llama : reuse compute graphs

ggml-ci

Author: ggerganov

common/arg.cpp

@@ -1464,6 +1464,14 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
             params.swa_full = true;
         }
     ).set_env("LLAMA_ARG_SWA_FULL"));
+    add_opt(common_arg(
+        {"--graph-reuse", "-gr"},
+        string_format("reuse previous compute graphs when possible (default: %s)"
+            "[(more info)](https://github.com/ggml-org/llama.cpp/pull/14482)", params.graph_reuse ? "true" : "false"),
+        [](common_params & params) {
+            params.graph_reuse = true;
+        }
+    ).set_env("LLAMA_ARG_GRAPH_REUSE"));
     add_opt(common_arg(
         {"--no-context-shift"},
         string_format("disables context shift on infinite text generation (default: %s)", params.ctx_shift ? "disabled" : "enabled"),

common/common.cpp

@@ -1157,6 +1157,7 @@ struct llama_context_params common_context_params_to_llama(const common_params &
     cparams.no_perf           = params.no_perf;
     cparams.op_offload        = !params.no_op_offload;
     cparams.swa_full          = params.swa_full;
+    cparams.graph_reuse       = params.graph_reuse;
 
     cparams.type_k = params.cache_type_k;
     cparams.type_v = params.cache_type_v;

common/common.h

@@ -330,6 +330,7 @@ struct common_params {
     bool no_perf           = false; // disable performance metrics
     bool ctx_shift         = true;  // context shift on inifinite text generation
     bool swa_full          = false; // use full-size SWA cache (https://github.com/ggml-org/llama.cpp/pull/13194#issuecomment-2868343055)
+    bool graph_reuse       = false; // reuse previous compute graphs when possible
 
     bool input_prefix_bos  = false; // prefix BOS to user inputs, preceding input_prefix
     bool use_mmap          = true;  // use mmap for faster loads

include/llama.h

@@ -374,6 +374,8 @@ extern "C" {
         bool swa_full;    // use full-size SWA cache (https://github.com/ggml-org/llama.cpp/pull/13194#issuecomment-2868343055)
                           // NOTE: setting to false when n_seq_max > 1 can cause bad performance in some cases
                           //       ref: https://github.com/ggml-org/llama.cpp/pull/13845#issuecomment-2924800573
+
+        bool graph_reuse; // reuse previous compute graphs when possible
     };
 
     // model quantization parameters
@@ -1429,6 +1431,7 @@ extern "C" {
 
         int32_t n_p_eval;
         int32_t n_eval;
+        int32_t n_reused;
     };
 
     struct llama_perf_sampler_data {

src/llama-batch.h

@@ -34,6 +34,31 @@ struct llama_ubatch {
     llama_seq_id *  seq_id_unq; // [n_seqs_unq]       | s   | seq_id
     int32_t      *  seq_idx;    // [LLAMA_MAX_SEQ]    | -   | seq_idx
     int8_t       *  output;     // [n_tokens]         | i   | -
+
+    bool is_same(const llama_ubatch & other) const {
+        bool res =
+            equal_seqs   == other.equal_seqs &&
+            n_tokens     == other.n_tokens &&
+            n_seq_tokens == other.n_seq_tokens &&
+            n_seqs       == other.n_seqs &&
+            n_seqs_unq   == other.n_seqs_unq &&
+            (
+                (!token && !other.token) ||
+                (!embd  && !other.embd)
+            );
+
+        if (!res) {
+            return false;
+        }
+
+        // TODO: this won't work because seq_id_unq ptr can point to an old balloc that has
+        //       been freed by this point. find a way to fix this
+        //for (uint32_t s = 0; s < n_seqs_unq; ++s) {
+        //    res &= seq_id_unq[s] == other.seq_id_unq[s];
+        //}
+
+        return res;
+    }
 };
 
 // a helper for sanitizing, fulfilling and splitting a batch

src/llama-context.cpp

@@ -101,7 +101,8 @@ llama_context::llama_context(
 
     cparams.n_ubatch = std::min(cparams.n_batch, params.n_ubatch == 0 ? params.n_batch : params.n_ubatch);
 
-    cparams.op_offload = params.op_offload;
+    cparams.op_offload  = params.op_offload;
+    cparams.graph_reuse = params.graph_reuse;
 
     const uint32_t n_ctx_per_seq = cparams.n_ctx / cparams.n_seq_max;
 
@@ -227,8 +228,8 @@ llama_context::llama_context(
 
         LLAMA_LOG_DEBUG("%s: max_nodes = %zu\n", __func__, max_nodes);
 
