summaryrefslogtreecommitdiff
path: root/src/cuda-sim
diff options
context:
space:
mode:
Diffstat (limited to 'src/cuda-sim')
-rw-r--r--src/cuda-sim/cuda-sim.cc5
-rw-r--r--src/cuda-sim/cuda_device_runtime.cc3
-rw-r--r--src/cuda-sim/instructions.cc5
3 files changed, 8 insertions, 5 deletions
diff --git a/src/cuda-sim/cuda-sim.cc b/src/cuda-sim/cuda-sim.cc
index ec68b5b..a51ea3e 100644
--- a/src/cuda-sim/cuda-sim.cc
+++ b/src/cuda-sim/cuda-sim.cc
@@ -1507,7 +1507,8 @@ static unsigned get_tex_datasize( const ptx_instruction *pI, ptx_thread_info *th
std::string texname = src1.name();
gpgpu_t *gpu = thread->get_gpu();
- const struct textureInfo* texInfo = gpu->get_texinfo(texname);
+ kernel_info_t& k = thread->get_kernel();
+ const struct textureInfo* texInfo = k.get_texinfo(texname);
unsigned data_size = texInfo->texel_size;
return data_size;
@@ -1918,7 +1919,7 @@ kernel_info_t *gpgpu_opencl_ptx_sim_init_grid(class function_info *entry,
struct dim3 blockDim,
gpgpu_t *gpu )
{
- kernel_info_t *result = new kernel_info_t(gridDim,blockDim,entry);
+ kernel_info_t *result = new kernel_info_t(gridDim,blockDim,entry,gpu->getNameArrayMapping(),gpu->getNameInfoMapping());
unsigned argcount=args.size();
unsigned argn=1;
for( gpgpu_ptx_sim_arg_list_t::iterator a = args.begin(); a != args.end(); a++ ) {
diff --git a/src/cuda-sim/cuda_device_runtime.cc b/src/cuda-sim/cuda_device_runtime.cc
index b399133..917e7a8 100644
--- a/src/cuda-sim/cuda_device_runtime.cc
+++ b/src/cuda-sim/cuda_device_runtime.cc
@@ -198,7 +198,8 @@ void gpgpusim_cuda_launchDeviceV2(const ptx_instruction * pI, ptx_thread_info *
memory_space *device_kernel_param_mem;
//create child kernel_info_t and index it with parameter_buffer address
- device_grid = new kernel_info_t(config.grid_dim, config.block_dim, device_kernel_entry);
+ gpgpu_t* gpu=thread->get_gpu();
+ device_grid = new kernel_info_t(config.grid_dim, config.block_dim, device_kernel_entry, gpu->getNameArrayMapping(), gpu->getNameInfoMapping());
device_grid->launch_cycle = gpu_sim_cycle + gpu_tot_sim_cycle;
kernel_info_t & parent_grid = thread->get_kernel();
DEV_RUNTIME_REPORT("child kernel launched by " << parent_grid.name() << ", cta (" <<
diff --git a/src/cuda-sim/instructions.cc b/src/cuda-sim/instructions.cc
index c85654c..d29fc8b 100644
--- a/src/cuda-sim/instructions.cc
+++ b/src/cuda-sim/instructions.cc
@@ -5260,9 +5260,10 @@ void tex_impl( const ptx_instruction *pI, ptx_thread_info *thread )
thread->get_vector_operand_values(src2, ptx_tex_regs, nelem); //ptx_reg should be 4 entry vector type...coordinates into texture
gpgpu_t *gpu = thread->get_gpu();
+ kernel_info_t &k = thread->get_kernel();
const struct textureReference* texref = gpu->get_texref(texname);
- const struct cudaArray* cuArray = gpu->get_texarray(texname);
- const struct textureInfo* texInfo = gpu->get_texinfo(texname);
+ const struct cudaArray* cuArray = k.get_texarray(texname);
+ const struct textureInfo* texInfo = k.get_texinfo(texname);
const struct textureReferenceAttr* texAttr = gpu->get_texattr(texname);
//assume always 2D f32 input