > > H1) The benchmark results are suspect.
> > > H2) The test equipment does not respond as expected to the chosen
> inputs.
> > > H3) Significant result differences are driven by other unidentified
> inputs
> > > or combinations thereof.
> > >
> > > Verification of
> > > H1: Run similar trials with another benchmark tool.
> > > H2: Perform similar trials on other equipment with the same benchmark
> > tool.
> > > H3: Choose other inputs or parameters and combinations thereof and
> re-run
> > > (hardware, op system, other opengl tweaks, etc.).
> > >
> > > One can draw no other conclusion from my analysis other than that "the
> > > benchmark provided no significant different results with the inputs
> chosen
> > > on that particular machine". At the least, my results must be
confirmed
> > > independently before any other conclusions can be drawn.
> > >
> > > Thanks Many,
> > > Jack
> > >
> > >
> > > "Ron Crain" wrote in message
> > > news:EF089B81EB15AD063E42823F17687CA6@in.WebX.maYIadrTaRb...
> > > > Very interesting.
> > > > I said it in Andy's thread and I'll say it here. I don't trust
these
> > > > numbers.
> > > >
> > > >
> > > > "Jack Brentner" wrote in message
> > > > news:1013E178544D2FC0DA171760A7AFCC88@in.WebX.maYIadrTaRb...
> > > > > Using Andy Shelter's excellent graphics benchmark, I ran it lots
of
> > > times
> > > > > with different nVidia drivers, OpenGl tweaks, suggested registry
> > hacks,
> > > 1
> > > > > and 2 monitors, and with/without the statistics posting onscreen
> hack.
> > > > >
> > > > > In all but one trial, I rebooted the machine between trials.
> > > > >
> > > > > Short of a Chi-Squared statistical analysis, my instincts tell me
> that
> > > > there
> > > > > is no significant difference in the results with ANY combination
of
> > > > > settings.
> > > > >
> > > > > See Excel spreadsheet posted in CF.
> > > > >
> > > > > -Jack
> > > > >
> > > > >
> > > >
> > > >
> > >
> > >
> >
> >
>
>
Message 8 of 13
Anonymous
in reply to: Anonymous

