Xpra: Ticket #2416: CUDA device selection by name

Follow up for #2415: we should be able to select the list of enabled / disabled devices using their name and / or PCI bus location (ie: 0000:1A:00.0) without doing a full CUDA context initialization.

load_device can get to the device name before calling make_context (which is slow).

Thu, 12 Sep 2019 08:41:49 GMT - Antoine Martin: status changed

We can't skip the make_context completely in select_device because driver.mem_get_info() requires a valid context.

Maybe we can have different policies for load balancing, and one may not require memory info to do its job.

Fri, 01 Nov 2019 12:22:14 GMT - Antoine Martin:


Fri, 01 Nov 2019 16:41:45 GMT - Antoine Martin:

r24340 adds a load-balancing attribute and a round-robin option, the memory policy is still the default. (chooses the GPU with the most free memory)

Still needs testing.

Thu, 21 Nov 2019 07:43:12 GMT - Antoine Martin:

Seems to have caused a bug: #2489.

See also initial load balancing code in #520.

Sat, 15 Feb 2020 13:03:53 GMT - Antoine Martin: status changed; resolution set

See examples in browser/xpra/trunk/src/etc/xpra/cuda.conf.

Sat, 23 Jan 2021 05:50:42 GMT - migration script:

this ticket has been moved to: https://github.com/Xpra-org/xpra/issues/2416