ggml : fix loongson compile warnings #7537

ggerganov · 2024-05-25T17:24:40Z

Adding a ggml-101-loongson node to ggml-ci - still some compile warnings left to fix

ggerganov · 2024-05-25T17:28:30Z

@junchao-loongson I've configured the loongson node to run ggml-ci. Currently, there are a few build errors:

https://github.com/ggml-org/ci/blob/results/llama.cpp/80/787c2a26c54998fff5f621e5aa7ae9866d0bfd/ggml-101-loongson/stdall

In file included from /home/loongson/work/llama.cpp/ggml-impl.h:464,
                 from /home/loongson/work/llama.cpp/ggml-quants.c:5:
/home/loongson/work/llama.cpp/ggml-quants.c: In function ‘ggml_vec_dot_q3_K_q8_K’:
/home/loongson/work/llama.cpp/ggml-quants.c:6841:90: error: invalid argument to built-in function
             const __m256i q3h_0 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvandn_v(hbits, __lasx_xvslli_h(mone, bit)), bit), 2);
                                                                                          ^~~~~~~~~~~~~~~
/home/loongson/work/llama.cpp/ggml-quants.c:6841:35: error: invalid argument to built-in function
             const __m256i q3h_0 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvandn_v(hbits, __lasx_xvslli_h(mone, bit)), bit), 2);
                                   ^~~~~~~~~~~~~~~
/home/loongson/work/llama.cpp/ggml-quants.c:6845:90: error: invalid argument to built-in function
             const __m256i q3h_1 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvandn_v(hbits, __lasx_xvslli_h(mone, bit)), bit), 2);
                                                                                          ^~~~~~~~~~~~~~~
/home/loongson/work/llama.cpp/ggml-quants.c:6845:35: error: invalid argument to built-in function
             const __m256i q3h_1 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvandn_v(hbits, __lasx_xvslli_h(mone, bit)), bit), 2);
                                   ^~~~~~~~~~~~~~~
/home/loongson/work/llama.cpp/ggml-quants.c:6849:90: error: invalid argument to built-in function
             const __m256i q3h_2 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvandn_v(hbits, __lasx_xvslli_h(mone, bit)), bit), 2);
                                                                                          ^~~~~~~~~~~~~~~
/home/loongson/work/llama.cpp/ggml-quants.c:6849:35: error: invalid argument to built-in function
             const __m256i q3h_2 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvandn_v(hbits, __lasx_xvslli_h(mone, bit)), bit), 2);
                                   ^~~~~~~~~~~~~~~
/home/loongson/work/llama.cpp/ggml-quants.c:6853:90: error: invalid argument to built-in function
             const __m256i q3h_3 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvandn_v(hbits, __lasx_xvslli_h(mone, bit)), bit), 2);
                                                                                          ^~~~~~~~~~~~~~~
/home/loongson/work/llama.cpp/ggml-quants.c:6853:35: error: invalid argument to built-in function
             const __m256i q3h_3 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvandn_v(hbits, __lasx_xvslli_h(mone, bit)), bit), 2);
                                   ^~~~~~~~~~~~~~~
/home/loongson/work/llama.cpp/ggml-quants.c: In function ‘ggml_vec_dot_q5_K_q8_K’:
/home/loongson/work/llama.cpp/ggml-quants.c:8051:35: error: invalid argument to built-in function
             const __m256i q5h_0 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvand_v(hbits, hmask), bit++), 4);
                                   ^~~~~~~~~~~~~~~
/home/loongson/work/llama.cpp/ggml-quants.c:8056:35: error: invalid argument to built-in function
             const __m256i q5h_1 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvand_v(hbits, hmask), bit++), 4);
                                   ^~~~~~~~~~~~~~~
make[2]: *** [CMakeFiles/ggml.dir/build.make:118：CMakeFiles/ggml.dir/ggml-quants.c.o] 错误 1
make[2]: *** 正在等待未完成的任务....
make[1]: *** [CMakeFiles/Makefile2:820：CMakeFiles/ggml.dir/all] 错误 2
make: *** [Makefile:146：all] 错误 2

Could you suggest a fix for these?

