ggml: riscv: add riscv spacemit backend (#15288)

* ggml: add spacemit backend

Change-Id: I249bdc043485d815a9c351867137bc1e27cc2e23

* add new line at end of file

Change-Id: I889ed1c85fb45e62350ecde0c06f70450cadfbe2

* add riscv zba extension limit

Change-Id: I321eb200f859751727afe5cae13074dfce2bb0ce

* fixed for review comments, file renamed and format

Change-Id: Ia20b6ec24a36638e62e0fe07cf100916a7cce3ce

* fixed for code format, after clang-format

Change-Id: I5dc33a0412da3d3f2d77075d8939185d3009eca2

* use _Float16 instead of __fp16

Change-Id: I039fb02bb95270e641bc4442204e658735859d43

* add ci for riscv64-spacemit-ime-native

Change-Id: I711c1033061df1a289ea77891b2997599dfe8279

* update debian-13-riscv64-spacemit-ime-native ci label

Change-Id: Ifb2b891e2fca57b5da604fce2ac255f27731179a

* remove license comment for spacemit ime

Change-Id: If0dc3ca30a958631ccca0a28b62e0b825f9fb0c3

* upgrade binutils for gcc ime

Change-Id: Ibf2fa74c1064408974cb5b45f044d40987e5fb45

* add spacemit ime cross jobs

Change-Id: I80d74909941d41cb9cd09e51d8baf01c985cbfc6

* remove native compile for riscv64-spacemit-ime

Change-Id: I01920afafdc73fa7424014fd648d243f8ec9e25e

* ci : add caching for spacemit ime cross toolchain

Change-Id: Ic54a192019a2fd982bbd58225ce3bbc38f4053de

* ci: bug fixed for cache path and env

Change-Id: I28c42e10b6fff053bb6580926ca2353448cb042a

* Update .github/workflows/build-linux-cross.yml for cache path

Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>

* bugfixed for  build-linux-cross.yml,  syntax error

Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>

---------

Co-authored-by: cailinxi <linxi.cai@spacemit.com>
Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>
This commit is contained in:
alex-spacemit
2025-09-29 22:50:44 +08:00
committed by GitHub
parent 2ddd3f2356
commit b77e6c18e1
11 changed files with 4501 additions and 0 deletions

View File

@@ -253,3 +253,47 @@ jobs:
-DCMAKE_FIND_ROOT_PATH_MODE_INCLUDE=BOTH -DCMAKE_FIND_ROOT_PATH_MODE_INCLUDE=BOTH
cmake --build build --config Release -j $(nproc) cmake --build build --config Release -j $(nproc)
ubuntu-24-riscv64-cpu-spacemit-ime-cross:
runs-on: ubuntu-24.04
env:
SPACEMIT_IME_TOOLCHAIN_VERSION: "1.1.2"
SPACEMIT_IME_TOOLCHAIN_PATH: "spacemit-toolchain-linux-glibc-x86_64"
steps:
- uses: actions/checkout@v4
- name: Cache Toolchain
uses: actions/cache@v4
id: cache-spacemit-ime-cross-toolchain
with:
path: ./${{ env.SPACEMIT_IME_TOOLCHAIN_PATH }}
key: ${{ runner.os }}-spacemit-ime-toolchain-v${{ env.SPACEMIT_IME_TOOLCHAIN_VERSION }}
- name: Setup Toolchain
if: steps.cache-spacemit-ime-cross-toolchain.outputs.cache-hit != 'true'
run: |
wget --quiet --no-check-certificate https://archive.spacemit.com/toolchain/spacemit-toolchain-linux-glibc-x86_64-v${{ env.SPACEMIT_IME_TOOLCHAIN_VERSION }}.tar.xz -O ${{ env.SPACEMIT_IME_TOOLCHAIN_PATH }}.tar.xz
rm -rf ${{ env.SPACEMIT_IME_TOOLCHAIN_PATH }}
mkdir -p ${{ env.SPACEMIT_IME_TOOLCHAIN_PATH }}
tar xf ${{ env.SPACEMIT_IME_TOOLCHAIN_PATH }}.tar.xz -C ${{ env.SPACEMIT_IME_TOOLCHAIN_PATH }} --strip-components=1
rm -rf ${{ env.SPACEMIT_IME_TOOLCHAIN_PATH }}.tar.xz
- name: Build
run: |
export RISCV_ROOT_PATH=${PWD}/${{ env.SPACEMIT_IME_TOOLCHAIN_PATH }}
cmake -B build -DLLAMA_CURL=OFF \
-DCMAKE_BUILD_TYPE=Release \
-DGGML_OPENMP=OFF \
-DLLAMA_BUILD_EXAMPLES=ON \
-DLLAMA_BUILD_TOOLS=ON \
-DLLAMA_BUILD_TESTS=OFF \
-DGGML_CPU_RISCV64_SPACEMIT=ON \
-DGGML_RVV=ON \
-DGGML_RV_ZFH=ON \
-DGGML_RV_ZICBOP=ON \
-DRISCV64_SPACEMIT_IME_SPEC=RISCV64_SPACEMIT_IME1 \
-DCMAKE_TOOLCHAIN_FILE=${PWD}/cmake/riscv64-spacemit-linux-gnu-gcc.cmake
cmake --build build --config Release -j $(nproc)

