| cbuffer cbuffer_u : register(b0, space0) { |
| uint4 u[32]; |
| }; |
| |
| matrix<float16_t, 3, 2> tint_symbol(uint4 buffer[32], uint offset) { |
| const uint scalar_offset = ((offset + 0u)) / 4; |
| uint ubo_load = buffer[scalar_offset / 4][scalar_offset % 4]; |
| const uint scalar_offset_1 = ((offset + 4u)) / 4; |
| uint ubo_load_1 = buffer[scalar_offset_1 / 4][scalar_offset_1 % 4]; |
| const uint scalar_offset_2 = ((offset + 8u)) / 4; |
| uint ubo_load_2 = buffer[scalar_offset_2 / 4][scalar_offset_2 % 4]; |
| return matrix<float16_t, 3, 2>(vector<float16_t, 2>(float16_t(f16tof32(ubo_load & 0xFFFF)), float16_t(f16tof32(ubo_load >> 16))), vector<float16_t, 2>(float16_t(f16tof32(ubo_load_1 & 0xFFFF)), float16_t(f16tof32(ubo_load_1 >> 16))), vector<float16_t, 2>(float16_t(f16tof32(ubo_load_2 & 0xFFFF)), float16_t(f16tof32(ubo_load_2 >> 16)))); |
| } |
| |
| [numthreads(1, 1, 1)] |
| void f() { |
| const matrix<float16_t, 2, 3> t = transpose(tint_symbol(u, 260u)); |
| uint ubo_load_3 = u[0].z; |
| const float16_t l = length(vector<float16_t, 2>(float16_t(f16tof32(ubo_load_3 & 0xFFFF)), float16_t(f16tof32(ubo_load_3 >> 16))).yx); |
| uint ubo_load_4 = u[0].z; |
| const float16_t a = abs(vector<float16_t, 2>(float16_t(f16tof32(ubo_load_4 & 0xFFFF)), float16_t(f16tof32(ubo_load_4 >> 16))).yx.x); |
| return; |
| } |