mikhail/llama.cpp-mtp-turboquant
1
0
Fork 0
Code Issues Pull Requests Actions 32 Packages Projects Releases Wiki Activity

logs : reduce (#23021)

Browse Source 
* logs : reduce

* args : fix envs

* server : fix build

* common : print verbosity level at start

* server : clean-up logs

* server : print prompt processing timings + sampling params

* minor : whitespaces
This commit is contained in:
Georgi Gerganov
2026-05-14 13:05:52 +03:00
committed by GitHub
parent 81b0d882ae
commit 67b2b7f2f2
17 changed files with 207 additions and 147 deletions
Download Patch File Download Diff File Expand all files Collapse all files
common/arg.cpp
+12 -8
View File
@@ -308,12 +308,14 @@ static bool common_params_handle_remote_preset(common_params & params, llama_exa
common_download_opts opts;
opts.bearer_token = params.hf_token;
opts.offline = params.offline;
LOG_TRC("%s: looking for remote preset at %s\n", __func__, preset_url.c_str());
const int status = common_download_file_single(preset_url, preset_path, opts);
const bool has_preset = status >= 200 && status < 400;
// remote preset is optional, so we don't error out if not found
if (has_preset) {
LOG_INF("applying remote preset from %s\n", preset_url.c_str());
LOG_TRC("%s: applying remote preset from %s\n", __func__, preset_url.c_str());
common_preset_context ctx(ex, /* only_remote_allowed */ true);
common_preset global;
auto remote_presets = ctx.load_from_ini(preset_path, global);
@@ -326,7 +328,7 @@ static bool common_params_handle_remote_preset(common_params & params, llama_exa
throw std::runtime_error("Remote preset.ini does not contain [" + std::string(hf_tag) + "] section");
}
} else {
LOG_INF("%s", "no remote preset found, skipping\n");
LOG_TRC("%s: no remote preset found, skipping\n", __func__);
}
return has_preset;
@@ -3301,18 +3303,20 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
).set_env("LLAMA_LOG_VERBOSITY"));
add_opt(common_arg(
{"--log-prefix"},
{"--no-log-prefix"},
"Enable prefix in log messages",
[](common_params &) {
common_log_set_prefix(common_log_main(), true);
[](common_params &, bool value) {
common_log_set_prefix(common_log_main(), value);
}
).set_env("LLAMA_LOG_PREFIX"));
).set_env("LLAMA_ARG_LOG_PREFIX"));
add_opt(common_arg(
{"--log-timestamps"},
{"--no-log-timestamps"},
"Enable timestamps in log messages",
[](common_params &) {
common_log_set_timestamps(common_log_main(), true);
[](common_params &, bool value) {
common_log_set_timestamps(common_log_main(), value);
}
).set_env("LLAMA_LOG_TIMESTAMPS"));
).set_env("LLAMA_ARG_LOG_TIMESTAMPS"));
//
// speculative parameters
common/common.cpp
+22 -7
View File
@@ -366,15 +366,29 @@ void common_init() {
SetConsoleCP(CP_UTF8);
#endif
llama_log_set(common_log_default_callback, NULL);
common_log_set_prefix(common_log_main(), true);
common_log_set_timestamps(common_log_main(), true);
llama_log_set(common_log_default_callback, NULL);
}
void common_params_print_info(const common_params & params) {
#ifdef NDEBUG
const char * build_type = "";
#else
const char * build_type = " (debug)";
#endif
LOG_TRC("%s: build %d (%s) with %s for %s%s\n", __func__, llama_build_number(), llama_commit(), llama_compiler(), llama_build_target(), build_type);
LOG_DBG("build: %d (%s) with %s for %s%s\n", llama_build_number(), llama_commit(), llama_compiler(), llama_build_target(), build_type);
