Bug: ggml/src/ggml-cpu/ggml-cpu.c:2373: error: undefined reference to 'getcpu'

[dspasyuk]

What happened?

I keep seeing this bug on various CentOS and Roku Linux distros with glibs > 2.28 but less than 2.33

Changing this line in ggml/src/ggml-cpu/ggml-cpu.c:2373

#if GLIBC > 2 || (GLIBC == 2 && GLIBC_MINOR > 28) || defined(COSMOPOLITAN)

to

#if GLIBC > 2 || (GLIBC == 2 && GLIBC_MINOR > 33) || defined(COSMOPOLITAN)

fixes the issue for all of my setups. Could we make this permanent?

Name and Version

ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
Device 0: NVIDIA A100 80GB PCIe, compute capability 8.0, VMM: yes
version: 4104 (0fff7fd)
built with cc (GCC) 9.3.0 for x86_64-pc-linux-gnu

What operating system are you seeing the problem on?

AlmaLinux release 8.6 (Sky Tiger)
Rocky Linux 8

Relevant log output

ggml/src/ggml-cpu/ggml-cpu.c:2373: error: undefined reference to 'getcpu'
collect2: error: ld returned 1 exit status
make: *** [Makefile:1240: llama-batched] Error 1
make: *** Waiting for unfinished jobs....
ggml/src/ggml-cpu/ggml-cpu.c:2373: error: undefined reference to 'getcpu'
collect2: error: ld returned 1 exit status
make: *** [Makefile:1285: llama-gguf] Error 1
ggml/src/ggml-cpu/ggml-cpu.c:2373: error: undefined reference to 'getcpu'
collect2: error: ld returned 1 exit status
make: *** [Makefile:1577: llama-vdot] Error 1
ggml/src/ggml-cpu/ggml-cpu.c:2373: error: undefined reference to 'getcpu'
collect2: error: ld returned 1 exit status
make: *** [Makefile:1582: llama-q8dot] Error 1
ggml/src/ggml-cpu/ggml-cpu.c:2373: error: undefined reference to 'getcpu'
collect2: error: ld returned 1 exit status
make: *** [Makefile:1327: llama-bench] Error 1
ggml/src/ggml-cpu/ggml-cpu.c:2373: error: undefined reference to 'getcpu'
collect2: error: ld returned 1 exit status
make: *** [Makefile:1245: llama-batched-bench] Error 1

[ggerganov]

Unless it somehow breaks some other environments, I don't see a problem to make this permanent. Let's merge it and see how it goes.

[FirstTimeEZ]

seems like it caused a test to fail

https://github.com/ggml-org/ci/tree/results/llama.cpp/a4/3178299c2f74f14bcfc659bfd9fd32d931d1f4/ggml-4-x86-cuda-v100

[ggerganov]

More likely caused by #10318

[FirstTimeEZ]

I made an issue about it #10356

[dspasyuk]

Many thanks guys!