ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-01-26 09:09:50 +00:00
Code Issues Packages Projects Releases Wiki Activity

Be able to set reduce op data type for split mode "graph" (#1087)

Browse Source 
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
This commit is contained in:
Kawrakow
2025-12-24 14:01:29 +01:00
committed by GitHub
parent 903377bc34
commit 5e64235d4c
7 changed files with 23 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
common/common.cpp
View File

@@ -1436,6 +1436,14 @@ bool gpt_params_find_arg(int argc, char ** argv, const std::string & arg, gpt_pa
params.split_mode_graph_scheduling = true;
return true;
}
if (arg == "-smf16" || arg == "--split-mode-f16") {
params.split_mode_f16 = true;
return true;
}
if (arg == "-smf32" || arg == "--split-mode-f32") {
params.split_mode_f16 = false;
return true;
}
if (arg == "--numa") {
CHECK_ARG
std::string value(argv[i]);
@@ -2122,6 +2130,8 @@ void gpt_params_print_usage(int /*argc*/, char ** argv, const gpt_params & param
options.push_back({ "*", "-ser, --smart-expert-reduction", "experts reduction (default: %d,%g)", params.min_experts, params.thresh_experts});
options.push_back({ "*", "-mqkv, --merge-qkv,", "merge Q,K,V (default: %d)", params.merge_qkv});
options.push_back({ "*", "-khad, --k-cache-hadamard,", "Use Hadamard transform for K-cache (default: %d)", params.k_cache_hadamard});
options.push_back({ "*", "-smf16, --split-mode-f16,", "Use f16 for data exchange between GPUs (default: %d)", params.split_mode_f16});
options.push_back({ "*", "-smf32, --split-mode-f32,", "Use f32 for data exchange between GPUs (default: %d)", !params.split_mode_f16});
options.push_back({ "*", "-smgs, --split-mode-graph-scheduling,", "Force Split Mode Graph Scheduling (default: %d)", params.split_mode_graph_scheduling});
options.push_back({ "*", "-vq, --validate-quants", "validate quantized data while loading the model (default: %d)", params.validate_quants});
options.push_back({ "*", "-p, --prompt PROMPT", "prompt to start generation with\n"
@@ -3156,6 +3166,7 @@ struct llama_context_params llama_context_params_from_gpt_params(const gpt_param
cparams.graph_reuse = params.graph_reuse;
cparams.k_cache_hadamard = params.k_cache_hadamard;
cparams.split_mode_graph_scheduling = params.split_mode_graph_scheduling;
cparams.split_mode_f16 = params.split_mode_f16;
cparams.min_experts = params.min_experts;
cparams.thresh_experts = params.thresh_experts;
cparams.only_active_experts = params.only_active_exps;
@@ -4138,6 +4149,7 @@ void yaml_dump_non_result_info(FILE * stream, const gpt_params & params, const l
fprintf(stream, "graph_reuse: %s # default: false\n", params.graph_reuse ? "true" : "false");
fprintf(stream, "k_cache_hadamard: %s # default: false\n", params.k_cache_hadamard ? "true" : "false");
fprintf(stream, "split_mode_graph_scheduling: %s # default: false\n", params.split_mode_graph_scheduling ? "true" : "false");
fprintf(stream, "split_mode_f16: %s # default: true\n", params.split_mode_f16 ? "true" : "false");
fprintf(stream, "ser: %d,%g # defaulr: -1,0\n", params.min_experts, params.thresh_experts);
fprintf(stream, "temp: %f # default: 0.8\n", sparams.temp);

1
common/common.h
View File

@@ -289,6 +289,7 @@ struct gpt_params {
bool merge_qkv = false; // if true, merge separate Q, K, V tensors into a single, contiguous tensor
bool k_cache_hadamard = false; // if true, use Hadamard transform for the K-cache (only makes sense with quantized cache)
bool split_mode_graph_scheduling = false; // if true, force split mode graph scheduling
bool split_mode_f16 = true; // if true, intermediate results will be cast to f16 before copying to other GPUs to perform reduce ops
std::string cache_type_k = "f16"; // KV cache data type for the K
std::string cache_type_v = "f16"; // KV cache data type for the V