diff options
| -rw-r--r-- | CHANGES | 3 | ||||
| -rw-r--r-- | libopencl/opencl_runtime_api.cc | 1 | ||||
| -rw-r--r-- | src/cuda-sim/instructions.cc | 59 |
3 files changed, 42 insertions, 21 deletions
@@ -5,6 +5,9 @@ Version 3.1.1+edits (development branch) versus 3.1.1 - Fixed a compile error that happens with newer gcc/g++ versions (4.7.1) - Fixed a bug with the association between cuobjdump output and cubin handles (Bug #7 external) + - Adding support for atomic operations with generic memory space. Before + this, atomic operations can only work on global memory space. (Bug #14 + external) Version 3.1.1 versus 3.1.0 - Add checks to top level makefile to ensure setup_environment is run and checks to diff --git a/libopencl/opencl_runtime_api.cc b/libopencl/opencl_runtime_api.cc index 0b00556..8c9a0d1 100644 --- a/libopencl/opencl_runtime_api.cc +++ b/libopencl/opencl_runtime_api.cc @@ -94,6 +94,7 @@ # endif # endif +#define CL_USE_DEPRECATED_OPENCL_1_0_APIS #include <CL/cl.h> #include <map> diff --git a/src/cuda-sim/instructions.cc b/src/cuda-sim/instructions.cc index 6976ba0..a7685aa 100644 --- a/src/cuda-sim/instructions.cc +++ b/src/cuda-sim/instructions.cc @@ -867,9 +867,6 @@ void atom_callback( const inst_t* inst, ptx_thread_info* thread ) { const ptx_instruction *pI = dynamic_cast<const ptx_instruction*>(inst); - // Check state space - assert( pI->get_space()==global_space ); - // "Decode" the output type unsigned to_type = pI->get_type(); size_t size; @@ -877,10 +874,10 @@ void atom_callback( const inst_t* inst, ptx_thread_info* thread ) type_info_key::type_decode(to_type, size, t); // Set up operand variables - ptx_reg_t data, // d - src1_data, // a - src2_data, // b - op_result; // temp variable to hold operation result + ptx_reg_t data; // d + ptx_reg_t src1_data; // a + ptx_reg_t src2_data; // b + ptx_reg_t op_result; // temp variable to hold operation result bool data_ready = false; @@ -890,18 +887,28 @@ void atom_callback( const inst_t* inst, ptx_thread_info* thread ) const operand_info &src2 = pI->src2(); // b // Get operand values + src1_data = thread->get_operand_value(src1, src1, to_type, thread, 1); // a if (dst.get_symbol()->type()){ - src1_data = thread->get_operand_value(src1, dst, to_type, thread, 1); // a - src2_data = thread->get_operand_value(src2, dst, to_type, thread, 1); // b + src2_data = thread->get_operand_value(src2, dst, to_type, thread, 1); // b } else { //This is the case whent he first argument (dest) is '_' - src1_data = thread->get_operand_value(src1, src1, to_type, thread, 1); // a - src2_data = thread->get_operand_value(src2, src1, to_type, thread, 1); // b + src2_data = thread->get_operand_value(src2, src1, to_type, thread, 1); // b } + // Check state space + addr_t effective_address = src1_data.u64; + memory_space_t space = pI->get_space(); + if (space == undefined_space) { + // generic space - determine space via address + space = whichspace(effective_address); + assert( space == global_space ); + effective_address = generic_to_global(effective_address); + } + assert( space == global_space ); + // Copy value pointed to in operand 'a' into register 'd' // (i.e. copy src1_data to dst) - thread->get_global_memory()->read(src1_data.u32,size/8,&data.s64); + thread->get_global_memory()->read(effective_address,size/8,&data.s64); if (dst.get_symbol()->type()){ thread->set_operand_value(dst, data, to_type, thread, pI); // Write value into register 'd' } @@ -1144,7 +1151,7 @@ void atom_callback( const inst_t* inst, ptx_thread_info* thread ) // Write operation result into global memory // (i.e. copy src1_data to dst) - thread->get_global_memory()->write(src1_data.u32,size/8,&op_result.s64,thread,pI); + thread->get_global_memory()->write(effective_address,size/8,&op_result.s64,thread,pI); } // atom_impl will now result in a callback being called in mem_ctrl_pop (gpu-sim.c) @@ -1153,23 +1160,33 @@ void atom_impl( const ptx_instruction *pI, ptx_thread_info *thread ) // SYNTAX // atom.space.operation.type d, a, b[, c]; (now read in callback) - // Check state space - assert( pI->get_space()== global_space ); + // obtain memory space of the operation + memory_space_t space = pI->get_space(); // get the memory address const operand_info &src1 = pI->src1(); - const operand_info &dst = pI->dst(); + // const operand_info &dst = pI->dst(); // not needed for effective address calculation unsigned i_type = pI->get_type(); ptx_reg_t src1_data; - if (dst.get_symbol()->type()){ - src1_data = thread->get_operand_value(src1, dst, i_type, thread, 1); + src1_data = thread->get_operand_value(src1, src1, i_type, thread, 1); + addr_t effective_address = src1_data.u64; + + addr_t effective_address_final; + + // handle generic memory space by converting it to global + if ( space == undefined_space ) { + assert( whichspace(effective_address) == global_space ); + effective_address_final = generic_to_global(effective_address); + space = global_space; } else { - src1_data = thread->get_operand_value(src1, src1, i_type, thread, 1); + assert( space == global_space ); + effective_address_final = effective_address; } - memory_space_t space = pI->get_space(); + // Check state space + assert( space == global_space ); - thread->m_last_effective_address = src1_data.u32; + thread->m_last_effective_address = effective_address_final; thread->m_last_memory_space = space; thread->m_last_dram_callback.function = atom_callback; thread->m_last_dram_callback.instruction = pI; |