View File

@@ -58,3 +58,63 @@ jobs:
-DCMAKE_FIND_ROOT_PATH_MODE_INCLUDE=BOTH -DCMAKE_FIND_ROOT_PATH_MODE_INCLUDE=BOTH
cmake --build build --config Release -j $(nproc) cmake --build build --config Release -j $(nproc)
# debian-13-riscv64-spacemit-ime-native: # Bianbu 2.2
# runs-on: [self-hosted, RISCV64]
# steps:
# - name: Install prerequisites
# run: |
# sudo apt-get update || true
# sudo apt-get install -y libatomic1
# - uses: actions/checkout@v4
# - name: Setup Riscv
# run: |
# sudo apt-get update || true
# sudo apt-get install -y --no-install-recommends \
# build-essential \
# gcc-14-riscv64-linux-gnu \
# g++-14-riscv64-linux-gnu \
# ccache \
# cmake
# sudo apt-get upgrade binutils -y
# - name: Setup ccache
# run: |
# mkdir -p $HOME/.ccache
# ccache -M 5G -d $HOME/.ccache
# export CCACHE_LOGFILE=/home/runneruser/ccache_debug/ccache.log
# export CCACHE_DEBUGDIR="/home/runneruser/ccache_debug"
# echo "$GITHUB_WORKSPACE"
# echo "CCACHE_LOGFILE=$CCACHE_LOGFILE" >> $GITHUB_ENV
# echo "CCACHE_DEBUGDIR=$CCACHE_DEBUGDIR" >> $GITHUB_ENV
# echo "CCACHE_BASEDIR=$GITHUB_WORKSPACE" >> $GITHUB_ENV
# echo "CCACHE_DIR=$HOME/.ccache" >> $GITHUB_ENV
# - name: Build
# run: |
# cmake -B build \
# -DLLAMA_CURL=OFF \
# -DCMAKE_BUILD_TYPE=Release \
# -DGGML_OPENMP=OFF \
# -DLLAMA_BUILD_EXAMPLES=ON \
# -DLLAMA_BUILD_TOOLS=ON \
# -DLLAMA_BUILD_TESTS=OFF \
# -DCMAKE_SYSTEM_NAME=Linux \
# -DCMAKE_SYSTEM_PROCESSOR=riscv64 \
# -DCMAKE_C_COMPILER=riscv64-linux-gnu-gcc-14 \
# -DCMAKE_CXX_COMPILER=riscv64-linux-gnu-g++-14 \
# -DCMAKE_C_COMPILER_LAUNCHER=ccache \
# -DCMAKE_CXX_COMPILER_LAUNCHER=ccache \
# -DCMAKE_POSITION_INDEPENDENT_CODE=ON \
# -DCMAKE_FIND_ROOT_PATH=/usr/lib/riscv64-linux-gnu \
# -DCMAKE_FIND_ROOT_PATH_MODE_PROGRAM=NEVER \
# -DCMAKE_FIND_ROOT_PATH_MODE_LIBRARY=ONLY \
# -DCMAKE_FIND_ROOT_PATH_MODE_INCLUDE=BOTH \
# -DGGML_RVV=ON \
# -DGGML_RV_ZFH=ON \
# -DGGML_RV_ZICBOP=ON \
# -DGGML_CPU_RISCV64_SPACEMIT=ON \
# -DRISCV64_SPACEMIT_IME_SPEC=RISCV64_SPACEMIT_IME1
# cmake --build build --config Release -j $(nproc)

