It's good enough to run whatever local model you want. 2x 80core GPU is no joke. Linking them together gives it effectively 1.6 TB/s of bandwidth. 1TB of total memory.
You can run the full Deepseek 671b q8 model at 40 tokens/s. Q4 model at 80 tokens/s. 37B active params at a time because R1 is MoE.
Linking 2 of these together let's you run a model more capable (R1) than GPT4o at a comfortable speed at home. That was simply fantasy a year ago.
> with a vanishingly small fraction of flops and a small fraction of memory bandwidth
Is it though?
Wikipedia says [1] an M3 Max can do 14 TFLOPS of FP32, so an M3 Ultra ought to do 28 TFLOPS. nVidia claims [2] a Blackwell GPU does 80 TFLOPs of FP32. So M3 Ultra is 1/3 the speed of a Blackwell.
Calling that "a vanishingly small fraction" seems like a bit of an exaggeration.
I mean, by that metric, a single Blackwell GPU only has "a vanishingly small fraction" of the memory of an M3 Ultra. And the M3 Ultra is only burning "a vanishingly small fraction" of a Blackwell's electrical power.
nVidia likes throwing around numbers like "20 petaFLOPs" for FP4, but that's not real floating point... it's just 1990's-vintage uLaw/aLaw integer math.
Edit: Further, most (all?) of the TFLOPs numbers you see on nVidia datasheets for "Tensor FLOPs" have a little asterisk next to them saying they are "effective" TFLOPs using the sparsity feature, where half the elements of the matrix multiplication are zeroed.
