Re: [Mesa-dev] [PATCH] radv: enable VK_AMD_gpu_shader_half_float
a-b now. On Mon, Apr 8, 2019 at 9:31 AM Samuel Pitoiset wrote: > > > On 3/21/19 11:09 AM, Bas Nieuwenhuizen wrote: > > Honestly the zero tests is worrying me. This is a pretty big extension > > and I have questions like: > > > > to 16-bit loads + 16-bit ALU actually work together or have we been > > silently relying on the fact there is always a ZExt cast after and > > that did not care about input size? > > With the two 16-bit fixes I have just sent, it should be now safe to > enable it. > > VK_KHR_shader_float16_int8 now works too, I'm just waiting for CTS > bugfixes before enabling it by default. > > Can you Rb this one or do you still think it's too fragile? > > > > > On Thu, Mar 21, 2019 at 10:03 AM Samuel Pitoiset > > wrote: > >> Should be safe to enable as all instructions seem to support 16-bit. > >> Unfortunately, there is no CTS test. > >> > >> Signed-off-by: Samuel Pitoiset > >> --- > >> src/amd/vulkan/radv_extensions.py | 1 + > >> 1 file changed, 1 insertion(+) > >> > >> diff --git a/src/amd/vulkan/radv_extensions.py > >> b/src/amd/vulkan/radv_extensions.py > >> index 421f8b926ea..23106765c2a 100644 > >> --- a/src/amd/vulkan/radv_extensions.py > >> +++ b/src/amd/vulkan/radv_extensions.py > >> @@ -122,6 +122,7 @@ EXTENSIONS = [ > >> Extension('VK_EXT_vertex_attribute_divisor', 3, True), > >> Extension('VK_AMD_draw_indirect_count', 1, True), > >> Extension('VK_AMD_gcn_shader',1, True), > >> +Extension('VK_AMD_gpu_shader_half_float', 1, > >> 'device->rad_info.chip_class >= VI && HAVE_LLVM >= 0x0800'), > >> Extension('VK_AMD_rasterization_order', 1, > >> 'device->has_out_of_order_rast'), > >> Extension('VK_AMD_shader_core_properties',1, True), > >> Extension('VK_AMD_shader_info', 1, True), > >> -- > >> 2.21.0 > >> > >> ___ > >> mesa-dev mailing list > >> mesa-dev@lists.freedesktop.org > >> https://lists.freedesktop.org/mailman/listinfo/mesa-dev ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
Re: [Mesa-dev] [PATCH] radv: enable VK_AMD_gpu_shader_half_float
On 3/21/19 11:09 AM, Bas Nieuwenhuizen wrote: Honestly the zero tests is worrying me. This is a pretty big extension and I have questions like: to 16-bit loads + 16-bit ALU actually work together or have we been silently relying on the fact there is always a ZExt cast after and that did not care about input size? With the two 16-bit fixes I have just sent, it should be now safe to enable it. VK_KHR_shader_float16_int8 now works too, I'm just waiting for CTS bugfixes before enabling it by default. Can you Rb this one or do you still think it's too fragile? On Thu, Mar 21, 2019 at 10:03 AM Samuel Pitoiset wrote: Should be safe to enable as all instructions seem to support 16-bit. Unfortunately, there is no CTS test. Signed-off-by: Samuel Pitoiset --- src/amd/vulkan/radv_extensions.py | 1 + 1 file changed, 1 insertion(+) diff --git a/src/amd/vulkan/radv_extensions.py b/src/amd/vulkan/radv_extensions.py index 421f8b926ea..23106765c2a 100644 --- a/src/amd/vulkan/radv_extensions.py +++ b/src/amd/vulkan/radv_extensions.py @@ -122,6 +122,7 @@ EXTENSIONS = [ Extension('VK_EXT_vertex_attribute_divisor', 3, True), Extension('VK_AMD_draw_indirect_count', 1, True), Extension('VK_AMD_gcn_shader',1, True), +Extension('VK_AMD_gpu_shader_half_float', 1, 'device->rad_info.chip_class >= VI && HAVE_LLVM >= 0x0800'), Extension('VK_AMD_rasterization_order', 1, 'device->has_out_of_order_rast'), Extension('VK_AMD_shader_core_properties',1, True), Extension('VK_AMD_shader_info', 1, True), -- 2.21.0 ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
Re: [Mesa-dev] [PATCH] radv: enable VK_AMD_gpu_shader_half_float
