Merge "Update Android.bp following XNNPACK rebase"
diff --git a/Android.bp b/Android.bp
index 6f514e2..ccdd1d6 100644
--- a/Android.bp
+++ b/Android.bp
@@ -45,8 +45,8 @@
     "src/f32-argmaxpool/4x-scalar-c1.c",
     "src/f32-argmaxpool/9p8x-scalar-c1.c",
     "src/f32-argmaxpool/9x-scalar-c1.c",
-    "src/f32-avgpool/9p8x-scalar-c1.c",
-    "src/f32-avgpool/9x-scalar-c1.c",
+    "src/f32-avgpool/9p8x-minmax-scalar-c1.c",
+    "src/f32-avgpool/9x-minmax-scalar-c1.c",
     "src/f32-clamp/gen/scalar-x1.c",
     "src/f32-clamp/gen/scalar-x2.c",
     "src/f32-clamp/gen/scalar-x4.c",
@@ -57,28 +57,44 @@
     "src/f32-dwconv-spchw/3x3s2p1-scalar.c",
     "src/f32-dwconv-spchw/5x5p2-scalar.c",
     "src/f32-dwconv-spchw/5x5s2p2-scalar.c",
-    "src/f32-dwconv/gen/up1x25-scalar-acc2.c",
-    "src/f32-dwconv/gen/up1x25-scalar.c",
     "src/f32-dwconv/gen/up1x4-scalar-acc2.c",
     "src/f32-dwconv/gen/up1x4-scalar.c",
     "src/f32-dwconv/gen/up1x9-scalar-acc2.c",
     "src/f32-dwconv/gen/up1x9-scalar.c",
-    "src/f32-dwconv/gen/up2x25-scalar-acc2.c",
-    "src/f32-dwconv/gen/up2x25-scalar.c",
+    "src/f32-dwconv/gen/up1x25-scalar-acc2.c",
+    "src/f32-dwconv/gen/up1x25-scalar.c",
     "src/f32-dwconv/gen/up2x4-scalar-acc2.c",
     "src/f32-dwconv/gen/up2x4-scalar.c",
     "src/f32-dwconv/gen/up2x9-scalar-acc2.c",
     "src/f32-dwconv/gen/up2x9-scalar.c",
+    "src/f32-dwconv/gen/up2x25-scalar-acc2.c",
+    "src/f32-dwconv/gen/up2x25-scalar.c",
+    "src/f32-dwconv/gen/up1x4-minmax-scalar-acc2.c",
+    "src/f32-dwconv/gen/up1x4-minmax-scalar.c",
+    "src/f32-dwconv/gen/up1x9-minmax-scalar-acc2.c",
+    "src/f32-dwconv/gen/up1x9-minmax-scalar.c",
+    "src/f32-dwconv/gen/up1x25-minmax-scalar-acc2.c",
+    "src/f32-dwconv/gen/up1x25-minmax-scalar.c",
+    "src/f32-dwconv/gen/up2x4-minmax-scalar-acc2.c",
+    "src/f32-dwconv/gen/up2x4-minmax-scalar.c",
+    "src/f32-dwconv/gen/up2x9-minmax-scalar-acc2.c",
+    "src/f32-dwconv/gen/up2x9-minmax-scalar.c",
+    "src/f32-dwconv/gen/up2x25-minmax-scalar-acc2.c",
+    "src/f32-dwconv/gen/up2x25-minmax-scalar.c",
     "src/f32-gavgpool-spchw/scalar-x1.c",
-    "src/f32-gavgpool/7p7x-scalar-c1.c",
-    "src/f32-gavgpool/7x-scalar-c1.c",
-    "src/f32-gemm/gen-inc/1x4-scalar.c",
-    "src/f32-gemm/gen-inc/2x4-scalar.c",
-    "src/f32-gemm/gen-inc/4x4-scalar.c",
+    "src/f32-gavgpool/7p7x-minmax-scalar-c1.c",
+    "src/f32-gavgpool/7x-minmax-scalar-c1.c",
+    "src/f32-gemm/gen-inc/1x4inc-minmax-scalar.c",
+    "src/f32-gemm/gen-inc/2x4inc-minmax-scalar.c",
+    "src/f32-gemm/gen-inc/4x4inc-minmax-scalar.c",
     "src/f32-gemm/gen/1x4-scalar.c",
     "src/f32-gemm/gen/2x4-scalar.c",
     "src/f32-gemm/gen/4x2-scalar.c",
     "src/f32-gemm/gen/4x4-scalar.c",
+    "src/f32-gemm/gen/1x4-minmax-scalar.c",
+    "src/f32-gemm/gen/2x4-minmax-scalar.c",
+    "src/f32-gemm/gen/4x2-minmax-scalar.c",
+    "src/f32-gemm/gen/4x4-minmax-scalar.c",
     "src/f32-hswish/gen/scalar-x1.c",
     "src/f32-hswish/gen/scalar-x2.c",
     "src/f32-hswish/gen/scalar-x4.c",
@@ -89,13 +105,17 @@
     "src/f32-igemm/gen/2x4-scalar.c",
     "src/f32-igemm/gen/4x2-scalar.c",
     "src/f32-igemm/gen/4x4-scalar.c",
-    "src/f32-maxpool/9p8x-scalar-c1.c",
-    "src/f32-pavgpool/9p8x-scalar-c1.c",
-    "src/f32-pavgpool/9x-scalar-c1.c",
-    "src/f32-ppmm/gen/2x4-scalar.c",
-    "src/f32-ppmm/gen/3x3-scalar.c",
-    "src/f32-ppmm/gen/4x2-scalar.c",
-    "src/f32-ppmm/gen/4x4-scalar.c",
+    "src/f32-igemm/gen/1x4-minmax-scalar.c",
+    "src/f32-igemm/gen/2x4-minmax-scalar.c",
+    "src/f32-igemm/gen/4x2-minmax-scalar.c",
+    "src/f32-igemm/gen/4x4-minmax-scalar.c",
+    "src/f32-maxpool/9p8x-minmax-scalar-c1.c",
+    "src/f32-pavgpool/9p8x-minmax-scalar-c1.c",
+    "src/f32-pavgpool/9x-minmax-scalar-c1.c",
+    "src/f32-ppmm/gen/2x4-minmax-scalar.c",
+    "src/f32-ppmm/gen/3x3-minmax-scalar.c",
+    "src/f32-ppmm/gen/4x2-minmax-scalar.c",
+    "src/f32-ppmm/gen/4x4-minmax-scalar.c",
     "src/f32-prelu/gen/scalar-2x1.c",
     "src/f32-prelu/gen/scalar-2x4.c",
     "src/f32-raddstoreexpminusmax/gen/scalar-lut64-p2-x1.c",
@@ -120,28 +140,28 @@
     "src/f32-sigmoid/gen/scalar-p5-div-x1.c",
     "src/f32-sigmoid/gen/scalar-p5-div-x2.c",
     "src/f32-sigmoid/gen/scalar-p5-div-x4.c",
-    "src/f32-spmm/gen/1x1-scalar-pipelined.c",
-    "src/f32-spmm/gen/1x1-scalar.c",
-    "src/f32-spmm/gen/2x1-scalar-pipelined.c",
-    "src/f32-spmm/gen/2x1-scalar.c",
-    "src/f32-spmm/gen/4x1-scalar-pipelined.c",
-    "src/f32-spmm/gen/4x1-scalar.c",
-    "src/f32-spmm/gen/8x1-scalar-pipelined.c",
-    "src/f32-spmm/gen/8x1-scalar.c",
-    "src/f32-spmm/gen/8x2-scalar.c",
-    "src/f32-spmm/gen/8x4-scalar.c",
-    "src/f32-vbinary/gen/vadd-scalar-x1.c",
-    "src/f32-vbinary/gen/vadd-scalar-x2.c",
-    "src/f32-vbinary/gen/vadd-scalar-x4.c",
-    "src/f32-vbinary/gen/vaddc-scalar-x1.c",
-    "src/f32-vbinary/gen/vaddc-scalar-x2.c",
-    "src/f32-vbinary/gen/vaddc-scalar-x4.c",
-    "src/f32-vbinary/gen/vdiv-scalar-x1.c",
-    "src/f32-vbinary/gen/vdiv-scalar-x2.c",
-    "src/f32-vbinary/gen/vdiv-scalar-x4.c",
-    "src/f32-vbinary/gen/vdivc-scalar-x1.c",
-    "src/f32-vbinary/gen/vdivc-scalar-x2.c",
-    "src/f32-vbinary/gen/vdivc-scalar-x4.c",
+    "src/f32-spmm/gen/1x1-minmax-scalar-pipelined.c",
+    "src/f32-spmm/gen/1x1-minmax-scalar.c",
+    "src/f32-spmm/gen/2x1-minmax-scalar-pipelined.c",
+    "src/f32-spmm/gen/2x1-minmax-scalar.c",
+    "src/f32-spmm/gen/4x1-minmax-scalar-pipelined.c",