LOG_INF("log_info: verbosity = %d (adjust with the `-lv N` CLI arg)\n", common_log_get_verbosity_thold());
LOG_INF("device_info:\n");
for (size_t i = 0; i < ggml_backend_dev_count(); ++i) {
auto * dev = ggml_backend_dev_get(i);
size_t free, total;
ggml_backend_dev_memory(dev, &free, &total);
LOG_INF(" - %-8s: %s (%zu MiB, %zu MiB free)\n", ggml_backend_dev_name(dev), ggml_backend_dev_description(dev), total / 1024 / 1024, free / 1024 / 1024);
}
LOG_INF("%s\n", common_params_get_system_info(params).c_str());
}
std::string common_params_get_system_info(const common_params & params) {
@@ -1147,7 +1161,8 @@ common_init_result::common_init_result(common_params & params) :
auto cparams = common_context_params_to_llama(params);
if (params.fit_params) {
LOG_INF("%s: fitting params to device memory, for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on\n", __func__);
LOG_INF("%s: fitting params to device memory ...\n", __func__);
LOG_INF("%s: (for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on)\n", __func__);
common_fit_params(params.model.path.c_str(), &mparams, &cparams,
params.tensor_split,
params.tensor_buft_overrides.data(),
@@ -1196,7 +1211,7 @@ common_init_result::common_init_result(common_params & params) :
// initialize once
for (llama_token i = 0; i < llama_vocab_n_tokens(vocab); i++) {
if (llama_vocab_is_eog(vocab, i)) {
LOG_INF("%s: added %s logit bias = %f\n", __func__, common_token_to_piece(vocab, i).c_str(), -INFINITY);
LOG_TRC("%s: added %s logit bias = %f\n", __func__, common_token_to_piece(vocab, i).c_str(), -INFINITY);
params.sampling.logit_bias_eog.push_back({i, -INFINITY});
}
}
@@ -1209,12 +1224,12 @@ common_init_result::common_init_result(common_params & params) :
}
//if (params.sampling.penalty_last_n == -1) {
// LOG_INF("%s: setting penalty_last_n to ctx_size = %d\n", __func__, llama_n_ctx(lctx));
// LOG_TRC("%s: setting penalty_last_n to ctx_size = %d\n", __func__, llama_n_ctx(lctx));
// params.sampling.penalty_last_n = llama_n_ctx(lctx);
//}
//if (params.sampling.dry_penalty_last_n == -1) {
// LOG_INF("%s: setting dry_penalty_last_n to ctx_size = %d\n", __func__, llama_n_ctx(lctx));
// LOG_TRC("%s: setting dry_penalty_last_n to ctx_size = %d\n", __func__, llama_n_ctx(lctx));
// params.sampling.dry_penalty_last_n = llama_n_ctx(lctx);
//}
@@ -1422,7 +1437,7 @@ common_context_seq_rm_type common_context_can_seq_rm(llama_context * ctx) {
// try to remove the last tokens
if (!llama_memory_seq_rm(mem, 0, 1, -1)) {
LOG_WRN("%s: the context does not support partial sequence removal\n", __func__);
LOG_TRC("%s: the context does not support partial sequence removal\n", __func__);
res = COMMON_CONTEXT_SEQ_RM_TYPE_FULL;
goto done;
}
common/common.h
+1
View File
@@ -686,6 +686,7 @@ struct common_params {
// initializes the logging system and prints info about the build
void common_init();
void common_params_print_info(const common_params & params);
std::string common_params_get_system_info(const common_params & params);
bool parse_cpu_range(const std::string & range, bool(&boolmask)[GGML_MAX_N_THREADS]);
common/download.cpp
+2 -2
View File
@@ -320,9 +320,9 @@ static int common_download_file_single_online(const std::string & url,
auto head = cli.Head(parts.path);
