Jump to content
Light-O-Rama Forums

A Specific Question For LOR Software Developers


Guest Guest

Recommended Posts

LOR will have it's work cut out for this summer. It has a great foundation but just needs a faster data spewing device. I have seen this with many software packages as they work fine for a while and then when they see triple the amount of data they suffer until they get the required optimization attention they need.

Link to comment
Share on other sites

Are you removing the clipboards from your saved list? Leaving them there reduces the amount of available memory.

Which tool are these clipboards coming from?

I usually have only 1 clipboard loaded. Is there a way to purge it once I have pasted? After a large cut/paste, I copied a single empty cell.

But no go, LOR will not save...

Link to comment
Share on other sites

Caught between a rock and a hard place.....

LSP can handle the sequence just fine, but can't play it (they have player issues.....), LSP can't even play my COTB file that comes in at 700MB but LR chugs right a long..

LOR can play everything so far just fine, but I can't run the large sequence to, well, sequence....

I know it's too late now, but I want to play this sequence, xlights???

Link to comment
Share on other sites

That stinks! I also had no luck with switching to unicast the result was exactly the same as multicast. It only lags during lots of fades and transitions across more than 22 universes. I went back in and earased everything and just used on off commands for all 65 universes and playback was smooth on my lights. So yes we have hit a glass cieling IMO. I also spoke with several users of other software and hardware and they are seeing the same problems if they fade in and out of colors for all universes. They just changed thier sequences to make it less demanding on the software. I dont want to do that, I really like some of the effects I created in S3 this year I would hate to have to delete them to be able to add more universes. I really believe LOR can handle whatever we throw at it but it may need some more memory allocated to it or utilize more cpu than it currently does.

Funny......

That sentence brings me back to my CCPixel debacle....

I was told that fades are the worst for bandwith.... Take out your fades......

time for my thinking cap....(don't hold your breathe, it's pretty small)

Link to comment
Share on other sites

I went over that with Andy. The problem with fades over DMX vs fades with LOR. DMX needs a command for every part of the fade, so 40 times a second its telling that specific channel what to do. Multiply that over say 1000 pixels, means at complete load your sending 3000 commands 40 times a second over the duration of the fade. So for 1 minute you have 7.2 million commands if they are all changing with the fade. Again this is worst case scenario. With LOR on the other hand on the CCR controllers of 120v setups. It send 1 command to fade that channel for 60 seconds from 100% to say 0% and nothing more until the channel has something else to do. Where we are starting to see the problems is in those cases where millions of codes have to be sent in those tiny time intervals. The problem comes in as is it the software choking, is the ethernet were using capable of that data stream, or are the controllers unable to buffer and play that many commands in that amount of time.

This is my point of view, not LOR's or any developers just what I see is a possibility to why were seeing those data issues. If anyone has the ability to confirm this I would be appreciative.

  • Like 1
Link to comment
Share on other sites

DMX is really not commands, but a series of frames of intensities. And it is always trying to send out 40+ frames a second.

From a network perspective, all on, or all fading is no difference.

From a SE or show player perspective, it is a computational difference. You have to calculate the intensity of each channel in each frame. If you have a bunch of channels that turned on or off, you just have to mark new values of 0 or 255 into the appropriate byte of the frame. But with a fade, for each frame, and each fade, you have to figure, I am 1 second into a 10 second fade from 0 to 255. Thus i need 25 for this channel in this frame.

This is where the delay comes from. This is why it is more computer intensive to do DMX than LOR.

  • Like 1
Link to comment
Share on other sites

DMX is really not commands, but a series of frames of intensities. And it is always trying to send out 40+ frames a second.

From a network perspective, all on, or all fading is no difference.

From a SE or show player perspective, it is a computational difference. You have to calculate the intensity of each channel in each frame. If you have a bunch of channels that turned on or off, you just have to mark new values of 0 or 255 into the appropriate byte of the frame. But with a fade, for each frame, and each fade, you have to figure, I am 1 second into a 10 second fade from 0 to 255. Thus i need 25 for this channel in this frame.

This is where the delay comes from. This is why it is more computer intensive to do DMX than LOR.

So on the controllers end that is reading the DMX signal its normal for it to be recieveing updates on the 40 frames per second, the software is where the bottleneck comes in.

Link to comment
Share on other sites

There can be bottlenecks on E1.31 controllers, where they have to process too many universes of multicast traffic to get their data, but the symptoms would most likely be similar with fades or on/off events.

If one wanted to seek better clarity, use 2 nics, and split your multicast across 2 networks and see if that changed the behavior.

Link to comment
Share on other sites

OMG!!!!!!!!!!!!!!!!!