+    "src/f32-spmm/gen/4x1-minmax-scalar.c",
+    "src/f32-spmm/gen/8x1-minmax-scalar-pipelined.c",
+    "src/f32-spmm/gen/8x1-minmax-scalar.c",
+    "src/f32-spmm/gen/8x2-minmax-scalar.c",
+    "src/f32-spmm/gen/8x4-minmax-scalar.c",
+    "src/f32-vbinary/gen/vadd-minmax-scalar-x1.c",
+    "src/f32-vbinary/gen/vadd-minmax-scalar-x2.c",
+    "src/f32-vbinary/gen/vadd-minmax-scalar-x4.c",
+    "src/f32-vbinary/gen/vaddc-minmax-scalar-x1.c",
+    "src/f32-vbinary/gen/vaddc-minmax-scalar-x2.c",
+    "src/f32-vbinary/gen/vaddc-minmax-scalar-x4.c",
+    "src/f32-vbinary/gen/vdiv-minmax-scalar-x1.c",
+    "src/f32-vbinary/gen/vdiv-minmax-scalar-x2.c",
+    "src/f32-vbinary/gen/vdiv-minmax-scalar-x4.c",
+    "src/f32-vbinary/gen/vdivc-minmax-scalar-x1.c",
+    "src/f32-vbinary/gen/vdivc-minmax-scalar-x2.c",
+    "src/f32-vbinary/gen/vdivc-minmax-scalar-x4.c",
     "src/f32-vbinary/gen/vmax-scalar-x1.c",
     "src/f32-vbinary/gen/vmax-scalar-x2.c",
     "src/f32-vbinary/gen/vmax-scalar-x4.c",
@@ -154,44 +174,44 @@
     "src/f32-vbinary/gen/vminc-scalar-x1.c",
     "src/f32-vbinary/gen/vminc-scalar-x2.c",
     "src/f32-vbinary/gen/vminc-scalar-x4.c",
-    "src/f32-vbinary/gen/vmul-scalar-x1.c",
-    "src/f32-vbinary/gen/vmul-scalar-x2.c",
-    "src/f32-vbinary/gen/vmul-scalar-x4.c",
-    "src/f32-vbinary/gen/vmulc-scalar-x1.c",
-    "src/f32-vbinary/gen/vmulc-scalar-x2.c",
-    "src/f32-vbinary/gen/vmulc-scalar-x4.c",
-    "src/f32-vbinary/gen/vrdivc-scalar-x1.c",
-    "src/f32-vbinary/gen/vrdivc-scalar-x2.c",
-    "src/f32-vbinary/gen/vrdivc-scalar-x4.c",
-    "src/f32-vbinary/gen/vrsubc-scalar-x1.c",
-    "src/f32-vbinary/gen/vrsubc-scalar-x2.c",
-    "src/f32-vbinary/gen/vrsubc-scalar-x4.c",
-    "src/f32-vbinary/gen/vsub-scalar-x1.c",
-    "src/f32-vbinary/gen/vsub-scalar-x2.c",
-    "src/f32-vbinary/gen/vsub-scalar-x4.c",
-    "src/f32-vbinary/gen/vsubc-scalar-x1.c",
-    "src/f32-vbinary/gen/vsubc-scalar-x2.c",
-    "src/f32-vbinary/gen/vsubc-scalar-x4.c",
-    "src/f32-vmulcaddc/gen/c1-scalar-2x.c",
-    "src/f32-vmulcaddc/gen/c2-scalar-2x.c",
-    "src/f32-vmulcaddc/gen/c4-scalar-2x.c",
+    "src/f32-vbinary/gen/vmul-minmax-scalar-x1.c",
+    "src/f32-vbinary/gen/vmul-minmax-scalar-x2.c",
+    "src/f32-vbinary/gen/vmul-minmax-scalar-x4.c",
+    "src/f32-vbinary/gen/vmulc-minmax-scalar-x1.c",
+    "src/f32-vbinary/gen/vmulc-minmax-scalar-x2.c",
+    "src/f32-vbinary/gen/vmulc-minmax-scalar-x4.c",
+    "src/f32-vbinary/gen/vrdivc-minmax-scalar-x1.c",
+    "src/f32-vbinary/gen/vrdivc-minmax-scalar-x2.c",
+    "src/f32-vbinary/gen/vrdivc-minmax-scalar-x4.c",
+    "src/f32-vbinary/gen/vrsubc-minmax-scalar-x1.c",
+    "src/f32-vbinary/gen/vrsubc-minmax-scalar-x2.c",
+    "src/f32-vbinary/gen/vrsubc-minmax-scalar-x4.c",
+    "src/f32-vbinary/gen/vsub-minmax-scalar-x1.c",
+    "src/f32-vbinary/gen/vsub-minmax-scalar-x2.c",
+    "src/f32-vbinary/gen/vsub-minmax-scalar-x4.c",
+    "src/f32-vbinary/gen/vsubc-minmax-scalar-x1.c",
+    "src/f32-vbinary/gen/vsubc-minmax-scalar-x2.c",
+    "src/f32-vbinary/gen/vsubc-minmax-scalar-x4.c",
+    "src/f32-vmulcaddc/gen/c1-minmax-scalar-2x.c",
+    "src/f32-vmulcaddc/gen/c2-minmax-scalar-2x.c",
+    "src/f32-vmulcaddc/gen/c4-minmax-scalar-2x.c",
     "src/math/expminus-scalar-lut2048-p1.c",
     "src/math/expminus-scalar-lut64-p2.c",
     "src/math/expminus-scalar-p5.c",
     "src/math/sigmoid-scalar-lut2048-p1-div.c",
     "src/math/sigmoid-scalar-lut64-p2-div.c",
     "src/math/sigmoid-scalar-p5-div.c",
-    "src/q8-avgpool/9p8x-scalar-c1.c",
-    "src/q8-avgpool/9x-scalar-c1.c",
-    "src/q8-dwconv/up1x9-scalar.c",
-    "src/q8-gavgpool/7p7x-scalar-c1.c",
-    "src/q8-gavgpool/7x-scalar-c1.c",
-    "src/q8-gemm/2x2-scalar.c",
-    "src/q8-igemm/2x2-scalar.c",
-    "src/q8-vadd/scalar.c",
+    "src/q8-avgpool/9p8x-minmax-scalar-c1.c",
+    "src/q8-avgpool/9x-minmax-scalar-c1.c",
+    "src/q8-dwconv/up1x9-minmax-scalar.c",
+    "src/q8-gavgpool/7p7x-minmax-scalar-c1.c",
+    "src/q8-gavgpool/7x-minmax-scalar-c1.c",
+    "src/q8-gemm/2x2-minmax-scalar.c",
+    "src/q8-igemm/2x2-minmax-scalar.c",
+    "src/q8-vadd/minmax-scalar.c",
     "src/u8-clamp/scalar-x4.c",
     "src/u8-lut32norm/scalar.c",
-    "src/u8-maxpool/9p8x-scalar-c1.c",
+    "src/u8-maxpool/9p8x-minmax-scalar-c1.c",
     "src/u8-rmax/scalar.c",
     "src/x32-packx/x2-scalar.c",
     "src/x32-packx/x3-scalar.c",
@@ -217,72 +237,72 @@
     "src/f32-argmaxpool/4x-psimd-c4.c",
     "src/f32-argmaxpool/9p8x-psimd-c4.c",
     "src/f32-argmaxpool/9x-psimd-c4.c",
-    "src/f32-avgpool/9p8x-psimd-c4.c",
-    "src/f32-avgpool/9x-psimd-c4.c",
+    "src/f32-avgpool/9p8x-minmax-psimd-c4.c",
+    "src/f32-avgpool/9x-minmax-psimd-c4.c",
     "src/f32-clamp/gen/psimd-x4.c",
     "src/f32-clamp/gen/psimd-x8.c",
-    "src/f32-dwconv/gen/up4x25-psimd-acc2.c",
-    "src/f32-dwconv/gen/up4x25-psimd.c",
-    "src/f32-dwconv/gen/up4x4-psimd-acc2.c",
-    "src/f32-dwconv/gen/up4x4-psimd.c",
-    "src/f32-dwconv/gen/up4x9-psimd-acc2.c",
-    "src/f32-dwconv/gen/up4x9-psimd.c",
-    "src/f32-dwconv/gen/up8x25-psimd-acc2.c",
-    "src/f32-dwconv/gen/up8x25-psimd.c",
-    "src/f32-dwconv/gen/up8x4-psimd-acc2.c",
-    "src/f32-dwconv/gen/up8x4-psimd.c",
-    "src/f32-dwconv/gen/up8x9-psimd-acc2.c",
-    "src/f32-dwconv/gen/up8x9-psimd.c",
-    "src/f32-gavgpool/7p7x-psimd-c4.c",
-    "src/f32-gavgpool/7x-psimd-c4.c",
-    "src/f32-gemm/gen/1x8-psimd-loadsplat.c",
-    "src/f32-gemm/gen/1x8-psimd-splat.c",
-    "src/f32-gemm/gen/1x8s4-psimd.c",
