Text archives Help

---

Rocky Rhodes wrote:

bin/manta -np 8 -imagedisplay null -bench -camera "pinhole( -eye 852.55
-380.453 114.667 -lookat 1044.98 -383.53 113.746 -up 0.110811 0.157369
0.981303 -fov 60 )" -scene "lib/libscene_boeing777.so( -file
/dev/shm/engine.v3c1 -np 8 -alpha 1.0)"

It appears that the order of the command line args passed to bin/manta matters. The following order produces the expected result.

bin/manta -np 8 -scene "lib/libscene_boeing777.so( -file /dev/shm/engine.v3c1 -np 8 -alpha 
1.0)" -camera "pinhole( -eye 852.55 -380.453 114.667 -lookat 1044.98 -383.53 113.746 
-up 0.110811 0.157369 0.981303 -fov 60 )" [-imagedisplay null] -bench

Why the need to set the FGL_MACRO_TILE_FB env variable?  It doesn't seem to
affect my results (after trying it once).

For some reason yeager running over vizserver wasn't using the graphics pipes (which was extremely expensive with vizserver) unless we set that variable. I don't know where else it is a problem so I include it in all of the run scripts.


Abe

-----Original Message-----
From: Abe Stephens [mailto:abe@sci.utah.edu]
Sent: Tuesday, October 04, 2005 10:14 AM
To: Rocky Rhodes
Cc: 'Steven Parker'; manta@sci.utah.edu
Subject: Re: [MANTA] "-bench -imagedisplay null" broken in itanium branch?

Please send me the command line and the hostname. I will investigate.

I ran the following two commands on fisher and only saw a 9% difference:

bin/manta -np 8 -bench 20 20 (6.1fps)
bin/manta -np 8 -imagedisplay null -bench 20 20 (6.7 fps)

Also be sure to "setenv FGL_MACRO_TILE_FB 1" in your environment.

Abe

Rocky Rhodes wrote:

That'd be great, but I guess I don't believe it is actually rendering

that

fast.  This is the engine of the Boeing with only 8 processors doing the
rendering.  Maybe Abe has been even more productive than I have been
assuming...

        Rocky

-----Original Message-----
From: Steven Parker [mailto:sparker@cs.utah.edu]
Sent: Tuesday, October 04, 2005 9:54 AM
To: Rocky Rhodes
Cc: manta@sci.utah.edu; Abe Stephens
Subject: Re: [MANTA] "-bench -imagedisplay null" broken in itanium

branch?

Is it possible that it is rendering at 79fps and the display of the
image is slowing it down to 4.0?

Steve


On Oct 4, 2005, at 10:50 AM, Rocky Rhodes wrote:

The "-bench" option seems to be busted in the itanium branch when
run with the "-imagedisplay null" option.  I ran a scene with the
UI and saw about 4.0 fps on the screen.  Then I ran with "-bench -
imagedisplay null" and saw 79 fps.  When I run without the null
imagedisplay I get about the right number.



          Rocky

---

• [MANTA] "-bench -imagedisplay null" broken in itanium branch?, Rocky Rhodes, 10/04/2005
  • Re: [MANTA] "-bench -imagedisplay null" broken in itanium branch?, Steven Parker, 10/04/2005
  • <Possible follow-up(s)>
  • RE: [MANTA] "-bench -imagedisplay null" broken in itanium branch?, Rocky Rhodes, 10/04/2005
    • Re: [MANTA] "-bench -imagedisplay null" broken in itanium branch?, Abe Stephens, 10/04/2005
  • RE: [MANTA] "-bench -imagedisplay null" broken in itanium branch?, Rocky Rhodes, 10/04/2005
    • Re: [MANTA] "-bench -imagedisplay null" broken in itanium branch?, Abe Stephens, 10/04/2005

---

Archive powered by MHonArc 2.6.16.