github-actions · 2024-05-25T17:53:40Z

📈 llama.cpp server for bench-server-baseline on Standard_NC4as_T4_v3 for phi-2-q4_0: 551 iterations 🚀

Expand details for performance related PR only

Concurrent users: 8, duration: 10m
HTTP request : avg=8488.66ms p(95)=19849.83ms fails=, finish reason: stop=503 truncated=48
Prompt processing (pp): avg=101.93tk/s p(95)=447.04tk/s
Token generation (tg): avg=33.84tk/s p(95)=46.25tk/s
ggml-org/models/phi-2/ggml-model-q4_0.gguf parallel=8 ctx-size=16384 ngl=33 batch-size=2048 ubatch-size=256 pp=1024 pp+tg=2048 branch=gg/ci-loongson commit=fd5de67bb7835011e3caa21ea21cc49657fb1198

More

---
config:
    xyChart:
        titleFontSize: 12
        width: 900
        height: 600
    themeVariables:
        xyChart:
            titleColor: "#000000"
---
xychart-beta
    title "llama.cpp bench-server-baseline on Standard_NC4as_T4_v3
 duration=10m 551 iterations"
    y-axis "llamacpp:prompt_tokens_seconds"
    x-axis "llamacpp:prompt_tokens_seconds" 1717069285 --> 1717069909
    line [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 648.52, 648.52, 648.52, 648.52, 648.52, 780.49, 780.49, 780.49, 780.49, 780.49, 787.89, 787.89, 787.89, 787.89, 787.89, 863.15, 863.15, 863.15, 863.15, 863.15, 868.0, 868.0, 868.0, 868.0, 868.0, 859.27, 859.27, 859.27, 859.27, 859.27, 854.47, 854.47, 854.47, 854.47, 854.47, 862.91, 862.91, 862.91, 862.91, 862.91, 874.94, 874.94, 874.94, 874.94, 874.94, 871.18, 871.18, 871.18, 871.18, 871.18, 892.56, 892.56, 892.56, 892.56, 892.56, 860.72, 860.72, 860.72, 860.72, 860.72, 857.66, 857.66, 857.66, 857.66, 857.66, 852.48, 852.48, 852.48, 852.48, 852.48, 854.81, 854.81, 854.81, 854.81, 854.81, 834.27, 834.27, 834.27, 834.27, 834.27, 834.68, 834.68, 834.68, 834.68, 834.68, 840.89, 840.89, 840.89, 840.89, 840.89, 846.97, 846.97, 846.97, 846.97, 846.97, 856.99, 856.99, 856.99, 856.99, 856.99, 855.72, 855.72, 855.72, 855.72, 855.72, 857.81, 857.81, 857.81, 857.81, 857.81, 868.3, 868.3, 868.3, 868.3, 868.3, 871.01, 871.01, 871.01, 871.01, 871.01, 872.83, 872.83, 872.83, 872.83, 872.83, 874.45, 874.45, 874.45, 874.45, 874.45, 860.3, 860.3, 860.3, 860.3, 860.3, 855.34, 855.34, 855.34, 855.34, 855.34, 857.02, 857.02, 857.02, 857.02, 857.02, 859.58, 859.58, 859.58, 859.58, 859.58, 858.8, 858.8, 858.8, 858.8, 858.8, 856.51, 856.51, 856.51, 856.51, 856.51, 846.74, 846.74, 846.74, 846.74, 846.74, 858.36, 858.36, 858.36, 858.36, 858.36, 858.28, 858.28, 858.28, 858.28, 858.28, 863.7, 863.7, 863.7, 863.7, 863.7, 861.4, 861.4, 861.4, 861.4, 861.4, 860.98, 860.98, 860.98, 860.98, 860.98, 862.57, 862.57, 862.57, 862.57, 862.57, 864.57, 864.57, 864.57, 864.57, 864.57, 864.52, 864.52, 864.52, 864.52, 864.52, 841.87, 841.87, 841.87, 841.87, 841.87, 833.7, 833.7, 833.7, 833.7, 833.7, 833.11, 833.11, 833.11, 833.11, 833.11, 831.92, 831.92, 831.92, 831.92, 831.92, 834.67, 834.67, 834.67, 834.67, 834.67, 834.42, 834.42, 834.42, 834.42, 834.42, 834.09, 834.09, 834.09, 834.09, 834.09, 839.92, 839.92, 839.92, 839.92, 839.92, 840.2, 840.2, 840.2, 840.2, 840.2, 846.75, 846.75, 846.75, 846.75, 846.75, 845.52, 845.52, 845.52, 845.52, 845.52, 845.72, 845.72, 845.72, 845.72, 845.72, 846.47, 846.47, 846.47, 846.47, 846.47, 846.89, 846.89, 846.89, 846.89, 846.89, 846.54, 846.54, 846.54, 846.54, 846.54, 847.54, 847.54, 847.54, 847.54, 847.54, 846.65, 846.65, 846.65, 846.65, 846.65, 847.55, 847.55, 847.55, 847.55, 847.55, 848.85, 848.85, 848.85, 848.85, 848.85, 849.0, 849.0, 849.0]

