|
Oh no, some typos:
x2 4200 goes 2200mhz default and used that for start of course.
Quadro FX 1500 vsync off
Textured 10.7 fps
I'm in flu, hope rest makes some sense.
On Thu, 10 May 2007 00:20:39 +0300, Mikko Ronkainen <noratio(at)kolumbus.fi>
wrote:
On Wed, 09 May 2007 18:08:00 +0300, Simon Pickard
<mail(at)simonpickard.co.uk> wrote:
Just wondering here.. but how much of this is gfx card related and how
much is cpu?
highly cpu, but seems still quadros beat badly game cards.
When running these tests is the cpu maxed out at 100%? I.e. the cpu is
the limiting factor?
Problem is core swapping xsi dropping to single threaded, like discussed
elsewhere. My other core peaks 100% in this test.
Proud to present results from weary freak section:
All tests linux fc3_32bit/blackbox Twinview 2560x1024
amd x2 4200 (at)defaults 2000mhz ram2gb 200(x2) quite moderate homebrew.
Quadro FX 1500 vsync off
Textured 10.7 17fps
Hiddenline 17fps
Wire 32fps
Depth 20.1
Shade 14.6
OC'd same box 2530mhz ram230mhz(x2) and eh, some other quick tweaks but
card on defaults, vsync off.
Tex 12.3
Hid 19.7
Wire 37.1
Depth 23.7
forgot to shade
And this is the wonder, best buy ever:
Quadro fx 500 other same specs homebrew all defaults vsync off:
tex 9.8
wire 16.5
hiddenline 9.6
shade 10.7
seems this time I forgot Depth
This test don't do justice to the cards, but seems I'm not buying game
cards near future. If I'd been smart would have bought quadro 3500,
everything I've heard it goes wild compared to my cards in direct
interaction. Only downside seems to be particles. I believe I've asked
some information of the supported cards from soft but cant find it,
might have changed my decisions in favor of 3500 or 1400. In my simple
tests shaded goes ~25% faster in 6.01 than 6.0 this is not cpu related.
Overall I shouldn't be dissapointed, I'm wiping off any game card on
market and performance don't seem that bad overall either.
Regards
Mikko
Regards,
Simon
On 09/05/07, Kris Rivel <krisrivel(at)gmail.com> wrote:
I haven't done particle stuff on this new box with the 4600 yet but
I'll
check it out. I don't suppose you have a test scene with some
benchmarks I
could compare to do you? I have a Dell laptop with a Geforce 6800
Ultra Go
which I could compare to but the two systems are so drastically
different, I
don't know if it would be fair. My laptop is a Centrino with 2GB Ram,
Win32. My BOXX is a dual quad-core 2.66 with 4GB Ram and Win64 with
the
Quadro 4600.
Kris
Morten Bartholdy wrote:
They should be very much the same - a peculiar difference is the
GeForce
needs two power plugs while the Quadro only needs one. Steve -
"flashing" -
what does your admin mean by that? I only know of Softquadro'ing, but
don't
know if it can still be done with these newer cards.
James, you can grab the test scene here:
http://colorshopvfx.dk/XSI/GFX_test_Array_Rot.rar - 18 MB
My tests with the Quadro 4600 were done with a dual quadcore XEON 2,66
GHz
box with 4 GB RAM on Win XP Pro 32bit. I maximized the camera viewport
1280
x 1024 rez (don't know how much that matters for the fps) and played
the
animation in a loop until fps stabilized - playback set to all frames.
The
driver was the latest Quadro 4600 version - fresh down from Nvidias
site.
The results are like this:
Wireframe: 30 fps (at NTSC framerate this was probably just maxed
out and
could have run faster - didn't check.)
Hidden Line: 8.6 fps
Depthcue: 22.2 fps
Shaded: 10 fps
Textured: 12 fps
The GeForce 8800 GTX on the same box gave me these numbers:
Wireframe: 8.6 fps
Hidden Line: 4.4 fps
Depthcue: 8.2 fps
Shaded: 5.5 fps
Textured: 4.3 fps
On my own workstation - a dual XEON 3,4 GHz single core with a GeForce
7800
GTX I got these numbers (on a 1920 x 1200 display):
Wireframe: 11.4 fps
Hidden Line: 5.5 fps
Depthcue: 6.8 fps
Shaded: 6.1 fps
Textured: 3.9 fps
These numbers compared to the performance of the 8800 suggest it might
not
have been running with the right driver!! My hardware pusher told me
it was
installed with the latest Nvidia version so I didn't check, but since
it was
actually slower than my 7800 I feel like going back and redo the test
of the
8800 making sure to run the right driver. Perhaps it can actually gain
on
the Quadro after all (hopes up higher).
