rocm-systems

2016-10-04 22:17:18 +05:30

channel_descriptor.h

2017-03-31 12:11:34 -05:00

code_object_bundle.hpp

This switches HIP from its currently convoluted macro + pfe based dispatch mechanism to a more natural one partially based on the existing module API. The basic idea is that HCC will always correctly emit __global__ functions: as empty-bodied stubs, on host, and as kernels, on device. It then becomes trivial to obtain the mangled name on host, at dispatch, from the function's address, and then to use the mangled name to retrieve the kernel. This should address all problems stemming from serialisation, dubious mismatches due to the manufactured functor, macro-isms et al. It also immediately enables support for generalised globals as a consequence of that being available in the module API. Finally, it will make debug much easier, since the actual names of the __global__ functions will automatically be used in traces etc. One detail is that due to how dispatch works now (hipLaunchKernel and hipLaunchKernelGGL are themselves variadic function templates which deduce the function type of the callee), in certain cases it may be necesssary to insert explicit casts to ensure that the variadic argument list selects a viable overload - this can be observed in some unit tests. Eventually we may be able to remove this limitation, but for now it does not appear terribly onerous. The code is not extremely HIPpie, nor is it fully optimised, but rather is intended as a starting point for the HIP team to make its own.

2017-11-01 15:09:59 +00:00

concepts.hpp

GGL update, fix for thread-safe access to streams (accelerator_views).

2017-04-01 14:50:39 -05:00

device_functions.h

fixed header names

2017-03-31 12:18:55 -05:00

driver_types.h

HIP Texture Support

2017-07-17 15:16:12 -04:00

grid_launch_GGL.hpp

Remove leftover comment.

2017-11-22 19:37:03 +00:00

helpers.hpp

fix GGL helper header file, reorder for C++17

2017-05-22 08:52:43 -05:00

hip_complex.h

add math.h to cover sqrtf function

2017-04-10 08:53:12 -05:00

hip_db.h

Add hipdbPrintMem - wrapper for hcc memory tracker.

2017-07-03 15:05:30 -05:00

hip_fp16.h

Add clang version guard so the hip_fp16.h header won't be picked up by gcc

2017-06-08 19:24:32 -05:00

hip_ldg.h

fixed header names

2017-03-31 12:18:55 -05:00

hip_runtime_api.h

Update docs.

2017-08-28 15:44:38 -05:00

hip_runtime.h

Correct ill-formed merge in earlier commit and adjust for differences with the new CUDA natural indexing mechanism.

2017-11-20 16:33:52 +00:00

hip_texture_types.h

HIP Texture Support

2017-07-17 15:16:12 -04:00

hip_vector_types.h

removed bad copy constructor

2017-06-16 09:02:26 -05:00

host_defines.h

This switches HIP from its currently convoluted macro + pfe based dispatch mechanism to a more natural one partially based on the existing module API. The basic idea is that HCC will always correctly emit __global__ functions: as empty-bodied stubs, on host, and as kernels, on device. It then becomes trivial to obtain the mangled name on host, at dispatch, from the function's address, and then to use the mangled name to retrieve the kernel. This should address all problems stemming from serialisation, dubious mismatches due to the manufactured functor, macro-isms et al. It also immediately enables support for generalised globals as a consequence of that being available in the module API. Finally, it will make debug much easier, since the actual names of the __global__ functions will automatically be used in traces etc. One detail is that due to how dispatch works now (hipLaunchKernel and hipLaunchKernelGGL are themselves variadic function templates which deduce the function type of the callee), in certain cases it may be necesssary to insert explicit casts to ensure that the variadic argument list selects a viable overload - this can be observed in some unit tests. Eventually we may be able to remove this limitation, but for now it does not appear terribly onerous. The code is not extremely HIPpie, nor is it fully optimised, but rather is intended as a starting point for the HIP team to make its own.

2017-11-01 15:09:59 +00:00

math_functions.h

Fix math ordering for --genco mode.

2017-10-02 21:52:16 +00:00

program_state.hpp

Refactor the __device__ versions of memset and memcpy to be less awkward i.e. not return nullptr as opposed to the destination pointer (it can only be assumed it was done for maximum confusion) and actually unroll as they claim to. Change all of the {to, from}Symbol functions to use hipModuleGetGlobal, as opposed to hc::accelerator::get_symbol_address which is no longer valid with module based dispatch.

2017-11-21 02:40:34 +00:00

texture_functions.h

HIP Texture Support

2017-07-17 15:16:12 -04:00

texture_types.h

HIP Texture Support

2017-07-17 15:16:12 -04:00