[open] Processing of a loop goes extremely slowly (expyriment)

irlijn

Dear all,

After a textdisplay with a duration 'keypress', I want Open Sesame to run two sequences in random order. To achieve this, I've made a loop with the variable 'order' with two cycles (1 and 2). If [order]=1 the first sequence should run, if [order]=2 the second sequence should run. This works, but the processing is very slowly. It takes more than 30 seconds to start the first or second sequence after the keypress...

The reaction time is measured in the sequence, so I use expyriment as back-end. When I choose legacy / PyGame, the experiment works fine, but the quality of the screen goes down (?) . I have no idea what the difference is between expyriment and PyGame and what I'm doing wrong... Does anyone has any idea?

Thanks in advance!

eduard

Hi Irlijn,

You have probably seen this already, but if you want to learn more about the different (dis)advantages of each backend, you can have a look on the documentation page.

Regarding xpyriment it says that you should use it, if timing is very important for your experiment and your stimuli aren't very complex. So, my first guess is that you do something quite complex within your sequences. that causes these long loading times. If timing isn't that crucial for you, I'd recommend you to use another backend, like legacy. However, if you need this temporal precision, you should try to simplify your sequences, as far as this is possible.

When I choose legacy / PyGame, the experiment works fine, but the quality of the screen goes down (?) .

Btw, what exactly do you mean with "the quality goes down" and the question mark. Does this really happen, or do you only expect it to happen? Did you try also different backends, such as psycho?

I hope I could help a bit.

Eduard

irlijn

Hi Eduard,

Thank you for your answer. The timing is very important, and the stimuli are normal images (768 x 576 pixels). The only 'complex' thing that happens in my experiment are the two sequences (with a loop) in a loop.
I don't know if I am correct by saying the screen has a lower quality, but xpyriment shows the text on the text display in a smaller size which makes it looks sharper compared to legacy. As if the resolution of the screen is lower when using legacy. This happens only for the text displays, the images have the same quality/sharpness. I've tried psycho as well as backend, but it gave this error:

"Unexpected error: line: 102; exception message: unsupported BMP header type (124); exception type: IOError"

So I think I have to follow your recommendation to simplify my experiment. Do you know if there's another way to randomize the order of two sequences? Because I think this is the biggest problem right now.

This is the body of my experiment:
http://i59.tinypic.com/mwun1f.jpg

Thank you for your time and help.

Irlijn

sebastiaan

Hi Irlijn,

I don't know if I am correct by saying the screen has a lower quality, but xpyriment shows the text on the text display in a smaller size which makes it looks sharper compared to legacy. As if the resolution of the screen is lower when using legacy.

The resolution that you specify in your experiment (1024 x 768 by default) has a different meaning for the legacy and xpyriment back-ends. The legacy back-end actually changes the monitor's resolution, whereas xpyriment (and psycho) creates a 'virtual monitor' in the center of the screen. This is explained by @lvanderlinden in this post:

• http://forum.cogsci.nl/index.php?p=/discussion/comment/1306#Comment_1306

Regarding the speed issue. The xpyriment back-end relies on the video card to render stimuli. This can be slow, especially with older/ slower video cards (the issue may disappear on other computers). However, I see in the screenshot that you're using the deprecated text_display plug-in, which is especially affected by this. So as a first step, you could replace these by the newer form_text_display plug-in, or a custom form if necessary.

• http://osdoc.cogsci.nl/forms/about/

Cheers!
Sebastiaan