vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

fixes for non-llvm builds

Browse source
This commit is contained in:
Faisal Zaghloul 2024-05-30 07:58:02 -04:00
parent 88dc99a20b
commit e8c3364387
2 changed files with 11 additions and 8 deletions
Download patch file Download diff file Expand all files Collapse all files

17
examples/llama-bench/llama-bench.cpp
View file

@ -191,13 +191,16 @@ struct cmd_params {
bool verbose; bool verbose;
output_formats output_format; output_formats output_format;
}; };
//
//static const cpu_params default_cpuparams(
// int32_t(std::thread::hardware_concurrency()),
// {false},
// false,
// 1,
// false,
// false
//);
int32_t n_threads = -1;
bool cpumask[GGML_N_CORES_MAX] = { false }; // CPU affinity mask.
bool mask_valid = false; // Default: any CPU
int32_t priority = 0; // Scheduling prio : (0 - normal, 1 - medium, 2 - high, 3 - realtime)
bool strict_cpu = false; // Use strict CPU placement
bool poll = false; // Use polling (busywait) to wait for work
static const cmd_params cmd_params_defaults = { static const cmd_params cmd_params_defaults = {
/* model */ {"models/7B/ggml-model-q4_0.gguf"}, /* model */ {"models/7B/ggml-model-q4_0.gguf"},
/* n_prompt */ {512}, /* n_prompt */ {512},
@ -217,7 +220,7 @@ static const cmd_params cmd_params_defaults = {
/* use_mmap */ {true}, /* use_mmap */ {true},
/* embeddings */ {false}, /* embeddings */ {false},
/* numa */ GGML_NUMA_STRATEGY_DISABLED, /* numa */ GGML_NUMA_STRATEGY_DISABLED,
/* cpuparams */ {int32_t(std::thread::hardware_concurrency()), {false}, false, 1, false, false}, /* cpuparams */ {},
/* reps */ 5, /* reps */ 5,
/* verbose */ false, /* verbose */ false,
/* output_format */ MARKDOWN /* output_format */ MARKDOWN

2
ggml.c
View file

@ -20062,7 +20062,7 @@ struct ggml_cplan ggml_graph_plan(
} }
cplan.threadpool = threadpool; cplan.threadpool = threadpool;
cplan.n_threads = n_threads; cplan.n_threads = MIN(max_tasks, n_threads);
cplan.work_size = work_size; cplan.work_size = work_size;
cplan.work_data = NULL; cplan.work_data = NULL;