Merge 'origin/master' into hipblas

AAbushady · May 20, 2023 · c66115b · c66115b
2 parents a0b2d5f + b8ee340
commit c66115b
Show file tree

Hide file tree

Showing 25 changed files with 1,277 additions and 518 deletions.
diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
@@ -165,7 +165,7 @@ jobs:
           - build: 'clblast'
             defines: '-DLLAMA_CLBLAST=ON -DCMAKE_PREFIX_PATH="$env:RUNNER_TEMP/clblast"'
           - build: 'openblas'
-            defines: '-DLLAMA_OPENBLAS=ON -DBLAS_LIBRARIES="/LIBPATH:$env:RUNNER_TEMP/openblas/lib" -DOPENBLAS_INC="$env:RUNNER_TEMP/openblas/include"'
+            defines: '-DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS -DBLAS_INCLUDE_DIRS="$env:RUNNER_TEMP/openblas/include"'
 
     steps:
       - name: Clone

diff --git a/BLIS.md b/BLIS.md
@@ -0,0 +1,67 @@
+BLIS Installation Manual
+------------------------
+
+BLIS is a portable software framework for high-performance BLAS-like dense linear algebra libraries. It has received awards and recognition, including the 2023 James H. Wilkinson Prize for Numerical Software and the 2020 SIAM Activity Group on Supercomputing Best Paper Prize. BLIS provides a new BLAS-like API and a compatibility layer for traditional BLAS routine calls. It offers features such as object-based API, typed API, BLAS and CBLAS compatibility layers.
+
+Project URL: https://github.com/flame/blis
+
+### Prepare:
+
+Compile BLIS:
+
+```bash
+git clone https://github.com/flame/blis
+cd blis
+./configure --enable-cblas -t openmp,pthreads auto
+# will install to /usr/local/ by default.
+make -j
+```
+
+Install BLIS:
+
+```bash
+sudo make install
+```
+
+We recommend using openmp since it's easier to modify the cores been used.
+
+### llama.cpp compilation
+
+Makefile:
+
+```bash
+make LLAMA_BLIS=1 -j
+# make LLAMA_BLIS=1 benchmark-matmult
+```
+
+CMake:
+
+```bash
+mkdir build
+cd build
+cmake -DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=FLAME ..
+make -j
+```
+
+### llama.cpp execution
+
+According to the BLIS documentation, we could set the following
+environment variables to modify the behavior of openmp:
+
+```
+export GOMP_GPU_AFFINITY="0-19"
+export BLIS_NUM_THREADS=14
+```
+
+And then run the binaries as normal.
+
+
+### Intel specific issue
+
+Some might get the error message saying that `libimf.so` cannot be found.
+Please follow this [stackoverflow page](https://stackoverflow.com/questions/70687930/intel-oneapi-2022-libimf-so-no-such-file-or-directory-during-openmpi-compila).
+
+### Reference:
+
+1. https://github.com/flame/blis#getting-started
+2. https://github.com/flame/blis/blob/master/docs/Multithreading.md
diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -65,7 +65,8 @@ endif()
 
 # 3rd party libs
 option(LLAMA_ACCELERATE             "llama: enable Accelerate framework"                    ON)
-option(LLAMA_OPENBLAS               "llama: use OpenBLAS"                                   OFF)
+option(LLAMA_BLAS                   "llama: use BLAS"                                       OFF)
+option(LLAMA_BLAS_VENDOR            "llama: BLA_VENDOR from https://cmake.org/cmake/help/latest/module/FindBLAS.html#blas-lapack-vendors" Generic)
 option(LLAMA_CUBLAS                 "llama: use cuBLAS"                                     OFF)
 option(LLAMA_CLBLAST                "llama: use CLBlast"                                    OFF)
 option(LLAMA_HIPBLAS                "llama: use hipBLAS"                                    OFF)
@@ -146,36 +147,28 @@ if (APPLE AND LLAMA_ACCELERATE)
     endif()
 endif()
 
-if (LLAMA_OPENBLAS)
+if (LLAMA_BLAS)
     if (LLAMA_STATIC)
         set(BLA_STATIC ON)
     endif()
-
-    set(BLA_VENDOR OpenBLAS)
+    if ($(CMAKE_VERSION) VERSION_GREATER_EQUAL 3.22)
+        set(BLA_SIZEOF_INTEGER 8)
+    endif()
+    set(BLA_VENDOR ${LLAMA_BLAS_VENDOR})
     find_package(BLAS)
     if (BLAS_FOUND)
-        message(STATUS "OpenBLAS found")
+        message(STATUS "BLAS found, Libraries: ${BLAS_LIBRARIES}")
 
