From f92994bb8373d7931ed88d7db202fc0265d5ef84 Mon Sep 17 00:00:00 2001 From: Rhys Perry Date: Wed, 29 Mar 2023 14:59:00 +0100 Subject: [PATCH] radv/gfx11: improve RT scratch allocation Signed-off-by: Rhys Perry Reviewed-by: Friedrich Vock Reviewed-by: Samuel Pitoiset Part-of: --- src/amd/vulkan/radv_cmd_buffer.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/src/amd/vulkan/radv_cmd_buffer.c b/src/amd/vulkan/radv_cmd_buffer.c index a511e4f4c28..1ea8e4800a8 100644 --- a/src/amd/vulkan/radv_cmd_buffer.c +++ b/src/amd/vulkan/radv_cmd_buffer.c @@ -9849,8 +9849,11 @@ radv_trace_rays(struct radv_cmd_buffer *cmd_buffer, const VkTraceRaysIndirectCom uint32_t scratch_bytes_per_wave = pipeline->base.scratch_bytes_per_wave; uint32_t wave_size = rt_prolog->info.wave_size; - /* The hardware register is specified as a multiple of 256 DWORDS. */ - scratch_bytes_per_wave += align(cmd_buffer->state.rt_stack_size * wave_size, 1024); + /* The hardware register is specified as a multiple of 64 or 256 DWORDS. */ + unsigned scratch_alloc_granule = + cmd_buffer->device->physical_device->rad_info.gfx_level >= GFX11 ? 256 : 1024; + scratch_bytes_per_wave += + align(cmd_buffer->state.rt_stack_size * wave_size, scratch_alloc_granule); cmd_buffer->compute_scratch_size_per_wave_needed = MAX2(cmd_buffer->compute_scratch_size_per_wave_needed, scratch_bytes_per_wave);