A new kind of workload scheduler. I think it's silly that it's so normal datacenters to sit at ~20% utilization all the time.
With the right scheduler I think we could get that above 90%. Would love to hear any feedback / thoughts. Here's a blog explaining: https://docs.burla.dev/blog/dynamic-hardware
It's cool that this is possible on a single node but I still think distributed is the way.
The point of these tools is productivity. What are you trying to accomplish and how long does it take to accomplish? This includes time spent writing code and fussing with configs. This would take <1min to run and <10 to write on a cluster. Happy to might make a demo to prove it.
Yes cost matters also, but running many machines for a short period of time is the same as one for a long time? Open to honest rebuttal.
Or basically a generic nestable `remote_parallel_map` for python functions over lists of objects.
I haven't had a chance to fully watch the video yet / I understand it focuses on lower levels of abstraction / GPU programming. But I'd love to know how this fit's into what the speaker is looking for / what it's missing (other than obviously it not being a way to program GPU's) (also full disclosure I am a co-founder).
With the right scheduler I think we could get that above 90%. Would love to hear any feedback / thoughts. Here's a blog explaining: https://docs.burla.dev/blog/dynamic-hardware