oops that's in FPS, sorry... so not quite as good as the 750s

--
Sean Dotson, PE
http://www.sdotson.com
Check the Inventor FAQ for most common questions
-----------------------------------------
"Sean Dotson" wrote in message
news:E832AE2740623D3A31E9F49583CF85A0@in.WebX.maYIadrTaRb...
> For what it's worth...I have a 700XGL and these are my results (P4 2.0GHz,
> 1/2 gig ram)
>
> Shaded = 22.1 FPS
> Hidden = 8.1 FPS
> Wireframe = 16.6 FPS
> Multiple Windows = 23.3 FPS
> Average = 17.5 FPS
>
> better than a lot of the 750s
>
> --
> Sean Dotson, PE
> http://www.sdotson.com
> Check the Inventor FAQ for most common questions
> -----------------------------------------
> "Jack Brentner" wrote in message
> news:64327BFACA760CBA23645078A68DFB35@in.WebX.maYIadrTaRb...
> > Yes I suspect as such. I'm not even sure the MB supports beyond AGP 1X.
I
> > purchased the 750 XGL at the time there was a promotion from PNY and
> > solidjerks for $399 - couldn't pass it up. The Voodoo5 card was pukin'
and
> > diein' on the XP operating system, and 3dfx underground drivers turned
out
> > to be vaporware.
> >
> > Would like to see a similar analysis for the 750 XGL in a worthy system,
> > like oh say a dual 2.0 Xeon 😉
> >
> > Again, the results indicate that a low-end machine (read P3) will not
> > respond to differing input.
> >
> > Thanks Many,
> > Jack
> >
> > "Andy Shelter" wrote in message
> > news:923BEBFEDA08B0AFE2D358AFAA819AEC@in.WebX.maYIadrTaRb...
> > > Jack,
> > > I'm excited about your extensive tests.
> > > After viewing your system specs I come to this conclusion:
> > > Your system seems to have a bottleneck with your P3-650 because your
> > results
> > > are so bad. On other systems the 750XGL gets about 30FPS in shaded
mode.
> > On
> > > your system only 10 FPS.
> > > You can not see any different results because your graphics board is
> > waiting
> > > on the CPU in EVERY case.
> > > The results on other systems are very different because of
> > > - different hardware (CPU, GPU, memory)
> > > - different Inventor graphics settings (I recommend to use IV default
> > > settings to assure we compare apples with apples)
> > > - different OS, different displaydriver and displaydriver settings
(this
> > is
> > > probably the smallest factor)
> > >
> > > Andy
> > >
> > >
> > >
> > > "Jack Brentner" wrote in message
> > > news:5199353292B5E6F4BEE877981CD1D6F6@in.WebX.maYIadrTaRb...
> > > > One can postulate any number of new hypotheses.
> > > >
> > > > To name a few:
> > > > H1) The benchmark results are suspect.
> > > > H2) The test equipment does not respond as expected to the chosen
> > inputs.
> > > > H3) Significant result differences are driven by other unidentified
> > inputs
> > > > or combinations thereof.
> > > >
> > > > Verification of
> > > > H1: Run similar trials with another benchmark tool.
> > > > H2: Perform similar trials on other equipment with the same
benchmark
> > > tool.
> > > > H3: Choose other inputs or parameters and combinations thereof and
> > re-run
> > > > (hardware, op system, other opengl tweaks, etc.).
> > > >
> > > > One can draw no other conclusion from my analysis other than that
"the
> > > > benchmark provided no significant different results with the inputs
> > chosen
> > > > on that particular machine". At the least, my results must be
> confirmed
> > > > independently before any other conclusions can be drawn.
> > > >
> > > > Thanks Many,
> > > > Jack
> > > >
> > > >
> > > > "Ron Crain" wrote in message
> > > > news:EF089B81EB15AD063E42823F17687CA6@in.WebX.maYIadrTaRb...
> > > > > Very interesting.
> > > > > I said it in Andy's thread and I'll say it here. I don't trust
> these
> > > > > numbers.
> > > > >
> > > > >
> > > > > "Jack Brentner" wrote in message
> > > > > news:1013E178544D2FC0DA171760A7AFCC88@in.WebX.maYIadrTaRb...
> > > > > > Using Andy Shelter's excellent graphics benchmark, I ran it lots
> of
> > > > times
> > > > > > with different nVidia drivers, OpenGl tweaks, suggested registry
> > > hacks,
> > > > 1
> > > > > > and 2 monitors, and with/without the statistics posting onscreen
> > hack.
> > > > > >
> > > > > > In all but one trial, I rebooted the machine between trials.
> > > > > >
> > > > > > Short of a Chi-Squared statistical analysis, my instincts tell
me
> > that
> > > > > there
> > > > > > is no significant difference in the results with ANY combination
> of
> > > > > > settings.
> > > > > >
> > > > > > See Excel spreadsheet posted in CF.
> > > > > >
> > > > > > -Jack
> > > > > >
> > > > > >
> > > > >
> > > > >
> > > >
> > > >
> > >
> > >
> >
> >
>
>
Message 9 of 13
Anonymous
in reply to: Anonymous

I have reason to suspect a lot of us are wondering what are the
corresponding improvements in performance due to factors such as choice of
driver, OpenGL tweaks, registry hacks, etc. and a system specific analysis
with a reliable benchmark tool is the only way this can be achieved. One
where numbers can be generated and you could feel reasonably certain these
numbers were realistic. It hasn't been disproven that the benchmark tool
before us isn't one we can feel comfortable basing these numbers on. Albeit
the exercise is time consuming for any individual.


"Sean Dotson" wrote in message
news:700056FCB51BCA417F8E3BC700EE0D07@in.WebX.maYIadrTaRb...
> oops that's in FPS, sorry... so not quite as good as the 750s
>
> --
> Sean Dotson, PE
> http://www.sdotson.com
> Check the Inventor FAQ for most common questions
> -----------------------------------------
> "Sean Dotson" wrote in message
> news:E832AE2740623D3A31E9F49583CF85A0@in.WebX.maYIadrTaRb...
> > For what it's worth...I have a 700XGL and these are my results (P4
2.0GHz,
> > 1/2 gig ram)
> >
> > Shaded = 22.1 FPS
> > Hidden = 8.1 FPS
> > Wireframe = 16.6 FPS
> > Multiple Windows = 23.3 FPS
> > Average = 17.5 FPS
> >
> > better than a lot of the 750s
> >
> > --
> > Sean Dotson, PE
> > http://www.sdotson.com
> > Check the Inventor FAQ for most common questions
> > -----------------------------------------
> > "Jack Brentner" wrote in message
> > news:64327BFACA760CBA23645078A68DFB35@in.WebX.maYIadrTaRb...
> > > Yes I suspect as such. I'm not even sure the MB supports beyond AGP
1X.
> I
> > > purchased the 750 XGL at the time there was a promotion from PNY and
> > > solidjerks for $399 - couldn't pass it up. The Voodoo5 card was pukin'
> and
> > > diein' on the XP operating system, and 3dfx underground drivers turned
> out
> > > to be vaporware.
> > >
> > > Would like to see a similar analysis for the 750 XGL in a worthy
system,
> > > like oh say a dual 2.0 Xeon 😉
> > >
> > > Again, the results indicate that a low-end machine (read P3) will not
> > > respond to differing input.
> > >
> > > Thanks Many,
> > > Jack
> > >
> > > "Andy Shelter" wrote in message
> > > news:923BEBFEDA08B0AFE2D358AFAA819AEC@in.WebX.maYIadrTaRb...
> > > > Jack,
> > > > I'm excited about your extensive tests.
> > > > After viewing your system specs I come to this conclusion:
> > > > Your system seems to have a bottleneck with your P3-650 because your
> > > results
> > > > are so bad. On other systems the 750XGL gets about 30FPS in shaded
> mode.
> > > On
> > > > your system only 10 FPS.
> > > > You can not see any different results because your graphics board is
> > > waiting
> > > > on the CPU in EVERY case.
> > > > The results on other systems are very different because of
> > > > - different hardware (CPU, GPU, memory)
> > > > - different Inventor graphics settings (I recommend to use IV
default
> > > > settings to assure we compare apples with apples)
> > > > - different OS, different displaydriver and displaydriver settings
> (this
> > > is
> > > > probably the smallest factor)
> > > >
> > > > Andy
> > > >
> > > >
> > > >
> > > > "Jack Brentner" wrote in message
> > > > news:5199353292B5E6F4BEE877981CD1D6F6@in.WebX.maYIadrTaRb...
> > > > > One can postulate any number of new hypotheses.
> > > > >
> > > > > To name a few:
> > > > > H1) The benchmark results are suspect.
> > > > > H2) The test equipment does not respond as expected to the chosen
> > > inputs.
> > > > > H3) Significant result differences are driven by other
unidentified
> > > inputs
> > > > > or combinations thereof.
> > > > >
> > > > > Verification of
> > > > > H1: Run similar trials with another benchmark tool.
> > > > > H2: Perform similar trials on other equipment with the same
> benchmark
> > > > tool.
> > > > > H3: Choose other inputs or parameters and combinations thereof and
> > > re-run
> > > > > (hardware, op system, other opengl tweaks, etc.).
> > > > >
> > > > > One can draw no other conclusion from my analysis other than that
> "the
> > > > > benchmark provided no significant different results with the
inputs
> > > chosen
> > > > > on that particular machine". At the least, my results must be
> > confirmed
> > > > > independently before any other conclusions can be drawn.
> > > > >
> > > > > Thanks Many,
> > > > > Jack
> > > > >
> > > > >
> > > > > "Ron Crain" wrote in message
> > > > > news:EF089B81EB15AD063E42823F17687CA6@in.WebX.maYIadrTaRb...
> > > > > > Very interesting.
> > > > > > I said it in Andy's thread and I'll say it here. I don't trust
> > these
> > > > > > numbers.
> > > > > >
> > > > > >
> > > > > > "Jack Brentner" wrote in message
> > > > > > news:1013E178544D2FC0DA171760A7AFCC88@in.WebX.maYIadrTaRb...
> > > > > > > Using Andy Shelter's excellent graphics benchmark, I ran it
lots
> > of
> > > > > times
> > > > > > > with different nVidia drivers, OpenGl tweaks, suggested
registry
> > > > hacks,
> > > > > 1
> > > > > > > and 2 monitors, and with/without the statistics posting
onscreen
> > > hack.
> > > > > > >
> > > > > > > In all but one trial, I rebooted the machine between trials.
> > > > > > >
> > > > > > > Short of a Chi-Squared statistical analysis, my instincts tell
> me
> > > that
> > > > > > there
> > > > > > > is no significant difference in the results with ANY
combination
> > of
> > > > > > > settings.
> > > > > > >
> > > > > > > See Excel spreadsheet posted in CF.
> > > > > > >
> > > > > > > -Jack
> > > > > > >
> > > > > > >
> > > > > >
> > > > > >
> > > > >
> > > > >
> > > >
> > > >
> > >
> > >
> >
> >
>
>
Message 10 of 13
Anonymous
in reply to: Anonymous