More

---
config:
    xyChart:
        titleFontSize: 12
        width: 900
        height: 600
    themeVariables:
        xyChart:
            titleColor: "#000000"
---
xychart-beta
    title "llama.cpp bench-server-baseline on Standard_NC4as_T4_v3
 duration=10m 551 iterations"
    y-axis "llamacpp:predicted_tokens_seconds"
    x-axis "llamacpp:predicted_tokens_seconds" 1717069285 --> 1717069909
    line [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 41.62, 41.62, 41.62, 41.62, 41.62, 34.24, 34.24, 34.24, 34.24, 34.24, 29.03, 29.03, 29.03, 29.03, 29.03, 31.37, 31.37, 31.37, 31.37, 31.37, 31.61, 31.61, 31.61, 31.61, 31.61, 32.35, 32.35, 32.35, 32.35, 32.35, 33.02, 33.02, 33.02, 33.02, 33.02, 33.55, 33.55, 33.55, 33.55, 33.55, 33.97, 33.97, 33.97, 33.97, 33.97, 34.13, 34.13, 34.13, 34.13, 34.13, 34.58, 34.58, 34.58, 34.58, 34.58, 34.12, 34.12, 34.12, 34.12, 34.12, 33.83, 33.83, 33.83, 33.83, 33.83, 33.31, 33.31, 33.31, 33.31, 33.31, 30.45, 30.45, 30.45, 30.45, 30.45, 30.34, 30.34, 30.34, 30.34, 30.34, 30.29, 30.29, 30.29, 30.29, 30.29, 30.53, 30.53, 30.53, 30.53, 30.53, 30.45, 30.45, 30.45, 30.45, 30.45, 30.51, 30.51, 30.51, 30.51, 30.51, 30.53, 30.53, 30.53, 30.53, 30.53, 30.55, 30.55, 30.55, 30.55, 30.55, 30.57, 30.57, 30.57, 30.57, 30.57, 30.34, 30.34, 30.34, 30.34, 30.34, 30.44, 30.44, 30.44, 30.44, 30.44, 30.63, 30.63, 30.63, 30.63, 30.63, 30.71, 30.71, 30.71, 30.71, 30.71, 30.89, 30.89, 30.89, 30.89, 30.89, 31.23, 31.23, 31.23, 31.23, 31.23, 31.29, 31.29, 31.29, 31.29, 31.29, 31.35, 31.35, 31.35, 31.35, 31.35, 31.37, 31.37, 31.37, 31.37, 31.37, 31.45, 31.45, 31.45, 31.45, 31.45, 31.45, 31.45, 31.45, 31.45, 31.45, 31.3, 31.3, 31.3, 31.3, 31.3, 31.04, 31.04, 31.04, 31.04, 31.04, 30.78, 30.78, 30.78, 30.78, 30.78, 30.64, 30.64, 30.64, 30.64, 30.64, 30.81, 30.81, 30.81, 30.81, 30.81, 30.97, 30.97, 30.97, 30.97, 30.97, 30.99, 30.99, 30.99, 30.99, 30.99, 31.1, 31.1, 31.1, 31.1, 31.1, 30.71, 30.71, 30.71, 30.71, 30.71, 30.24, 30.24, 30.24, 30.24, 30.24, 30.2, 30.2, 30.2, 30.2, 30.2, 29.88, 29.88, 29.88, 29.88, 29.88, 29.83, 29.83, 29.83, 29.83, 29.83, 29.82, 29.82, 29.82, 29.82, 29.82, 29.84, 29.84, 29.84, 29.84, 29.84, 29.96, 29.96, 29.96, 29.96, 29.96, 30.03, 30.03, 30.03, 30.03, 30.03, 29.96, 29.96, 29.96, 29.96, 29.96, 29.94, 29.94, 29.94, 29.94, 29.94, 29.86, 29.86, 29.86, 29.86, 29.86, 29.86, 29.86, 29.86, 29.86, 29.86, 30.0, 30.0, 30.0, 30.0, 30.0, 30.07, 30.07, 30.07, 30.07, 30.07, 30.13, 30.13, 30.13, 30.13, 30.13, 30.19, 30.19, 30.19, 30.19, 30.19, 30.29, 30.29, 30.29, 30.29, 30.29, 30.37, 30.37, 30.37]