View File

@@ -50,6 +50,7 @@
/ggml/src/ggml-blas/ @slaren /ggml/src/ggml-blas/ @slaren
/ggml/src/ggml-common.h @ggerganov @slaren /ggml/src/ggml-common.h @ggerganov @slaren
/ggml/src/ggml-cpu/ @ggerganov @slaren /ggml/src/ggml-cpu/ @ggerganov @slaren
/ggml/src/ggml-cpu/spacemit/ @alex-spacemit
/ggml/src/ggml-cuda/common.cuh @slaren /ggml/src/ggml-cuda/common.cuh @slaren
/ggml/src/ggml-cuda/fattn* @JohannesGaessler /ggml/src/ggml-cuda/fattn* @JohannesGaessler
/ggml/src/ggml-cuda/ggml-cuda.cu @slaren /ggml/src/ggml-cuda/ggml-cuda.cu @slaren

View File

@@ -0,0 +1,29 @@
set(CMAKE_SYSTEM_NAME Linux)
set(CMAKE_SYSTEM_PROCESSOR riscv64)
set(CMAKE_SYSTEM_VERSION 1)
if (CMAKE_HOST_SYSTEM_PROCESSOR MATCHES "^(riscv)")
message(STATUS "HOST SYSTEM ${CMAKE_HOST_SYSTEM_PROCESSOR}")
else()
set(GNU_MACHINE riscv64-unknown-linux-gnu CACHE STRING "GNU compiler triple")
if (DEFINED ENV{RISCV_ROOT_PATH})
file(TO_CMAKE_PATH $ENV{RISCV_ROOT_PATH} RISCV_ROOT_PATH)
else()
message(FATAL_ERROR "RISCV_ROOT_PATH env must be defined")
endif()
set(RISCV_ROOT_PATH ${RISCV_ROOT_PATH} CACHE STRING "root path to riscv toolchain")
set(CMAKE_C_COMPILER ${RISCV_ROOT_PATH}/bin/riscv64-unknown-linux-gnu-gcc)
set(CMAKE_CXX_COMPILER ${RISCV_ROOT_PATH}/bin/riscv64-unknown-linux-gnu-g++)
set(CMAKE_STRIP ${RISCV_ROOT_PATH}/bin/riscv64-unknown-linux-gnu-strip)
set(CMAKE_FIND_ROOT_PATH "${RISCV_ROOT_PATH}/riscv64-unknown-linux-gnu")
set(CMAKE_SYSROOT "${RISCV_ROOT_PATH}/sysroot")
endif()
set(CMAKE_FIND_ROOT_PATH_MODE_PROGRAM NEVER)
set(CMAKE_FIND_ROOT_PATH_MODE_LIBRARY ONLY)
set(CMAKE_FIND_ROOT_PATH_MODE_INCLUDE ONLY)
set(CMAKE_FIND_ROOT_PATH_MODE_PACKAGE ONLY)
set(CMAKE_C_FLAGS "-march=rv64gcv_zfh_zba_zicbop -mabi=lp64d ${CMAKE_C_FLAGS}")
set(CMAKE_CXX_FLAGS "-march=rv64gcv_zfh_zba_zicbop -mabi=lp64d ${CXX_FLAGS}")
set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -latomic")