if (!head || head->status < 200 || head->status >= 300) {
LOG_WRN("%s: HEAD failed, status: %d\n", __func__, head ? head->status : -1);
LOG_TRC("%s: HEAD failed, status: %d\n", __func__, head ? head->status : -1);
if (file_exists) {
LOG_INF("%s: using cached file (HEAD failed): %s\n", __func__, path.c_str());
LOG_TRC("%s: using cached file (HEAD failed): %s\n", __func__, path.c_str());
return 304; // 304 Not Modified - fake cached response
}
return head ? head->status : -1;
common/fit.cpp
+44 -44
View File
@@ -168,7 +168,7 @@ static void common_params_fit_impl(
// step 1: get data for default parameters and check whether any changes are necessary in the first place
LOG_INF("%s: getting device memory data for initial parameters:\n", __func__);
LOG_TRC("%s: getting device memory data for initial parameters:\n", __func__);
const dmds_t dmds_full = common_get_device_memory_data(path_model, mparams, cparams, devs, hp_ngl, hp_nct, hp_nex, log_level);
const size_t nd = devs.size(); // number of devices
@@ -213,13 +213,13 @@ static void common_params_fit_impl(
LOG_INF("%s: projected to use %" PRId64 " MiB of host memory vs. %" PRId64 " MiB of total host memory\n",
__func__, sum_projected_used/MiB, sum_free/MiB);
if (sum_projected_free >= margins[0]) {
LOG_INF("%s: will leave %" PRId64 " >= %" PRId64 " MiB of system memory, no changes needed\n",
LOG_TRC("%s: will leave %" PRId64 " >= %" PRId64 " MiB of system memory, no changes needed\n",
__func__, sum_projected_free/MiB, margins[0]/MiB);
return;
}
} else {
if (nd > 1) {
LOG_INF("%s: projected memory use with initial parameters [MiB]:\n", __func__);
LOG_TRC("%s: projected memory use with initial parameters [MiB]:\n", __func__);
}
for (size_t id = 0; id < nd; id++) {
const llama_device_memory_data & dmd = dmds_full[id];
@@ -234,16 +234,16 @@ static void common_params_fit_impl(
sum_projected_model += dmd.mb.model;
if (nd > 1) {
LOG_INF("%s: - %s: %6" PRId64 " total, %6" PRId64 " used, %6" PRId64 " free vs. target of %6" PRId64 "\n",
LOG_TRC("%s: - %s: %6" PRId64 " total, %6" PRId64 " used, %6" PRId64 " free vs. target of %6" PRId64 "\n",
__func__, dev_names[id].c_str(), dmd.total/MiB, projected_used/MiB, projected_free/MiB, margins[id]/MiB);
}
}
assert(sum_free >= 0 && sum_projected_used >= 0);
LOG_INF("%s: projected to use %" PRId64 " MiB of device memory vs. %" PRId64 " MiB of free device memory\n",
LOG_TRC("%s: projected to use %" PRId64 " MiB of device memory vs. %" PRId64 " MiB of free device memory\n",
__func__, sum_projected_used/MiB, sum_free/MiB);
if (nd == 1) {
if (projected_free_per_device[0] >= margins[0]) {
LOG_INF("%s: will leave %" PRId64 " >= %" PRId64 " MiB of free device memory, no changes needed\n",
LOG_TRC("%s: will leave %" PRId64 " >= %" PRId64 " MiB of free device memory, no changes needed\n",
__func__, projected_free_per_device[0]/MiB, margins[0]/MiB);
return;
}
@@ -256,7 +256,7 @@ static void common_params_fit_impl(
}
}
if (!changes_needed) {
LOG_INF("%s: targets for free memory can be met on all devices, no changes needed\n", __func__);
LOG_TRC("%s: targets for free memory can be met on all devices, no changes needed\n", __func__);
return;
}
}
@@ -275,10 +275,10 @@ static void common_params_fit_impl(
}
if (global_surplus < 0) {
if (nd <= 1) {
