Cycles: add HIP device support for AMD GPUs

NOTE: this feature is not ready for user testing, and not yet enabled in daily
builds. It is being merged now for easier collaboration on development.

HIP is a heterogenous compute interface allowing C++ code to be executed on
GPUs similar to CUDA. It is intended to bring back AMD GPU rendering support
on Windows and Linux.

https://github.com/ROCm-Developer-Tools/HIP.

As of the time of writing, it should compile and run on Linux with existing
HIP compilers and driver runtimes. Publicly available compilers and drivers
for Windows will come later.

See task T91571 for more details on the current status and work remaining
to be done.

Credits:

Sayak Biswas (AMD)
Arya Rafii (AMD)
Brian Savery (AMD)

Differential Revision: https://developer.blender.org/D12578

1 files changed, 9 insertions, 4 deletions

diff --git a/intern/cycles/blender/addon/properties.py b/intern/cycles/blender/addon/properties.py
index e2b671848d0..67207874431 100644
--- a/intern/cycles/blender/addon/properties.py
+++ b/intern/cycles/blender/addon/properties.py
@@ -111,6 +111,7 @@ enum_device_type = (
     ('CPU', "CPU", "CPU", 0),
     ('CUDA', "CUDA", "CUDA", 1),
     ('OPTIX', "OptiX", "OptiX", 3),
+    ("HIP", "HIP", "HIP", 4)
 )
 
 enum_texture_limit = (
@@ -1266,12 +1267,16 @@ class CyclesPreferences(bpy.types.AddonPreferences):
 
     def get_device_types(self, context):
         import _cycles
-        has_cuda, has_optix = _cycles.get_device_types()
+        has_cuda, has_optix, has_hip = _cycles.get_device_types()
+
         list = [('NONE', "None", "Don't use compute device", 0)]
         if has_cuda:
             list.append(('CUDA', "CUDA", "Use CUDA for GPU acceleration", 1))
         if has_optix:
             list.append(('OPTIX', "OptiX", "Use OptiX for GPU acceleration", 3))
+        if has_hip:
+            list.append(('HIP', "HIP", "Use HIP for GPU acceleration", 4))
+
         return list
 
     compute_device_type: EnumProperty(
@@ -1296,7 +1301,7 @@ class CyclesPreferences(bpy.types.AddonPreferences):
 
     def update_device_entries(self, device_list):
         for device in device_list:
-            if not device[1] in {'CUDA', 'OPTIX', 'CPU'}:
+            if not device[1] in {'CUDA', 'OPTIX', 'CPU', 'HIP'}:
                 continue
             # Try to find existing Device entry
             entry = self.find_existing_device_entry(device)
@@ -1330,7 +1335,7 @@ class CyclesPreferences(bpy.types.AddonPreferences):
             elif entry.type == 'CPU':
                 cpu_devices.append(entry)
         # Extend all GPU devices with CPU.
-        if compute_device_type != 'CPU':
+        if compute_device_type != 'CPU' and compute_device_type != 'HIP':
             devices.extend(cpu_devices)
         return devices
 
@@ -1340,7 +1345,7 @@ class CyclesPreferences(bpy.types.AddonPreferences):
         import _cycles
         # Ensure `self.devices` is not re-allocated when the second call to
         # get_devices_for_type is made, freeing items from the first list.
-        for device_type in ('CUDA', 'OPTIX', 'OPENCL'):
+        for device_type in ('CUDA', 'OPTIX', 'HIP'):
             self.update_device_entries(_cycles.available_devices(device_type))
 
     # Deprecated: use refresh_devices instead.