-    "src/f32-gemm/gen/4x2c4-psimd.c",
-    "src/f32-gemm/gen/4x8-psimd-loadsplat.c",
-    "src/f32-gemm/gen/4x8-psimd-splat.c",
-    "src/f32-gemm/gen/4x8s4-psimd.c",
-    "src/f32-gemm/gen/6x8-psimd-loadsplat.c",
-    "src/f32-gemm/gen/6x8-psimd-splat.c",
-    "src/f32-gemm/gen/6x8s4-psimd.c",
-    "src/f32-gemm/gen-inc/1x8-psimd-loadsplat.c",
-    "src/f32-gemm/gen-inc/1x8-psimd-splat.c",
-    "src/f32-gemm/gen-inc/1x8s4-psimd.c",
-    "src/f32-gemm/gen-inc/4x8-psimd-loadsplat.c",
-    "src/f32-gemm/gen-inc/4x8-psimd-splat.c",
-    "src/f32-gemm/gen-inc/4x8s4-psimd.c",
-    "src/f32-gemm/gen-inc/6x8-psimd-loadsplat.c",
-    "src/f32-gemm/gen-inc/6x8-psimd-splat.c",
-    "src/f32-gemm/gen-inc/6x8s4-psimd.c",
+    "src/f32-dwconv/gen/up4x25-minmax-psimd-acc2.c",
+    "src/f32-dwconv/gen/up4x25-minmax-psimd.c",
+    "src/f32-dwconv/gen/up4x4-minmax-psimd-acc2.c",
+    "src/f32-dwconv/gen/up4x4-minmax-psimd.c",
+    "src/f32-dwconv/gen/up4x9-minmax-psimd-acc2.c",
+    "src/f32-dwconv/gen/up4x9-minmax-psimd.c",
+    "src/f32-dwconv/gen/up8x25-minmax-psimd-acc2.c",
+    "src/f32-dwconv/gen/up8x25-minmax-psimd.c",
+    "src/f32-dwconv/gen/up8x4-minmax-psimd-acc2.c",
+    "src/f32-dwconv/gen/up8x4-minmax-psimd.c",
+    "src/f32-dwconv/gen/up8x9-minmax-psimd-acc2.c",
+    "src/f32-dwconv/gen/up8x9-minmax-psimd.c",
+    "src/f32-gavgpool/7p7x-minmax-psimd-c4.c",
+    "src/f32-gavgpool/7x-minmax-psimd-c4.c",
+    "src/f32-gemm/gen/1x8-minmax-psimd-loadsplat.c",
+    "src/f32-gemm/gen/1x8-minmax-psimd-splat.c",
+    "src/f32-gemm/gen/1x8s4-minmax-psimd.c",
+    "src/f32-gemm/gen/4x2c4-minmax-psimd.c",
+    "src/f32-gemm/gen/4x8-minmax-psimd-loadsplat.c",
+    "src/f32-gemm/gen/4x8-minmax-psimd-splat.c",
+    "src/f32-gemm/gen/4x8s4-minmax-psimd.c",
+    "src/f32-gemm/gen/6x8-minmax-psimd-loadsplat.c",
+    "src/f32-gemm/gen/6x8-minmax-psimd-splat.c",
+    "src/f32-gemm/gen/6x8s4-minmax-psimd.c",
+    "src/f32-gemm/gen-inc/1x8inc-minmax-psimd-loadsplat.c",
+    "src/f32-gemm/gen-inc/1x8inc-minmax-psimd-splat.c",
+    "src/f32-gemm/gen-inc/1x8s4inc-minmax-psimd.c",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-psimd-loadsplat.c",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-psimd-splat.c",
+    "src/f32-gemm/gen-inc/4x8s4inc-minmax-psimd.c",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-psimd-loadsplat.c",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-psimd-splat.c",
+    "src/f32-gemm/gen-inc/6x8s4inc-minmax-psimd.c",
     "src/f32-hswish/gen/psimd-x4.c",
     "src/f32-hswish/gen/psimd-x8.c",
     "src/f32-ibilinear/gen/psimd-c4.c",
     "src/f32-ibilinear/gen/psimd-c8.c",
-    "src/f32-igemm/gen/1x8-psimd-loadsplat.c",
-    "src/f32-igemm/gen/1x8-psimd-splat.c",
-    "src/f32-igemm/gen/1x8s4-psimd.c",
-    "src/f32-igemm/gen/4x2c4-psimd.c",
-    "src/f32-igemm/gen/4x8-psimd-loadsplat.c",
-    "src/f32-igemm/gen/4x8-psimd-splat.c",
-    "src/f32-igemm/gen/4x8s4-psimd.c",
-    "src/f32-igemm/gen/6x8-psimd-loadsplat.c",
-    "src/f32-igemm/gen/6x8-psimd-splat.c",
-    "src/f32-igemm/gen/6x8s4-psimd.c",
-    "src/f32-maxpool/9p8x-psimd-c4.c",
-    "src/f32-pavgpool/9p8x-psimd-c4.c",
-    "src/f32-pavgpool/9x-psimd-c4.c",
-    "src/f32-ppmm/gen/4x8-psimd.c",
+    "src/f32-igemm/gen/1x8-minmax-psimd-loadsplat.c",
+    "src/f32-igemm/gen/1x8-minmax-psimd-splat.c",
+    "src/f32-igemm/gen/1x8s4-minmax-psimd.c",
+    "src/f32-igemm/gen/4x2c4-minmax-psimd.c",
+    "src/f32-igemm/gen/4x8-minmax-psimd-loadsplat.c",
+    "src/f32-igemm/gen/4x8-minmax-psimd-splat.c",
+    "src/f32-igemm/gen/4x8s4-minmax-psimd.c",
+    "src/f32-igemm/gen/6x8-minmax-psimd-loadsplat.c",
+    "src/f32-igemm/gen/6x8-minmax-psimd-splat.c",
+    "src/f32-igemm/gen/6x8s4-minmax-psimd.c",
+    "src/f32-maxpool/9p8x-minmax-psimd-c4.c",
+    "src/f32-pavgpool/9p8x-minmax-psimd-c4.c",
+    "src/f32-pavgpool/9x-minmax-psimd-c4.c",
+    "src/f32-ppmm/gen/4x8-minmax-psimd.c",
     "src/f32-prelu/gen/psimd-2x4.c",
     "src/f32-prelu/gen/psimd-2x8.c",
     "src/f32-rmax/psimd.c",
-    "src/f32-vbinary/gen/vadd-psimd-x4.c",
-    "src/f32-vbinary/gen/vadd-psimd-x8.c",
-    "src/f32-vbinary/gen/vaddc-psimd-x4.c",
-    "src/f32-vbinary/gen/vaddc-psimd-x8.c",
-    "src/f32-vbinary/gen/vdiv-psimd-x4.c",
-    "src/f32-vbinary/gen/vdiv-psimd-x8.c",
-    "src/f32-vbinary/gen/vdivc-psimd-x4.c",
-    "src/f32-vbinary/gen/vdivc-psimd-x8.c",
+    "src/f32-vbinary/gen/vadd-minmax-psimd-x4.c",
+    "src/f32-vbinary/gen/vadd-minmax-psimd-x8.c",
+    "src/f32-vbinary/gen/vaddc-minmax-psimd-x4.c",
+    "src/f32-vbinary/gen/vaddc-minmax-psimd-x8.c",
+    "src/f32-vbinary/gen/vdiv-minmax-psimd-x4.c",
+    "src/f32-vbinary/gen/vdiv-minmax-psimd-x8.c",
+    "src/f32-vbinary/gen/vdivc-minmax-psimd-x4.c",
+    "src/f32-vbinary/gen/vdivc-minmax-psimd-x8.c",
     "src/f32-vbinary/gen/vmax-psimd-x4.c",
     "src/f32-vbinary/gen/vmax-psimd-x8.c",
     "src/f32-vbinary/gen/vmaxc-psimd-x4.c",
@@ -291,20 +311,20 @@
     "src/f32-vbinary/gen/vmin-psimd-x8.c",
     "src/f32-vbinary/gen/vminc-psimd-x4.c",
     "src/f32-vbinary/gen/vminc-psimd-x8.c",
-    "src/f32-vbinary/gen/vmul-psimd-x4.c",
-    "src/f32-vbinary/gen/vmul-psimd-x8.c",
-    "src/f32-vbinary/gen/vmulc-psimd-x4.c",
-    "src/f32-vbinary/gen/vmulc-psimd-x8.c",
-    "src/f32-vbinary/gen/vrdivc-psimd-x4.c",
-    "src/f32-vbinary/gen/vrdivc-psimd-x8.c",
-    "src/f32-vbinary/gen/vrsubc-psimd-x4.c",
-    "src/f32-vbinary/gen/vrsubc-psimd-x8.c",
-    "src/f32-vbinary/gen/vsub-psimd-x4.c",
-    "src/f32-vbinary/gen/vsub-psimd-x8.c",
-    "src/f32-vbinary/gen/vsubc-psimd-x4.c",
-    "src/f32-vbinary/gen/vsubc-psimd-x8.c",
-    "src/f32-vmulcaddc/gen/c4-psimd-2x.c",
-    "src/f32-vmulcaddc/gen/c8-psimd-2x.c",
+    "src/f32-vbinary/gen/vmul-minmax-psimd-x4.c",
+    "src/f32-vbinary/gen/vmul-minmax-psimd-x8.c",
+    "src/f32-vbinary/gen/vmulc-minmax-psimd-x4.c",
+    "src/f32-vbinary/gen/vmulc-minmax-psimd-x8.c",
+    "src/f32-vbinary/gen/vrdivc-minmax-psimd-x4.c",
+    "src/f32-vbinary/gen/vrdivc-minmax-psimd-x8.c",
+    "src/f32-vbinary/gen/vrsubc-minmax-psimd-x4.c",
+    "src/f32-vbinary/gen/vrsubc-minmax-psimd-x8.c",
+    "src/f32-vbinary/gen/vsub-minmax-psimd-x4.c",
+    "src/f32-vbinary/gen/vsub-minmax-psimd-x8.c",
+    "src/f32-vbinary/gen/vsubc-minmax-psimd-x4.c",
+    "src/f32-vbinary/gen/vsubc-minmax-psimd-x8.c",
+    "src/f32-vmulcaddc/gen/c4-minmax-psimd-2x.c",
+    "src/f32-vmulcaddc/gen/c8-minmax-psimd-2x.c",
     "src/x32-packx/x4-psimd.c",
     "src/x32-pad/x2-psimd.c",
     "src/x32-unpool/psimd.c",
@@ -340,73 +360,73 @@
 
 // ISA-specific micro-kernels
 NEON_UKERNELS = [
-    "src/f32-avgpool/9p8x-neon-c4.c",
-    "src/f32-avgpool/9x-neon-c4.c",
+    "src/f32-avgpool/9p8x-minmax-neon-c4.c",
+    "src/f32-avgpool/9x-minmax-neon-c4.c",
     "src/f32-clamp/gen/neon-x4.c",
     "src/f32-clamp/gen/neon-x8.c",
-    "src/f32-dwconv/gen/up4x9-neon.c",
-    "src/f32-dwconv/gen/up4x9-neon-acc2.c",
-    "src/f32-dwconv/gen/up8x9-neon.c",
-    "src/f32-dwconv/gen/up8x9-neon-acc2.c",
+    "src/f32-dwconv/gen/up4x9-minmax-neon.c",
+    "src/f32-dwconv/gen/up4x9-minmax-neon-acc2.c",
+    "src/f32-dwconv/gen/up8x9-minmax-neon.c",
+    "src/f32-dwconv/gen/up8x9-minmax-neon-acc2.c",
     "src/f32-gavgpool-spchw/neon-x4.c",
-    "src/f32-gavgpool/7p7x-neon-c4.c",
-    "src/f32-gavgpool/7x-neon-c4.c",
-    "src/f32-gemm/gen/1x8-neon-lane-ld64.c",
-    "src/f32-gemm/gen/4x2-neon-lane-ld64.c",
-    "src/f32-gemm/gen/4x8-neon-lane-ld128.c",
-    "src/f32-gemm/gen/4x8-neon-lane-ld64.c",
-    "src/f32-gemm/gen/5x8-neon-lane-ld64.c",
-    "src/f32-gemm/gen/6x8-neon-lane-ld64.c",
-    "src/f32-gemm/gen/6x8-neon-lane-ld128.c",
-    "src/f32-gemm/gen/1x8-neon-dup-ld64.c",
-    "src/f32-gemm/gen/4x8-neon-dup-ld128.c",
-    "src/f32-gemm/gen/4x8-neon-dup-ld64.c",
-    "src/f32-gemm/gen/6x8-neon-dup-ld64.c",
-    "src/f32-gemm/gen/6x8-neon-dup-ld128.c",
-    "src/f32-gemm/gen/1x8s4-neon.c",
-    "src/f32-gemm/gen/4x8s4-neon.c",
-    "src/f32-gemm/gen/6x8s4-neon.c",
-    "src/f32-gemm/gen/8x8s4-neon.c",
-    "src/f32-gemm/gen-inc/1x8-neon-lane-ld64.c",
-    "src/f32-gemm/gen-inc/4x8-neon-lane-ld128.c",
-    "src/f32-gemm/gen-inc/4x8-neon-lane-ld64.c",
-    "src/f32-gemm/gen-inc/5x8-neon-lane-ld64.c",
-    "src/f32-gemm/gen-inc/6x8-neon-lane-ld64.c",
-    "src/f32-gemm/gen-inc/6x8-neon-lane-ld128.c",
-    "src/f32-gemm/gen-inc/1x8-neon-dup-ld64.c",
-    "src/f32-gemm/gen-inc/4x8-neon-dup-ld128.c",
-    "src/f32-gemm/gen-inc/4x8-neon-dup-ld64.c",
-    "src/f32-gemm/gen-inc/6x8-neon-dup-ld64.c",
-    "src/f32-gemm/gen-inc/6x8-neon-dup-ld128.c",
-    "src/f32-gemm/gen-inc/1x8s4-neon.c",
-    "src/f32-gemm/gen-inc/4x8s4-neon.c",
-    "src/f32-gemm/gen-inc/6x8s4-neon.c",
-    "src/f32-gemm/gen-inc/8x8s4-neon.c",
+    "src/f32-gavgpool/7p7x-minmax-neon-c4.c",
+    "src/f32-gavgpool/7x-minmax-neon-c4.c",
+    "src/f32-gemm/gen/1x8-minmax-neon-lane-ld64.c",
+    "src/f32-gemm/gen/4x2-minmax-neon-lane-ld64.c",
+    "src/f32-gemm/gen/4x8-minmax-neon-lane-ld128.c",
+    "src/f32-gemm/gen/4x8-minmax-neon-lane-ld64.c",
+    "src/f32-gemm/gen/5x8-minmax-neon-lane-ld64.c",
+    "src/f32-gemm/gen/6x8-minmax-neon-lane-ld64.c",
+    "src/f32-gemm/gen/6x8-minmax-neon-lane-ld128.c",
+    "src/f32-gemm/gen/1x8-minmax-neon-dup-ld64.c",
+    "src/f32-gemm/gen/4x8-minmax-neon-dup-ld128.c",
+    "src/f32-gemm/gen/4x8-minmax-neon-dup-ld64.c",
+    "src/f32-gemm/gen/6x8-minmax-neon-dup-ld64.c",
+    "src/f32-gemm/gen/6x8-minmax-neon-dup-ld128.c",
+    "src/f32-gemm/gen/1x8s4-minmax-neon.c",
+    "src/f32-gemm/gen/4x8s4-minmax-neon.c",
+    "src/f32-gemm/gen/6x8s4-minmax-neon.c",
+    "src/f32-gemm/gen/8x8s4-minmax-neon.c",
+    "src/f32-gemm/gen-inc/1x8inc-minmax-neon-lane-ld64.c",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-neon-lane-ld128.c",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-neon-lane-ld64.c",
+    "src/f32-gemm/gen-inc/5x8inc-minmax-neon-lane-ld64.c",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-neon-lane-ld64.c",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-neon-lane-ld128.c",
+    "src/f32-gemm/gen-inc/1x8inc-minmax-neon-dup-ld64.c",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-neon-dup-ld128.c",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-neon-dup-ld64.c",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-neon-dup-ld64.c",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-neon-dup-ld128.c",
+    "src/f32-gemm/gen-inc/1x8s4inc-minmax-neon.c",
+    "src/f32-gemm/gen-inc/4x8s4inc-minmax-neon.c",
+    "src/f32-gemm/gen-inc/6x8s4inc-minmax-neon.c",
+    "src/f32-gemm/gen-inc/8x8s4inc-minmax-neon.c",
     "src/f32-hswish/gen/neon-x4.c",
     "src/f32-hswish/gen/neon-x8.c",
     "src/f32-ibilinear/gen/neon-c4.c",
     "src/f32-ibilinear/gen/neon-c8.c",
-    "src/f32-igemm/gen/1x8-neon-lane-ld64.c",
-    "src/f32-igemm/gen/4x2-neon-lane-ld64.c",
-    "src/f32-igemm/gen/4x4-neon-lane-ld64.c",
-    "src/f32-igemm/gen/4x8-neon-lane-ld128.c",
-    "src/f32-igemm/gen/4x8-neon-lane-ld64.c",
