dear geoff
thanks for the reports. i'm cc-ing to the psychtoolbox to inform other
users.
1. I'm very surprised that ScreenTest crashes on the Apple AGP
NV11/GeForce2MX video card, but this is obviously important to know. I
will try to fix this as soon as I can get access to a computer with that
card.
2. The ScreenTest results on your two computers have only one significant
surprise, as you note: The CopyBits rate to your Rage 128 card is a mere
45 MB/s on your fast G4, which is less than the 66 MB/s that you achieve
with the same card on your slower G3. I don't have any explanation for
that. ScreenTest is simply documenting the performance of the driver. If
you want to explore, you might try running ScreenTest again when only one
Rage 128 is installed on your G4, since it's conceivable that the
presence of two identical cards affects the operation of the driver. Note
that the G4 is running Mac OS 9.1 and has a newer version of the driver
(1.0.1 vs 1.0), so the NEW driver seems to be slower. It is conceivable
that downgrading the MacOS to 9.0 would give you the old driver and the
higher speed, but I don't know if 9.0 supports your G4. In principle it
might be possible to grab the old driver from Mac OS 9.0 and install it
in the new system, but i don't know how to do that.
3. The warnings in your ScreenTest about SetClut waiting or not waiting
for blanking are merely warnings. They are of interest to me, in making
ScreenTest produce the most informative results, but you can ignore them,
since you can always synch to interrupts instead of SetClut.
best
denis
p.s. you may want to upgrade MATLAB from 5.2.0 to 5.2.1 on your G3 since
the bug fixes are worth having. however, this won't affect ScreenTest.
---------------- Begin Forwarded Message ----------------
Date: 8/2/01 7:42 PM
Received: 8/2/01 7:50 PM
From: Geoff Loftus, gloftus@...
To: David Brainard, brainard@...
pelli Denis Pelli, denis@...
Hi David: I was just about to write you guys. It turns out that
Screentest will
just not run, i.e., it always crashes when the Apple AGP NV11/GeForce2MX
video
card is installed; there need not be a monitor attached to it. It was the
card
that came standard with my G4 and is used to run the apple flat-screen
displays,
and it happened with three separate cards on two computers.
So I took the card out leaving two previous auxiliary video cards. I
attach the
results of screentest with them. As you'll see, there's something goofy
going on
and screentest requested that I send you guys the results...I'm puzzled
as to
why the copybits rate is so low. For comparison, I'm also sending you
screentest
results from a G3 wherein copybits is considerably faster. I guess I don't
completely understand the degree to which it's the computer vs the video
card
that's making the difference. Anyway, when you've had a chance to look
this
over, maybe I can contact one of you and pick your brains a bit.
Geoff
Geoffrey R. Loftus, PhD Office: (206) 543-8874
Department of Psychology Home: (206) 547-6969
Guthrie Hall, Room 131 Mobile: (206) 605-1974
University of Washington Fax: (206) 685-3157
Box 351525 Web:
http://faculty.washington.edu/gloftus
Seattle, WA 98195
----------
thanks for the reports. i'm cc-ing to the psychtoolbox to inform other
users.
1. I'm very surprised that ScreenTest crashes on the Apple AGP
NV11/GeForce2MX video card, but this is obviously important to know. I
will try to fix this as soon as I can get access to a computer with that
card.
2. The ScreenTest results on your two computers have only one significant
surprise, as you note: The CopyBits rate to your Rage 128 card is a mere
45 MB/s on your fast G4, which is less than the 66 MB/s that you achieve
with the same card on your slower G3. I don't have any explanation for
that. ScreenTest is simply documenting the performance of the driver. If
you want to explore, you might try running ScreenTest again when only one
Rage 128 is installed on your G4, since it's conceivable that the
presence of two identical cards affects the operation of the driver. Note
that the G4 is running Mac OS 9.1 and has a newer version of the driver
(1.0.1 vs 1.0), so the NEW driver seems to be slower. It is conceivable
that downgrading the MacOS to 9.0 would give you the old driver and the
higher speed, but I don't know if 9.0 supports your G4. In principle it
might be possible to grab the old driver from Mac OS 9.0 and install it
in the new system, but i don't know how to do that.