On 3/21/19 11:09 AM, Bas Nieuwenhuizen wrote: Honestly the zero tests is worrying me. This is a pretty big extension and I have questions like: to 16-bit loads + 16-bit ALU actually work together or have we been silently relying on the fact there is always a ZExt cast after and that did not care about input size? Yeah... It's clearly untested, I think it works but it's just a guess. Maybe we should implement crucible tests? On Thu, Mar 21, 2019 at 10:03 AM Samuel Pitoiset wrote: Should be safe to enable as all instructions seem to support 16-bit. Unfortunately, there is no CTS test. Signed-off-by: Samuel Pitoiset --- src/amd/vulkan/radv_extensions.py | 1 + 1 file changed, 1 insertion(+) diff --git a/src/amd/vulkan/radv_extensions.py b/src/amd/vulkan/radv_extensions.py index 421f8b926ea..23106765c2a 100644 --- a/src/amd/vulkan/radv_extensions.py +++ b/src/amd/vulkan/radv_extensions.py @@ -122,6 +122,7 @@ EXTENSIONS = [ Extension('VK_EXT_vertex_attribute_divisor', 3, True), Extension('VK_AMD_draw_indirect_count', 1, True), Extension('VK_AMD_gcn_shader',1, True), +Extension('VK_AMD_gpu_shader_half_float', 1, 'device->rad_info.chip_class >= VI && HAVE_LLVM >= 0x0800'), Extension('VK_AMD_rasterization_order', 1, 'device->has_out_of_order_rast'), Extension('VK_AMD_shader_core_properties',1, True), Extension('VK_AMD_shader_info', 1, True), -- 2.21.0 ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
Re: [Mesa-dev] [PATCH] radv: enable VK_AMD_gpu_shader_half_float
Honestly the zero tests is worrying me. This is a pretty big extension and I have questions like: to 16-bit loads + 16-bit ALU actually work together or have we been silently relying on the fact there is always a ZExt cast after and that did not care about input size? On Thu, Mar 21, 2019 at 10:03 AM Samuel Pitoiset wrote: > > Should be safe to enable as all instructions seem to support 16-bit. > Unfortunately, there is no CTS test. > > Signed-off-by: Samuel Pitoiset > --- > src/amd/vulkan/radv_extensions.py | 1 + > 1 file changed, 1 insertion(+) > > diff --git a/src/amd/vulkan/radv_extensions.py > b/src/amd/vulkan/radv_extensions.py > index 421f8b926ea..23106765c2a 100644 > --- a/src/amd/vulkan/radv_extensions.py > +++ b/src/amd/vulkan/radv_extensions.py > @@ -122,6 +122,7 @@ EXTENSIONS = [ > Extension('VK_EXT_vertex_attribute_divisor', 3, True), > Extension('VK_AMD_draw_indirect_count', 1, True), > Extension('VK_AMD_gcn_shader',1, True), > +Extension('VK_AMD_gpu_shader_half_float', 1, > 'device->rad_info.chip_class >= VI && HAVE_LLVM >= 0x0800'), > Extension('VK_AMD_rasterization_order', 1, > 'device->has_out_of_order_rast'), > Extension('VK_AMD_shader_core_properties',1, True), > Extension('VK_AMD_shader_info', 1, True), > -- > 2.21.0 > > ___ > mesa-dev mailing list > mesa-dev@lists.freedesktop.org > https://lists.freedesktop.org/mailman/listinfo/mesa-dev ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [PATCH] radv: enable VK_AMD_gpu_shader_half_float
Should be safe to enable as all instructions seem to support 16-bit. Unfortunately, there is no CTS test. Signed-off-by: Samuel Pitoiset --- src/amd/vulkan/radv_extensions.py | 1 + 1 file changed, 1 insertion(+) diff --git a/src/amd/vulkan/radv_extensions.py b/src/amd/vulkan/radv_extensions.py index 421f8b926ea..23106765c2a 100644 --- a/src/amd/vulkan/radv_extensions.py +++ b/src/amd/vulkan/radv_extensions.py @@ -122,6 +122,7 @@ EXTENSIONS = [ Extension('VK_EXT_vertex_attribute_divisor', 3, True), Extension('VK_AMD_draw_indirect_count', 1, True), Extension('VK_AMD_gcn_shader',1, True), +Extension('VK_AMD_gpu_shader_half_float', 1, 'device->rad_info.chip_class >= VI && HAVE_LLVM >= 0x0800'), Extension('VK_AMD_rasterization_order', 1, 'device->has_out_of_order_rast'), Extension('VK_AMD_shader_core_properties',1, True), Extension('VK_AMD_shader_info', 1, True), -- 2.21.0 ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev