mtmd: fit_params now take into account mmproj (ggml-org#21489)

* mtmd: fit_params now take into account mmproj

* rename alloc_compute_meta to reserve_compute_meta

* rm unused functions

* add ggml_backend_dev_t support

* add debug log

Author: ngxson

tools/mtmd/clip.cpp

@@ -162,8 +162,14 @@ struct clip_ctx {
 
     bool debug_output_embeddings = false;
 
+    // for measuring memory usage
+    bool no_alloc = false;
+    std::map<ggml_backend_dev_t, size_t> mem_usage;
+    std::map<ggml_backend_dev_t, size_t> mem_compute;
+
     clip_ctx(clip_context_params & ctx_params) {
         flash_attn_type = ctx_params.flash_attn_type;
+        no_alloc = ctx_params.no_alloc;
         backend_cpu = ggml_backend_init_by_type(GGML_BACKEND_DEVICE_TYPE_CPU, nullptr);
         if (!backend_cpu) {
             throw std::runtime_error("failed to initialize CPU backend");
@@ -1688,6 +1694,8 @@ struct clip_model_loader {
                 ggml_set_name(data_tensor, cur->name);
                 loaded_tensor_names.insert(name);
                 cur = data_tensor;
+                // add to weight memory counter
+                ctx_clip.mem_usage[ggml_backend_get_device(ctx_clip.backend)] += ggml_nbytes(cur);
             }
             return cur;
         };
@@ -2602,7 +2610,7 @@ struct clip_model_loader {
         }
 
         // load data
-        {
+        if (!ctx_clip.no_alloc) {
             std::vector<uint8_t> read_buf;
 
             // alloc memory and offload data
@@ -2676,7 +2684,7 @@ struct clip_model_loader {
         if (ctx_clip.flash_attn_type == CLIP_FLASH_ATTN_TYPE_AUTO) {
             // try to enable flash attention to see if it's supported
             ctx_clip.flash_attn_type = CLIP_FLASH_ATTN_TYPE_ENABLED;
-            info = alloc_compute_meta(ctx_clip, batch);
+            info = reserve_compute_meta(ctx_clip, batch);
             if (!info.fattn && info.fattn_op) {
                 auto op = info.fattn_op;
                 LOG_WRN("%s: *****************************************************************\n", __func__);
@@ -2695,10 +2703,10 @@ struct clip_model_loader {
                 LOG_WRN("%s: please report this on github as an issue\n", __func__);
                 LOG_WRN("%s: *****************************************************************\n", __func__);
                 ctx_clip.flash_attn_type = CLIP_FLASH_ATTN_TYPE_DISABLED;
-                alloc_compute_meta(ctx_clip, batch);
+                reserve_compute_meta(ctx_clip, batch);
             }
         } else {
-            info = alloc_compute_meta(ctx_clip, batch);
+            info = reserve_compute_meta(ctx_clip, batch);
             if (!info.fattn && ctx_clip.flash_attn_type == CLIP_FLASH_ATTN_TYPE_ENABLED) {
                 LOG_WRN("%s: flash attention is not supported by the current backend; falling back to CPU (performance will be degraded)\n", __func__);
             }
@@ -2737,12 +2745,14 @@ struct clip_model_loader {
         }
     }
 
-    static support_info_graph alloc_compute_meta(clip_ctx & ctx_clip, const clip_image_f32_batch & batch) {
+    // only initialize backend buffers, but do not allocate them yet
+    static support_info_graph reserve_compute_meta(clip_ctx & ctx_clip, const clip_image_f32_batch & batch) {
         ctx_clip.buf_compute_meta.resize(ctx_clip.max_nodes * ggml_tensor_overhead() + ggml_graph_overhead());
 
         ggml_cgraph * gf = clip_image_build_graph(&ctx_clip, batch);
         ggml_backend_sched_reserve(ctx_clip.sched.get(), gf);
 
+        ctx_clip.mem_compute.clear();
         for (size_t i = 0; i < ctx_clip.backend_ptrs.size(); ++i) {
             ggml_backend_t backend = ctx_clip.backend_ptrs[i];
             ggml_backend_buffer_type_t buft = ctx_clip.backend_buft[i];
@@ -2752,6 +2762,7 @@ struct clip_model_loader {
                         ggml_backend_buft_name(buft),
                         size / 1024.0 / 1024.0);
             }
+            ctx_clip.mem_compute[ggml_backend_get_device(backend)] += size;
         }
 
         const int n_splits = ggml_backend_sched_get_n_splits(ctx_clip.sched.get());
@@ -4266,22 +4277,6 @@ int clip_n_mmproj_embd(const struct clip_ctx * ctx) {
     }
 }
 
-int clip_is_minicpmv(const struct clip_ctx * ctx) {
-    // TODO: remove this function
-    if (ctx->proj_type() == PROJECTOR_TYPE_MINICPMV) {
-        return ctx->model.hparams.minicpmv_version;
-    }
-    if (ctx->proj_type() == PROJECTOR_TYPE_MINICPMV4_6) {
-        return 46;
-    }
-    return 0;
-}
-
-bool clip_is_glm(const struct clip_ctx * ctx) {
-    // TODO: remove this function
-    return ctx->proj_type() == PROJECTOR_TYPE_GLM_EDGE;
-}
-
 bool clip_is_llava(const struct clip_ctx * ctx) {
     return ctx->model.hparams.has_llava_projector;
 }
@@ -4330,6 +4325,14 @@ const clip_hparams * clip_get_hparams(const struct clip_ctx * ctx) {
     return &ctx->model.hparams;
 }
 
+std::map<ggml_backend_dev_t, size_t> clip_get_mem_usage(const struct clip_ctx * ctx) {
+    std::map<ggml_backend_dev_t, size_t> result = ctx->mem_usage;
+    for (auto & [dev, size] : ctx->mem_compute) {
+        result[dev] += size;
+    }
+    return result;
+}
+
 //
 // API for debugging
 //

tools/mtmd/clip.h

@@ -6,6 +6,8 @@
 #include <stddef.h>
 #include <stdint.h>
 
+#include <map>
+
 // !!! Internal header, to be used by mtmd only !!!
 
 #define MTMD_INTERNAL_HEADER
@@ -40,6 +42,7 @@ struct clip_context_params {
     bool warmup;
     ggml_backend_sched_eval_callback cb_eval;
     void * cb_eval_user_data;
+    bool no_alloc;
 };
 
 struct clip_init_result {
@@ -102,8 +105,6 @@ struct ggml_tensor * clip_get_newline_tensor(const struct clip_ctx * ctx);
 bool clip_image_encode      (struct clip_ctx * ctx, int n_threads, struct clip_image_f32 * img, float * vec);
 bool clip_image_batch_encode(struct clip_ctx * ctx, int n_threads, const struct clip_image_f32_batch * imgs, float * vec);
 
-int clip_is_minicpmv(const struct clip_ctx * ctx);
-bool clip_is_glm(const struct clip_ctx * ctx);
 bool clip_is_llava(const struct clip_ctx * ctx);
 // note for contributor: this clip_is_(model) pattern is deprecated
 //                       do NOT add new functions like this
@@ -116,6 +117,8 @@ void clip_image_f32_batch_add_mel(struct clip_image_f32_batch * batch, int n_mel
 bool clip_has_vision_encoder(const struct clip_ctx * ctx);
 bool clip_has_audio_encoder(const struct clip_ctx * ctx);
 
+std::map<ggml_backend_dev_t, size_t> clip_get_mem_usage(const struct clip_ctx * ctx);
+
 struct clip_cap {
     bool has_vision;
     bool has_audio;