diff --git a/Makefile b/Makefile
index cb683db..1c31a0d 100644
--- a/Makefile
+++ b/Makefile
@@ -159,19 +159,19 @@ test_gpt2: test_gpt2.c
 	$(CC) $(CFLAGS) $(INCLUDES) $(LDFLAGS) $< $(LDLIBS) $(OUTPUT_FILE)
 
 train_gpt2cu: train_gpt2.cu
-	$(NVCC) $(NVCC_FLAGS) $< $(NVCC_LDFLAGS) $(NVCC_INCLUDES) $(NVCC_LDLIBS) $(NVCC_LDFLAGS) $(CUDA_OUTPUT_FILE)
+	$(NVCC) $(NVCC_FLAGS) $(PFLAGS) $< $(NVCC_LDFLAGS) $(NVCC_INCLUDES) $(NVCC_LDLIBS) $(NVCC_LDFLAGS) $(CUDA_OUTPUT_FILE)
 
 train_gpt2fp32cu: train_gpt2_fp32.cu
 	$(NVCC) $(NVCC_FLAGS) $< $(NVCC_LDFLAGS) $(NVCC_INCLUDES) $(NVCC_LDLIBS) $(NVCC_LDFLAGS) $(CUDA_OUTPUT_FILE)
 
 test_gpt2cu: test_gpt2.cu
-	$(NVCC) $(NVCC_FLAGS) $< $(NVCC_LDFLAGS) $(NVCC_INCLUDES) $(NVCC_LDLIBS) $(NVCC_LDFLAGS) $(CUDA_OUTPUT_FILE)
+	$(NVCC) $(NVCC_FLAGS) $(PFLAGS) $< $(NVCC_LDFLAGS) $(NVCC_INCLUDES) $(NVCC_LDLIBS) $(NVCC_LDFLAGS) $(CUDA_OUTPUT_FILE)
 
 test_gpt2fp32cu: test_gpt2_fp32.cu
 	$(NVCC) $(NVCC_FLAGS) $< $(NVCC_LDFLAGS) $(NVCC_INCLUDES) $(NVCC_LDLIBS) $(NVCC_LDFLAGS) $(CUDA_OUTPUT_FILE)
 
 profile_gpt2cu: profile_gpt2.cu
-	$(NVCC) $(NVCC_FLAGS) -lineinfo $< $(NVCC_LDFLAGS) $(CUDA_OUTPUT_FILE)
+	$(NVCC) $(NVCC_FLAGS) $(PFLAGS) -lineinfo $< $(NVCC_LDFLAGS) $(CUDA_OUTPUT_FILE)
 
 clean:
 	$(REMOVE_FILES) $(TARGETS)