LOG_INF("%s: cannot meet free memory target of %" PRId64 " MiB, need to reduce device memory by %" PRId64 " MiB\n",
LOG_TRC("%s: cannot meet free memory target of %" PRId64 " MiB, need to reduce device memory by %" PRId64 " MiB\n",
__func__, margins[0]/MiB, -global_surplus/MiB);
} else {
LOG_INF(
LOG_TRC(
"%s: cannot meet free memory targets on all devices, need to use %" PRId64 " MiB less in total\n",
__func__, -global_surplus/MiB);
}
@@ -320,28 +320,28 @@ static void common_params_fit_impl(
const int64_t bytes_per_ctx = (sum_projected_used - sum_projected_used_min_ctx) / (hp_nct - n_ctx_min);
const int64_t memory_reduction = (hp_nct - cparams->n_ctx) * bytes_per_ctx;
LOG_INF("%s: context size reduced from %" PRIu32 " to %" PRIu32 " -> need %" PRId64 " MiB less memory in total\n",
LOG_TRC("%s: context size reduced from %" PRIu32 " to %" PRIu32 " -> need %" PRId64 " MiB less memory in total\n",
__func__, hp_nct, cparams->n_ctx, memory_reduction/MiB);
if (nd <= 1) {
LOG_INF("%s: entire model can be fit by reducing context\n", __func__);
LOG_TRC("%s: entire model can be fit by reducing context\n", __func__);
return;
}
LOG_INF("%s: entire model should be fit across devices by reducing context\n", __func__);
LOG_TRC("%s: entire model should be fit across devices by reducing context\n", __func__);
} else {
const int64_t memory_reduction = sum_projected_used - sum_projected_used_min_ctx;
LOG_INF("%s: context size reduced from %" PRIu32 " to %" PRIu32 " -> need %" PRId64 " MiB less memory in total\n",
LOG_TRC("%s: context size reduced from %" PRIu32 " to %" PRIu32 " -> need %" PRId64 " MiB less memory in total\n",
__func__, hp_nct, cparams->n_ctx, memory_reduction/MiB);
}
} else {
if (n_ctx_min == UINT32_MAX) {
LOG_INF("%s: user has requested full context size of %" PRIu32 " -> no change\n", __func__, hp_nct);
LOG_TRC("%s: user has requested full context size of %" PRIu32 " -> no change\n", __func__, hp_nct);
} else {
LOG_INF("%s: default model context size is %" PRIu32 " which is <= the min. context size of %" PRIu32 " -> no change\n",
LOG_TRC("%s: default model context size is %" PRIu32 " which is <= the min. context size of %" PRIu32 " -> no change\n",
__func__, hp_nct, n_ctx_min);
}
}
} else {
LOG_INF("%s: context size set by user to %" PRIu32 " -> no change\n", __func__, cparams->n_ctx);
LOG_TRC("%s: context size set by user to %" PRIu32 " -> no change\n", __func__, cparams->n_ctx);
}
}
}
@@ -485,10 +485,10 @@ static void common_params_fit_impl(
const dmds_t dmd_nl = common_get_device_memory_data(
path_model, &mparams_copy, cparams, devs, hp_ngl, hp_nct, hp_nex, log_level);
LOG_INF("%s: memory for test allocation by device:\n", func_name);
LOG_TRC("%s: memory for test allocation by device:\n", func_name);
for (size_t id = 0; id < nd; id++) {
const ngl_t & n = ngl_per_device[id];
LOG_INF(
LOG_TRC(
"%s: id=%zu, n_layer=%2" PRIu32 ", n_part=%2" PRIu32 ", overflow_type=%d, mem=%6" PRId64 " MiB\n",
func_name, id, n.n_layer, n.n_part, int(n.overflow_type), dmd_nl[id].mb.total()/MiB);
}
@@ -509,7 +509,7 @@ static void common_params_fit_impl(
tensor_buft_overrides[1] = {nullptr, nullptr};
mparams->tensor_buft_overrides = tensor_buft_overrides;
LOG_INF("%s: getting device memory data with all MoE tensors moved to system memory:\n", __func__);
LOG_TRC("%s: getting device memory data with all MoE tensors moved to system memory:\n", __func__);
