feat: show usage on unknown arg (#767)

Some terminals have slow display latency, so frequent output
during model loading can actually slow down the process.

Also, since tensor loading times can vary a lot, the progress
display now shows the average across past iterations instead
of just the last one.

.github/workflows/build.yml

@@ -163,7 +163,7 @@ jobs:
           - build: "avx512"
             defines: "-DGGML_NATIVE=OFF -DGGML_AVX512=ON -DGGML_AVX=ON -DGGML_AVX2=ON -DSD_BUILD_SHARED_LIBS=ON"
           - build: "cuda12"
-            defines: "-DSD_CUDA=ON -DSD_BUILD_SHARED_LIBS=ON -DCMAKE_CUDA_ARCHITECTURES=90;89;80;75"
+            defines: "-DSD_CUDA=ON -DSD_BUILD_SHARED_LIBS=ON -DCMAKE_CUDA_ARCHITECTURES=90;89;86;80;75"
           # - build: "rocm5.5"
           #   defines: '-G Ninja -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ -DSD_HIPBLAS=ON -DCMAKE_BUILD_TYPE=Release -DAMDGPU_TARGETS="gfx1100;gfx1102;gfx1030" -DSD_BUILD_SHARED_LIBS=ON'
           - build: 'vulkan'

README.md

@@ -129,7 +129,7 @@ This provides BLAS acceleration using the ROCm cores of your AMD GPU. Make sure
 Windows User Refer to [docs/hipBLAS_on_Windows.md](docs%2FhipBLAS_on_Windows.md) for a comprehensive guide.
 
 ```
-cmake .. -G "Ninja" -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ -DSD_HIPBLAS=ON -DCMAKE_BUILD_TYPE=Release -DAMDGPU_TARGETS=gfx1100
+cmake .. -G "Ninja" -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ -DSD_HIPBLAS=ON -DCMAKE_BUILD_TYPE=Release -DAMDGPU_TARGETS=gfx1100 -DCMAKE_BUILD_WITH_INSTALL_RPATH=ON
 cmake --build . --config Release
 ```
 

docs/chroma.md

@@ -24,7 +24,7 @@ You can download the preconverted gguf weights from [silveroxides/Chroma-GGUF](h
 For example:
 
 ```
- .\bin\Release\sd.exe -diffusion-model  ..\models\chroma-unlocked-v40-q8_0.gguf --vae ..\models\ae.sft --t5xxl ..\models\t5xxl_fp16.safetensors  -p "a lovely cat holding a sign says 'chroma.cpp'" --cfg-scale 4.0 --sampling-method euler -v --chroma-disable-dit-mask
+ .\bin\Release\sd.exe --diffusion-model  ..\models\chroma-unlocked-v40-q8_0.gguf --vae ..\models\ae.sft --t5xxl ..\models\t5xxl_fp16.safetensors  -p "a lovely cat holding a sign says 'chroma.cpp'" --cfg-scale 4.0 --sampling-method euler -v --chroma-disable-dit-mask
 ```
 
 ![](../assets/flux/chroma_v40.png)

examples/cli/main.cpp

@@ -297,10 +297,12 @@ bool parse_options(int argc, const char** argv, ArgOptions& options) {
     bool invalid_arg = false;
     std::string arg;
     for (int i = 1; i < argc; i++) {
+        bool found_arg = false;
         arg = argv[i];
 
         for (auto& option : options.string_options) {
             if ((option.short_name.size() > 0 && arg == option.short_name) || (option.long_name.size() > 0 && arg == option.long_name)) {
+                found_arg = true;
                 if (++i >= argc) {
                     invalid_arg = true;
                     break;
@@ -314,6 +316,7 @@ bool parse_options(int argc, const char** argv, ArgOptions& options) {
 
