No output after commit 84d9015 on Android

[gjmulder]

Discussed in https://github.com/ggerganov/llama.cpp/discussions/234

^{Originally posted by ShouNichi March 17, 2023} When git checkout 84d9015 and make, there will be no output (only the model loading message) in termux. git checkout 63fd76f will produce a fully-functional binary.

I've moved this to issues. Please provide sample output from the working build and the non-working build.

[gjmulder]

From #234:

[ShouNichi]

On another android device it will spit out error messages (at commit 84d9015) like

ggml.c:1364:25: warning: call to undeclared function 'vdotq_s32'; ISO C99 and later do not support implicit function declarations [-Wimplicit-function-declaration]
        int32x4_t p_0 = vdotq_s32(vdupq_n_s32(0), v0_0ls, v1_0ls);
                        ^
ggml.c:1364:19: error: initializing 'int32x4_t' (vector of 4 'int32_t' values) with an expression of incompatible type 'int'
        int32x4_t p_0 = vdotq_s32(vdupq_n_s32(0), v0_0ls, v1_0ls);
                  ^     ~~~~~~~~~
ggml.c:1365:19: error: initializing 'int32x4_t' (vector of 4 'int32_t' values) with an expression of incompatible type 'int'
        int32x4_t p_1 = vdotq_s32(vdupq_n_s32(0), v0_1ls, v1_1ls);
                  ^     ~~~~~~~~~
ggml.c:1367:13: error: assigning to 'int32x4_t' (vector of 4 'int32_t' values) from incompatible type 'int'
        p_0 = vdotq_s32(p_0, v0_0hs, v1_0hs);
            ^ ~~~~~~
ggml.c:1368:13: error: assigning to 'int32x4_t' (vector of 4 'int32_t' values) from incompatible type 'int'
        p_1 = vdotq_s32(p_1, v0_1hs, v1_1hs);
            ^ ~~~~~~
1 warning and 4 errors generated.
make: *** [Makefile:182: ggml.o] Error 1

and the latest commit would work just file, so I assume it is a device specific issue. On the original device, even with the fix/latest commit, it refuses to work. The not-working output is terminal output.txt

and the normal output is... just normal output, with things generating after the loading process.

[j-f1]

The latest commit should include the changes from https://github.com/ggerganov/llama.cpp/pull/139 which are supposed to avoid the usage of vdotq_s32 if it isn’t available, but maybe a usage snuck in somewhere?

[ShouNichi]

The latest commit should include the changes from #139 which are supposed to avoid the usage of vdotq_s32 if it isn’t available, but maybe a usage snuck in somewhere?

~Probably. Well, the problematic device is a Samsung Note 9. Don't know whether this info helps.~ No idea about how to help debugging. Any suggestions?

system_info: n_threads = 4 / 8 | AVX = 0 | AVX2 = 0 | AVX512 = 0 | FMA = 0 | NEON = 1 | ARM_FMA = 1 | F16C = 0 | FP16_VA = 1 | WASM_SIMD = 0 | BLAS = 0 | SSE3 = 0 | VSX = 0 | freshly cloned.txt