const dmds_t dmds_cpu_moe = common_get_device_memory_data(
path_model, mparams, cparams, devs, hp_ngl, hp_nct, hp_nex, log_level);
@@ -519,10 +519,10 @@ static void common_params_fit_impl(
}
if (global_surplus_cpu_moe > 0) {
LOG_INF("%s: with only dense weights in device memory there is a total surplus of %" PRId64 " MiB\n",
LOG_TRC("%s: with only dense weights in device memory there is a total surplus of %" PRId64 " MiB\n",
__func__, global_surplus_cpu_moe/MiB);
} else {
LOG_INF("%s: with only dense weights in device memory there is still a total deficit of %" PRId64 " MiB\n",
LOG_TRC("%s: with only dense weights in device memory there is still a total deficit of %" PRId64 " MiB\n",
__func__, -global_surplus_cpu_moe/MiB);
}
@@ -535,7 +535,7 @@ static void common_params_fit_impl(
targets.reserve(nd);
for (size_t id = 0; id < nd; id++) {
targets.push_back(dmds_full[id].free - margins[id]);
LOG_INF("%s: id=%zu, target=%" PRId64 " MiB\n", __func__, id, targets[id]/MiB);
LOG_TRC("%s: id=%zu, target=%" PRId64 " MiB\n", __func__, id, targets[id]/MiB);
}
std::vector<ggml_backend_buffer_type_t> overflow_bufts; // which bufts the first partial layer of a device overflows to:
@@ -555,9 +555,9 @@ static void common_params_fit_impl(
// - once we only have a difference of a single layer, stop and return the lower bound that just barely still fits
// - the last device has the output layer, which cannot be a partial layer
if (hp_nex == 0) {
LOG_INF("%s: filling dense layers back-to-front:\n", __func__);
LOG_TRC("%s: filling dense layers back-to-front:\n", __func__);
} else {
LOG_INF("%s: filling dense-only layers back-to-front:\n", __func__);
LOG_TRC("%s: filling dense-only layers back-to-front:\n", __func__);
}
for (int id = nd - 1; id >= 0; id--) {
uint32_t n_unassigned = hp_ngl + 1;
@@ -576,7 +576,7 @@ static void common_params_fit_impl(
if (mem_high[id] > targets[id]) {
assert(ngl_per_device_high[id].n_layer > ngl_per_device[id].n_layer);
uint32_t delta = ngl_per_device_high[id].n_layer - ngl_per_device[id].n_layer;
LOG_INF("%s: start filling device %" PRIu32 ", delta=%" PRIu32 "\n", __func__, id, delta);
LOG_TRC("%s: start filling device %" PRIu32 ", delta=%" PRIu32 "\n", __func__, id, delta);
while (delta > 1) {
uint32_t step_size = int64_t(delta) * (targets[id] - mem[id]) / (mem_high[id] - mem[id]);
step_size = std::max(step_size, uint32_t(1));
@@ -593,11 +593,11 @@ static void common_params_fit_impl(
if (mem_test[id] <= targets[id]) {
ngl_per_device = ngl_per_device_test;
mem = mem_test;
LOG_INF("%s: set ngl_per_device[%d].n_layer=%" PRIu32 "\n", __func__, id, ngl_per_device[id].n_layer);
LOG_TRC("%s: set ngl_per_device[%d].n_layer=%" PRIu32 "\n", __func__, id, ngl_per_device[id].n_layer);
} else {
ngl_per_device_high = ngl_per_device_test;
mem_high = mem_test;
LOG_INF("%s: set ngl_per_device_high[%d].n_layer=%" PRIu32 "\n", __func__, id, ngl_per_device_high[id].n_layer);
LOG_TRC("%s: set ngl_per_device_high[%d].n_layer=%" PRIu32 "\n", __func__, id, ngl_per_device_high[id].n_layer);
}
delta = ngl_per_device_high[id].n_layer - ngl_per_device[id].n_layer;
}
@@ -605,12 +605,12 @@ static void common_params_fit_impl(
assert(ngl_per_device_high[id].n_layer == n_unassigned);
ngl_per_device = ngl_per_device_high;