So at work yesterday I did a lot of googling. And when you have a memory issue, you go to gamers and gamer friends.

a guy at tomshardware.com (yes, I am thebaronn there also) mentioned a that can enable a 32bit program to take advantage of more memory if running on a 64bit operation system.

I changed S3 to allow it to use more than the 32bit 2GB memory limit.

So far so good. I was at a max of saving Pixel panels 1, 2, and 5 and half of pixel panel 3. (I have 5 universes on each panel of video rendered at 15fps) and the file max was 1.1GB

My filie is now 1.3GB and I am finished with 1, 2, 3, 5 and half of 4.

So now, if I get lucky and finish the sequence (you can see in the photo that I am already nearing the 4GB memory limit) playing it will be interesting.

More to come, this has me ver excited to see what a 64bit version could do. I do know that in LSP the 32bit version always had memory limits and whne the 64bit version came out, they were gone and I was able to render some amazing stuff!

Cross your fingers for me that I can finish this sequence and sleep!

post-953-0-80621000-1357083551_thumb.jpg

post-953-0-93178000-1357083560_thumb.jpg

post-953-0-16009900-1357084243_thumb.jpg

Edited by thebaronn
Link to comment
Share on other sites

Just realized I was so caught up in the madness I didn't realize 1 key piece of info....

Even when S3 gives the error: "unable to save sequence, would you like to save under another name?"

S3 actually did save the sequence in the temp file, I believe it just ran out of memory when it tried to refresh the sequence.

I only noticed it because when I went to unload the temp file I noticed that the LORTEMP3 was over 1.4GB (bigger than any sequence I had) so I opened it, it "looks" okay.

So, since I reached the GB memory ceiling pretty fast (only 2:37 sec to go in 5 universes) I am going to assume that it is okay and sequence from there.

I haven't tried playing it yet, I will cross that bridge soon enough.

Link to comment
Share on other sites

Sorry, I was caught up and wanted to finish it.

Now befroe I go on, I want to stress the possible issues, TOASTING YOUR COMPUTER!!!

When you change anything or introduce random code tweaker tools, you could have issues. Beofre W.O.W Blizzard ahad another fanstasy online gaming thing called NeverWinter Nights or something like that (before I becamse a LightAholic) and we change the code in the game all the time. Granted, I just followed instructions, like I did here.

I googled "can I get a 32bit program to access more RAm with 64bit Windows 7" and seahced forever (do you know there is more than 1 seach page? LOL! I actually looked this time) Seval places said you could change the code with a code editor and Microsoft even gives a list of known programs that can benifit from doing so. It is called "large Address Aware" and when you allow a program to realize more of its potential because systems now are awesome, the resuts can be amazing! But, you know, I don't really like to access code on my computer because I sometimes get excited and miss things (hence why I do not really like the DIY lighting stuff) so I searched for a program to do it for me. I found 1 but didn't really like it. Over at Phantasy Star Online I found the answer. An actual program than works like windows explorer. GUI? Muy Bien!

So I went tp http://www.ntcore.com/exsuite.php and looked at it, it seemed pretty legit. And well, I didn't care. On my monster PC, I have a 120GB SSD, not a lot of room. I have 2 programs.....LSP and LOR. That's it. So worse case, I toast it and have to blank out the hard drive and re-install windows (and trust me, I was a HUGE music sharer back in the day, I have jacked up my share of computers)

I installed it. I made a copy of the sequence editor (just in case).

Then I do the following:

Run CFF Explorer Suite as an administraitor.

Open the LOR SequenceEditor.exe

Click on the "+" in front of Nt headers (It may already be open, always is n mine)

Click on the box that is in the Characteristics row and Meaning Column then click on the box next to "app can handle >2GB address space"

Then save all the way.

BOOM!!! 4GB of memory!! Now there is also a patch t allow more than 4GB, but that was super intensive and you have to alter parts of windows so I didn't bother with that. But I read that the 4GB ceiling my son become a thing of the past because new programs are just becoming to memory hungry.

post-953-0-02791600-1357091327_thumb.jpg

post-953-0-28752500-1357091339_thumb.jpg

Link to comment
Share on other sites

So I played the sequence......

Sound came on, no lights.. Great..

I had a problem earlier with my sequencer not playing the E1.31 (yes the winsock was open...) but the scheduler would...

Re-installed and the problem disappeared, but now it was back.

I set up the scheduler.

ERROR!!!!! CANNOT ALLOCATE MEMORY FOR EVENT MODES!!!

i have a pic but I guess I am at my limit....

So I have to allocate more momory for the LORMonitor.exe

It started and sound came out!! woohoo! The scheduler was now able to play a 1.6GB file!

But no lights.......

