By: Matthias Waldhauer (M.Waldhauer.delete@this.youknowwhattodo.gmx.de), November 2, 2015 11:39 am
Room: Moderated Discussions
juanrga (nospam.delete@this.juanrga.com) on October 30, 2015 1:00 pm wrote:
> The fact AMD Zen provides 256-bit AVX compatibility via fusing 2x 128bit units
> shows that they consider 256-bit software is relevant enough to be supported.
Even Jaguar does this. So this likely is not a significant power efficiency consideration. AVX (128b/256b) allows to use 3 operands, which might help creating more efficient code.
>
> The real reason why AMD Zen is stuck with 128bit units can be cache/memory bottlenecks or Skybridge
> (pin and inside compatibility of both K12 and Zen) or simply saving die space or .
128b has less to do with cache/mem bottlenecks (why?), but more with power consumption. It's L1 cache, buffered by L2, buffered by L3.
> The fact AMD Zen provides 256-bit AVX compatibility via fusing 2x 128bit units
> shows that they consider 256-bit software is relevant enough to be supported.
Even Jaguar does this. So this likely is not a significant power efficiency consideration. AVX (128b/256b) allows to use 3 operands, which might help creating more efficient code.
>
> The real reason why AMD Zen is stuck with 128bit units can be cache/memory bottlenecks or Skybridge
> (pin and inside compatibility of both K12 and Zen) or simply saving die space or .
128b has less to do with cache/mem bottlenecks (why?), but more with power consumption. It's L1 cache, buffered by L2, buffered by L3.
