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
This commit is contained in:
Brian Savery
2021-09-28 16:51:14 +02:00
committed by Brecht Van Lommel
parent 262b211856
commit 044a77352f
45 changed files with 4854 additions and 19 deletions

View File

@@ -59,12 +59,23 @@ DebugFlags::CUDA::CUDA() : adaptive_compile(false)
reset();
}
DebugFlags::HIP::HIP() : adaptive_compile(false)
{
reset();
}
void DebugFlags::CUDA::reset()
{
if (getenv("CYCLES_CUDA_ADAPTIVE_COMPILE") != NULL)
adaptive_compile = true;
}
void DebugFlags::HIP::reset()
{
if (getenv("CYCLES_HIP_ADAPTIVE_COMPILE") != NULL)
adaptive_compile = true;
}
DebugFlags::OptiX::OptiX()
{
reset();
@@ -103,6 +114,10 @@ std::ostream &operator<<(std::ostream &os, DebugFlagsConstRef debug_flags)
os << "OptiX flags:\n"
<< " Debug : " << string_from_bool(debug_flags.optix.use_debug) << "\n";
os << "HIP flags:\n"
<< " HIP streams : " << string_from_bool(debug_flags.hip.adaptive_compile) << "\n";
return os;
}