mem = mem_high;
LOG_INF("%s: set ngl_per_device[%d].n_layer=%" PRIu32 "\n", __func__, id, ngl_per_device[id].n_layer);
LOG_TRC("%s: set ngl_per_device[%d].n_layer=%" PRIu32 "\n", __func__, id, ngl_per_device[id].n_layer);
}
}
const int64_t projected_margin = dmds_full[id].free - mem[id];
LOG_INF(
LOG_TRC(
"%s: - %s: %2" PRIu32 " layers, %6" PRId64 " MiB used, %6" PRId64 " MiB free\n",
__func__, dev_names[id].c_str(), ngl_per_device[id].n_layer, mem[id]/MiB, projected_margin/MiB);
}
@@ -634,7 +634,7 @@ static void common_params_fit_impl(
}
assert(id_dense_start < nd);
LOG_INF("%s: converting dense-only layers to full layers and filling them front-to-back with overflow to next device/system memory:\n", __func__);
LOG_TRC("%s: converting dense-only layers to full layers and filling them front-to-back with overflow to next device/system memory:\n", __func__);
for (size_t id = 0; id <= id_dense_start && id_dense_start < nd; id++) {
std::vector<ngl_t> ngl_per_device_high = ngl_per_device;
for (size_t jd = id_dense_start; jd < nd; jd++) {
@@ -674,13 +674,13 @@ static void common_params_fit_impl(
ngl_per_device = ngl_per_device_test;
mem = mem_test;
id_dense_start = id_dense_start_test;
LOG_INF("%s: set ngl_per_device[%zu].(n_layer, n_part)=(%" PRIu32 ", %" PRIu32 "), id_dense_start=%zu\n",
LOG_TRC("%s: set ngl_per_device[%zu].(n_layer, n_part)=(%" PRIu32 ", %" PRIu32 "), id_dense_start=%zu\n",
__func__, id, ngl_per_device[id].n_layer, ngl_per_device[id].n_part, id_dense_start);
} else {
ngl_per_device_high = ngl_per_device_test;
mem_high = mem_test;
id_dense_start_high = id_dense_start_test;
LOG_INF("%s: set ngl_per_device_high[%zu].(n_layer, n_part)=(%" PRIu32 ", %" PRIu32 "), id_dense_start_high=%zu\n",
LOG_TRC("%s: set ngl_per_device_high[%zu].(n_layer, n_part)=(%" PRIu32 ", %" PRIu32 "), id_dense_start_high=%zu\n",
__func__, id, ngl_per_device_high[id].n_layer, ngl_per_device_high[id].n_part, id_dense_start_high);
}
assert(ngl_per_device_high[id].n_full() >= ngl_per_device[id].n_full());
@@ -690,7 +690,7 @@ static void common_params_fit_impl(
ngl_per_device = ngl_per_device_high;
mem = mem_high;
id_dense_start = id_dense_start_high;
LOG_INF("%s: set ngl_per_device[%zu].(n_layer, n_part)=(%" PRIu32 ", %" PRIu32 "), id_dense_start=%zu\n",
LOG_TRC("%s: set ngl_per_device[%zu].(n_layer, n_part)=(%" PRIu32 ", %" PRIu32 "), id_dense_start=%zu\n",
__func__, id, ngl_per_device[id].n_layer, ngl_per_device[id].n_part, id_dense_start);
}
@@ -710,44 +710,44 @@ static void common_params_fit_impl(
if (id < nd - 1) {
overflow_bufts_test[id] = ggml_backend_dev_buffer_type(devs[id + 1]);
}
LOG_INF("%s: trying to fit one extra layer with overflow_type=LAYER_FRACTION_UP\n", __func__);
LOG_TRC("%s: trying to fit one extra layer with overflow_type=LAYER_FRACTION_UP\n", __func__);
std::vector<int64_t> mem_test = get_memory_for_layers(__func__, ngl_per_device_test, overflow_bufts_test);
if (mem_test[id] < targets[id] && (id + 1 == nd || mem_test[id + 1] < targets[id + 1])) {
ngl_per_device = ngl_per_device_test;
overflow_bufts = overflow_bufts_test;
mem = mem_test;
id_dense_start = id_dense_start_test;
LOG_INF("%s: set ngl_per_device[%zu].(n_layer, n_part, overflow_type)=(%" PRIu32 ", %" PRIu32 ", UP), id_dense_start=%zu\n",
