Oh hey, I wrote this!
Thanks for sharing it.
Thanks for sharing it.
[1] https://en.wikipedia.org/wiki/List_of_Nvidia_graphics_proces...
For example, "compute capability" sounds like it'd be what you need, but it's actually more of a software versioning index :(
Was thinking of splitting the difference by collecting up the quoted arithmetic (FLOP/s) and memory bandwidths from the manufacturer datasheets. But there's caveats there too, e.g. the dreaded "With sparsity" asterisk on the Tensor Core FLOP/s of recent generations.
Any chance you could just make it a single long webpage (as opposed to making me click through one page at a time)?
For some reason on my iPad the links don’t always work the first time I click them.
"These groups of threads, known as warps , are switched out on a per clock cycle basis — roughly one nanosecond. CPU thread context switches, on the other hand, take few hundred to a few thousand clock cycles"
I would note that intels SMT does do something very similar (2 hw threads). Other like the xeon phi would round robin 4 threads on a single core.
SMT allows for concurrent execution of both threads (thus independent front-end for fetch, decode especially) and certain core resources are statically partitioned unlike a warp being scheduled on SM.
I'm not a graphics expert but warps seem closer to run-time/dynamic VLIW than SMT.
This mapping is so close that translation from GPU to CPU relatively easy and performant.
> intels SMT does do something very similar (2 hw threads)
Yeah that's a good point. One thing I learned from looking at both hardware stacks more closely was that they aren't as different as they seem at first -- lots of the same ideas or techniques get are used, but in different ways.