idk ask the poor cuDNN devs lmao
You can also wait till Graphcore shrinks their vaporware for real many dumb xtors@mm^2 goodness.
New TCs are very green, and very mean.
I doubt it hits 400W workload power in generic non-GEMM FP32/64/you name it.
Yeah the client Volta3 is a...
Yeah, the latter is more of a Qualcomm-tier happening, and we all know what happened to QC GPU lead.
Like 3% server and some 4-5% client over 3 years is about as unimpressive as it gets.
Good margin expansion and debt control, yeah, but still....
Also, AMD's CPU division has proven that it is possible to topple goliath will a well timed shot.
Idk, AMD isn't making any money while Intel is...
Depends on what you want from it.
Current AMD is a very tightly ran ship yeah.
But it's somewhere Q3 launch so soon(TM).
Eh GEMM cores now go brrrrrrr even in FP64 proper + it is a solid bandwidth uptick.
Arcturus is yeah, a far meaner GPGPU actual part, but it has...
Yeah A100 is a funky GEMM machine, but nothing snazzy besides that.
Which works for intended market, but may or may not piss the wide and varied...
ODI is very not soon(TM).
Also Co-EMIB is not ODI.
Yeah because nothing works without lotsa DTCO these days.
~Q3 or so.
Depends on how it goes now lmao, with both customers and supply chains being lowkey on fire.
Either console silicon or N21.
Only some SRAM macros iirc.
That's not even remotely relevant to their design methodologies but still.
Yeah RDNA2 is the first time they've been applying their super-sleek CPU design methodologies to GPUs.