mirror of
https://github.com/gentoo-mirror/gentoo.git
synced 2026-08-06 00:48:18 -07:00
960578c12e
To prevent such issues in the future, ebuild now relies on AMDGPU_TARGETS USE expand. As a bonus, now it can be built for Strix Point/Halo Point APUs and passes tests there, even though 2.5.x has no official support for them (while Blender trunk still uses 2.5.x). Also added a small backport from upstream 3.1.0 for hip-7 compatibility (same reason: Blender still uses 2.5.x). Closes: https://bugs.gentoo.org/970995 Signed-off-by: Sv. Lockal <lockalsash@gmail.com> Part-of: https://github.com/gentoo/gentoo/pull/45973 Signed-off-by: Sam James <sam@gentoo.org>
69 lines
2.9 KiB
Diff
69 lines
2.9 KiB
Diff
Partial backport from upstream for hip-7 compatibility (no member named 'data' in 'HIP_vector_type...)
|
|
|
|
Backports https://github.com/GPUOpen-LibrariesAndSDKs/HIPRT/commit/592a92b5afcc6de16c5371b1f41daec4ac59b77b
|
|
--- a/hiprt/impl/hiprt_device_impl.h
|
|
+++ b/hiprt/impl/hiprt_device_impl.h
|
|
@@ -362,10 +362,10 @@ TraversalBase<Stack>::testInternalNode( const hiprtRay& ray, const float3& invD,
|
|
auto result = __builtin_amdgcn_image_bvh_intersect_ray_l(
|
|
encodeBaseAddr( nodes, nodeIndex ),
|
|
ray.maxT,
|
|
- float4{ ray.origin.x, ray.origin.y, ray.origin.z, 0.0f }.data,
|
|
- float4{ ray.direction.x, ray.direction.y, ray.direction.z, 0.0f }.data,
|
|
- float4{ invD.x, invD.y, invD.z, 0.0f }.data,
|
|
- m_descriptor.data );
|
|
+ { ray.origin.x, ray.origin.y, ray.origin.z, 0.0f },
|
|
+ { ray.direction.x, ray.direction.y, ray.direction.z, 0.0f },
|
|
+ { invD.x, invD.y, invD.z, 0.0f },
|
|
+ { m_descriptor.x, m_descriptor.y, m_descriptor.z, m_descriptor.w } );
|
|
#endif
|
|
if ( m_stack.vacancy() < 3 )
|
|
{
|
|
@@ -397,11 +397,13 @@ HIPRT_DEVICE bool TraversalBase<Stack>::testTriangleNode(
|
|
hit.normal = node.m_triPair.fetchTriangle( leafIndex & 1 ).normal( node.m_flags >> ( ( leafIndex & 1 ) * 8 ) );
|
|
}
|
|
#else
|
|
- const float4 origin = float4{ ray.origin.x, ray.origin.y, ray.origin.z, 0.0f };
|
|
- const float4 direction = float4{ ray.direction.x, ray.direction.y, ray.direction.z, 0.0f };
|
|
- const float4 invDir = float4{ invD.x, invD.y, invD.z, 0.0f };
|
|
- auto result = __builtin_amdgcn_image_bvh_intersect_ray_l(
|
|
- encodeBaseAddr( nodes, leafIndex ), ray.maxT, origin.data, direction.data, invDir.data, m_descriptor.data );
|
|
+ auto result = __builtin_amdgcn_image_bvh_intersect_ray_l(
|
|
+ encodeBaseAddr( nodes, leafIndex ),
|
|
+ ray.maxT,
|
|
+ { ray.origin.x, ray.origin.y, ray.origin.z, 0.0f },
|
|
+ { ray.direction.x, ray.direction.y, ray.direction.z, 0.0f },
|
|
+ { invD.x, invD.y, invD.z, 0.0f },
|
|
+ { m_descriptor.x, m_descriptor.y, m_descriptor.z, m_descriptor.w } );
|
|
float invDenom = __ocml_native_recip_f32( __int_as_float( result[1] ) );
|
|
float t = __int_as_float( result[0] ) * invDenom;
|
|
hasHit = ray.minT <= t && t <= ray.maxT;
|
|
--- a/hiprt/hiprt_common.h
|
|
+++ b/hiprt/hiprt_common.h
|
|
@@ -110,15 +110,25 @@
|
|
#endif
|
|
|
|
#if defined( __KERNELCC_RTC__ )
|
|
+#if defined( __CUDACC_RTC__ ) || HIP_VERSION_MAJOR < 7
|
|
using int8_t = char;
|
|
using uint8_t = unsigned char;
|
|
using int16_t = short;
|
|
using uint16_t = unsigned short;
|
|
-#if defined( __CUDACC_RTC__ )
|
|
+
|
|
using int32_t = int;
|
|
using uint32_t = unsigned int;
|
|
using int64_t = long long;
|
|
using uint64_t = unsigned long long;
|
|
+#else
|
|
+using int8_t = __hip_internal::int8_t;
|
|
+using uint8_t = __hip_internal::uint8_t;
|
|
+using int16_t = __hip_internal::int16_t;
|
|
+using uint16_t = __hip_internal::uint16_t;
|
|
+using int32_t = __hip_internal::int32_t;
|
|
+using uint32_t = __hip_internal::uint32_t;
|
|
+using int64_t = __hip_internal::int64_t;
|
|
+using uint64_t = __hip_internal::uint64_t;
|
|
#endif
|
|
#endif
|
|
|