-    "src/f32-igemm/gen/6x8-neon-lane-ld64.c",
-    "src/f32-igemm/gen/6x8-neon-lane-ld128.c",
-    "src/f32-igemm/gen/1x8-neon-dup-ld64.c",
-    "src/f32-igemm/gen/4x8-neon-dup-ld128.c",
-    "src/f32-igemm/gen/4x8-neon-dup-ld64.c",
-    "src/f32-igemm/gen/6x8-neon-dup-ld64.c",
-    "src/f32-igemm/gen/6x8-neon-dup-ld128.c",
-    "src/f32-igemm/gen/1x8s4-neon.c",
-    "src/f32-igemm/gen/4x8s4-neon.c",
-    "src/f32-igemm/gen/6x8s4-neon.c",
-    "src/f32-igemm/gen/8x8s4-neon.c",
-    "src/f32-maxpool/9p8x-neon-c4.c",
-    "src/f32-pavgpool/9p8x-neon-c4.c",
-    "src/f32-pavgpool/9x-neon-c4.c",
-    "src/f32-ppmm/gen/4x8-neon.c",
-    "src/f32-ppmm/gen/8x8-neon.c",
+    "src/f32-igemm/gen/1x8-minmax-neon-lane-ld64.c",
+    "src/f32-igemm/gen/4x2-minmax-neon-lane-ld64.c",
+    "src/f32-igemm/gen/4x4-minmax-neon-lane-ld64.c",
+    "src/f32-igemm/gen/4x8-minmax-neon-lane-ld128.c",
+    "src/f32-igemm/gen/4x8-minmax-neon-lane-ld64.c",
+    "src/f32-igemm/gen/6x8-minmax-neon-lane-ld64.c",
+    "src/f32-igemm/gen/6x8-minmax-neon-lane-ld128.c",
+    "src/f32-igemm/gen/1x8-minmax-neon-dup-ld64.c",
+    "src/f32-igemm/gen/4x8-minmax-neon-dup-ld128.c",
+    "src/f32-igemm/gen/4x8-minmax-neon-dup-ld64.c",
+    "src/f32-igemm/gen/6x8-minmax-neon-dup-ld64.c",
+    "src/f32-igemm/gen/6x8-minmax-neon-dup-ld128.c",
+    "src/f32-igemm/gen/1x8s4-minmax-neon.c",
+    "src/f32-igemm/gen/4x8s4-minmax-neon.c",
+    "src/f32-igemm/gen/6x8s4-minmax-neon.c",
+    "src/f32-igemm/gen/8x8s4-minmax-neon.c",
+    "src/f32-maxpool/9p8x-minmax-neon-c4.c",
+    "src/f32-pavgpool/9p8x-minmax-neon-c4.c",
+    "src/f32-pavgpool/9x-minmax-neon-c4.c",
+    "src/f32-ppmm/gen/4x8-minmax-neon.c",
+    "src/f32-ppmm/gen/8x8-minmax-neon.c",
     "src/f32-prelu/gen/neon-2x4.c",
     "src/f32-prelu/gen/neon-2x8.c",
     "src/f32-raddstoreexpminusmax/gen/neon-p5-x4.c",
@@ -453,10 +473,10 @@
     "src/f32-sigmoid/gen/neon-rr2-lut2048-p1-nr2recps-x16.c",
     "src/f32-sigmoid/gen/neon-rr2-lut2048-p1-nr2recps-x20.c",
     "src/f32-sigmoid/gen/neon-rr2-lut2048-p1-nr2recps-x24.c",
-    "src/f32-vbinary/gen/vadd-neon-x4.c",
-    "src/f32-vbinary/gen/vadd-neon-x8.c",
-    "src/f32-vbinary/gen/vaddc-neon-x4.c",
-    "src/f32-vbinary/gen/vaddc-neon-x8.c",
+    "src/f32-vbinary/gen/vadd-minmax-neon-x4.c",
+    "src/f32-vbinary/gen/vadd-minmax-neon-x8.c",
+    "src/f32-vbinary/gen/vaddc-minmax-neon-x4.c",
+    "src/f32-vbinary/gen/vaddc-minmax-neon-x8.c",
     "src/f32-vbinary/gen/vmax-neon-x4.c",
     "src/f32-vbinary/gen/vmax-neon-x8.c",
     "src/f32-vbinary/gen/vmaxc-neon-x4.c",
@@ -465,30 +485,30 @@
     "src/f32-vbinary/gen/vmin-neon-x8.c",
     "src/f32-vbinary/gen/vminc-neon-x4.c",
     "src/f32-vbinary/gen/vminc-neon-x8.c",
-    "src/f32-vbinary/gen/vmul-neon-x4.c",
-    "src/f32-vbinary/gen/vmul-neon-x8.c",
-    "src/f32-vbinary/gen/vmulc-neon-x4.c",
-    "src/f32-vbinary/gen/vmulc-neon-x8.c",
-    "src/f32-vbinary/gen/vrsubc-neon-x4.c",
-    "src/f32-vbinary/gen/vrsubc-neon-x8.c",
-    "src/f32-vbinary/gen/vsub-neon-x4.c",
-    "src/f32-vbinary/gen/vsub-neon-x8.c",
-    "src/f32-vbinary/gen/vsubc-neon-x4.c",
-    "src/f32-vbinary/gen/vsubc-neon-x8.c",
-    "src/f32-vmulcaddc/gen/c4-neon-2x.c",
-    "src/f32-vmulcaddc/gen/c8-neon-2x.c",
-    "src/q8-avgpool/9p8x-neon-c8.c",
-    "src/q8-avgpool/9x-neon-c8.c",
-    "src/q8-dwconv/up8x9-neon.c",
-    "src/q8-gavgpool/7p7x-neon-c8.c",
-    "src/q8-gavgpool/7x-neon-c8.c",
-    "src/q8-gemm/4x8-neon.c",
-    "src/q8-gemm/8x8-neon.c",
-    "src/q8-igemm/4x8-neon.c",
-    "src/q8-igemm/8x8-neon.c",
-    "src/q8-vadd/neon.c",
+    "src/f32-vbinary/gen/vmul-minmax-neon-x4.c",
+    "src/f32-vbinary/gen/vmul-minmax-neon-x8.c",
+    "src/f32-vbinary/gen/vmulc-minmax-neon-x4.c",
+    "src/f32-vbinary/gen/vmulc-minmax-neon-x8.c",
+    "src/f32-vbinary/gen/vrsubc-minmax-neon-x4.c",
+    "src/f32-vbinary/gen/vrsubc-minmax-neon-x8.c",
+    "src/f32-vbinary/gen/vsub-minmax-neon-x4.c",
+    "src/f32-vbinary/gen/vsub-minmax-neon-x8.c",
+    "src/f32-vbinary/gen/vsubc-minmax-neon-x4.c",
+    "src/f32-vbinary/gen/vsubc-minmax-neon-x8.c",
+    "src/f32-vmulcaddc/gen/c4-minmax-neon-2x.c",
+    "src/f32-vmulcaddc/gen/c8-minmax-neon-2x.c",
+    "src/q8-avgpool/9p8x-minmax-neon-c8.c",
+    "src/q8-avgpool/9x-minmax-neon-c8.c",
+    "src/q8-dwconv/up8x9-minmax-neon.c",
+    "src/q8-gavgpool/7p7x-minmax-neon-c8.c",
+    "src/q8-gavgpool/7x-minmax-neon-c8.c",
+    "src/q8-gemm/4x8-minmax-neon.c",
+    "src/q8-gemm/8x8-minmax-neon.c",
+    "src/q8-igemm/4x8-minmax-neon.c",
+    "src/q8-igemm/8x8-minmax-neon.c",
+    "src/q8-vadd/minmax-neon.c",
     "src/u8-clamp/neon-x64.c",
-    "src/u8-maxpool/9p8x-neon-c16.c",
+    "src/u8-maxpool/9p8x-minmax-neon-c16.c",
     "src/u8-rmax/neon.c",
     "src/x32-packx/x4-neon-st4.c",
     "src/x32-pad/x2-neon.c",
@@ -516,41 +536,41 @@
 NEONFMA_UKERNELS = [
     "src/f32-ibilinear/gen/neonfma-c4.c",
     "src/f32-ibilinear/gen/neonfma-c8.c",
-    "src/f32-igemm/gen/1x8-neonfma-dup-ld64.c",
-    "src/f32-igemm/gen/4x8-neonfma-dup-ld128.c",
-    "src/f32-igemm/gen/4x8-neonfma-dup-ld64.c",
-    "src/f32-igemm/gen/6x8-neonfma-dup-ld64.c",
-    "src/f32-igemm/gen/6x8-neonfma-dup-ld128.c",
-    "src/f32-igemm/gen/1x8s4-neonfma.c",
-    "src/f32-igemm/gen/4x8s4-neonfma.c",
-    "src/f32-igemm/gen/6x8s4-neonfma.c",
-    "src/f32-igemm/gen/8x8s4-neonfma.c",
-    "src/f32-dwconv/gen/up4x9-neonfma.c",
-    "src/f32-dwconv/gen/up4x9-neonfma-acc2.c",
-    "src/f32-dwconv/gen/up8x9-neonfma.c",
