diff --git a/hiprt/hiprt_libpath.h b/hiprt/hiprt_libpath.h
index 58daa96..88cbcf2 100644
--- a/hiprt/hiprt_libpath.h
+++ b/hiprt/hiprt_libpath.h
@@ -34,7 +34,7 @@
 #ifdef HIPRT_PREFER_HIP_5
 const char* g_hip_paths[] = { "amdhip64.dll", "amdhip64_6.dll", NULL };
 #else
-const char* g_hip_paths[] = { "amdhip64_6.dll", "amdhip64.dll", NULL };
+const char* g_hip_paths[] = { "amdhip64_7.dll", "amdhip64_6.dll", "amdhip64.dll", NULL };
 #endif
 
 const char* g_hiprtc_paths[] = {
@@ -72,9 +72,13 @@ const char* g_hip_paths[] = {
 	"/opt/rocm/lib/libamdhip64.so",
 	"/opt/rocm/hip/lib/libamdhip64.so",
 
-	"libamdhip64.so.5",
-	"/opt/rocm/lib/libamdhip64.so.5",
-	"/opt/rocm/hip/lib/libamdhip64.so.5",
+	"libamdhip64.so.7",
+	"/opt/rocm/lib/libamdhip64.so.7",
+	"/opt/rocm/hip/lib/libamdhip64.so.7",
+
+	"libamdhip64.so.6",
+	"/opt/rocm/lib/libamdhip64.so.6",
+	"/opt/rocm/hip/lib/libamdhip64.so.6",
 	NULL };
 #endif
 
diff --git a/hiprt/impl/hiprt_device_impl.h b/hiprt/impl/hiprt_device_impl.h
index 0687234..272594d 100644
--- a/hiprt/impl/hiprt_device_impl.h
+++ b/hiprt/impl/hiprt_device_impl.h
@@ -37,7 +37,8 @@
 #include <hiprt/hiprt_device.h>

 #if __gfx1030__ || __gfx1031__ || __gfx1032__ || __gfx1033__ || __gfx1034__ || __gfx1035__ || __gfx1036__ || __gfx1100__ || \
-	__gfx1101__ || __gfx1102__ || __gfx1103__ || __gfx1150__ || __gfx1151__ || __gfx1152__ || __gfx1200__ || __gfx1201__
+	__gfx1101__ || __gfx1102__ || __gfx1103__ || __gfx1150__ || __gfx1151__ || __gfx1152__ || __gfx1153__ || \
+	__gfx1200__ || __gfx1201__
 #ifndef __USE_HWI__
 #define __USE_HWI__
 #endif
@@ -362,10 +363,10 @@ 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 +398,13 @@ 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;
diff --git a/scripts/bitcodes/common_tools.py b/scripts/bitcodes/common_tools.py
index 10604e5..bd46b5f 100644
--- a/scripts/bitcodes/common_tools.py
+++ b/scripts/bitcodes/common_tools.py
@@ -58,6 +58,7 @@ def getAMDGPUArchs(hip_sdk_version_num):

     if hip_sdk_version_num >= 63:
         gpus_archs.append('gfx1152')
+        gpus_archs.append('gfx1153')

     if hip_sdk_version_num >= 62: # Navi4 supported from 6.2
         gpus_archs.append('gfx1200')