I’ve had a couple problems using Touchin/out chops and tops in different instances of Touch on different GPU’s. The dats connect fine but both the chops and tops don’t get their clients consistently. Sometimes I restart the projects I have to delete the originals and then add new ins/outs to get them to connect again. When I do that sometimes they’ll even connect having different port numbers. One thing I haven’t tried is using 127.0.0.1 instead of localhost for the address I’ll try that next.
Also I do have quite a bit of other data coming in from Ableton which might not be helping.
One time I got the chops to connect by using Use synced ports setting. Should I be using this setting? If so will I be able to use it across three instances of Touch?
Hey,
One bug I just found was that the TouchIn/Out TOP and CHOPs didn’t work for ports under 5000. So in the file you gave me that’s why the Touch In/Out TOPs arn’t working (they are set the 3000). Could this be the issue you are running into or do you think it’s something ontop of that?
I had a feeling certain ports weren’t working. I was just trying to stay away from the ableton sync ports and lemur ports. No problem I’ll just use ports 30000 and up.
Hi Matt, I’m running two instances of Touch each on one of two Quadro 5000’s on the same machine with Windows 7. It’s working great. In the performance monitor it’s only taking about a 1 ms to cook the TouchIn Dat with a resolution of 1920 by 480. It was even working nicely at 3072 by 768.
I haven’t tried the sharedmem in/out chops yet but I did just try the TOPS. As far as cooking times go its seems the Touchin/out’s are quite a bit faster than the sharedmem’s.
touchout - .08 ms
sharedmemout - 3.1 ms
touchin - 1.2 ms
sharedmemin - 2.6 ms
Maybe there are other factors to consider? In the long run could they be more stable?
Could they be faster because that data is just being streamed and not written to memory?
I just thought that looking at the cook times doesn’t take into account the time after the first instance writes the frame and second instance starts to read it. Could that be faster with the sharedmem?
Tomorrow I’ll checkout the sharedmem CHOPs.
Edit:
I just did a test with moviein TOP playing the countdown movie and found that the touchin is 4 or 5 frames behind the sharedmem. I also did another test where I stop the movie playing with a button and found the touchin took some time to stop while the sharedmemin perceptibly stopped at the same time.
I guess a couple of ms of cook time is a small price to pay for nearly instant data transfer!
Seeing this I think I’ll switch to the sharedmem Chops also. I always thought the sharedmem shared graphics memory not cpu memory.