Since this benchmark is based on uncompiled VBA like mine was I suspect it
has the same shortfalls. While both of these tools are useful for "shades
of gray" comparisons I don't think they can give hard and fast numbers with
any degree of accuracy. I'd be curious to run the same benchmark 1000 times
and see how much the #s differ. I know that my benchmark will drift up to
10% if run in succession without rebooting. I suspect the same is true of
this one.

--
Sean Dotson, PE
http://www.sdotson.com
Check the Inventor FAQ for most common questions
-----------------------------------------
"Jack Brentner" wrote in message
news:A545F1C425BC3DA7115354DD35B05D6F@in.WebX.maYIadrTaRb...
> I have reason to suspect a lot of us are wondering what are the
> corresponding improvements in performance due to factors such as choice of
> driver, OpenGL tweaks, registry hacks, etc. and a system specific analysis
> with a reliable benchmark tool is the only way this can be achieved. One
> where numbers can be generated and you could feel reasonably certain these
> numbers were realistic. It hasn't been disproven that the benchmark tool
> before us isn't one we can feel comfortable basing these numbers on.
Albeit
> the exercise is time consuming for any individual.
>
>
> "Sean Dotson" wrote in message
> news:700056FCB51BCA417F8E3BC700EE0D07@in.WebX.maYIadrTaRb...
> > oops that's in FPS, sorry... so not quite as good as the 750s
> >
> > --
> > Sean Dotson, PE
> > http://www.sdotson.com
> > Check the Inventor FAQ for most common questions
> > -----------------------------------------
> > "Sean Dotson" wrote in message
> > news:E832AE2740623D3A31E9F49583CF85A0@in.WebX.maYIadrTaRb...
> > > For what it's worth...I have a 700XGL and these are my results (P4
> 2.0GHz,
> > > 1/2 gig ram)
> > >
> > > Shaded = 22.1 FPS
> > > Hidden = 8.1 FPS
> > > Wireframe = 16.6 FPS
> > > Multiple Windows = 23.3 FPS
> > > Average = 17.5 FPS
> > >
> > > better than a lot of the 750s
> > >
> > > --
> > > Sean Dotson, PE
> > > http://www.sdotson.com
> > > Check the Inventor FAQ for most common questions
> > > -----------------------------------------
> > > "Jack Brentner" wrote in message
> > > news:64327BFACA760CBA23645078A68DFB35@in.WebX.maYIadrTaRb...
> > > > Yes I suspect as such. I'm not even sure the MB supports beyond AGP
> 1X.
> > I
> > > > purchased the 750 XGL at the time there was a promotion from PNY and
> > > > solidjerks for $399 - couldn't pass it up. The Voodoo5 card was
pukin'
> > and
> > > > diein' on the XP operating system, and 3dfx underground drivers
turned
> > out
> > > > to be vaporware.
> > > >
> > > > Would like to see a similar analysis for the 750 XGL in a worthy
> system,
> > > > like oh say a dual 2.0 Xeon 😉
> > > >
> > > > Again, the results indicate that a low-end machine (read P3) will
not
> > > > respond to differing input.
> > > >
> > > > Thanks Many,
> > > > Jack
> > > >
> > > > "Andy Shelter" wrote in message
> > > > news:923BEBFEDA08B0AFE2D358AFAA819AEC@in.WebX.maYIadrTaRb...
> > > > > Jack,
> > > > > I'm excited about your extensive tests.
> > > > > After viewing your system specs I come to this conclusion:
> > > > > Your system seems to have a bottleneck with your P3-650 because
your
> > > > results
> > > > > are so bad. On other systems the 750XGL gets about 30FPS in shaded
> > mode.
> > > > On
> > > > > your system only 10 FPS.
> > > > > You can not see any different results because your graphics board
is
> > > > waiting
> > > > > on the CPU in EVERY case.
> > > > > The results on other systems are very different because of
> > > > > - different hardware (CPU, GPU, memory)
> > > > > - different Inventor graphics settings (I recommend to use IV
> default
> > > > > settings to assure we compare apples with apples)
> > > > > - different OS, different displaydriver and displaydriver settings
> > (this
> > > > is
> > > > > probably the smallest factor)
> > > > >
> > > > > Andy
> > > > >
> > > > >
> > > > >
> > > > > "Jack Brentner" wrote in message
> > > > > news:5199353292B5E6F4BEE877981CD1D6F6@in.WebX.maYIadrTaRb...
> > > > > > One can postulate any number of new hypotheses.
> > > > > >
> > > > > > To name a few:
> > > > > > H1) The benchmark results are suspect.
> > > > > > H2) The test equipment does not respond as expected to the
chosen
> > > > inputs.
> > > > > > H3) Significant result differences are driven by other
> unidentified
> > > > inputs
> > > > > > or combinations thereof.
> > > > > >
> > > > > > Verification of
> > > > > > H1: Run similar trials with another benchmark tool.
> > > > > > H2: Perform similar trials on other equipment with the same
> > benchmark
> > > > > tool.
> > > > > > H3: Choose other inputs or parameters and combinations thereof
and
> > > > re-run
> > > > > > (hardware, op system, other opengl tweaks, etc.).
> > > > > >
> > > > > > One can draw no other conclusion from my analysis other than
that
> > "the
> > > > > > benchmark provided no significant different results with the
> inputs
> > > > chosen
> > > > > > on that particular machine". At the least, my results must be
> > > confirmed
> > > > > > independently before any other conclusions can be drawn.
> > > > > >
> > > > > > Thanks Many,
> > > > > > Jack
> > > > > >
> > > > > >
> > > > > > "Ron Crain" wrote in message
> > > > > > news:EF089B81EB15AD063E42823F17687CA6@in.WebX.maYIadrTaRb...
> > > > > > > Very interesting.
> > > > > > > I said it in Andy's thread and I'll say it here. I don't
trust
> > > these
> > > > > > > numbers.
> > > > > > >
> > > > > > >
> > > > > > > "Jack Brentner" wrote in message
> > > > > > > news:1013E178544D2FC0DA171760A7AFCC88@in.WebX.maYIadrTaRb...
> > > > > > > > Using Andy Shelter's excellent graphics benchmark, I ran it
> lots
> > > of
> > > > > > times
> > > > > > > > with different nVidia drivers, OpenGl tweaks, suggested
> registry
> > > > > hacks,
> > > > > > 1
> > > > > > > > and 2 monitors, and with/without the statistics posting
> onscreen
> > > > hack.
> > > > > > > >
> > > > > > > > In all but one trial, I rebooted the machine between trials.
> > > > > > > >
> > > > > > > > Short of a Chi-Squared statistical analysis, my instincts
tell
> > me
> > > > that
> > > > > > > there
> > > > > > > > is no significant difference in the results with ANY
> combination
> > > of
> > > > > > > > settings.
> > > > > > > >
> > > > > > > > See Excel spreadsheet posted in CF.
> > > > > > > >
> > > > > > > > -Jack
> > > > > > > >
> > > > > > > >
> > > > > > >
> > > > > > >
> > > > > >
> > > > > >
> > > > >
> > > > >
> > > >
> > > >
> > >
> > >
> >
> >
>
>
Message 11 of 13
Anonymous
in reply to: Anonymous

