ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-01-26 17:20:01 +00:00
Code Issues Packages Projects Releases Wiki Activity

Disable split mode "row" (#987)

Browse Source 
* Disable split mode "row"

* Also llama-bench

---------

Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
This commit is contained in:
Kawrakow
2025-11-19 16:15:50 +01:00
committed by GitHub
parent bacb8fb79f
commit 0f6986a33c
2 changed files with 11 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
common/common.cpp
View File

@@ -1240,10 +1240,10 @@ bool gpt_params_find_arg(int argc, char ** argv, const std::string & arg, gpt_pa
params.split_mode = LLAMA_SPLIT_MODE_LAYER;
}
else if (arg_next == "row") {
#ifdef GGML_USE_SYCL
fprintf(stderr, "warning: The split mode value:[row] is not supported by llama.cpp with SYCL. It's developing.\nExit!\n");
exit(1);
#endif // GGML_USE_SYCL
fprintf(stderr, "\n\n=====================================================================================\n");
fprintf(stderr, " Split mode row is no longer supported\n");
fprintf(stderr, "=====================================================================================\n\n\n");
GGML_ABORT("fatal error");
params.split_mode = LLAMA_SPLIT_MODE_ROW;
}
else {
@@ -2217,8 +2217,7 @@ void gpt_params_print_usage(int /*argc*/, char ** argv, const gpt_params & param
options.push_back({ "*", "-sm, --split-mode SPLIT_MODE",
"how to split the model across multiple GPUs, one of:\n"
" - none: use one GPU only\n"
" - layer (default): split layers and KV across GPUs\n"
" - row: split rows across GPUs" });
" - layer (default): split layers and KV across GPUs\n" });
options.push_back({ "*", "-ts, --tensor-split SPLIT",
"fraction of the model to offload to each GPU, comma-separated list of proportions, e.g. 3,1" });
options.push_back({ "*", "-dev, --device dev1,dev2",

8
examples/llama-bench/llama-bench.cpp
View File

@@ -334,7 +334,7 @@ static void print_usage(int /* argc */, char ** argv) {
printf(" -ngl, --n-gpu-layers <n> (default: %s)\n", join(cmd_params_defaults.n_gpu_layers, ",").c_str());
printf(" --n-cpu-moe <n> (default: none)\n");
printf(" -rpc, --rpc <rpc_servers> (default: %s)\n", join(cmd_params_defaults.rpc_servers, ",").c_str());
printf(" -sm, --split-mode <none|layer|row> (default: %s)\n", join(transform_to_str(cmd_params_defaults.split_mode, split_mode_str), ",").c_str());
printf(" -sm, --split-mode <none|layer> (default: %s)\n", join(transform_to_str(cmd_params_defaults.split_mode, split_mode_str), ",").c_str());
printf(" -mg, --main-gpu <i> (default: %s)\n", join(cmd_params_defaults.main_gpu, ",").c_str());
printf(" -nkvo, --no-kv-offload <0|1> (default: %s)\n", join(cmd_params_defaults.no_kv_offload, ",").c_str());
printf(" -fa, --flash-attn <0|1> (default: %s)\n", join(cmd_params_defaults.flash_attn, ",").c_str());
@@ -631,7 +631,11 @@ static cmd_params parse_cmd_params(int argc, char ** argv) {
} else if (m == "layer") {
mode = LLAMA_SPLIT_MODE_LAYER;
} else if (m == "row") {
mode = LLAMA_SPLIT_MODE_ROW;
fprintf(stderr, "\n\n=======================================================================\n");
fprintf(stderr, "Split mode 'row' is no longer supported\n");
fprintf(stderr, "=======================================================================\n\n\n");
invalid_param = true;
break;
} else {
invalid_param = true;
break;