Details

More

---
config:
    xyChart:
        titleFontSize: 12
        width: 900
        height: 600
    themeVariables:
        xyChart:
            titleColor: "#000000"
---
xychart-beta
    title "llama.cpp bench-server-baseline on Standard_NC4as_T4_v3
 duration=10m 551 iterations"
    y-axis "llamacpp:kv_cache_usage_ratio"
    x-axis "llamacpp:kv_cache_usage_ratio" 1717069285 --> 1717069909
    line [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.19, 0.19, 0.19, 0.19, 0.19, 0.37, 0.37, 0.37, 0.37, 0.37, 0.15, 0.15, 0.15, 0.15, 0.15, 0.17, 0.17, 0.17, 0.17, 0.17, 0.23, 0.23, 0.23, 0.23, 0.23, 0.24, 0.24, 0.24, 0.24, 0.24, 0.16, 0.16, 0.16, 0.16, 0.16, 0.16, 0.16, 0.16, 0.16, 0.16, 0.17, 0.17, 0.17, 0.17, 0.17, 0.11, 0.11, 0.11, 0.11, 0.11, 0.1, 0.1, 0.1, 0.1, 0.1, 0.33, 0.33, 0.33, 0.33, 0.33, 0.22, 0.22, 0.22, 0.22, 0.22, 0.4, 0.4, 0.4, 0.4, 0.4, 0.31, 0.31, 0.31, 0.31, 0.31, 0.23, 0.23, 0.23, 0.23, 0.23, 0.19, 0.19, 0.19, 0.19, 0.19, 0.15, 0.15, 0.15, 0.15, 0.15, 0.17, 0.17, 0.17, 0.17, 0.17, 0.19, 0.19, 0.19, 0.19, 0.19, 0.22, 0.22, 0.22, 0.22, 0.22, 0.25, 0.25, 0.25, 0.25, 0.25, 0.25, 0.25, 0.25, 0.25, 0.25, 0.11, 0.11, 0.11, 0.11, 0.11, 0.1, 0.1, 0.1, 0.1, 0.1, 0.19, 0.19, 0.19, 0.19, 0.19, 0.18, 0.18, 0.18, 0.18, 0.18, 0.12, 0.12, 0.12, 0.12, 0.12, 0.16, 0.16, 0.16, 0.16, 0.16, 0.15, 0.15, 0.15, 0.15, 0.15, 0.19, 0.19, 0.19, 0.19, 0.19, 0.17, 0.17, 0.17, 0.17, 0.17, 0.14, 0.14, 0.14, 0.14, 0.14, 0.24, 0.24, 0.24, 0.24, 0.24, 0.3, 0.3, 0.3, 0.3, 0.3, 0.44, 0.44, 0.44, 0.44, 0.44, 0.2, 0.2, 0.2, 0.2, 0.2, 0.12, 0.12, 0.12, 0.12, 0.12, 0.14, 0.14, 0.14, 0.14, 0.14, 0.17, 0.17, 0.17, 0.17, 0.17, 0.2, 0.2, 0.2, 0.2, 0.2, 0.35, 0.35, 0.35, 0.35, 0.35, 0.54, 0.54, 0.54, 0.54, 0.54, 0.33, 0.33, 0.33, 0.33, 0.33, 0.33, 0.33, 0.33, 0.33, 0.33, 0.19, 0.19, 0.19, 0.19, 0.19, 0.26, 0.26, 0.26, 0.26, 0.26, 0.14, 0.14, 0.14, 0.14, 0.14, 0.17, 0.17, 0.17, 0.17, 0.17, 0.17, 0.17, 0.17, 0.17, 0.17, 0.31, 0.31, 0.31, 0.31, 0.31, 0.19, 0.19, 0.19, 0.19, 0.19, 0.22, 0.22, 0.22, 0.22, 0.22, 0.14, 0.14, 0.14, 0.14, 0.14, 0.09, 0.09, 0.09, 0.09, 0.09, 0.16, 0.16, 0.16, 0.16, 0.16, 0.14, 0.14, 0.14, 0.14, 0.14, 0.13, 0.13, 0.13, 0.13, 0.13, 0.16, 0.16, 0.16, 0.16, 0.16, 0.19, 0.19, 0.19, 0.19, 0.19, 0.18, 0.18, 0.18]