I saw the same upward drift without reboot as well with this tool, so I made
sure to reboot between trials to exclude this factor in my data. With the
one driver 29.42, I ran the macro 3 different times and noticed the Average
FPS range from 7.4 to 7.6. I maintain that a similar battery of trials with
a faster test machine could yield beneficial information regarding the
driver and tweak factors - provided you reboot between trials. Without it,
any claims to improvement by these factors is merely just that.



"Sean Dotson" wrote in message
news:D3DD95CC90BC322E8BE16D7D1332FAD8@in.WebX.maYIadrTaRb...
> Since this benchmark is based on uncompiled VBA like mine was I suspect it
> has the same shortfalls. While both of these tools are useful for "shades
> of gray" comparisons I don't think they can give hard and fast numbers
with
> any degree of accuracy. I'd be curious to run the same benchmark 1000
times
> and see how much the #s differ. I know that my benchmark will drift up to
> 10% if run in succession without rebooting. I suspect the same is true of
> this one.
>
> --
> Sean Dotson, PE
> http://www.sdotson.com
> Check the Inventor FAQ for most common questions
> -----------------------------------------
> "Jack Brentner" wrote in message
> news:A545F1C425BC3DA7115354DD35B05D6F@in.WebX.maYIadrTaRb...
> > I have reason to suspect a lot of us are wondering what are the
> > corresponding improvements in performance due to factors such as choice
of
> > driver, OpenGL tweaks, registry hacks, etc. and a system specific
analysis
> > with a reliable benchmark tool is the only way this can be achieved. One
> > where numbers can be generated and you could feel reasonably certain
these
> > numbers were realistic. It hasn't been disproven that the benchmark tool
> > before us isn't one we can feel comfortable basing these numbers on.
> Albeit
> > the exercise is time consuming for any individual.
> >
> >
> > "Sean Dotson" wrote in message
> > news:700056FCB51BCA417F8E3BC700EE0D07@in.WebX.maYIadrTaRb...
> > > oops that's in FPS, sorry... so not quite as good as the 750s
> > >
> > > --
> > > Sean Dotson, PE
> > > http://www.sdotson.com
> > > Check the Inventor FAQ for most common questions
> > > -----------------------------------------
> > > "Sean Dotson" wrote in message
> > > news:E832AE2740623D3A31E9F49583CF85A0@in.WebX.maYIadrTaRb...
> > > > For what it's worth...I have a 700XGL and these are my results (P4
> > 2.0GHz,
> > > > 1/2 gig ram)
> > > >
> > > > Shaded = 22.1 FPS
> > > > Hidden = 8.1 FPS
> > > > Wireframe = 16.6 FPS
> > > > Multiple Windows = 23.3 FPS
> > > > Average = 17.5 FPS
> > > >
> > > > better than a lot of the 750s
> > > >
> > > > --
> > > > Sean Dotson, PE
> > > > http://www.sdotson.com
> > > > Check the Inventor FAQ for most common questions
> > > > -----------------------------------------
> > > > "Jack Brentner" wrote in message
> > > > news:64327BFACA760CBA23645078A68DFB35@in.WebX.maYIadrTaRb...
> > > > > Yes I suspect as such. I'm not even sure the MB supports beyond
AGP
> > 1X.
> > > I
> > > > > purchased the 750 XGL at the time there was a promotion from PNY
and
> > > > > solidjerks for $399 - couldn't pass it up. The Voodoo5 card was
> pukin'
> > > and
> > > > > diein' on the XP operating system, and 3dfx underground drivers
> turned
> > > out
> > > > > to be vaporware.
> > > > >
> > > > > Would like to see a similar analysis for the 750 XGL in a worthy
> > system,
> > > > > like oh say a dual 2.0 Xeon 😉
> > > > >
> > > > > Again, the results indicate that a low-end machine (read P3) will
> not
> > > > > respond to differing input.
> > > > >
> > > > > Thanks Many,
> > > > > Jack
> > > > >
> > > > > "Andy Shelter" wrote in message
> > > > > news:923BEBFEDA08B0AFE2D358AFAA819AEC@in.WebX.maYIadrTaRb...
> > > > > > Jack,