LOG_TRC("%s: set ngl_per_device[%zu].(n_layer, n_part, overflow_type)=(%" PRIu32 ", %" PRIu32 ", UP), id_dense_start=%zu\n",
__func__, id, ngl_per_device[id].n_layer, ngl_per_device[id].n_part, id_dense_start);
ngl_per_device_test[id].overflow_type = LAYER_FRACTION_GATE;
LOG_INF("%s: trying to fit one extra layer with overflow_type=LAYER_FRACTION_GATE\n", __func__);
LOG_TRC("%s: trying to fit one extra layer with overflow_type=LAYER_FRACTION_GATE\n", __func__);
mem_test = get_memory_for_layers(__func__, ngl_per_device_test, overflow_bufts_test);
if (mem_test[id] < targets[id] && (id + 1 == nd || mem_test[id + 1] < targets[id + 1])) {
ngl_per_device = ngl_per_device_test;
overflow_bufts = overflow_bufts_test;
mem = mem_test;
id_dense_start = id_dense_start_test;
LOG_INF("%s: set ngl_per_device[%zu].(n_layer, n_part, overflow_type)=(%" PRIu32 ", %" PRIu32 ", GATE), id_dense_start=%zu\n",
LOG_TRC("%s: set ngl_per_device[%zu].(n_layer, n_part, overflow_type)=(%" PRIu32 ", %" PRIu32 ", GATE), id_dense_start=%zu\n",
__func__, id, ngl_per_device[id].n_layer, ngl_per_device[id].n_part, id_dense_start);
}
} else {
ngl_per_device_test[id].overflow_type = LAYER_FRACTION_ATTN;
LOG_INF("%s: trying to fit one extra layer with overflow_type=LAYER_FRACTION_ATTN\n", __func__);
LOG_TRC("%s: trying to fit one extra layer with overflow_type=LAYER_FRACTION_ATTN\n", __func__);
mem_test = get_memory_for_layers(__func__, ngl_per_device_test, overflow_bufts_test);
if (mem_test[id] < targets[id] && (id + 1 == nd || mem_test[id + 1] < targets[id + 1])) {
ngl_per_device = ngl_per_device_test;
overflow_bufts = overflow_bufts_test;
mem = mem_test;
id_dense_start = id_dense_start_test;
LOG_INF("%s: set ngl_per_device[%zu].(n_layer, n_part, overflow_type)=(%" PRIu32 ", %" PRIu32 ", ATTN), id_dense_start=%zu\n",
LOG_TRC("%s: set ngl_per_device[%zu].(n_layer, n_part, overflow_type)=(%" PRIu32 ", %" PRIu32 ", ATTN), id_dense_start=%zu\n",
__func__, id, ngl_per_device[id].n_layer, ngl_per_device[id].n_part, id_dense_start);
}
}
}
const int64_t projected_margin = dmds_full[id].free - mem[id];
LOG_INF(
LOG_TRC(
"%s: - %s: %2" PRIu32 " layers (%2" PRIu32 " overflowing), %6" PRId64 " MiB used, %6" PRId64 " MiB free\n",
__func__, dev_names[id].c_str(), ngl_per_device[id].n_layer, ngl_per_device[id].n_part, mem[id]/MiB, projected_margin/MiB);
}
@@ -755,7 +755,7 @@ static void common_params_fit_impl(
// print info for devices that were not changed during the conversion from dense only to full layers:
for (size_t id = id_dense_start + 1; id < nd; id++) {
const int64_t projected_margin = dmds_full[id].free - mem[id];
LOG_INF(
LOG_TRC(
"%s: - %s: %2" PRIu32 " layers (%2" PRIu32 " overflowing), %6" PRId64 " MiB used, %6" PRId64 " MiB free\n",
__func__, dev_names[id].c_str(), ngl_per_device[id].n_layer, ngl_per_device[id].n_part, mem[id]/MiB, projected_margin/MiB);
}
@@ -776,7 +776,7 @@ enum common_params_fit_status common_fit_params(
common_params_fit_status status = COMMON_PARAMS_FIT_STATUS_SUCCESS;
try {
common_params_fit_impl(path_model, mparams, cparams, tensor_split, tensor_buft_overrides, margins, n_ctx_min, log_level);
LOG_INF("%s: successfully fit params to free device memory\n", __func__);
LOG_TRC("%s: successfully fit params to free device memory\n", __func__);
} catch (const common_params_fit_exception & e) {
