fixed some build errors on linux, changed icon resolution, added more error printing

.gitignore

@@ -34,7 +34,6 @@ models/*
 /vdot
 /Pipfile
 
-build-info.h
 arm_neon.h
 compile_commands.json
 

Makefile

@@ -266,7 +266,7 @@ gpttype_adapter_clblast.o: gpttype_adapter.cpp
 	$(CXX) $(CXXFLAGS) $(CLBLAST_FLAGS) -c $< -o $@
 
 clean:
-	rm -vf *.o main quantize_llama quantize_gpt2 quantize_gptj quantize_neox quantize-stats perplexity embedding benchmark-matmult save-load-state build-info.h main.exe quantize_llama.exe quantize_gptj.exe quantize_gpt2.exe quantize_neox.exe koboldcpp.dll koboldcpp_openblas.dll koboldcpp_noavx2.dll koboldcpp_openblas_noavx2.dll koboldcpp_clblast.dll koboldcpp.so koboldcpp_openblas.so koboldcpp_noavx2.so koboldcpp_openblas_noavx2.so koboldcpp_clblast.so gptj.exe gpt2.exe
+	rm -vf *.o main quantize_llama quantize_gpt2 quantize_gptj quantize_neox quantize-stats perplexity embedding benchmark-matmult save-load-state main.exe quantize_llama.exe quantize_gptj.exe quantize_gpt2.exe quantize_neox.exe koboldcpp.dll koboldcpp_openblas.dll koboldcpp_noavx2.dll koboldcpp_openblas_noavx2.dll koboldcpp_clblast.dll koboldcpp.so koboldcpp_openblas.so koboldcpp_noavx2.so koboldcpp_openblas_noavx2.so koboldcpp_clblast.so gptj.exe gpt2.exe
 
 #
 # Examples
@@ -319,13 +319,8 @@ libllama.so: llama.o ggml.o $(OBJS)
 save-load-state: examples/save-load-state/save-load-state.cpp build-info.h ggml.o llama.o common.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
 
-build-info.h: $(wildcard .git/index) scripts/build-info.sh
+build-info.h: 
-	@sh scripts/build-info.sh > $@.tmp
+	$(DONOTHING)
-	@if ! cmp -s $@.tmp $@; then \
-		mv $@.tmp $@; \
-	else \
-		rm $@.tmp; \
-	fi
 
 #
 # Tests

build-info.h

@@ -0,0 +1,7 @@
+#ifndef BUILD_INFO_H
+#define BUILD_INFO_H
+
+#define BUILD_NUMBER 999
+#define BUILD_COMMIT "KOBOLDCPP"
+
+#endif // BUILD_INFO_H

ggml-opencl.cpp

@@ -557,7 +557,7 @@ static void ggml_cl_mul_mat_f32(const ggml_tensor * src0, const ggml_tensor * sr
                                             &queue, &ev_sgemm);
 
             if (status != clblast::StatusCode::kSuccess) {
- 				printf("\nF32 Matmul Failed (%d): You may be out of VRAM. Please check if you have enough.\n",status);    
+ 				printf("\nF32 Matmul Failed (%d): [dims: %lld,%lld,%lld,%lld] You may be out of VRAM. Please check if you have enough.\n",status,ne00,ne01,ne10,ne11);
                 GGML_ASSERT(false);
             }
 
@@ -656,7 +656,7 @@ static void ggml_cl_mul_mat_f16(const ggml_tensor * src0, const ggml_tensor * sr
                                             &queue, &ev_sgemm);
 
             if (status != clblast::StatusCode::kSuccess) {
-				printf("\nF16 Matmul Failed (%d): You may be out of VRAM. Please check if you have enough.\n",status);
+				printf("\nF16 Matmul Failed (%d): [dims: %lld,%lld,%lld,%lld] You may be out of VRAM. Please check if you have enough.\n",status,ne00,ne01,ne10,ne11);
            		GGML_ASSERT(false);
             }
 
@@ -764,7 +764,7 @@ static void ggml_cl_mul_mat_q_f32(const ggml_tensor * src0, const ggml_tensor *
                                             &queue, &ev_sgemm);
 
                 if (status != clblast::StatusCode::kSuccess) {
-					printf("\nQF32 Matmul Failed (%d): You may be out of VRAM. Please check if you have enough.\n",status);
+					printf("\nQF32 Matmul Failed (%d): [dims: %lld,%lld,%lld,%lld] You may be out of VRAM. Please check if you have enough.\n",status,ne00,ne01,ne10,ne11);
                     GGML_ASSERT(false);
                 }
             }

koboldcpp.py

@@ -208,7 +208,7 @@ maxctx = 2048
 maxlen = 128
 modelbusy = False
 defaultport = 5001
-KcppVersion = "1.24"
+KcppVersion = "1.25"
 
 class ServerRequestHandler(http.server.SimpleHTTPRequestHandler):
     sys_version = ""

otherarch/ggml_v2-opencl.cpp

@@ -573,7 +573,7 @@ static void ggml_v2_cl_mul_mat_f32(const ggml_v2_tensor * src0, const ggml_v2_te
                                             &queue, &ev_sgemm);
 
             if (status != clblast::StatusCode::kSuccess) {
-                printf("\nF32 Matmul Failed (%d): You may be out of VRAM. Please check if you have enough.\n",status);
+                printf("\nF32 Matmul Failed (%d): [dims: %lld,%lld,%lld,%lld] You may be out of VRAM. Please check if you have enough.\n",status,ne00,ne01,ne10,ne11);
                 GGML_V2_ASSERT(false);
             }
 
@@ -672,7 +672,7 @@ static void ggml_v2_cl_mul_mat_f16(const ggml_v2_tensor * src0, const ggml_v2_te
                                             &queue, &ev_sgemm);
 
             if (status != clblast::StatusCode::kSuccess) {
-                printf("\nF16 Matmul Failed (%d): You may be out of VRAM. Please check if you have enough.\n",status);
+                printf("\nF16 Matmul Failed (%d): [dims: %lld,%lld,%lld,%lld] You may be out of VRAM. Please check if you have enough.\n",status,ne00,ne01,ne10,ne11);
                 GGML_V2_ASSERT(false);
             }
 
@@ -780,7 +780,7 @@ static void ggml_v2_cl_mul_mat_q_f32(const ggml_v2_tensor * src0, const ggml_v2_
                                             &queue, &ev_sgemm);
 
                 if (status != clblast::StatusCode::kSuccess) {
-                    printf("\nQF32 Matmul Failed (%d): You may be out of VRAM. Please check if you have enough.\n",status);
+                    printf("\nQF32 Matmul Failed (%d): [dims: %lld,%lld,%lld,%lld] You may be out of VRAM. Please check if you have enough.\n",status,ne00,ne01,ne10,ne11);
                     GGML_V2_ASSERT(false);
                 }
             }