> > > > > > I'm excited about your extensive tests.
> > > > > > After viewing your system specs I come to this conclusion:
> > > > > > Your system seems to have a bottleneck with your P3-650 because
> your
> > > > > results
> > > > > > are so bad. On other systems the 750XGL gets about 30FPS in
shaded
> > > mode.
> > > > > On
> > > > > > your system only 10 FPS.
> > > > > > You can not see any different results because your graphics
board
> is
> > > > > waiting
> > > > > > on the CPU in EVERY case.
> > > > > > The results on other systems are very different because of
> > > > > > - different hardware (CPU, GPU, memory)
> > > > > > - different Inventor graphics settings (I recommend to use IV
> > default
> > > > > > settings to assure we compare apples with apples)
> > > > > > - different OS, different displaydriver and displaydriver
settings
> > > (this
> > > > > is
> > > > > > probably the smallest factor)
> > > > > >
> > > > > > Andy
> > > > > >
> > > > > >
> > > > > >
> > > > > > "Jack Brentner" wrote in message
> > > > > > news:5199353292B5E6F4BEE877981CD1D6F6@in.WebX.maYIadrTaRb...
> > > > > > > One can postulate any number of new hypotheses.
> > > > > > >
> > > > > > > To name a few:
> > > > > > > H1) The benchmark results are suspect.
> > > > > > > H2) The test equipment does not respond as expected to the
> chosen
> > > > > inputs.
> > > > > > > H3) Significant result differences are driven by other
> > unidentified
> > > > > inputs
> > > > > > > or combinations thereof.
> > > > > > >
> > > > > > > Verification of
> > > > > > > H1: Run similar trials with another benchmark tool.
> > > > > > > H2: Perform similar trials on other equipment with the same
> > > benchmark
> > > > > > tool.
> > > > > > > H3: Choose other inputs or parameters and combinations thereof
> and
> > > > > re-run
> > > > > > > (hardware, op system, other opengl tweaks, etc.).
> > > > > > >
> > > > > > > One can draw no other conclusion from my analysis other than
> that
> > > "the
> > > > > > > benchmark provided no significant different results with the
> > inputs
> > > > > chosen
> > > > > > > on that particular machine". At the least, my results must be
> > > > confirmed
> > > > > > > independently before any other conclusions can be drawn.
> > > > > > >
> > > > > > > Thanks Many,
> > > > > > > Jack
> > > > > > >
> > > > > > >
> > > > > > > "Ron Crain" wrote in message
> > > > > > > news:EF089B81EB15AD063E42823F17687CA6@in.WebX.maYIadrTaRb...
> > > > > > > > Very interesting.
> > > > > > > > I said it in Andy's thread and I'll say it here. I don't
> trust
> > > > these
> > > > > > > > numbers.
> > > > > > > >
> > > > > > > >
> > > > > > > > "Jack Brentner" wrote in message
> > > > > > > > news:1013E178544D2FC0DA171760A7AFCC88@in.WebX.maYIadrTaRb...
> > > > > > > > > Using Andy Shelter's excellent graphics benchmark, I ran
it
> > lots
> > > > of
> > > > > > > times
> > > > > > > > > with different nVidia drivers, OpenGl tweaks, suggested
> > registry
> > > > > > hacks,
> > > > > > > 1
> > > > > > > > > and 2 monitors, and with/without the statistics posting
> > onscreen
> > > > > hack.
> > > > > > > > >
> > > > > > > > > In all but one trial, I rebooted the machine between
trials.
> > > > > > > > >
> > > > > > > > > Short of a Chi-Squared statistical analysis, my instincts
> tell
> > > me
> > > > > that
> > > > > > > > there
> > > > > > > > > is no significant difference in the results with ANY
> > combination
> > > > of
> > > > > > > > > settings.
> > > > > > > > >
> > > > > > > > > See Excel spreadsheet posted in CF.
> > > > > > > > >
> > > > > > > > > -Jack
> > > > > > > > >
> > > > > > > > >
> > > > > > > >
> > > > > > > >
> > > > > > >
> > > > > > >
> > > > > >
> > > > > >
> > > > >
> > > > >
> > > >
> > > >
> > >
> > >
> >
> >
>
>
Message 12 of 13
rllthomas
in reply to: Anonymous