-        // buffer used to store the computation graph and the tensor meta data
-        buf_compute_meta.resize(ggml_tensor_overhead()*max_nodes + ggml_graph_overhead_custom(max_nodes, false));
+        gf_res_prev.reset(new llm_graph_result(max_nodes));
+        gf_res_reserve.reset(new llm_graph_result(max_nodes));
 
         // TODO: move these checks to ggml_backend_sched
         // enabling pipeline parallelism in the scheduler increases memory usage, so it is only done when necessary
@@ -388,10 +389,6 @@ ggml_backend_sched_t llama_context::get_sched() const {
     return sched.get();
 }
 
-ggml_context * llama_context::get_ctx_compute() const {
-    return ctx_compute.get();
-}
-
 uint32_t llama_context::n_ctx() const {
     return cparams.n_ctx;
 }
@@ -678,38 +675,52 @@ bool llama_context::apply_adapter_cvec(
     return cvec.apply(model, data, len, n_embd, il_start, il_end);
 }
 
-llm_graph_result_ptr llama_context::process_ubatch(const llama_ubatch & ubatch, llm_graph_type gtype, llama_memory_context_i * mctx, ggml_status & ret) {
+llm_graph_result_i * llama_context::process_ubatch(const llama_ubatch & ubatch, llm_graph_type gtype, llama_memory_context_i * mctx, ggml_status & ret) {
     if (mctx && !mctx->apply()) {
         LLAMA_LOG_ERROR("%s: failed to apply memory context\n", __func__);
         ret = GGML_STATUS_FAILED;
         return nullptr;
     }
 
-    auto * gf = graph_init();
-    if (!gf) {
-        LLAMA_LOG_ERROR("%s: failed to initialize graph\n", __func__);
-        ret = GGML_STATUS_FAILED;
-        return nullptr;
-    }
+    auto * res = gf_res_prev.get();
+    auto * gf  = res->get_gf();
 
-    auto res = graph_build(ctx_compute.get(), gf, ubatch, gtype, mctx);
-    if (!res) {
-        LLAMA_LOG_ERROR("%s: failed to build graph\n", __func__);
-        ret = GGML_STATUS_FAILED;
-        return nullptr;
-    }
+    // the new graph parameters
+    // in order to correctly reuse a graph, it's full topology has to be uniquely determined by these parameters
+    const auto gparams = graph_params(res, ubatch, mctx, gtype);
 
-    // LLAMA_LOG_INFO("graph build time: %.3f ms (%d nodes, %d leafs)\n", (ggml_time_us() - t_start_us)/1000.0, gf->n_nodes, gf->n_leafs);
+    const bool can_reuse = cparams.graph_reuse && res->update(gparams);
+    if (can_reuse) {
+        LLAMA_LOG_DEBUG("%s: reusing previous graph\n", __func__);
+        n_reused++;
+    } else {
+        res->reset();
 
-    if (!ggml_backend_sched_alloc_graph(sched.get(), gf)) {
-        LLAMA_LOG_ERROR("%s: failed to allocate graph\n", __func__);
-        ret = GGML_STATUS_ALLOC_FAILED;
-        return nullptr;
+        ggml_backend_sched_reset(sched.get());
+        ggml_backend_sched_set_eval_callback(sched.get(), cparams.cb_eval, cparams.cb_eval_user_data);
+
+        //const auto t_start_us = ggml_time_us();
+
+        gf = model.build_graph(gparams);
+
+        //LLAMA_LOG_INFO("graph build time: %.3f ms\n", (ggml_time_us() - t_start_us)/1000.0);
+
+        if (!gf) {
+            LLAMA_LOG_ERROR("%s: failed to initialize graph\n", __func__);
+            ret = GGML_STATUS_FAILED;
+            return nullptr;
+        }
+
+        if (!ggml_backend_sched_alloc_graph(sched.get(), gf)) {
+            LLAMA_LOG_ERROR("%s: failed to allocate graph\n", __func__);
+            ret = GGML_STATUS_ALLOC_FAILED;
+            return nullptr;
+        }
     }
 
     res->set_inputs(&ubatch);
 
-    const auto status = graph_compute(gf, ubatch.n_tokens > 1);
+    const auto status = graph_compute(res->get_gf(), ubatch.n_tokens > 1);
     if (status != GGML_STATUS_SUCCESS) {
         LLAMA_LOG_ERROR("%s: failed to compute graph, compute status: %d\n", __func__, status);
         ret = status;
@@ -767,9 +778,6 @@ int llama_context::encode(const llama_batch & batch_inp) {
 
     n_outputs = n_tokens;
 
-    ggml_backend_sched_reset(sched.get());
-    ggml_backend_sched_set_eval_callback(sched.get(), cparams.cb_eval, cparams.cb_eval_user_data);
-
     const auto causal_attn_org = cparams.causal_attn;
 