More

---
config:
    xyChart:
        titleFontSize: 12
        width: 900
        height: 600
    themeVariables:
        xyChart:
            titleColor: "#000000"
---
xychart-beta
    title "llama.cpp bench-server-baseline on Standard_NC4as_T4_v3
 duration=10m 551 iterations"
    y-axis "llamacpp:requests_processing"
    x-axis "llamacpp:requests_processing" 1717069285 --> 1717069909
    line [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 4.0, 4.0, 4.0, 4.0, 4.0, 7.0, 7.0, 7.0, 7.0, 7.0, 1.0, 1.0, 1.0, 1.0, 1.0, 3.0, 3.0, 3.0, 3.0, 3.0, 5.0, 5.0, 5.0, 5.0, 5.0, 4.0, 4.0, 4.0, 4.0, 4.0, 4.0, 4.0, 4.0, 4.0, 4.0, 4.0, 4.0, 4.0, 4.0, 4.0, 6.0, 6.0, 6.0, 6.0, 6.0, 5.0, 5.0, 5.0, 5.0, 5.0, 1.0, 1.0, 1.0, 1.0, 1.0, 4.0, 4.0, 4.0, 4.0, 4.0, 8.0, 8.0, 8.0, 8.0, 8.0, 8.0, 8.0, 8.0, 8.0, 8.0, 3.0, 3.0, 3.0, 3.0, 3.0, 6.0, 6.0, 6.0, 6.0, 6.0, 4.0, 4.0, 4.0, 4.0, 4.0, 7.0, 7.0, 7.0, 7.0, 7.0, 1.0, 1.0, 1.0, 1.0, 1.0, 7.0, 7.0, 7.0, 7.0, 7.0, 6.0, 6.0, 6.0, 6.0, 6.0, 8.0, 8.0, 8.0, 8.0, 8.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 4.0, 4.0, 4.0, 4.0, 4.0, 8.0, 8.0, 8.0, 8.0, 8.0, 3.0, 3.0, 3.0, 3.0, 3.0, 3.0, 3.0, 3.0, 3.0, 3.0, 7.0, 7.0, 7.0, 7.0, 7.0, 4.0, 4.0, 4.0, 4.0, 4.0, 5.0, 5.0, 5.0, 5.0, 5.0, 3.0, 3.0, 3.0, 3.0, 3.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 4.0, 4.0, 4.0, 4.0, 4.0, 5.0, 5.0, 5.0, 5.0, 5.0, 6.0, 6.0, 6.0, 6.0, 6.0, 5.0, 5.0, 5.0, 5.0, 5.0, 7.0, 7.0, 7.0, 7.0, 7.0, 8.0, 8.0, 8.0, 8.0, 8.0, 5.0, 5.0, 5.0, 5.0, 5.0, 7.0, 7.0, 7.0, 7.0, 7.0, 2.0, 2.0, 2.0, 2.0, 2.0, 4.0, 4.0, 4.0, 4.0, 4.0, 2.0, 2.0, 2.0, 2.0, 2.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 8.0, 8.0, 8.0, 8.0, 8.0, 6.0, 6.0, 6.0, 6.0, 6.0, 4.0, 4.0, 4.0, 4.0, 4.0, 5.0, 5.0, 5.0, 5.0, 5.0, 8.0, 8.0, 8.0, 8.0, 8.0, 8.0, 8.0, 8.0, 8.0, 8.0, 6.0, 6.0, 6.0, 6.0, 6.0, 2.0, 2.0, 2.0]

