Adding a check for copmliance between the runtime simulation config and MAX_THREAD_PER_SM. Also decreased LOCAL_MEM_SIZE_MAX to 8kB to make it fit within our allotted memory space (otherwise the simulator may mistaken global memory access as local memory accesses).

[git-p4: depot-paths = "//depot/gpgpu_sim_research/fermi/distribution/": change = 11520]

2 files changed, 6 insertions, 1 deletions

diff --git a/src/abstract_hardware_model.h b/src/abstract_hardware_model.h
index 6c2b190..2fdebfc 100644
--- a/src/abstract_hardware_model.h
+++ b/src/abstract_hardware_model.h
@@ -268,7 +268,7 @@ protected:
 #define GLOBAL_HEAP_START 0x80000000
    // start allocating from this address (lower values used for allocating globals in .ptx file)
 #define SHARED_MEM_SIZE_MAX (64*1024)
-#define LOCAL_MEM_SIZE_MAX (16*1024)
+#define LOCAL_MEM_SIZE_MAX (8*1024)
 #define MAX_STREAMING_MULTIPROCESSORS 64
 #define MAX_THREAD_PER_SM 2048
 #define TOTAL_LOCAL_MEM_PER_SM (MAX_THREAD_PER_SM*LOCAL_MEM_SIZE_MAX)
diff --git a/src/gpgpu-sim/shader.h b/src/gpgpu-sim/shader.h
index fbecac3..f8b0e58 100644
--- a/src/gpgpu-sim/shader.h
+++ b/src/gpgpu-sim/shader.h
@@ -918,6 +918,11 @@ struct shader_core_config : public core_config
            printf("GPGPU-Sim uArch: error while parsing configuration string gpgpu_shader_core_pipeline_opt\n");
            abort();
         }
+        if (n_thread_per_shader > MAX_THREAD_PER_SM) {
+           printf("GPGPU-Sim uArch: Error ** increase MAX_THREAD_PER_SM in abstract_hardware_model.h from %u to %u\n", 
+                  MAX_THREAD_PER_SM, n_thread_per_shader);
+           abort();
+        }
         max_warps_per_shader =  n_thread_per_shader/warp_size;
         assert( !(n_thread_per_shader % warp_size) );
         max_sfu_latency = 32;