-    "src/f32-dwconv/gen/up8x9-neonfma-acc2.c",
-    "src/f32-gemm/gen/1x8-neonfma-dup-ld64.c",
-    "src/f32-gemm/gen/4x8-neonfma-dup-ld128.c",
-    "src/f32-gemm/gen/4x8-neonfma-dup-ld64.c",
-    "src/f32-gemm/gen/6x8-neonfma-dup-ld64.c",
-    "src/f32-gemm/gen/6x8-neonfma-dup-ld128.c",
-    "src/f32-gemm/gen/1x8s4-neonfma.c",
-    "src/f32-gemm/gen/4x8s4-neonfma.c",
-    "src/f32-gemm/gen/6x8s4-neonfma.c",
-    "src/f32-gemm/gen/8x8s4-neonfma.c",
-    "src/f32-gemm/gen-inc/1x8-neonfma-dup-ld64.c",
-    "src/f32-gemm/gen-inc/4x8-neonfma-dup-ld128.c",
-    "src/f32-gemm/gen-inc/4x8-neonfma-dup-ld64.c",
-    "src/f32-gemm/gen-inc/6x8-neonfma-dup-ld64.c",
-    "src/f32-gemm/gen-inc/6x8-neonfma-dup-ld128.c",
-    "src/f32-gemm/gen-inc/1x8s4-neonfma.c",
-    "src/f32-gemm/gen-inc/4x8s4-neonfma.c",
-    "src/f32-gemm/gen-inc/6x8s4-neonfma.c",
-    "src/f32-gemm/gen-inc/8x8s4-neonfma.c",
+    "src/f32-igemm/gen/1x8-minmax-neonfma-dup-ld64.c",
+    "src/f32-igemm/gen/4x8-minmax-neonfma-dup-ld128.c",
+    "src/f32-igemm/gen/4x8-minmax-neonfma-dup-ld64.c",
+    "src/f32-igemm/gen/6x8-minmax-neonfma-dup-ld64.c",
+    "src/f32-igemm/gen/6x8-minmax-neonfma-dup-ld128.c",
+    "src/f32-igemm/gen/1x8s4-minmax-neonfma.c",
+    "src/f32-igemm/gen/4x8s4-minmax-neonfma.c",
+    "src/f32-igemm/gen/6x8s4-minmax-neonfma.c",
+    "src/f32-igemm/gen/8x8s4-minmax-neonfma.c",
+    "src/f32-dwconv/gen/up4x9-minmax-neonfma.c",
+    "src/f32-dwconv/gen/up4x9-minmax-neonfma-acc2.c",
+    "src/f32-dwconv/gen/up8x9-minmax-neonfma.c",
+    "src/f32-dwconv/gen/up8x9-minmax-neonfma-acc2.c",
+    "src/f32-gemm/gen/1x8-minmax-neonfma-dup-ld64.c",
+    "src/f32-gemm/gen/4x8-minmax-neonfma-dup-ld128.c",
+    "src/f32-gemm/gen/4x8-minmax-neonfma-dup-ld64.c",
+    "src/f32-gemm/gen/6x8-minmax-neonfma-dup-ld64.c",
+    "src/f32-gemm/gen/6x8-minmax-neonfma-dup-ld128.c",
+    "src/f32-gemm/gen/1x8s4-minmax-neonfma.c",
+    "src/f32-gemm/gen/4x8s4-minmax-neonfma.c",
+    "src/f32-gemm/gen/6x8s4-minmax-neonfma.c",
+    "src/f32-gemm/gen/8x8s4-minmax-neonfma.c",
+    "src/f32-gemm/gen-inc/1x8inc-minmax-neonfma-dup-ld64.c",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-neonfma-dup-ld128.c",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-neonfma-dup-ld64.c",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-neonfma-dup-ld64.c",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-neonfma-dup-ld128.c",
+    "src/f32-gemm/gen-inc/1x8s4inc-minmax-neonfma.c",
+    "src/f32-gemm/gen-inc/4x8s4inc-minmax-neonfma.c",
+    "src/f32-gemm/gen-inc/6x8s4inc-minmax-neonfma.c",
+    "src/f32-gemm/gen-inc/8x8s4inc-minmax-neonfma.c",
     "src/f32-hswish/gen/neonfma-x4.c",
     "src/f32-hswish/gen/neonfma-x8.c",
-    "src/f32-ppmm/gen/4x8-neonfma.c",
-    "src/f32-ppmm/gen/8x8-neonfma.c",
+    "src/f32-ppmm/gen/4x8-minmax-neonfma.c",
+    "src/f32-ppmm/gen/8x8-minmax-neonfma.c",
     "src/f32-raddstoreexpminusmax/gen/neonfma-p5-x4.c",
     "src/f32-raddstoreexpminusmax/gen/neonfma-p5-x8.c",
     "src/f32-raddstoreexpminusmax/gen/neonfma-p5-x8-acc2.c",
@@ -629,8 +649,8 @@
     "src/f32-sigmoid/gen/neonfma-rr1-lut2048-p1-nr2recps-x16.c",
     "src/f32-sigmoid/gen/neonfma-rr1-lut2048-p1-nr2recps-x20.c",
     "src/f32-sigmoid/gen/neonfma-rr1-lut2048-p1-nr2recps-x24.c",
-    "src/f32-vmulcaddc/gen/c4-neonfma-2x.c",
-    "src/f32-vmulcaddc/gen/c8-neonfma-2x.c",
+    "src/f32-vmulcaddc/gen/c4-minmax-neonfma-2x.c",
+    "src/f32-vmulcaddc/gen/c8-minmax-neonfma-2x.c",
     "src/math/exp-neonfma-lut64-p2.c",
     "src/math/exp-neonfma-p5.c",
     "src/math/expminus-neonfma-lut2048-p1.c",
@@ -657,32 +677,32 @@
 ]
 
 AARCH64_NEONFMA_UKERNELS = [
-    "src/f32-vbinary/gen/vdiv-neon-x4.c",
-    "src/f32-vbinary/gen/vdiv-neon-x8.c",
-    "src/f32-vbinary/gen/vdivc-neon-x4.c",
-    "src/f32-vbinary/gen/vdivc-neon-x8.c",
-    "src/f32-vbinary/gen/vrdivc-neon-x4.c",
-    "src/f32-vbinary/gen/vrdivc-neon-x8.c",
-    "src/f32-gemm/gen/1x8-neonfma-lane-ld64.c",
-    "src/f32-gemm/gen/4x2-neonfma-lane-ld64.c",
-    "src/f32-gemm/gen/4x8-neonfma-lane-ld128.c",
-    "src/f32-gemm/gen/4x8-neonfma-lane-ld64.c",
-    "src/f32-gemm/gen/5x8-neonfma-lane-ld64.c",
-    "src/f32-gemm/gen/6x8-neonfma-lane-ld64.c",
-    "src/f32-gemm/gen/6x8-neonfma-lane-ld128.c",
-    "src/f32-gemm/gen-inc/1x8-neonfma-lane-ld64.c",
-    "src/f32-gemm/gen-inc/4x8-neonfma-lane-ld128.c",
-    "src/f32-gemm/gen-inc/4x8-neonfma-lane-ld64.c",
-    "src/f32-gemm/gen-inc/5x8-neonfma-lane-ld64.c",
-    "src/f32-gemm/gen-inc/6x8-neonfma-lane-ld64.c",
-    "src/f32-gemm/gen-inc/6x8-neonfma-lane-ld128.c",
-    "src/f32-igemm/gen/1x8-neonfma-lane-ld64.c",
-    "src/f32-igemm/gen/4x2-neonfma-lane-ld64.c",
-    "src/f32-igemm/gen/4x4-neonfma-lane-ld64.c",
-    "src/f32-igemm/gen/4x8-neonfma-lane-ld128.c",
-    "src/f32-igemm/gen/4x8-neonfma-lane-ld64.c",
-    "src/f32-igemm/gen/6x8-neonfma-lane-ld64.c",
-    "src/f32-igemm/gen/6x8-neonfma-lane-ld128.c",
+    "src/f32-vbinary/gen/vdiv-minmax-neon-x4.c",
+    "src/f32-vbinary/gen/vdiv-minmax-neon-x8.c",
+    "src/f32-vbinary/gen/vdivc-minmax-neon-x4.c",
+    "src/f32-vbinary/gen/vdivc-minmax-neon-x8.c",
+    "src/f32-vbinary/gen/vrdivc-minmax-neon-x4.c",
+    "src/f32-vbinary/gen/vrdivc-minmax-neon-x8.c",
+    "src/f32-gemm/gen/1x8-minmax-neonfma-lane-ld64.c",
+    "src/f32-gemm/gen/4x2-minmax-neonfma-lane-ld64.c",
+    "src/f32-gemm/gen/4x8-minmax-neonfma-lane-ld128.c",
+    "src/f32-gemm/gen/4x8-minmax-neonfma-lane-ld64.c",
+    "src/f32-gemm/gen/5x8-minmax-neonfma-lane-ld64.c",
