@nprofile1q... and then on the sending side in /etc/pulse/default.pa

load-module module-null-sink sink_name=rtp
load-module module-rtp-send source=rtp.monitor source_ip=192.168.42.69 destination_ip=192.168.2.3 port=12345
set-default-sink rtp

@nprofile1q...

#!/bin/sh
gst-launch-1.0 -v udpsrc port=12345 ! "application/x-rtp,media=(string)audio, clock-rate=(int)44100, encoding-name=(string)L16, channels=(int)2" ! rtpL16depay ! audioconvert ! pulsesink sync=true buffer-time=2500

Now you are

@nprofile1q... personally when i have that problem (e.g. to get sound out of a VM running on a rackmount server across the building from the machine I want to listen to the audio at) i just run raw pcm audio over rtp...

@nprofile1q... @nprofile1q... i have am exhaust fan system that i can run during the winter but I don't think it moves nearly enough air to reject the amount of heat i need.

And yes a normal heat pump won't work you specifically need a VRF unit designed for energy harvesting

@nprofile1q... I would absolutely not plan a new, large design with a titanium today based on my current level of understanding of the ecosystem, I would stick with a ku+.

But the only way I'm going to get that level of familiarity is to start doing test projects

@nprofile1q... I have concerns about the current state of the software but it will mature over time and if I get in as an early adopter hopefully i can help give them feedback to guide things

@nprofile1q... I have disagreements with efinix about some architectural details most notably the lack of lutram, but i can make it work and ts16ff is fast enough compared to ts28hpl that I expect to be at least competitive with 7 series if not ultrascale for most stuff

@nprofile1q... I don't like the SoC parts or their descendents (versal, zynq, su+, ku+ gen2).

I will continue to use spartan/artix/kintex7 and au+/ku+ for as long as I have easy access to cheap components and a vivado 2025.1 seat and/or if the FOSS tools mature.

But the newer generation products have been dead to me for a while, I was already looking at efinix because xilinx seemed to be losing their way and going all-in on acceleration and socs rather than pure play FPGAs. The licensing BS is just accelerating an already-planned migration.

Efinix's stuff looks to be cheaper and also potentially larger at least than usable with free vivado (e.g. ku5p tops out at 16 transceivers matching the Ti375, but planned larger titanium parts will have up to 40x 25G transceivers)

@nprofile1q... @nprofile1q... yeah i do nightly incrementals of a selected subset of important stuff (e.g. not including VM disk images or full home dirs, but hopefully all of my really important stuff) off site over my comcast pipe.

Fiber is not available here unless I want to shell out $85K for them to trench yards the entire length of the neighborhood out to the nearest splice point (there are no above ground utility poles, and no empty conduit to pull through, all existing utilities are direct burial).

For everything else, I just rely on Ceph's e2e checksumming, 3-way replication, and occasional snapshotting.

@nprofile1q... @nprofile1q... (it's "only" about 4 kW plus whatever losses there are in UPS, lighting, HVAC, etc... my PUE is somewhere between 1.00 and 1.56 but I don't have the instrumentation/visibility to separate house power from lab overhead currently, all I know is whole-building draw and output from the UPS)

@nprofile1q... @nprofile1q... I am seriously looking at energy harvesting heat pumps to heat the house with waste heat from the lab in the winter, right now I run the air conditioner in January and feel like I'm wasting a ton of energy doing so.

Because I am.

@nprofile1q... Adding in the whole office/lab I have another 512GB of RAM at my desk, 256 on the soldering bench, 128 on the microscope bench so ~1.8 TB plus whatever is inside the oscilloscopes, random raspberry pi's, laptops, etc.