     // always use non-causal attention for encoder graphs
@@ -778,7 +786,7 @@ int llama_context::encode(const llama_batch & batch_inp) {
     cparams.causal_attn = false;
 
     ggml_status status;
-    const auto res = process_ubatch(ubatch, LLM_GRAPH_TYPE_ENCODER, nullptr, status);
+    const auto * res = process_ubatch(ubatch, LLM_GRAPH_TYPE_ENCODER, nullptr, status);
 
     cparams.causal_attn = causal_attn_org;
 
@@ -846,7 +854,9 @@ int llama_context::encode(const llama_batch & batch_inp) {
 
     // Reset state for the next token before backend sync, to allow the CPU activities in the reset to
     // overlap with device computation.
-    ggml_backend_sched_reset(sched.get());
+    if (!cparams.graph_reuse) {
+        ggml_backend_sched_reset(sched.get());
+    }
 
     // TODO: hacky solution
     if (model.arch == LLM_ARCH_T5 && t_embd) {
@@ -1005,11 +1015,8 @@ int llama_context::decode(const llama_batch & batch_inp) {
             n_outputs = n_outputs_new;
         }
 
-        ggml_backend_sched_reset(sched.get());
-        ggml_backend_sched_set_eval_callback(sched.get(), cparams.cb_eval, cparams.cb_eval_user_data);
-
         ggml_status status;
-        const auto res = process_ubatch(ubatch, LLM_GRAPH_TYPE_DECODER, mctx.get(), status);
+        const auto * res = process_ubatch(ubatch, LLM_GRAPH_TYPE_DECODER, mctx.get(), status);
 
         if (!res) {
             // the last ubatch failed or was aborted -> remove all positions of that ubatch from the KV cache
@@ -1192,7 +1199,9 @@ int llama_context::decode(const llama_batch & batch_inp) {
 
     // Reset state for the next token before backend sync, to allow the CPU activities in the reset to
     // overlap with device computation.
-    ggml_backend_sched_reset(sched.get());
+    if (!cparams.graph_reuse) {
+        ggml_backend_sched_reset(sched.get());
+    }
 
     return 0;
 }
@@ -1275,20 +1284,8 @@ uint32_t llama_context::output_reserve(int32_t n_outputs) {
 // graph
 //
 
-int32_t llama_context::graph_max_nodes() const {
-    return std::max<int32_t>(65536, 5*model.n_tensors());
-}
-
-ggml_cgraph * llama_context::graph_init() {
-    ggml_init_params params = {
-        /*.mem_size   =*/ buf_compute_meta.size(),
-        /*.mem_buffer =*/ buf_compute_meta.data(),
-        /*.no_alloc   =*/ true,
-    };
-
-    ctx_compute.reset(ggml_init(params));
-
-    return ggml_new_graph_custom(ctx_compute.get(), graph_max_nodes(), false);
+uint32_t llama_context::graph_max_nodes() const {
+    return std::max<uint32_t>(65536u, 5u*model.n_tensors());
 }
 
 ggml_cgraph * llama_context::graph_reserve(uint32_t n_tokens, uint32_t n_seqs, uint32_t n_outputs, const llama_memory_context_i * mctx) {
@@ -1301,6 +1298,9 @@ ggml_cgraph * llama_context::graph_reserve(uint32_t n_tokens, uint32_t n_seqs, u
         LLAMA_LOG_DEBUG("%s: making n_tokens a multiple of n_seqs - n_tokens = %u, n_seqs = %u, n_outputs = %u\n", __func__, n_tokens, n_seqs, n_outputs);
     }
 
+    gf_res_prev->reset();
+    ggml_backend_sched_reset(sched.get());
+
     // store the n_outputs as it is, and restore it afterwards
     // TODO: not sure if needed, might simplify in the future by removing this
     const auto save_n_outputs = this->n_outputs;
@@ -1310,17 +1310,15 @@ ggml_cgraph * llama_context::graph_reserve(uint32_t n_tokens, uint32_t n_seqs, u
     llama_batch_allocr balloc(model.hparams.n_pos_per_embd());
     llama_ubatch ubatch = balloc.ubatch_reserve(n_tokens/n_seqs, n_seqs);
 
-    auto * gf = graph_init();
-    auto res = graph_build(ctx_compute.get(), gf, ubatch, LLM_GRAPH_TYPE_DEFAULT, mctx);
+    auto * res = gf_res_reserve.get();
 
-    this->n_outputs = save_n_outputs;
+    const auto gparams = graph_params(res, ubatch, mctx, LLM_GRAPH_TYPE_DEFAULT);
 
-    if (!res) {
-        LLAMA_LOG_ERROR("%s: failed to build worst-case graph\n", __func__);
-        return nullptr;
-    }
+    res->reset();
 