+    "src/f32-gemm/gen/6x8-minmax-neonfma-lane-ld64.c",
+    "src/f32-gemm/gen/6x8-minmax-neonfma-lane-ld128.c",
+    "src/f32-gemm/gen-inc/1x8inc-minmax-neonfma-lane-ld64.c",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-neonfma-lane-ld128.c",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-neonfma-lane-ld64.c",
+    "src/f32-gemm/gen-inc/5x8inc-minmax-neonfma-lane-ld64.c",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-neonfma-lane-ld64.c",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-neonfma-lane-ld128.c",
+    "src/f32-igemm/gen/1x8-minmax-neonfma-lane-ld64.c",
+    "src/f32-igemm/gen/4x2-minmax-neonfma-lane-ld64.c",
+    "src/f32-igemm/gen/4x4-minmax-neonfma-lane-ld64.c",
+    "src/f32-igemm/gen/4x8-minmax-neonfma-lane-ld128.c",
+    "src/f32-igemm/gen/4x8-minmax-neonfma-lane-ld64.c",
+    "src/f32-igemm/gen/6x8-minmax-neonfma-lane-ld64.c",
+    "src/f32-igemm/gen/6x8-minmax-neonfma-lane-ld128.c",
     "src/f32-conv-hwc/3x3s2p1c3x4-neonfma-2x2.c",
     "src/f32-conv-hwc/3x3s2p1c3x8-neonfma-2x2.c",
     "src/f32-conv-hwc2spchw/3x3s2p1c3x4-neonfma-2x2.c",
@@ -708,24 +728,24 @@
     "src/f32-sigmoid/gen/neonfma-rr1-lut2048-p1-div-x16.c",
     "src/f32-sigmoid/gen/neonfma-rr1-lut2048-p1-div-x20.c",
     "src/f32-sigmoid/gen/neonfma-rr1-lut2048-p1-div-x24.c",
-    "src/f32-spmm/gen/12x1-neonfma.c",
-    "src/f32-spmm/gen/12x2-neonfma.c",
-    "src/f32-spmm/gen/12x4-neonfma.c",
-    "src/f32-spmm/gen/16x1-neonfma-pipelined.c",
-    "src/f32-spmm/gen/16x1-neonfma-unroll2.c",
-    "src/f32-spmm/gen/16x1-neonfma.c",
-    "src/f32-spmm/gen/16x2-neonfma.c",
-    "src/f32-spmm/gen/16x4-neonfma.c",
-    "src/f32-spmm/gen/4x1-neonfma-pipelined.c",
-    "src/f32-spmm/gen/4x1-neonfma-unroll2.c",
-    "src/f32-spmm/gen/4x1-neonfma.c",
-    "src/f32-spmm/gen/4x2-neonfma.c",
-    "src/f32-spmm/gen/4x4-neonfma.c",
-    "src/f32-spmm/gen/8x1-neonfma-pipelined.c",
-    "src/f32-spmm/gen/8x1-neonfma-unroll2.c",
-    "src/f32-spmm/gen/8x1-neonfma.c",
-    "src/f32-spmm/gen/8x2-neonfma.c",
-    "src/f32-spmm/gen/8x4-neonfma.c",
+    "src/f32-spmm/gen/12x1-minmax-neonfma.c",
+    "src/f32-spmm/gen/12x2-minmax-neonfma.c",
+    "src/f32-spmm/gen/12x4-minmax-neonfma.c",
+    "src/f32-spmm/gen/16x1-minmax-neonfma-pipelined.c",
+    "src/f32-spmm/gen/16x1-minmax-neonfma-unroll2.c",
+    "src/f32-spmm/gen/16x1-minmax-neonfma.c",
+    "src/f32-spmm/gen/16x2-minmax-neonfma.c",
+    "src/f32-spmm/gen/16x4-minmax-neonfma.c",
+    "src/f32-spmm/gen/4x1-minmax-neonfma-pipelined.c",
+    "src/f32-spmm/gen/4x1-minmax-neonfma-unroll2.c",
+    "src/f32-spmm/gen/4x1-minmax-neonfma.c",
+    "src/f32-spmm/gen/4x2-minmax-neonfma.c",
+    "src/f32-spmm/gen/4x4-minmax-neonfma.c",
+    "src/f32-spmm/gen/8x1-minmax-neonfma-pipelined.c",
+    "src/f32-spmm/gen/8x1-minmax-neonfma-unroll2.c",
+    "src/f32-spmm/gen/8x1-minmax-neonfma.c",
+    "src/f32-spmm/gen/8x2-minmax-neonfma.c",
+    "src/f32-spmm/gen/8x4-minmax-neonfma.c",
     "src/math/sigmoid-neonfma-rr1-lut2048-p1-div.c",
     "src/math/sigmoid-neonfma-rr1-lut64-p2-div.c",
     "src/math/sigmoid-neonfma-rr1-p5-div.c",
@@ -735,72 +755,81 @@
 ]
 
 AARCH64_NEONFP16ARITH_UKERNELS = [
+    "src/f16-gemm/gen/1x8-neonfp16arith-ld64.c",
     "src/f16-gemm/gen/4x8-neonfp16arith-ld64.c",
     "src/f16-gemm/gen/6x8-neonfp16arith-ld64.c",
     "src/f16-gemm/gen/8x8-neonfp16arith-ld64.c",
+    "src/f16-spmm/gen/8x1-minmax-neonfp16arith.c",
+    "src/f16-spmm/gen/8x1-minmax-neonfp16arith-unroll2.c",
+    "src/f16-spmm/gen/16x1-minmax-neonfp16arith.c",
+    "src/f16-spmm/gen/16x1-minmax-neonfp16arith-unroll2.c",
+    "src/f16-spmm/gen/24x1-minmax-neonfp16arith.c",
+    "src/f16-spmm/gen/24x1-minmax-neonfp16arith-unroll2.c",
+    "src/f16-spmm/gen/32x1-minmax-neonfp16arith.c",
+    "src/f16-spmm/gen/32x1-minmax-neonfp16arith-unroll2.c",
 ]
 
 SSE_UKERNELS = [
-    "src/f32-avgpool/9p8x-sse-c4.c",
-    "src/f32-avgpool/9x-sse-c4.c",
+    "src/f32-avgpool/9p8x-minmax-sse-c4.c",
+    "src/f32-avgpool/9x-minmax-sse-c4.c",
     "src/f32-clamp/gen/sse-x4.c",
     "src/f32-clamp/gen/sse-x8.c",
     "src/f32-dwconv-spchw/3x3p1-sse.c",
     "src/f32-dwconv-spchw/3x3s2p1-sse.c",
-    "src/f32-dwconv/gen/up4x25-sse-acc2.c",
-    "src/f32-dwconv/gen/up4x25-sse.c",
-    "src/f32-dwconv/gen/up4x4-sse-acc2.c",
-    "src/f32-dwconv/gen/up4x4-sse.c",
-    "src/f32-dwconv/gen/up4x9-sse-acc2.c",
-    "src/f32-dwconv/gen/up4x9-sse.c",
-    "src/f32-dwconv/gen/up8x25-sse-acc2.c",
-    "src/f32-dwconv/gen/up8x25-sse.c",
-    "src/f32-dwconv/gen/up8x4-sse-acc2.c",
-    "src/f32-dwconv/gen/up8x4-sse.c",
-    "src/f32-dwconv/gen/up8x9-sse-acc2.c",
-    "src/f32-dwconv/gen/up8x9-sse.c",
+    "src/f32-dwconv/gen/up4x25-minmax-sse-acc2.c",
+    "src/f32-dwconv/gen/up4x25-minmax-sse.c",
+    "src/f32-dwconv/gen/up4x4-minmax-sse-acc2.c",
+    "src/f32-dwconv/gen/up4x4-minmax-sse.c",
+    "src/f32-dwconv/gen/up4x9-minmax-sse-acc2.c",
+    "src/f32-dwconv/gen/up4x9-minmax-sse.c",
+    "src/f32-dwconv/gen/up8x25-minmax-sse-acc2.c",
+    "src/f32-dwconv/gen/up8x25-minmax-sse.c",
+    "src/f32-dwconv/gen/up8x4-minmax-sse-acc2.c",
+    "src/f32-dwconv/gen/up8x4-minmax-sse.c",
+    "src/f32-dwconv/gen/up8x9-minmax-sse-acc2.c",
+    "src/f32-dwconv/gen/up8x9-minmax-sse.c",
     "src/f32-gavgpool-spchw/sse-x4.c",
-    "src/f32-gavgpool/7p7x-sse-c4.c",
-    "src/f32-gavgpool/7x-sse-c4.c",
-    "src/f32-gemm/gen/1x8-sse-dup.c",
-    "src/f32-gemm/gen/1x8-sse-load1.c",
-    "src/f32-gemm/gen/1x8s4-sse.c",