+        add_compile_options(${BLAS_LINKER_FLAGS})
         add_compile_definitions(GGML_USE_OPENBLAS)
-        add_link_options(${BLAS_LIBRARIES})
-        set(LLAMA_EXTRA_LIBS ${LLAMA_EXTRA_LIBS} openblas)
-
-        # find header file
-        set(OPENBLAS_INCLUDE_SEARCH_PATHS
-            /usr/include
-            /usr/include/openblas
-            /usr/include/openblas-base
-            /usr/local/include
-            /usr/local/include/openblas
-            /usr/local/include/openblas-base
-            /opt/OpenBLAS/include
-            $ENV{OpenBLAS_HOME}
-            $ENV{OpenBLAS_HOME}/include
-            )
-        find_path(OPENBLAS_INC NAMES cblas.h PATHS ${OPENBLAS_INCLUDE_SEARCH_PATHS})
-        add_compile_options(-I${OPENBLAS_INC})
+        set(LLAMA_EXTRA_LIBS ${LLAMA_EXTRA_LIBS} ${BLAS_LIBRARIES})
+
+        message("${BLAS_LIBRARIES} ${BLAS_INCLUDE_DIRS}")
+        include_directories(${BLAS_INCLUDE_DIRS})
     else()
-        message(WARNING "OpenBLAS not found")
+        message(WARNING "BLAS not found, please refer to "
+        "https://cmake.org/cmake/help/latest/module/FindBLAS.html#blas-lapack-vendors"
+        " to set correct LLAMA_BLAS_VENDOR")
     endif()
 endif()
 

diff --git a/Makefile b/Makefile
@@ -122,6 +122,10 @@ ifdef LLAMA_OPENBLAS
 		LDFLAGS += -lopenblas
 	endif
 endif
+ifdef LLAMA_BLIS
+	CFLAGS += -DGGML_USE_OPENBLAS -I/usr/local/include/blis -I/usr/include/blis
+	LDFLAGS += -lblis -L/usr/local/lib
+endif
 ifdef LLAMA_CUBLAS
 	CFLAGS    += -DGGML_USE_CUBLAS -I/usr/local/cuda/include -I/opt/cuda/include -I$(CUDA_PATH)/targets/x86_64-linux/include
 	CXXFLAGS  += -DGGML_USE_CUBLAS -I/usr/local/cuda/include -I/opt/cuda/include -I$(CUDA_PATH)/targets/x86_64-linux/include

diff --git a/README.md b/README.md
@@ -9,6 +9,7 @@ Inference of [LLaMA](https://arxiv.org/abs/2302.13971) model in pure C/C++
 
 **Hot topics:**
 
+- Quantization formats `Q4` and `Q8` have changed again (19 May) - [(info)](https://github.com/ggerganov/llama.cpp/pull/1508)
 - Quantization formats `Q4` and `Q5` have changed - requantize any old models [(info)](https://github.com/ggerganov/llama.cpp/pull/1405)
 - [Roadmap May 2023](https://github.com/ggerganov/llama.cpp/discussions/1220)
 
@@ -55,7 +56,7 @@ The main goal of `llama.cpp` is to run the LLaMA model using 4-bit integer quant
 - Mixed F16 / F32 precision
 - 4-bit, 5-bit and 8-bit integer quantization support
 - Runs on the CPU
-- OpenBLAS support
+- Supports OpenBLAS/Apple BLAS/ARM Performance Lib/ATLAS/BLIS/Intel MKL/NVHPC/ACML/SCSL/SGIMATH and [more](https://cmake.org/cmake/help/latest/module/FindBLAS.html#blas-lapack-vendors) in BLAS
 - cuBLAS and CLBlast support
 
 The original implementation of `llama.cpp` was [hacked in an evening](https://github.com/ggerganov/llama.cpp/issues/33#issuecomment-1465108022).
@@ -80,6 +81,7 @@ as the main playground for developing new features for the [ggml](https://github
 - [X] [Koala](https://bair.berkeley.edu/blog/2023/04/03/koala/)
 - [X] [OpenBuddy 🐶 (Multilingual)](https://github.com/OpenBuddy/OpenBuddy)
 - [X] [Pygmalion 7B / Metharme 7B](#using-pygmalion-7b--metharme-7b)
+- [X] [WizardLM](https://github.com/nlpxucan/WizardLM)
 
 **Bindings:**
 
@@ -272,10 +274,25 @@ Building the program with BLAS support may lead to some performance improvements
       ```bash
       mkdir build
       cd build
-      cmake .. -DLLAMA_OPENBLAS=ON
+      cmake .. -DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS
       cmake --build . --config Release
       ```
 
+- BLIS
+
+  Check [BLIS.md](BLIS.md) for more information.
+
+- Intel MKL
+
+  By default, `LLAMA_BLAS_VENDOR` is set to `Generic`, so if you already sourced intel environment script and assign `-DLLAMA_BLAS=ON` in cmake, the mkl version of Blas will automatically been selected. You may also specify it by:
+
+  ```bash
+  mkdir build
+  cd build
+  cmake .. -DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=Intel10_64lp -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx
+  cmake --build . -config Release
+  ```
+
 - cuBLAS
 