View File

@@ -0,0 +1,89 @@
> [!IMPORTANT]
> This build documentation is specific only to RISC-V SpacemiT SOCs.
## Build llama.cpp locally (for riscv64)
1. Prepare Toolchain For RISCV
~~~
wget https://archive.spacemit.com/toolchain/spacemit-toolchain-linux-glibc-x86_64-v1.1.2.tar.xz
~~~
2. Build
Below is the build script: it requires utilizing RISC-V vector instructions for acceleration. Ensure the `GGML_CPU_RISCV64_SPACEMIT` compilation option is enabled. The currently supported optimization version is `RISCV64_SPACEMIT_IME1`, corresponding to the `RISCV64_SPACEMIT_IME_SPEC` compilation option. Compiler configurations are defined in the `riscv64-spacemit-linux-gnu-gcc.cmake` file. Please ensure you have installed the RISC-V compiler and set the environment variable via `export RISCV_ROOT_PATH={your_compiler_path}`.
```bash
cmake -B build \
-DCMAKE_BUILD_TYPE=Release \
-DGGML_CPU_RISCV64_SPACEMIT=ON \
-DLLAMA_CURL=OFF \
-DGGML_RVV=ON \
-DGGML_RV_ZFH=ON \
-DGGML_RV_ZICBOP=ON \
-DRISCV64_SPACEMIT_IME_SPEC=RISCV64_SPACEMIT_IME1 \
-DCMAKE_TOOLCHAIN_FILE=${PWD}/cmake/riscv64-spacemit-linux-gnu-gcc.cmake \
-DCMAKE_INSTALL_PREFIX=build/installed
cmake --build build --parallel $(nproc) --config Release
pushd build
make install
popd
```
## Simulation
You can use QEMU to perform emulation on non-RISC-V architectures.
1. Download QEMU
~~~
wget https://archive.spacemit.com/spacemit-ai/qemu/jdsk-qemu-v0.0.14.tar.gz
~~~
2. Run Simulation
After build your llama.cpp, you can run the executable file via QEMU for simulation, for example:
~~~
export QEMU_ROOT_PATH={your QEMU file path}
export RISCV_ROOT_PATH_IME1={your RISC-V compiler path}
${QEMU_ROOT_PATH}/bin/qemu-riscv64 -L ${RISCV_ROOT_PATH_IME1}/sysroot -cpu max,vlen=256,elen=64,vext_spec=v1.0 ${PWD}/build/bin/llama-cli -m ${PWD}/models/Qwen2.5-0.5B-Instruct-Q4_0.gguf -t 1
~~~
## Performance
#### Quantization Support For Matrix
~~~
model name : Spacemit(R) X60
isa : rv64imafdcv_zicbom_zicboz_zicntr_zicond_zicsr_zifencei_zihintpause_zihpm_zfh_zfhmin_zca_zcd_zba_zbb_zbc_zbs_zkt_zve32f_zve32x_zve64d_zve64f_zve64x_zvfh_zvfhmin_zvkt_sscofpmf_sstc_svinval_svnapot_svpbmt
mmu : sv39
uarch : spacemit,x60
mvendorid : 0x710
marchid : 0x8000000058000001
~~~
Q4_0
| Model | Size | Params | backend | threads | test | t/s |
| -----------| -------- | ------ | ------- | ------- | ---- |------|
Qwen2.5 0.5B |403.20 MiB|630.17 M| cpu | 4 | pp512|64.12 ± 0.26|
Qwen2.5 0.5B |403.20 MiB|630.17 M| cpu | 4 | tg128|10.03 ± 0.01|
Qwen2.5 1.5B |1011.16 MiB| 1.78 B | cpu | 4 | pp512|24.16 ± 0.02|
Qwen2.5 1.5B |1011.16 MiB| 1.78 B | cpu | 4 | tg128|3.83 ± 0.06|
Qwen2.5 3B | 1.86 GiB | 3.40 B | cpu | 4 | pp512|12.08 ± 0.02|
Qwen2.5 3B | 1.86 GiB | 3.40 B | cpu | 4 | tg128|2.23 ± 0.02|
Q4_1
| Model | Size | Params | backend | threads | test | t/s |
| -----------| -------- | ------ | ------- | ------- | ---- |------|
Qwen2.5 0.5B |351.50 MiB|494.03 M| cpu | 4 | pp512|62.07 ± 0.12|
Qwen2.5 0.5B |351.50 MiB|494.03 M| cpu | 4 | tg128|9.91 ± 0.01|
Qwen2.5 1.5B |964.06 MiB| 1.54 B | cpu | 4 | pp512|22.95 ± 0.25|
Qwen2.5 1.5B |964.06 MiB| 1.54 B | cpu | 4 | tg128|4.01 ± 0.15|
Qwen2.5 3B | 1.85 GiB | 3.09 B | cpu | 4 | pp512|11.55 ± 0.16|
Qwen2.5 3B | 1.85 GiB | 3.09 B | cpu | 4 | tg128|2.25 ± 0.04|
Q4_K
| Model | Size | Params | backend | threads | test | t/s |
| -----------| -------- | ------ | ------- | ------- | ---- |------|
Qwen2.5 0.5B |462.96 MiB|630.17 M| cpu | 4 | pp512|9.29 ± 0.05|
Qwen2.5 0.5B |462.96 MiB|630.17 M| cpu | 4 | tg128|5.67 ± 0.04|
Qwen2.5 1.5B | 1.04 GiB | 1.78 B | cpu | 4 | pp512|10.38 ± 0.10|
Qwen2.5 1.5B | 1.04 GiB | 1.78 B | cpu | 4 | tg128|3.17 ± 0.08|
Qwen2.5 3B | 1.95 GiB | 3.40 B | cpu | 4 | pp512|4.23 ± 0.04|
Qwen2.5 3B | 1.95 GiB | 3.40 B | cpu | 4 | tg128|1.73 ± 0.00|