3. The warnings in your ScreenTest about SetClut waiting or not waiting
for blanking are merely warnings. They are of interest to me, in making
ScreenTest produce the most informative results, but you can ignore them,
since you can always synch to interrupts instead of SetClut.
best
denis
p.s. you may want to upgrade MATLAB from 5.2.0 to 5.2.1 on your G3 since
the bug fixes are worth having. however, this won't affect ScreenTest.
---------------- Begin Forwarded Message ----------------
Date: 8/2/01 7:42 PM
Received: 8/2/01 7:50 PM
From: Geoff Loftus, gloftus@...
To: David Brainard, brainard@...
pelli Denis Pelli, denis@...
Hi David: I was just about to write you guys. It turns out that
Screentest will
just not run, i.e., it always crashes when the Apple AGP NV11/GeForce2MX
video
card is installed; there need not be a monitor attached to it. It was the
card
that came standard with my G4 and is used to run the apple flat-screen
displays,
and it happened with three separate cards on two computers.
So I took the card out leaving two previous auxiliary video cards. I
attach the
results of screentest with them. As you'll see, there's something goofy
going on
and screentest requested that I send you guys the results...I'm puzzled
as to
why the copybits rate is so low. For comparison, I'm also sending you
screentest
results from a G3 wherein copybits is considerably faster. I guess I don't
completely understand the degree to which it's the computer vs the video
card
that's making the difference. Anyway, when you've had a chance to look
this
over, maybe I can contact one of you and pick your brains a bit.
Geoff
Geoffrey R. Loftus, PhD Office: (206) 543-8874
Department of Psychology Home: (206) 547-6969
Guthrie Hall, Room 131 Mobile: (206) 605-1974
University of Washington Fax: (206) 685-3157
Box 351525 Web:
http://faculty.washington.edu/gloftus
Seattle, WA 98195
----------
>From: David Brainard <brainard@...>from
>To: "Geoff Loftus" <gloftus@...>
>Subject: Re: screentest crash
>Date: Thu, Aug 2, 2001, 3:23 PM
>
>
> Geoff,
>
> I don't have any insight as to why ScreenTest crashes. I know for
> sure it runs OK with many G4s running OS 9. Denis may have
> more insight.
>
> DB
>
>>Dear Denis and David: This is a followup to the conversation I had yesterday
>>with (currently very busy) David.
>>
>>The problem started when vbl didn't seem to work with an LCD projector in the
>>lab. Computer is a G4 running OS 9.1, quite new.
>>
>>ScreenTest then crashed when I tried it: it started as usual, darkened the
>>screen, returned with a normal screen, but the screen was frozee
>>with the usual
>>vertical cursor. Opt-Cmd-Escape wouldn't even work to free it up.
>>
>>In successive attempts to get ScreenTest working, I did the following
>>
>>1. Removed all video cards, leaving only on-board video.
>>
>>2. Reinstalled MATLAB and got the latest version of your
>>psychophysics toolbox.
>>
>>And ScreenTest still crashes just as it did before.
>>
>>Some of your other tests do decidedly un-standard things as well, ranging
>>freezing the system to not quite crashing it, but changing the screen toweird
>>colors (apparently there's some kind of clut issue). I haven't systematically----------------- End Forwarded Message -----------------
>>gone through all tests noting their results although I could.
>>
>>I haven't reinstalled the system. That is of course the obvious next step
>>although at least superficially all system functions appear to be
>>fine. Because
>>we have some custom USB keypads, reinstalling the system is somewhat of a
>>hassle, so I haven't done it, in hopes of maybe getting some advice from you
>>guys.
>>
>>thanks much for any help...Geoff
>>
>>Geoffrey R. Loftus, PhD Office: (206) 543-8874
>>Department of Psychology Home: (206) 547-6969
>>Guthrie Hall, Room 131 Mobile: (206) 605-1974
>>University of Washington Fax: (206) 685-3157
>>Box 351525 Web: http://faculty.washington.edu/gloftus
>>Seattle, WA 98195
>