   This provides BLAS acceleration using the CUDA cores of your Nvidia GPU. Make sure to have the CUDA toolkit installed. You can download it from your Linux distro's package manager or from here: [CUDA Toolkit](https://developer.nvidia.com/cuda-downloads).
@@ -333,16 +350,16 @@ Several quantization methods are supported. They differ in the resulting model d
 
 | Model | Measure      | F16    | Q4_0   | Q4_1   | Q5_0   | Q5_1   | Q8_0   |
 |------:|--------------|-------:|-------:|-------:|-------:|-------:|-------:|
-|    7B | perplexity   | 5.9066 | 6.1565 | 6.0910 | 5.9862 | 5.9481 | 5.9069 |
-|    7B | file size    |  13.0G |   4.0G |   4.8G |   4.4G |   4.8G |   7.1G |
-|    7B | ms/tok @ 4th |    128 |     50 |     54 |     75 |     83 |     75 |
-|    7B | ms/tok @ 8th |    123 |     44 |     52 |     53 |     58 |     72 |
-|    7B | bits/weight  |   16.0 |    5.0 |    6.0 |    5.5 |    6.0 |    9.0 |
-|   13B | perplexity   | 5.2543 | 5.3860 | 5.3607 | 5.2856 | 5.2706 | 5.2548 |
-|   13B | file size    |  25.0G |   7.6G |   9.1G |   8.4G |   9.1G |    14G |
-|   13B | ms/tok @ 4th |    239 |     93 |    101 |    150 |    164 |    141 |
-|   13B | ms/tok @ 8th |    240 |     81 |     96 |     96 |    104 |    136 |
-|   13B | bits/weight  |   16.0 |    5.0 |    6.0 |    5.5 |    6.0 |    9.0 |
+|    7B | perplexity   | 5.9066 | 6.1565 | 6.0912 | 5.9862 | 5.9481 | 5.9070 |
+|    7B | file size    |  13.0G |   3.5G |   3.9G |   4.3G |   4.7G |   6.7G |
+|    7B | ms/tok @ 4th |    127 |     55 |     54 |     76 |     83 |     72 |
+|    7B | ms/tok @ 8th |    122 |     43 |     45 |     52 |     56 |     67 |
+|    7B | bits/weight  |   16.0 |    4.5 |    5.0 |    5.5 |    6.0 |    8.5 |
+|   13B | perplexity   | 5.2543 | 5.3860 | 5.3608 | 5.2856 | 5.2706 | 5.2548 |
+|   13B | file size    |  25.0G |   6.8G |   7.6G |   8.3G |   9.1G |    13G |
+|   13B | ms/tok @ 4th |      - |    103 |    105 |    148 |    160 |    131 |
+|   13B | ms/tok @ 8th |      - |     73 |     82 |     98 |    105 |    128 |
+|   13B | bits/weight  |   16.0 |    4.5 |    5.0 |    5.5 |    6.0 |    8.5 |
 
 ### Perplexity (measuring model quality)
 

diff --git a/convert.py b/convert.py
@@ -121,7 +121,6 @@ def make_tensors_list() -> List[str]:
             f'layers.{i}.feed_forward.w1.weight',
             f'layers.{i}.feed_forward.w2.weight',
             f'layers.{i}.feed_forward.w3.weight',
-            f'layers.{i}.atttention_norm.weight',
             f'layers.{i}.ffn_norm.weight',
         ]
     return ret
@@ -1055,7 +1054,7 @@ def load_some_model(path: Path) -> ModelPlus:
         files = list(path.glob("model-00001-of-*.safetensors"))
         if not files:
             # Try the PyTorch patterns too, with lower priority
-            globs = ["consolidated.00.pth", "pytorch_model-00001-of-*.bin", "*.pt"]
+            globs = ["consolidated.00.pth", "pytorch_model-00001-of-*.bin", "*.pt", "pytorch_model.bin" ]
             files = [file for glob in globs for file in path.glob(glob)]
         if not files:
             # Try GGML too, but with lower priority, since if both a non-GGML

diff --git a/examples/benchmark/benchmark-matmult.cpp b/examples/benchmark/benchmark-matmult.cpp
@@ -1,6 +1,7 @@
-#include <locale.h>
 #include "ggml.h"
 #include "build-info.h"
+
+#include <locale.h>
 #include <assert.h>
 #include <math.h>
 #include <cstring>
@@ -211,6 +212,7 @@ int main(int argc, char ** argv)  {
     printf("Iteration;NThreads; SizeX; SizeY; SizeZ; Required_FLOPS; Elapsed_u_Seconds; gigaFLOPS\n");
     printf("=====================================================================================\n");
 
+    double  gflops_sum = 0;
     for (int i=0;i<benchmark_params.n_iterations ;i++) {
 
         long long int start = ggml_time_us();
@@ -219,6 +221,7 @@ int main(int argc, char ** argv)  {
         long long int stop = ggml_time_us();
         long long int usec = stop-start;
         double gflops = (double)(flops_per_matrix)/usec/1000.0;
+        gflops_sum += gflops;
         printf("%9i;%8i;%6i;%6i;%6i;%15lli;%18lli;%10.2f\n",
             i,
             gf31.n_threads,
@@ -248,4 +251,7 @@ int main(int argc, char ** argv)  {
         // Running a different graph computation to make sure we override the CPU cache lines
         ggml_graph_compute(ctx, &gf32);
     }
+    printf("\n");
+    printf("Average%78.2f\n",gflops_sum/((double)benchmark_params.n_iterations));
+    printf("=====================================================================================\n");
 }