LOG_WRN("%s: failed to fit params to free device memory: %s\n", __func__, e.what());
status = COMMON_PARAMS_FIT_STATUS_FAILURE;
@@ -785,7 +785,7 @@ enum common_params_fit_status common_fit_params(
status = COMMON_PARAMS_FIT_STATUS_ERROR;
}
const int64_t t1_us = llama_time_us();
LOG_INF("%s: fitting params to free memory took %.2f seconds\n", __func__, (t1_us - t0_us) * 1e-6);
LOG_TRC("%s: fitting params to free memory took %.2f seconds\n", __func__, (t1_us - t0_us) * 1e-6);
return status;
}
@@ -925,7 +925,7 @@ void common_memory_breakdown_print(const struct llama_context * ctx) {
}
}
for (const auto & td : table_data) {
LOG_INF(td[0].c_str(),
LOG_TRC(td[0].c_str(),
__func__, td[1].c_str(), td[2].c_str(), td[3].c_str(), td[4].c_str(), td[5].c_str(),
td[6].c_str(), td[7].c_str(), td[8].c_str());
}
common/log.cpp
+2 -2
View File
@@ -435,10 +435,10 @@ void common_log_flush(struct common_log * log) {
static int common_get_verbosity(enum ggml_log_level level) {
switch (level) {
case GGML_LOG_LEVEL_DEBUG: return LOG_LEVEL_DEBUG;
case GGML_LOG_LEVEL_INFO: return LOG_LEVEL_INFO;
case GGML_LOG_LEVEL_INFO: return LOG_LEVEL_TRACE;
case GGML_LOG_LEVEL_WARN: return LOG_LEVEL_WARN;
case GGML_LOG_LEVEL_ERROR: return LOG_LEVEL_ERROR;
case GGML_LOG_LEVEL_CONT: return LOG_LEVEL_INFO; // same as INFO
case GGML_LOG_LEVEL_CONT: return LOG_LEVEL_TRACE;
case GGML_LOG_LEVEL_NONE:
default:
return LOG_LEVEL_OUTPUT;
common/log.h
+5 -2
View File
@@ -21,7 +21,8 @@
# define LOG_ATTRIBUTE_FORMAT(...) __attribute__((format(printf, __VA_ARGS__)))
#endif
#define LOG_LEVEL_DEBUG 4
#define LOG_LEVEL_DEBUG 5
#define LOG_LEVEL_TRACE 4
#define LOG_LEVEL_INFO 3
#define LOG_LEVEL_WARN 2
#define LOG_LEVEL_ERROR 1
@@ -111,13 +112,15 @@ void common_log_flush (struct common_log * log); // f
#define LOGV(verbosity, ...) LOG_TMPL(GGML_LOG_LEVEL_NONE, verbosity, __VA_ARGS__)
#define LOG_DBG(...) LOG_TMPL(GGML_LOG_LEVEL_DEBUG, LOG_LEVEL_DEBUG, __VA_ARGS__)
#define LOG_TRC(...) LOG_TMPL(GGML_LOG_LEVEL_INFO, LOG_LEVEL_TRACE, __VA_ARGS__)
#define LOG_INF(...) LOG_TMPL(GGML_LOG_LEVEL_INFO, LOG_LEVEL_INFO, __VA_ARGS__)
#define LOG_WRN(...) LOG_TMPL(GGML_LOG_LEVEL_WARN, LOG_LEVEL_WARN, __VA_ARGS__)
#define LOG_ERR(...) LOG_TMPL(GGML_LOG_LEVEL_ERROR, LOG_LEVEL_ERROR, __VA_ARGS__)
#define LOG_CNT(...) LOG_TMPL(GGML_LOG_LEVEL_CONT, LOG_LEVEL_INFO, __VA_ARGS__) // same as INFO
#define LOG_DBGV(verbosity, ...) LOG_TMPL(GGML_LOG_LEVEL_DEBUG, verbosity, __VA_ARGS__)
#define LOG_TRCV(verbosity, ...) LOG_TMPL(GGML_LOG_LEVEL_TRACE, verbosity, __VA_ARGS__)
#define LOG_INFV(verbosity, ...) LOG_TMPL(GGML_LOG_LEVEL_INFO, verbosity, __VA_ARGS__)
#define LOG_WRNV(verbosity, ...) LOG_TMPL(GGML_LOG_LEVEL_WARN, verbosity, __VA_ARGS__)
#define LOG_ERRV(verbosity, ...) LOG_TMPL(GGML_LOG_LEVEL_ERROR, verbosity, __VA_ARGS__)
#define LOG_DBGV(verbosity, ...) LOG_TMPL(GGML_LOG_LEVEL_DEBUG, verbosity, __VA_ARGS__)
#define LOG_CNTV(verbosity, ...) LOG_TMPL(GGML_LOG_LEVEL_CONT, verbosity, __VA_ARGS__)
common/speculative.cpp
+1 -1
View File
@@ -984,7 +984,7 @@ common_speculative * common_speculative_init(common_params_speculative & params,
}
if (impls.empty()) {
LOG_WRN("%s", "no implementations specified for speculative decoding\n");
LOG_WRN("%s: no implementations specified for speculative decoding\n", __func__);
return nullptr;
}