         for (auto& option : options.int_options) {
             if ((option.short_name.size() > 0 && arg == option.short_name) || (option.long_name.size() > 0 && arg == option.long_name)) {
+                found_arg = true;
                 if (++i >= argc) {
                     invalid_arg = true;
                     break;
@@ -327,6 +330,7 @@ bool parse_options(int argc, const char** argv, ArgOptions& options) {
 
         for (auto& option : options.float_options) {
             if ((option.short_name.size() > 0 && arg == option.short_name) || (option.long_name.size() > 0 && arg == option.long_name)) {
+                found_arg = true;
                 if (++i >= argc) {
                     invalid_arg = true;
                     break;
@@ -340,6 +344,7 @@ bool parse_options(int argc, const char** argv, ArgOptions& options) {
 
         for (auto& option : options.bool_options) {
             if ((option.short_name.size() > 0 && arg == option.short_name) || (option.long_name.size() > 0 && arg == option.long_name)) {
+                found_arg = true;
                 if (option.keep_true) {
                     *option.target = true;
                 } else {
@@ -353,6 +358,7 @@ bool parse_options(int argc, const char** argv, ArgOptions& options) {
 
         for (auto& option : options.manual_options) {
             if ((option.short_name.size() > 0 && arg == option.short_name) || (option.long_name.size() > 0 && arg == option.long_name)) {
+                found_arg = true;
                 int ret = option.cb(argc, argv, i);
                 if (ret < 0) {
                     invalid_arg = true;
@@ -364,6 +370,10 @@ bool parse_options(int argc, const char** argv, ArgOptions& options) {
         if (invalid_arg) {
             break;
         }
+        if (!found_arg) {
+            fprintf(stderr, "error: unknown argument: %s\n", arg.c_str());    
+            return false;
+        }
     }
     if (invalid_arg) {
         fprintf(stderr, "error: invalid parameter for argument: %s\n", arg.c_str());

model.cpp

@@ -1942,8 +1942,11 @@ bool ModelLoader::load_tensors(on_new_tensor_cb_t on_new_tensor_cb, ggml_backend
             return true;
         };
         int tensor_count = 0;
-        int64_t t1       = ggml_time_ms();
-        bool partial     = false;
+        int64_t t0       = ggml_time_ms();
+        int64_t t1       = t0;
+        bool partial     = true;
+        int tensor_max   = (int)processed_tensor_storages.size();
+        pretty_progress(0, tensor_max, 0.0f);
         for (auto& tensor_storage : processed_tensor_storages) {
             if (tensor_storage.file_index != file_index) {
                 ++tensor_count;
@@ -2046,21 +2049,29 @@ bool ModelLoader::load_tensors(on_new_tensor_cb_t on_new_tensor_cb, ggml_backend
                     ggml_backend_tensor_set(dst_tensor, convert_buffer.data(), 0, ggml_nbytes(dst_tensor));
                 }
             }
-            size_t tensor_max = processed_tensor_storages.size();
-            int64_t t2        = ggml_time_ms();
-            pretty_progress(++tensor_count, tensor_max, (t2 - t1) / 1000.0f);
-            t1      = t2;
-            partial = tensor_count != tensor_max;
+            ++tensor_count;
+            int64_t t2 = ggml_time_ms();
+            if ((t2 - t1) >= 200) {
+                t1 = t2;
+                pretty_progress(tensor_count, tensor_max, (t1 - t0) / (1000.0f * tensor_count));
+                partial = tensor_count != tensor_max;
+            }
+        }
+
+        if (partial) {
+            if (tensor_count >= 1) {
+                t1 = ggml_time_ms();
+                pretty_progress(tensor_count, tensor_max, (t1 - t0) / (1000.0f * tensor_count));
+            }
+            if (tensor_count < tensor_max) {
+                printf("\n");
+            }
         }
 
         if (zip != NULL) {
             zip_close(zip);
         }
 
-        if (partial) {
-            printf("\n");
-        }
-
         if (!success) {
             break;
         }