Re: [VirtualGL-Users] Improving performance on very large displays

2013-06-15 Thread DRC
On 6/15/13 11:13 AM, Paul Melis wrote: I've done some similar tests, mostly at 3840x2160 over a 10G link. I seemed to hit a bottleneck of about 10 fps, but so far assumed it was the readback from the GPU that was the problem (GTX680 in a PCI Express x16 slot), judging from the vglrun +pr

Re: [VirtualGL-Users] Improving performance on very large displays

2013-06-15 Thread DRC
Well, it's important to understand that VNC will usually just send the portions of the display that have changed (incremental framebuffer update.) It will send a full framebuffer update upon first connecting and when specifically requested by the viewer (using the Refresh or Lossless Refresh

Re: [VirtualGL-Users] Improving performance on very large displays

2013-06-15 Thread Arthur HUILLET
On Sat, Jun 15, 2013 at 12:50:01PM -0500, DRC wrote: I dunno about that. My modest Quadro 600 can read back about a gigapixel/second. I don't think that's likely to ever be the bottleneck on a reasonably modern system. I have seen some slides that suggest that nVidia cripples the readback

Re: [VirtualGL-Users] Improving performance on very large displays

2013-06-15 Thread DRC
Ah, well for the purposes of simulating the full-size display, client-side scaling (which is already implemented in both the Windows and Java viewers) is probably what you want. When you use client-side scaling, the server still sends full-sized updates, and they are decompressed on the

Re: [VirtualGL-Users] Improving performance on very large displays

2013-06-15 Thread DRC
That doesn't ring true, but I have no way of verifying it, either. The only Linux machine I have with a GeForce is slow in general (it won't even memcpy() at more than about 200 Mpixels/sec), so it's not a good basis for comparison. On 6/15/13 3:32 PM, Arthur HUILLET wrote: On Sat, Jun 15,

Re: [VirtualGL-Users] Improving performance on very large displays

2013-06-15 Thread Vadim Peretokin
I have a GTX 560 that I could run a test on, if there are any easily available and anyone wants to look at the number. -- This SF.net email is sponsored by Windows: Build for Windows Store.

Re: [VirtualGL-Users] Improving performance on very large displays

2013-06-15 Thread DRC
glreadtest in the VirtualGL source. On Jun 15, 2013, at 7:22 PM, Vadim Peretokin vpereto...@gmail.com wrote: I have a GTX 560 that I could run a test on, if there are any easily available and anyone wants to look at the number.

[VirtualGL-Users] Building 2.3.2 on Ubuntu 12.04

2013-06-15 Thread Vadim Peretokin
Hi, I'm having an issue building VirtualGL sources (2.3.2) on Ubuntu 12.04. The turbojpeg library isn't where the sources expect, so I update the variable to locate it - but it still fails with a non-obvious error: vadi@gooseberry:~/Programs/VirtualGL-2.3.2$ cmake .

Re: [VirtualGL-Users] Building 2.3.2 on Ubuntu 12.04

2013-06-15 Thread Vadim Peretokin
I'd like to run the glreadtest binary, which doesn't seem to come with the .deb and is integrated into the make cmake files for compilation. -- This SF.net email is sponsored by Windows: Build for Windows Store.

Re: [VirtualGL-Users] Building 2.3.2 on Ubuntu 12.04

2013-06-15 Thread Vadim Peretokin
The likely issue was that I was using Ubuntu's libjpeg-turbo. Using the .deb off the website + libjpeg-turbo sources built VGL okay. Sorry about the noise! -- This SF.net email is sponsored by Windows: Build for Windows

Re: [VirtualGL-Users] Improving performance on very large displays

2013-06-15 Thread Vadim Peretokin
Here is the data for a GTX 560 (not Ti), 310.14 drivers, Ubuntu 12.04 64bit: http://pastebin.com/raw.php?i=xW8v7UpF -- This SF.net email is sponsored by Windows: Build for Windows Store.

Re: [VirtualGL-Users] Building 2.3.2 on Ubuntu 12.04

2013-06-15 Thread Vadim Peretokin
I think that would be a good idea. It's not that big and it helps - afaik it's not something that is easily available (unless one digs around in the tests Phoronix Test Suite offers). -- This SF.net email is sponsored by