The difference in price between the GeForce 8800 GTX and the Quadro
4600 is
a factor 2.6 BTW, which has to be taken into account.
My remaining worry about buying the Quadro is actually the mentioning
of
extremely slow fps on particle sim/ogl display. I had a weird
observation
when I worked with the test scene cube. Before freezing the merged
object
(probably keeping the opstack for texture projections for all subparts,
about 1300 objects) the Quadro displayed extremely low fps for textured
playback - like 0.4 fps!! Could be something similar to the mentioned
Quadro
3500 and particle scene problem. Does anyone have similar experience
with
Quadros - Kris?
Best regards
Morten Bartholdy
3D & VFX Artist
----- Original Message -----
From: Steven Caron
To: XSI(at)Softimage.COM
Sent: Tuesday, May 08, 2007 6:41 PM
Subject: Re: hardware - which card is certified?
our desktop admin paul was mentioning to me these are pretty much the
same
card. i hope i am representing our conversation properly but the
geforce has
something inhibiting it. he mentioned flashing it?
steven
On 5/8/07, Morten Bartholdy <xsi(at)colorshopvfx.dk> wrote:
> I just tested a GeForce 8800 GTX vs. a Quadro 4600 on the same
machine
> today. These cards are supposedly based on the same gpu and have the
same
> amount of texture memory - 768 MB RAM. I used the latest GeForce
standard
> driver and the correct Quadro set to XSI ogl, and found a baffling
> performance difference in favour of the Quadro!! It was 2-3 times as
fast
as
> the GeForce in most tests - wireframe, hidden line, depth cue,
shaded and
> textured. I made a test with about 3000 objects and 1,6 million
triangles
> and one with one object and the same triangle amount - a merged
version af
> the first scene. Actually the Quadro was relative faster at handling
the
> scene with many objects thna the merged one. I am guessing it might
be
> possible to find a better OGL setting for the GeForce driver and
perhaps
> increase the performance some, but still, the difference is
remarkable.
>
> I wonder if someone here has had a chance to make more thorough
camparisons
> between GeForce and Quadro hardware twins, and can offer some
insight? I
am
> really beginning to wonder if the Quadro isn't worth the extra
pricetag -
> actually it makes sense - it is 2-3 times faster and cost about 3
times as
> much as the GeForce.
>
> Best regards
>
> Morten Bartholdy
> 3D & VFX Artist
>
>
>
>
>
> ----- Original Message -----
> From: "Todd Alan Peleg" <todd(at)theholdingcell.com>
> To: <XSI(at)Softimage.COM>
> Sent: Monday, May 07, 2007 11:24 PM
> Subject: Re: hardware - which card is certified?
>
>
> >i have a bad experience with 2 fx3500's and the fx3000's
> >
> > there are 2 of them here at the place i am at and the performance
of
them
> > has been whipped by my 5 year old laptop's nvidia
> >
> > (i have yet to explain it)
> >
> > some other people have offered MANY options to fix what seems to
be for
no
> > reason.. i haven't had time to investigate it.. but.. for it to be
the
> > case in multiple machines... i doubt its a rare thing..
> >
> > for what its worth
> >
> > todd
> >
> >
> >
> > Jeff McFall wrote:
> >
> >>Trying to pick cards for new machines and we are looking for some
> >>clarification regarding the QuadroFx3500.
> >>
> >>On the Softimage support site the QuadroFx3500 is not listed as
> >>"certified"
> >>
http://www.softimage.com/support/xsi/certified_graphics_cards/xp32/default.aspx
> >>
> >>However on the workstation page
> >>
http://www.softimage.com/support/xsi/certified_workstations/default.aspx
> >>The 3500 is listed as certified (at least for 5.11) in a Dell 690
and
Dell
> >>490
> >>
> >>Trying to get some solid info as to why this is.
> >>
> >>Is this just an oversight or something that should be considered
given
> >>changes in XSI 6. ?
> >>
> >>
> >>Also curious if there are any major performance differences
between the
> >>two cards.
> >>>From the specs it appears the 3500 may be more adept at handling
> >>>textures.
> >>Any other points to consider if we went with the lesser card?
> >>
> >>
> >>Thanks for any feedback or suggestions
> >>
> >>Jeff
> >>
> >>
> >>
> >>
> >>---
> >>Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text
in
body:
> >>unsubscribe xsi
> >>
> >>
> > ---
> > Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text
in
body:
> > unsubscribe xsi
> >
> >
>
> ---
> Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in
body:
> unsubscribe xsi
>
---
Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in
body:
unsubscribe xsi
---
Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in
body:
unsubscribe xsi
---
Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in body:
unsubscribe xsi
|
