[libre-riscv-dev] [isa-dev] Re: FP transcendentals (trigonometry, root/exp/log) proposal

lkcl luke.leighton at gmail.com
Wed Sep 11 22:16:00 BST 2019


On Thursday, September 12, 2019 at 3:23:28 AM UTC+8, Jacob Lifshay wrote:

> https://gpuopen.com/wp-content/uploads/2019/08/RDNA_Shader_ISA_5August2019.pdf

> looking through it again, the sin/cos instructions for Radeon are actually sin(x * 2 * pi) and cos(x * 2 * pi), reinforcing my idea to just implement sinpi/cospi for GPU applications, though there is a V_TRIG_PREOP_F64 instruction to accelerate argument reduction for f64 sin/cos.

Made a quick note of that, by changing "recommended" subsets and noting this in the insights.

Bear in mind Midgard has cos1pi and sin1pi, not cos2pi sin2pi.

L.



More information about the libre-riscv-dev mailing list