-    ggml_backend_sched_reset(sched.get());
+    auto * gf = model.build_graph(gparams);
+
+    this->n_outputs = save_n_outputs;
 
     // initialize scheduler with the specified graph
     if (!ggml_backend_sched_reserve(sched.get(), gf)) {
@@ -1331,28 +1329,27 @@ ggml_cgraph * llama_context::graph_reserve(uint32_t n_tokens, uint32_t n_seqs, u
     return gf;
 }
 
-llm_graph_result_ptr llama_context::graph_build(
-                      ggml_context * ctx,
-                       ggml_cgraph * gf,
-                const llama_ubatch & ubatch,
-                    llm_graph_type   gtype,
-      const llama_memory_context_i * mctx) {
-    return model.build_graph(
-            {
-                /*.ctx         =*/ ctx,
-                /*.arch        =*/ model.arch,
-                /*.hparams     =*/ model.hparams,
-                /*.cparams     =*/ cparams,
-                /*.ubatch      =*/ ubatch,
-                /*.sched       =*/ sched.get(),
-                /*.backend_cpu =*/ backend_cpu,
-                /*.cvec        =*/ &cvec,
-                /*.loras       =*/ &loras,
-                /*.mctx        =*/ mctx,
-                /*.cross       =*/ &cross,
-                /*.n_outputs   =*/ n_outputs,
-                /*.cb          =*/ graph_get_cb(),
-            }, gf, gtype);
+llm_graph_params llama_context::graph_params(
+                      llm_graph_result_i * res,
+                      const llama_ubatch & ubatch,
+            const llama_memory_context_i * mctx,
+            llm_graph_type   gtype) const {
+    return {
+        /*.arch        =*/ model.arch,
+        /*.hparams     =*/ model.hparams,
+        /*.cparams     =*/ cparams,
+        /*.ubatch      =*/ ubatch,
+        /*.gtype       =*/ gtype,
+        /*.sched       =*/ sched.get(),
+        /*.backend_cpu =*/ backend_cpu,
+        /*.cvec        =*/ &cvec,
+        /*.loras       =*/ &loras,
+        /*.mctx        =*/ mctx,
+        /*.cross       =*/ &cross,
+        /*.n_outputs   =*/ n_outputs,
+        /*.cb          =*/ graph_get_cb(),
+        /*.res         =*/ res,
+    };
 }
 
 ggml_status llama_context::graph_compute(
@@ -1930,6 +1927,7 @@ llama_perf_context_data llama_context::perf_get_data() const {
     data.t_eval_ms   = 1e-3 * t_eval_us;
     data.n_p_eval    = std::max(1, n_p_eval);
     data.n_eval      = std::max(1, n_eval);
+    data.n_reused    = std::max(0, n_reused);
 
     return data;
 }
@@ -1938,6 +1936,7 @@ void llama_context::perf_reset() {
     t_start_us  = ggml_time_us();
     t_eval_us   = n_eval = 0;
     t_p_eval_us = n_p_eval = 0;
+    n_reused    = 0;
 }
 
 //
@@ -2064,8 +2063,13 @@ void llama_context::opt_epoch_iter(
                 break;
             }
 
-            auto * gf = graph_init();
-            auto res = graph_build(ctx_compute.get(), gf, ubatch, LLM_GRAPH_TYPE_DEFAULT, mctx.get());
+            auto * res = gf_res_prev.get();
+
+            const auto gparams = graph_params(res, ubatch, mctx.get(), LLM_GRAPH_TYPE_DEFAULT);
+
+            res->reset();
+
+            auto * gf = model.build_graph(gparams);
 
             struct ggml_context * ctx_compute_opt;
             {
@@ -2187,6 +2191,7 @@ llama_context_params llama_context_default_params() {
         /*.no_perf                     =*/ true,
         /*.op_offload                  =*/ true,
         /*.swa_full                    =*/ true,
+        /*.graph_reuse                 =*/ false,
     };
 
     return result;
@@ -2807,6 +2812,7 @@ void llama_perf_context_print(const llama_context * ctx) {
     LLAMA_LOG_INFO("%s:        eval time = %10.2f ms / %5d runs   (%8.2f ms per token, %8.2f tokens per second)\n",
             __func__, data.t_eval_ms, data.n_eval, data.t_eval_ms / data.n_eval, 1e3 / data.t_eval_ms * data.n_eval);
     LLAMA_LOG_INFO("%s:       total time = %10.2f ms / %5d tokens\n", __func__, (t_end_ms - data.t_start_ms), (data.n_p_eval + data.n_eval));
+    LLAMA_LOG_INFO("%s:    graphs reused = %10d\n", __func__, data.n_reused);
 }
 
 void llama_perf_context_reset(llama_context * ctx) {