Reset all the E682, no lights. Great. I went to 192.168.0.101 expecting t see a bunch of errors......

I forgot, I plugged the computer to the garage ethernet jack to connect to the internet. It help to plug the computer into the router where the E682s are, LOL!

(I do not connect the show computer and router to the internet like others so)

And it works!!! 4.23 minutes!!! 3984 RGB Pixels over E1.31 with 26 universes. 12K logical channels, 1.6GB!!

The majority is rendered video at 15fps. I do not detect any lag. the videos start on cue.

I didn't make new years, but I feel freaking awesome!

Link to comment
Share on other sites

Going back to the multi-cast vs unicast discussion a couple dozen posts ago, remember that there is a rather hard limit to network capacity. Our LANs are far faster than RS-485, but there is still a limit. Depending on your network configuration, there are a couple things that can really slow that capacity down. Since E1.31 is sending levels for every channel on every universe configured about 40 times every second (even when you are not running a show, but have the DMX comm listener running), there is quite a bit of network traffic. Are you REALLY running at the speed you think you are? For example if you are using a fairly dumb hub on your network, and there is a single device that only supports 10Base-T, everything is now running at 10Base-T. Did you remember that old desktop in the corner that is driving a printer, or maybe an old access point that is 10Base-10? Currently I am running my E1.31 on same network as the rest of the home PCs etc through an intelligent managed switch and it will handle 10 or 100Base-T just fine. I do have one device that is 10Base-T only and the switch slows down the data for the one device just fine. It is interesting to look at the network utilization for that one port on the switch. I will shortly be adding a second network card to the show PC to get the multi-cast traffic onto it's own network that will have nothing but the show PC and E1.31 devices.

Now for those running really huge numbers of universes, I'm wondering if going to multiple E1.31 networks will be possible. If able to use uni-cast, it should not be a problem, but it might be harder to setup with multi-cast. Have to give that one some thought... Then to see if Jim can make the E682 handle uni-cast...

  • Like 1
Link to comment
Share on other sites

You ARE awesome! and brave. I dont want to make that change. Will hold out for S4. Maybe a 64bit version will come out some day.

Link to comment
Share on other sites

I see we are back to the hardware discussion. Its not hardware, I converted the 65 universe sequence to an xlights sequence last night. The result was a file size less than half of the original LOR sequence and it played back perfectly outside with no lag. Same exact hardware, hooked up the same exact way.

Edited by harrison0550
Link to comment
Share on other sites

Going back to the multi-cast vs unicast discussion a couple dozen posts ago, remember that there is a rather hard limit to network capacity. Our LANs are far faster than RS-485, but there is still a limit. Depending on your network configuration, there are a couple things that can really slow that capacity down. Since E1.31 is sending levels for every channel on every universe configured about 40 times every second (even when you are not running a show, but have the DMX comm listener running), there is quite a bit of network traffic. Are you REALLY running at the speed you think you are? For example if you are using a fairly dumb hub on your network, and there is a single device that only supports 10Base-T, everything is now running at 10Base-T. Did you remember that old desktop in the corner that is driving a printer, or maybe an old access point that is 10Base-10? Currently I am running my E1.31 on same network as the rest of the home PCs etc through an intelligent managed switch and it will handle 10 or 100Base-T just fine. I do have one device that is 10Base-T only and the switch slows down the data for the one device just fine. It is interesting to look at the network utilization for that one port on the switch. I will shortly be adding a second network card to the show PC to get the multi-cast traffic onto it's own network that will have nothing but the show PC and E1.31 devices.

Now for those running really huge numbers of universes, I'm wondering if going to multiple E1.31 networks will be possible. If able to use uni-cast, it should not be a problem, but it might be harder to setup with multi-cast. Have to give that one some thought... Then to see if Jim can make the E682 handle uni-cast...

I have the computer and E682s on their own network, no internet connection. I am only running 26 universes but obviously a lot of data. Is there a way to calculate the amount of data that LOR sends out vs the amount of data in transit vs amount of data received/processed?

All I know is I am playing actual music videos on my screen at 15fps and looks good to me, and I am the biggest critic. Although when it take LOR 5 minutes to open a sequence and 8minutes to save it, maybe my perception is influenced!

You ARE awesome! and brave. I dont want to make that change. Will hold out for S4. Maybe a 64bit version will come out some day.

I only did this because I am sick and love to push the limits. I pushed it in LSP and hit a brick wall with their scheduler.

I hit a wall in S3 but somehow I managed to jump over it.

I also apologize for my typing. I am terrible and go to fast and my double checks are bad as well and I know sometimes you guys are reading and asking if I am drunk or something.

Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
×
×
  • Create New...