junchao-loongson · 2024-05-29T12:09:34Z

Hello. The second argument of the __lasx_xvsrli_h instruction needs to be an immediate number, and this code relies on the compiler's -O2 optimization. I am going to specify the function here for O2 optimization. Thank you very much for your work on building the loongarch architecture ci!

…

-----原始邮件----- 发件人:"Georgi Gerganov" ***@***.***> 发送时间:2024-05-26 01:28:51 (星期日) 收件人: "ggerganov/llama.cpp" ***@***.***> 抄送: junchao-loongson ***@***.***>, Mention ***@***.***> 主题: Re: [ggerganov/llama.cpp] ggml : fix loongson compile warnings (PR #7537) @junchao-loongson I've configured the loongson node to run ggml-ci. Currently, there are a few build errors: https://github.com/ggml-org/ci/blob/results/llama.cpp/80/787c2a26c54998fff5f621e5aa7ae9866d0bfd/ggml-101-loongson/stdall In file included from /home/loongson/work/llama.cpp/ggml-impl.h:464, from /home/loongson/work/llama.cpp/ggml-quants.c:5: /home/loongson/work/llama.cpp/ggml-quants.c: In function ‘ggml_vec_dot_q3_K_q8_K’: /home/loongson/work/llama.cpp/ggml-quants.c:6841:90: error: invalid argument to built-in function const __m256i q3h_0 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvandn_v(hbits, __lasx_xvslli_h(mone, bit)), bit), 2); ^~~~~~~~~~~~~~~ /home/loongson/work/llama.cpp/ggml-quants.c:6841:35: error: invalid argument to built-in function const __m256i q3h_0 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvandn_v(hbits, __lasx_xvslli_h(mone, bit)), bit), 2); ^~~~~~~~~~~~~~~ /home/loongson/work/llama.cpp/ggml-quants.c:6845:90: error: invalid argument to built-in function const __m256i q3h_1 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvandn_v(hbits, __lasx_xvslli_h(mone, bit)), bit), 2); ^~~~~~~~~~~~~~~ /home/loongson/work/llama.cpp/ggml-quants.c:6845:35: error: invalid argument to built-in function const __m256i q3h_1 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvandn_v(hbits, __lasx_xvslli_h(mone, bit)), bit), 2); ^~~~~~~~~~~~~~~ /home/loongson/work/llama.cpp/ggml-quants.c:6849:90: error: invalid argument to built-in function const __m256i q3h_2 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvandn_v(hbits, __lasx_xvslli_h(mone, bit)), bit), 2); ^~~~~~~~~~~~~~~ /home/loongson/work/llama.cpp/ggml-quants.c:6849:35: error: invalid argument to built-in function const __m256i q3h_2 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvandn_v(hbits, __lasx_xvslli_h(mone, bit)), bit), 2); ^~~~~~~~~~~~~~~ /home/loongson/work/llama.cpp/ggml-quants.c:6853:90: error: invalid argument to built-in function const __m256i q3h_3 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvandn_v(hbits, __lasx_xvslli_h(mone, bit)), bit), 2); ^~~~~~~~~~~~~~~ /home/loongson/work/llama.cpp/ggml-quants.c:6853:35: error: invalid argument to built-in function const __m256i q3h_3 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvandn_v(hbits, __lasx_xvslli_h(mone, bit)), bit), 2); ^~~~~~~~~~~~~~~ /home/loongson/work/llama.cpp/ggml-quants.c: In function ‘ggml_vec_dot_q5_K_q8_K’: /home/loongson/work/llama.cpp/ggml-quants.c:8051:35: error: invalid argument to built-in function const __m256i q5h_0 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvand_v(hbits, hmask), bit++), 4); ^~~~~~~~~~~~~~~ /home/loongson/work/llama.cpp/ggml-quants.c:8056:35: error: invalid argument to built-in function const __m256i q5h_1 = __lasx_xvslli_h(__lasx_xvsrli_h(__lasx_xvand_v(hbits, hmask), bit++), 4); ^~~~~~~~~~~~~~~ make[2]: *** [CMakeFiles/ggml.dir/build.make:118：CMakeFiles/ggml.dir/ggml-quants.c.o] 错误 1 make[2]: *** 正在等待未完成的任务.... make[1]: *** [CMakeFiles/Makefile2:820：CMakeFiles/ggml.dir/all] 错误 2 make: *** [Makefile:146：all] 错误 2 Could you suggest a fix for these? — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: ***@***.***>