View File

@@ -439,6 +439,15 @@ function(ggml_add_cpu_backend_variant_impl tag_name)
ggml-cpu/arch/riscv/quants.c ggml-cpu/arch/riscv/quants.c
ggml-cpu/arch/riscv/repack.cpp ggml-cpu/arch/riscv/repack.cpp
) )
if (GGML_CPU_RISCV64_SPACEMIT)
target_compile_definitions(${GGML_CPU_NAME} PRIVATE GGML_USE_CPU_RISCV64_SPACEMIT ${RISCV64_SPACEMIT_IME_SPEC})
list(APPEND GGML_CPU_SOURCES
ggml-cpu/spacemit/ime.cpp
ggml-cpu/spacemit/ime.h
ggml-cpu/spacemit/ime1_kernels.cpp
ggml-cpu/spacemit/ime_kernels.h
)
endif()
set(MARCH_STR "rv64gc") set(MARCH_STR "rv64gc")
if (GGML_RV_ZFH) if (GGML_RV_ZFH)
string(APPEND MARCH_STR "_zfh") string(APPEND MARCH_STR "_zfh")

View File

@@ -18,6 +18,10 @@
# include "kleidiai/kleidiai.h" # include "kleidiai/kleidiai.h"
#endif #endif
#ifdef GGML_USE_CPU_RISCV64_SPACEMIT
# include "spacemit/ime.h"
#endif
#if defined(_WIN32) #if defined(_WIN32)
# define WIN32_LEAN_AND_MEAN # define WIN32_LEAN_AND_MEAN
# ifndef NOMINMAX # ifndef NOMINMAX
@@ -45,6 +49,12 @@ std::vector<ggml_backend_buffer_type_t> & ggml_backend_cpu_get_extra_buffer_type
} }
#endif #endif
#ifdef GGML_USE_CPU_RISCV64_SPACEMIT
if (ggml_backend_cpu_riscv64_spacemit_buffer_type()) {
bufts.push_back(ggml_backend_cpu_riscv64_spacemit_buffer_type());
}
#endif
#ifdef GGML_USE_CPU_KLEIDIAI #ifdef GGML_USE_CPU_KLEIDIAI
if (ggml_backend_cpu_kleidiai_buffer_type()) { if (ggml_backend_cpu_kleidiai_buffer_type()) {
bufts.push_back(ggml_backend_cpu_kleidiai_buffer_type()); bufts.push_back(ggml_backend_cpu_kleidiai_buffer_type());

File diff suppressed because it is too large Load Diff

View File

@@ -0,0 +1,13 @@
#pragma once
#include "ggml-alloc.h"
#ifdef __cplusplus
extern "C" {
#endif
ggml_backend_buffer_type_t ggml_backend_cpu_riscv64_spacemit_buffer_type(void);
#ifdef __cplusplus
}
#endif

File diff suppressed because it is too large Load Diff

View File

@@ -0,0 +1,26 @@
#pragma once
#include <cstddef>
namespace sqnbitgemm_spacemit_ime {
namespace ime1 {
size_t gemm_kernel_i8i4(size_t blk_len,
const std::byte * quant_a_ptr,
const std::byte * quant_b_data,
const float * quant_b_scale,
const std::byte * quant_b_zp,
float * c_ptr,
size_t count_m,
size_t count_n,
size_t count_k,
size_t block_count_k,
size_t ldc,
const float * bias,
const size_t scale_stride);
void quantize_a_row_i8(size_t blk_len, const float * a_ptr, size_t count_k, std::byte * quant_a_ptr);
void quantize_a_4row_i8(size_t blk_len, const float * a_ptr, size_t count_k, std::byte * quant_a_ptr);
} // namespace ime1
} // namespace sqnbitgemm_spacemit_ime