From a07c88e686fb9b65924876d472a8184f1f1849df Mon Sep 17 00:00:00 2001 From: Fernando Sahmkow Date: Tue, 22 Aug 2023 12:28:25 +0200 Subject: Query Cache: Simplify Prefix Sum compute shader --- .../host_shaders/queries_prefix_scan_sum.comp | 168 +++++++++++++-------- 1 file changed, 104 insertions(+), 64 deletions(-) (limited to 'src/video_core/host_shaders/queries_prefix_scan_sum.comp') diff --git a/src/video_core/host_shaders/queries_prefix_scan_sum.comp b/src/video_core/host_shaders/queries_prefix_scan_sum.comp index dce1279fe..8f10e248e 100644 --- a/src/video_core/host_shaders/queries_prefix_scan_sum.comp +++ b/src/video_core/host_shaders/queries_prefix_scan_sum.comp @@ -1,26 +1,24 @@ -// SPDX-FileCopyrightText: Copyright 2015 Graham Sellers, Richard Wright Jr. and Nicholas Haemel -// SPDX-License-Identifier: MIT - -// Code obtained from OpenGL SuperBible, Seventh Edition by Graham Sellers, Richard Wright Jr. and -// Nicholas Haemel. Modified to suit needs and optimize for subgroup +// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project +// SPDX-License-Identifier: GPL-3.0-or-later #version 460 core +#extension GL_KHR_shader_subgroup_basic : require +#extension GL_KHR_shader_subgroup_shuffle : require +#extension GL_KHR_shader_subgroup_shuffle_relative : require +#extension GL_KHR_shader_subgroup_arithmetic : require + #ifdef VULKAN -#extension GL_KHR_shader_subgroup_arithmetic : enable #define HAS_EXTENDED_TYPES 1 #define BEGIN_PUSH_CONSTANTS layout(push_constant) uniform PushConstants { -#define END_PUSH_CONSTANTS \ - } \ - ; +#define END_PUSH_CONSTANTS }; #define UNIFORM(n) #define BINDING_INPUT_BUFFER 0 #define BINDING_OUTPUT_IMAGE 1 #else // ^^^ Vulkan ^^^ // vvv OpenGL vvv -#extension GL_KHR_shader_subgroup_arithmetic : enable #extension GL_NV_gpu_shader5 : enable #ifdef GL_NV_gpu_shader5 #define HAS_EXTENDED_TYPES 1 @@ -43,19 +41,20 @@ END_PUSH_CONSTANTS layout(local_size_x = 32) in; layout(std430, binding = 0) readonly buffer block1 { - uvec2 input_data[gl_WorkGroupSize.x]; + uvec2 input_data[]; }; -layout(std430, binding = 1) writeonly coherent buffer block2 { - uvec2 output_data[gl_WorkGroupSize.x]; +layout(std430, binding = 1) coherent buffer block2 { + uvec2 output_data[]; }; layout(std430, binding = 2) coherent buffer block3 { uvec2 accumulated_data; }; -shared uvec2 shared_data[gl_WorkGroupSize.x * 2]; +shared uvec2 shared_data[2]; +// Simple Uint64 add that uses 2 uint variables for GPUs that don't support uint64 uvec2 AddUint64(uvec2 value_1, uvec2 value_2) { uint carry = 0; uvec2 result; @@ -64,61 +63,102 @@ uvec2 AddUint64(uvec2 value_1, uvec2 value_2) { return result; } -void main(void) { - uint id = gl_LocalInvocationID.x; - uvec2 base_value_1 = (id * 2) < max_accumulation_base ? accumulated_data : uvec2(0); - uvec2 base_value_2 = (id * 2 + 1) < max_accumulation_base ? accumulated_data : uvec2(0); - uint work_size = gl_WorkGroupSize.x; - uint rd_id; - uint wr_id; - uint mask; - uvec2 input_1 = input_data[id * 2]; - uvec2 input_2 = input_data[id * 2 + 1]; - // The number of steps is the log base 2 of the - // work group size, which should be a power of 2 - const uint steps = uint(log2(work_size)) + 1; - uint step = 0; - - // Each invocation is responsible for the content of - // two elements of the output array - shared_data[id * 2] = input_1; - shared_data[id * 2 + 1] = input_2; - // Synchronize to make sure that everyone has initialized - // their elements of shared_data[] with data loaded from - // the input arrays +// do subgroup Prefix Sum using Hillis and Steele's algorithm +uvec2 subgroupInclusiveAddUint64(uvec2 value) { + uvec2 result = value; + for (uint i = 1; i < gl_SubgroupSize; i *= 2) { + if (i <= gl_SubgroupInvocationID) { + uvec2 other = subgroupShuffleUp(result, i); // get value from subgroup_inv_id - i; + result = AddUint64(result, other); + } + } + return result; +} + +// Writes down the results to the output buffer and to the accumulation buffer +void WriteResults(uvec2 result) { + uint current_global_id = gl_GlobalInvocationID.x; + uvec2 base_data = current_global_id < max_accumulation_base ? accumulated_data : uvec2(0); + output_data[current_global_id] = result + base_data; + if (max_accumulation_base >= accumulation_limit + 1) { + if (current_global_id == accumulation_limit) { + accumulated_data = result; + } + return; + } + // We have that ugly case in which the accumulation data is reset in the middle somewhere. barrier(); - memoryBarrierShared(); - // For each step... - for (step = 0; step < steps; step++) { - // Calculate the read and write index in the - // shared array - mask = (1 << step) - 1; - rd_id = ((id >> step) << (step + 1)) + mask; - wr_id = rd_id + 1 + (id & mask); - // Accumulate the read data into our element - - shared_data[wr_id] = AddUint64(shared_data[rd_id], shared_data[wr_id]); - // Synchronize again to make sure that everyone - // has caught up with us - barrier(); - memoryBarrierShared(); + groupMemoryBarrier(); + if (current_global_id == accumulation_limit) { + uvec2 value_1 = output_data[max_accumulation_base]; + accumulated_data = AddUint64(result, -value_1); } - // Add the accumulation - shared_data[id * 2] = AddUint64(shared_data[id * 2], base_value_1); - shared_data[id * 2 + 1] = AddUint64(shared_data[id * 2 + 1], base_value_2); +} + +void main() { + uint subgroup_inv_id = gl_SubgroupInvocationID; + uint subgroup_id = gl_SubgroupID; + uint last_subgroup_id = subgroupMax(subgroup_inv_id); + uint current_global_id = gl_GlobalInvocationID.x; + uint total_work = gl_NumWorkGroups.x * gl_WorkGroupSize.x; + uvec2 data = input_data[current_global_id]; + // make sure all input data has been loaded + subgroupBarrier(); + subgroupMemoryBarrier(); + + uvec2 result = subgroupInclusiveAddUint64(data); + + // if we had less queries than our subgroup, just write down the results. + if (total_work <= gl_SubgroupSize) { // This condition is constant per dispatch. + WriteResults(result); + return; + } + + // We now have more, so lets write the last result into shared memory. + // Only pick the last subgroup. + if (subgroup_inv_id == last_subgroup_id) { + shared_data[subgroup_id] = result; + } + // wait until everyone loaded their stuffs barrier(); memoryBarrierShared(); - // Finally write our data back to the output buffer - output_data[id * 2] = shared_data[id * 2]; - output_data[id * 2 + 1] = shared_data[id * 2 + 1]; - if (id == 0) { - if (max_accumulation_base >= accumulation_limit + 1) { - accumulated_data = shared_data[accumulation_limit]; - return; + // Case 1: the total work for the grouped results can be calculated in a single subgroup + // operation (about 1024 queries). + uint total_extra_work = gl_NumSubgroups * gl_NumWorkGroups.x; + if (total_extra_work <= gl_SubgroupSize) { // This condition is constant per dispatch. + if (subgroup_id != 0) { + uvec2 tmp = shared_data[subgroup_inv_id]; + subgroupBarrier(); + subgroupMemoryBarrierShared(); + tmp = subgroupInclusiveAddUint64(tmp); + result = AddUint64(result, subgroupShuffle(tmp, subgroup_id - 1)); + } + + WriteResults(result); + return; + } + + // Case 2: our work amount is huge, so lets do it in O(log n) steps. + const uint extra = (total_extra_work ^ (total_extra_work - 1)) != 0 ? 1 : 0; + const uint steps = 1 << (findMSB(total_extra_work) + extra); + uint step; + // Hillis and Steele's algorithm + for (step = 1; step < steps; step *= 2) { + if (current_global_id < steps && current_global_id >= step) { + uvec2 current = shared_data[current_global_id]; + uvec2 other = shared_data[current_global_id - step]; + shared_data[current_global_id] = AddUint64(current, other); } - uvec2 value_1 = shared_data[max_accumulation_base]; - uvec2 value_2 = shared_data[accumulation_limit]; - accumulated_data = AddUint64(value_1, -value_2); + // steps is constant, so this will always execute in ever workgroup's thread. + barrier(); + memoryBarrierShared(); + } + // Only add results for groups higher than 0 + if (subgroup_id != 0) { + result = AddUint64(result, shared_data[subgroup_id - 1]); } + + // Just write the final results. We are done + WriteResults(result); } \ No newline at end of file -- cgit v1.2.3