My first comment is thanks for the tool it was a good idea. Certainly a lot easier than running the debug tools again and again. Secondly from some of the things I have read I suspect that with a modern video card on the high end of the spectrum the CPU has a lot to do with your performance. Resolution also makes an obvious difference. 3rd, as for the validity of the test I will say this, how valid is a SPECPerf etc. compared to real world results. This test at least attempts to measure Inventor performance. With everyone having a different configuration and some machines having webshots/hotsync/IM programs etc. running at startup it gets difficult to compare machine A to machine B. Using the same machine with the same OS configuration etc. and swapping cards around along with drivers would most likely yield some useful results to compare best performance and also cost benefits. Is that latest and greatest really worth quadruple the price of something else. Finally, this is CAD not a game does 15 fps really matter compared to 30? As long as I get a good image (which some competetive products can't do no matter what the video card) and I can control the motion I am happy. - Rich Thomas
Message 13 of 13
loop29
in reply to: Anonymous

Here are my results on Athlon XP 1800+ 512 MBDDRAM GF 4 Ti 4400 Softquadroed to Quadro 4 750 XGL, OS Win2K. resolution 1280 x 1024 32 bit:

Inventor6 Grapics Benchmark

03.12.2002 00:43:15

Shaded = 45 FPS
Hidden = 18 FPS
Wireframe = 30 FPS
Multiple Windows = 45 FPS
Average = 34 FPS

I tested 3 times and had same results for this setting.

regards

Can't find what you're looking for? Ask the community or share your knowledge.

Post to forums  

Autodesk Design & Make Report