junchao-loongson · 2024-05-30T09:29:26Z

Hello~ @ggerganov
I have fixed this issue in PR #7636.
You can extract code for handling bit variables from PR #7636 and add it to the #7537

Using cmake DCMAKE-BUILD-TYPE=Debug Compilation passed

ggml-ci

ggerganov · 2024-05-30T10:01:17Z

@junchao-loongson Thanks. There are some connection issues from this node - I just sent you an e-mail. PTAL

junchao-loongson · 2024-05-31T01:25:01Z

In the first PR, I made a mistake when doing the rebase code that caused the unit test to fail
Fix in #7646

Fix unexpected error introduced during rebase code.

ggml-ci

github-actions bot added the ggml changes relating to the ggml tensor library for machine learning label May 25, 2024

mofosyne added the Review Complexity : Medium Generally require more time to grok but manageable by beginner to medium expertise level label May 27, 2024

junchao-loongson mentioned this pull request May 30, 2024

Fix loongarch code dependency on O2 compilation options in ggml-quants #7636

Merged

ggml : fix loongson compile warnings

fd5de67

ggml-ci

ggerganov force-pushed the gg/ci-loongson branch from 80787c2 to fd5de67 Compare May 30, 2024 09:31

Fix loongarch quantize test fail.

50fb3d3

Fix unexpected error introduced during rebase code.

github-actions bot added the testing Everything test related label May 31, 2024

ggerganov marked this pull request as ready for review May 31, 2024 11:16

tests : disable json test due to lack of python on the CI node

77c16ee

ggml-ci

ggerganov force-pushed the gg/ci-loongson branch from 98f4c12 to 77c16ee Compare May 31, 2024 11:17

ggerganov merged commit 0c27e6f into master May 31, 2024
7 of 15 checks passed

ggerganov mentioned this pull request May 31, 2024

Fix loongarch quantize test fail. #7646

Closed

HougeLangley mentioned this pull request Jun 15, 2024

Add LoongArch64 ISA Support ollama/ollama#5067

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ggml : fix loongson compile warnings #7537

ggml : fix loongson compile warnings #7537

ggerganov commented May 25, 2024 •

edited

Loading

ggerganov commented May 25, 2024

github-actions bot commented May 25, 2024 •

edited

Loading

junchao-loongson commented May 29, 2024 via email

junchao-loongson commented May 30, 2024

ggerganov commented May 30, 2024

junchao-loongson commented May 31, 2024 •

edited

Loading

ggml : fix loongson compile warnings #7537

ggml : fix loongson compile warnings #7537

Conversation

ggerganov commented May 25, 2024 • edited Loading

ggerganov commented May 25, 2024

github-actions bot commented May 25, 2024 • edited Loading

junchao-loongson commented May 29, 2024 via email

junchao-loongson commented May 30, 2024

ggerganov commented May 30, 2024

junchao-loongson commented May 31, 2024 • edited Loading

ggerganov commented May 25, 2024 •

edited

Loading

github-actions bot commented May 25, 2024 •

edited

Loading

junchao-loongson commented May 31, 2024 •

edited

Loading