Merge pull request karpathy#283 from ChrisDryden/configurationRewrite

Changed ordering of type configuration to easily see unchanged values

train_gpt2.cu

@@ -63,15 +63,20 @@ enum PrecisionMode {
     PRECISION_BF16
 };
 
-// fp32
+// Default Properties
+typedef float floatN;
+#define CUBLAS_LOWP_COMPUTE cublas_compute_type
+#ifdef MULTI_GPU
+const ncclDataType_t ncclFloatN = ncclFloat;
+#endif
+
+// Specific configurations based on the enabled precision
 #if defined(ENABLE_FP32)
 typedef float floatX;
 #define CUBLAS_LOWP CUDA_R_32F
-#define CUBLAS_LOWP_COMPUTE cublas_compute_type // auto-select FP32 vs TF32
-const char* load_filename = "gpt2_124M.bin"; // fp32 weights
-PrecisionMode PRECISION_MODE = PRECISION_FP32;
+#define PRECISION_MODE PRECISION_FP32
+const char* load_filename = "gpt2_124M.bin";
 const char* precision_mode_str = "fp32";
-
 #ifdef MULTI_GPU
 const ncclDataType_t ncclFloatX = ncclFloat;
 #endif
@@ -80,24 +85,19 @@ const ncclDataType_t ncclFloatX = ncclFloat;
 #elif defined(ENABLE_FP16)
 typedef half floatX;
 #define CUBLAS_LOWP CUDA_R_16F
-#define CUBLAS_LOWP_COMPUTE CUBLAS_COMPUTE_32F
-const char* load_filename = "gpt2_124M.bin"; // fp32 weights
-PrecisionMode PRECISION_MODE = PRECISION_FP16;
+#define PRECISION_MODE PRECISION_FP16
+const char* load_filename = "gpt2_124M.bin";
 const char* precision_mode_str = "fp16";
-
 #ifdef MULTI_GPU
 const ncclDataType_t ncclFloatX = ncclHalf;
 #endif
 
-// bfloat16 (default!)
-#else
+#else // Default to bfloat16
 typedef __nv_bfloat16 floatX;
 #define CUBLAS_LOWP CUDA_R_16BF
-#define CUBLAS_LOWP_COMPUTE CUBLAS_COMPUTE_32F
-const char* load_filename = "gpt2_124M_bf16.bin"; // bf16 weights
-PrecisionMode PRECISION_MODE = PRECISION_BF16;
+#define PRECISION_MODE PRECISION_BF16
+const char* load_filename = "gpt2_124M_bf16.bin"; // bf16 weights specific filename
 const char* precision_mode_str = "bf16";
-
 #ifdef MULTI_GPU
 const ncclDataType_t ncclFloatX = ncclBfloat16;
 #endif