radv: increase gfx1100/gfx1101 physical vgprs

https://reviews.llvm.org/D134522

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18825>

diff --git a/src/amd/vulkan/radv_shader.c b/src/amd/vulkan/radv_shader.c
index 0c328e4..94275b7 100644 (file)
--- a/src/amd/vulkan/radv_shader.c
+++ b/src/amd/vulkan/radv_shader.c
@@ -2801,6 +2801,7 @@ radv_get_max_waves(const struct radv_device *device, struct radv_shader *shader,
 {
    struct radeon_info *info = &device->physical_device->rad_info;
    enum amd_gfx_level gfx_level = info->gfx_level;
+   enum radeon_family family = info->family;
    uint8_t wave_size = shader->info.wave_size;
    struct ac_shader_config *conf = &shader->config;
    unsigned max_simd_waves;
@@ -2827,7 +2828,9 @@ radv_get_max_waves(const struct radv_device *device, struct radv_shader *shader,
    if (conf->num_vgprs) {
       unsigned physical_vgprs = info->num_physical_wave64_vgprs_per_simd * (64 / wave_size);
       unsigned vgprs = align(conf->num_vgprs, wave_size == 32 ? 8 : 4);
-      if (gfx_level == GFX10_3)
+      if (family == CHIP_GFX1100 || family == CHIP_GFX1101)
+         vgprs = util_align_npot(vgprs, wave_size == 32 ? 24 : 12);
+      else if (gfx_level == GFX10_3)
          vgprs = align(vgprs, wave_size == 32 ? 16 : 8);
       max_simd_waves = MIN2(max_simd_waves, physical_vgprs / vgprs);
    }

diff --git a/src/amd/vulkan/winsys/null/radv_null_winsys.c b/src/amd/vulkan/winsys/null/radv_null_winsys.c
index c7bc1f3..0a3f7e2 100644 (file)
--- a/src/amd/vulkan/winsys/null/radv_null_winsys.c
+++ b/src/amd/vulkan/winsys/null/radv_null_winsys.c
@@ -127,7 +127,12 @@ radv_null_winsys_query_info(struct radeon_winsys *rws, struct radeon_info *info)
    else
       info->num_physical_sgprs_per_simd = 512;
 
-   info->num_physical_wave64_vgprs_per_simd = info->gfx_level >= GFX10 ? 512 : 256;
+   if (info->family == CHIP_GFX1100 || info->family == CHIP_GFX1101)
+      info->num_physical_wave64_vgprs_per_simd = 768;
+   else if (info->gfx_level >= GFX10)
+      info->num_physical_wave64_vgprs_per_simd = 512;
+   else
+      info->num_physical_wave64_vgprs_per_simd = 256;
    info->num_simd_per_compute_unit = info->gfx_level >= GFX10 ? 2 : 4;
    info->lds_size_per_workgroup = info->gfx_level >= GFX10 ? 128 * 1024 : 64 * 1024;
    info->lds_encode_granularity = info->gfx_level >= GFX7 ? 128 * 4 : 64 * 4;