fix potential divide by zero fault when bf16s / fp16s enabled, fix #2125

src/command.cpp

@@ -2513,7 +2513,7 @@ void VkTransfer::record_upload(const Mat& src, VkMat& dst, const Option& opt, bo
     //     NCNN_LOGE("record_upload src = %d | %d %d %d @ %d", src.dims, src.w, src.h, src.c, src.elempack);
 
     // NOTE keep the hack here ?
-    if (src.elemsize == src.elempack * 4u)
+    if (src.elembits() == 32)
     {
         if (opt.use_fp16_storage || (opt.use_fp16_packed && src.elempack % 4 == 0))
         {

src/net.cpp

@@ -1150,13 +1150,13 @@ int Net::forward_layer(int layer_index, std::vector<Mat>& blob_mats, const Optio
 #if NCNN_ARM82
         if (opt.use_fp16_storage && cpu_support_arm_asimdhp())
         {
-            if (bottom_blob.elemsize / bottom_blob.elempack == 4u && layer->support_fp16_storage)
+            if (bottom_blob.elembits() == 32 && layer->support_fp16_storage)
             {
                 Mat bottom_blob_fp16;
                 cast_float32_to_float16(bottom_blob, bottom_blob_fp16, opt);
                 bottom_blob = bottom_blob_fp16;
             }
-            if (bottom_blob.elemsize / bottom_blob.elempack == 2u && !layer->support_fp16_storage)
+            if (bottom_blob.elembits() == 16 && !layer->support_fp16_storage)
             {
                 Mat bottom_blob_fp32;
                 cast_float16_to_float32(bottom_blob, bottom_blob_fp32, opt);
@@ -1167,13 +1167,13 @@ int Net::forward_layer(int layer_index, std::vector<Mat>& blob_mats, const Optio
 #endif // NCNN_ARM82
         if (opt.use_bf16_storage)
         {
-            if (bottom_blob.elemsize / bottom_blob.elempack == 4u && layer->support_bf16_storage)
+            if (bottom_blob.elembits() == 32 && layer->support_bf16_storage)
             {
                 Mat bottom_blob_bf16;
                 cast_float32_to_bfloat16(bottom_blob, bottom_blob_bf16, opt);
                 bottom_blob = bottom_blob_bf16;
             }
-            if (bottom_blob.elemsize / bottom_blob.elempack == 2u && !layer->support_bf16_storage)
+            if (bottom_blob.elembits() == 16 && !layer->support_bf16_storage)
             {
                 Mat bottom_blob_fp32;
                 cast_bfloat16_to_float32(bottom_blob, bottom_blob_fp32, opt);
@@ -1283,13 +1283,13 @@ int Net::forward_layer(int layer_index, std::vector<Mat>& blob_mats, const Optio
 #if NCNN_ARM82
             if (opt.use_fp16_storage && cpu_support_arm_asimdhp())
             {
-                if (bottom_blobs[i].elemsize / bottom_blobs[i].elempack == 4u && layer->support_fp16_storage)
+                if (bottom_blobs[i].elembits() == 32 && layer->support_fp16_storage)
                 {
                     Mat bottom_blob_fp16;
                     cast_float32_to_float16(bottom_blobs[i], bottom_blob_fp16, opt);
                     bottom_blobs[i] = bottom_blob_fp16;
                 }
-                if (bottom_blobs[i].elemsize / bottom_blobs[i].elempack == 2u && !layer->support_fp16_storage)
+                if (bottom_blobs[i].elembits() == 16 && !layer->support_fp16_storage)
                 {
                     Mat bottom_blob_fp32;
                     cast_float16_to_float32(bottom_blobs[i], bottom_blob_fp32, opt);
@@ -1300,13 +1300,13 @@ int Net::forward_layer(int layer_index, std::vector<Mat>& blob_mats, const Optio
 #endif // NCNN_ARM82
             if (opt.use_bf16_storage)
             {
-                if (bottom_blobs[i].elemsize / bottom_blobs[i].elempack == 4u && layer->support_bf16_storage)
+                if (bottom_blobs[i].elembits() == 32 && layer->support_bf16_storage)
                 {
                     Mat bottom_blob_bf16;
                     cast_float32_to_bfloat16(bottom_blobs[i], bottom_blob_bf16, opt);
                     bottom_blobs[i] = bottom_blob_bf16;
                 }
-                if (bottom_blobs[i].elemsize / bottom_blobs[i].elempack == 2u && !layer->support_bf16_storage)
+                if (bottom_blobs[i].elembits() == 16 && !layer->support_bf16_storage)
                 {
                     Mat bottom_blob_fp32;
                     cast_bfloat16_to_float32(bottom_blobs[i], bottom_blob_fp32, opt);
@@ -2779,7 +2779,7 @@ int Extractor::extract(int blob_index, Mat& feat)
 #if NCNN_ARM82
     if (opt.use_fp16_storage && cpu_support_arm_asimdhp())
     {
-        if (feat.elemsize / feat.elempack == 2u)
+        if (feat.elembits() == 16)
         {
             Mat feat_fp32;
             cast_float16_to_float32(feat, feat_fp32, opt);
@@ -2790,7 +2790,7 @@ int Extractor::extract(int blob_index, Mat& feat)
 #endif // NCNN_ARM82
     if (opt.use_bf16_storage)
     {
-        if (feat.elemsize / feat.elempack == 2u)
+        if (feat.elembits() == 16)
         {
             Mat feat_fp32;
             cast_bfloat16_to_float32(feat, feat_fp32, opt);