OnsetDelay query

[Peter Quain]

Hi

This problem could reflect poorly performing graphics system (or 
other) on the computer. Agree with David - 1000 ms load times are 
ridiculous, and even 400 ms, on the laptop, seems like a very long 
time, whether it is happening during a trial sequence, or not. On an 
old p4 2.8Ghz laptop, garden variety graphics, 768Mb RAM I got load 
times in e-prime of 20 ms for a 2.3Mb BMP, 7 ms for 567Kb pic. 
E-prime Users guide cites old Pentium 450MHz 30ms load times, bitmap 
unspecified file size.

You should cache your image files.

- What are the file sizes? What are these machine's configurations?
Load times of 1000 ms out of order, unless your files are huge .. 
+100Mb maybe ..., or your computer is old (and thus very slow), or ill.

-Your computer (graphics card / drivers?; have you a g-Force, N-vidia 
card?? if so, try ATI... CPU?; RAM?; HDD?) mightn't be operating too 
well when it comes to discussing images with e-prime... or there 
could be something to do with the programming of the paradigm that is 
screwing things up between fixation and stim. The wait object? can't 
see how it could influence after fixation executes...

Maybe to find the more about cause of problem try 
logging  'ActionDelay' on stim presentation pbject, with no 
prerelease on fixation. Action delay will tell you how long it takes 
for stim object to do all prep for drawing to screen. 1) If loading 
is the problem, ActionDelay will be the same as onsetDelay. 2) if 
they aren't the same then something else in the program is possibly 
delaying execution of stim object. To check whether prerelease is 
doing anything, set fixation pr to maybe 200 ms, then run and log 
action delay and onset delay on stim object again. Action delay 
should be the same as with no pr, and onset delay *should* be reduced 
by 200 ms, i think. (but you report no influence of fixation 
prerelease on onset time..??)

Reduction in load time of 400 ms (from 1000 to 600 ms with no 
prerelease) with image dimension (and thus size) reduction (on the 
same computer) would suggest that it is your computer graphics (and 
or RAM, CPU, HDD) performance, not odd code, that is causing the 
delay. A 1024*768*32bit image is about 5/8 file size of 
1280*1024*32bit image (I think). Interestingly, but perhaps 
coincidentally , 5/8 of 1000ms = 625ms. However, I can't work out how 
this would add up with what you note about prerelease behaviour.

Same conclusion from the delay reduction on another machine (the 
laptop). Don't think it would have anything to do with screen *size*, 
provided that its screen could run native at the appropriate 
resolution. Different machine, different graphics hardware, CPU etc = 
different load times. Still long though ...

Anyway, try pre-load the image files, and if you still have timing 
issues, change your video card - or your computer, if you have a 
better specced 1 around.

Best
~Peter


At 01:54 AM 29/08/2009, you wrote:

>River,
>
> >So therefore the OnsetDelay in this instance describes the delay
> >between the end of the fixation duration and the start of the image
> >duration. Would this be correct?
>
>Just to be clear, there is practically *no* delay between the
>*actual* end of the fixation duration and the *actual* start of the
>image duration.  OnsetDelay refers rather to the delay from the
>*specified* start time of an object to the *actual* start time of
>that object.  This is illustrated in a very complex diagram in
>Appendix E of the User's Guide that came with E-Prime.
>
>Beyond that, I would guess that your high-resolution images present a
>problem.  You could test that further by trying some good old
>6402x480 16-bit images.  Offhand, OnsetDelays of over 400 ms seem
>*way* too long, and I hope others weigh in with their own experiences here.
>
>You might also resort to pre-loading or caching the images before
>display.  You may download PST's
>"<http://www.pstnet.com/e-prime/support/samples.asp?Mode=View&SampleID=23>Pre-loading 
>
>images without the use of Canvas" sample from their web site, or you
>may try the TimingParadigm5 example mentioned in Chapter 3 of the
>User's Guide that came with E-Prime (PST does not provide
>TimingParadigm5 for download, you must request it through e-mail or
>Web Support as I did).
>
>-- David McFarlane, Professional Faultfinder
>
>
>

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "E-Prime" group.
To post to this group, send email to e-prime at googlegroups.com
To unsubscribe from this group, send email to e-prime+unsubscribe at googlegroups.com
For more options, visit this group at http://groups.google.com/group/e-prime?hl=en
-~----------~----~----~----~------~----~------~--~---