-    "src/f32-gemm/gen/4x2c4-sse.c",
-    "src/f32-gemm/gen/4x8-sse-dup.c",
-    "src/f32-gemm/gen/4x8-sse-load1.c",
-    "src/f32-gemm/gen/4x8s4-sse.c",
-    "src/f32-gemm/gen-inc/1x8-sse-dup.c",
-    "src/f32-gemm/gen-inc/1x8-sse-load1.c",
-    "src/f32-gemm/gen-inc/1x8s4-sse.c",
-    "src/f32-gemm/gen-inc/4x8-sse-dup.c",
-    "src/f32-gemm/gen-inc/4x8-sse-load1.c",
-    "src/f32-gemm/gen-inc/4x8s4-sse.c",
+    "src/f32-gavgpool/7p7x-minmax-sse-c4.c",
+    "src/f32-gavgpool/7x-minmax-sse-c4.c",
+    "src/f32-gemm/gen/1x8-minmax-sse-dup.c",
+    "src/f32-gemm/gen/1x8-minmax-sse-load1.c",
+    "src/f32-gemm/gen/1x8s4-minmax-sse.c",
+    "src/f32-gemm/gen/4x2c4-minmax-sse.c",
+    "src/f32-gemm/gen/4x8-minmax-sse-dup.c",
+    "src/f32-gemm/gen/4x8-minmax-sse-load1.c",
+    "src/f32-gemm/gen/4x8s4-minmax-sse.c",
+    "src/f32-gemm/gen-inc/1x8inc-minmax-sse-dup.c",
+    "src/f32-gemm/gen-inc/1x8inc-minmax-sse-load1.c",
+    "src/f32-gemm/gen-inc/1x8s4inc-minmax-sse.c",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-sse-dup.c",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-sse-load1.c",
+    "src/f32-gemm/gen-inc/4x8s4inc-minmax-sse.c",
     "src/f32-hswish/gen/sse-x4.c",
     "src/f32-hswish/gen/sse-x8.c",
     "src/f32-ibilinear/gen/sse-c4.c",
     "src/f32-ibilinear/gen/sse-c8.c",
-    "src/f32-igemm/gen/1x8-sse-dup.c",
-    "src/f32-igemm/gen/1x8-sse-load1.c",
-    "src/f32-igemm/gen/1x8s4-sse.c",
-    "src/f32-igemm/gen/4x2c4-sse.c",
-    "src/f32-igemm/gen/4x8-sse-dup.c",
-    "src/f32-igemm/gen/4x8-sse-load1.c",
-    "src/f32-igemm/gen/4x8s4-sse.c",
-    "src/f32-maxpool/9p8x-sse-c4.c",
-    "src/f32-pavgpool/9p8x-sse-c4.c",
-    "src/f32-pavgpool/9x-sse-c4.c",
-    "src/f32-ppmm/gen/4x8-sse.c",
+    "src/f32-igemm/gen/1x8-minmax-sse-dup.c",
+    "src/f32-igemm/gen/1x8-minmax-sse-load1.c",
+    "src/f32-igemm/gen/1x8s4-minmax-sse.c",
+    "src/f32-igemm/gen/4x2c4-minmax-sse.c",
+    "src/f32-igemm/gen/4x8-minmax-sse-dup.c",
+    "src/f32-igemm/gen/4x8-minmax-sse-load1.c",
+    "src/f32-igemm/gen/4x8s4-minmax-sse.c",
+    "src/f32-maxpool/9p8x-minmax-sse-c4.c",
+    "src/f32-pavgpool/9p8x-minmax-sse-c4.c",
+    "src/f32-pavgpool/9x-minmax-sse-c4.c",
+    "src/f32-ppmm/gen/4x8-minmax-sse.c",
     "src/f32-rmax/sse.c",
-    "src/f32-spmm/gen/4x1-sse.c",
-    "src/f32-spmm/gen/8x1-sse.c",
-    "src/f32-vbinary/gen/vadd-sse-x4.c",
-    "src/f32-vbinary/gen/vadd-sse-x8.c",
-    "src/f32-vbinary/gen/vaddc-sse-x4.c",
-    "src/f32-vbinary/gen/vaddc-sse-x8.c",
-    "src/f32-vbinary/gen/vdiv-sse-x4.c",
-    "src/f32-vbinary/gen/vdiv-sse-x8.c",
-    "src/f32-vbinary/gen/vdivc-sse-x4.c",
-    "src/f32-vbinary/gen/vdivc-sse-x8.c",
+    "src/f32-spmm/gen/4x1-minmax-sse.c",
+    "src/f32-spmm/gen/8x1-minmax-sse.c",
+    "src/f32-vbinary/gen/vadd-minmax-sse-x4.c",
+    "src/f32-vbinary/gen/vadd-minmax-sse-x8.c",
+    "src/f32-vbinary/gen/vaddc-minmax-sse-x4.c",
+    "src/f32-vbinary/gen/vaddc-minmax-sse-x8.c",
+    "src/f32-vbinary/gen/vdiv-minmax-sse-x4.c",
+    "src/f32-vbinary/gen/vdiv-minmax-sse-x8.c",
+    "src/f32-vbinary/gen/vdivc-minmax-sse-x4.c",
+    "src/f32-vbinary/gen/vdivc-minmax-sse-x8.c",
     "src/f32-vbinary/gen/vmax-sse-x4.c",
     "src/f32-vbinary/gen/vmax-sse-x8.c",
     "src/f32-vbinary/gen/vmaxc-sse-x4.c",
@@ -809,20 +838,20 @@
     "src/f32-vbinary/gen/vmin-sse-x8.c",
     "src/f32-vbinary/gen/vminc-sse-x4.c",
     "src/f32-vbinary/gen/vminc-sse-x8.c",
-    "src/f32-vbinary/gen/vmul-sse-x4.c",
-    "src/f32-vbinary/gen/vmul-sse-x8.c",
-    "src/f32-vbinary/gen/vmulc-sse-x4.c",
-    "src/f32-vbinary/gen/vmulc-sse-x8.c",
-    "src/f32-vbinary/gen/vrdivc-sse-x4.c",
-    "src/f32-vbinary/gen/vrdivc-sse-x8.c",
-    "src/f32-vbinary/gen/vrsubc-sse-x4.c",
-    "src/f32-vbinary/gen/vrsubc-sse-x8.c",
-    "src/f32-vbinary/gen/vsub-sse-x4.c",
-    "src/f32-vbinary/gen/vsub-sse-x8.c",
-    "src/f32-vbinary/gen/vsubc-sse-x4.c",
-    "src/f32-vbinary/gen/vsubc-sse-x8.c",
-    "src/f32-vmulcaddc/gen/c4-sse-2x.c",
-    "src/f32-vmulcaddc/gen/c8-sse-2x.c",
+    "src/f32-vbinary/gen/vmul-minmax-sse-x4.c",
+    "src/f32-vbinary/gen/vmul-minmax-sse-x8.c",
+    "src/f32-vbinary/gen/vmulc-minmax-sse-x4.c",
+    "src/f32-vbinary/gen/vmulc-minmax-sse-x8.c",
+    "src/f32-vbinary/gen/vrdivc-minmax-sse-x4.c",
+    "src/f32-vbinary/gen/vrdivc-minmax-sse-x8.c",
+    "src/f32-vbinary/gen/vrsubc-minmax-sse-x4.c",
+    "src/f32-vbinary/gen/vrsubc-minmax-sse-x8.c",
+    "src/f32-vbinary/gen/vsub-minmax-sse-x4.c",
+    "src/f32-vbinary/gen/vsub-minmax-sse-x8.c",
+    "src/f32-vbinary/gen/vsubc-minmax-sse-x4.c",
+    "src/f32-vbinary/gen/vsubc-minmax-sse-x8.c",
+    "src/f32-vmulcaddc/gen/c4-minmax-sse-2x.c",
+    "src/f32-vmulcaddc/gen/c8-minmax-sse-2x.c",
     "src/x32-packx/x4-sse.c",
 ]
 
@@ -850,17 +879,17 @@
     "src/f32-sigmoid/gen/sse2-p5-div-x16.c",
     "src/f32-sigmoid/gen/sse2-p5-div-x20.c",
     "src/f32-sigmoid/gen/sse2-p5-div-x24.c",
-    "src/q8-avgpool/9p8x-sse2-c8.c",
-    "src/q8-avgpool/9x-sse2-c8.c",
-    "src/q8-igemm/4x4c2-sse2.c",
-    "src/q8-dwconv/up8x9-sse2.c",
-    "src/q8-gavgpool/7p7x-sse2-c8.c",
-    "src/q8-gavgpool/7x-sse2-c8.c",
-    "src/q8-gemm/2x4c8-sse2.c",
-    "src/q8-gemm/4x4c2-sse2.c",
-    "src/q8-vadd/sse2.c",
+    "src/q8-avgpool/9p8x-minmax-sse2-c8.c",
+    "src/q8-avgpool/9x-minmax-sse2-c8.c",
+    "src/q8-igemm/4x4c2-minmax-sse2.c",
+    "src/q8-dwconv/up8x9-minmax-sse2.c",
+    "src/q8-gavgpool/7p7x-minmax-sse2-c8.c",
+    "src/q8-gavgpool/7x-minmax-sse2-c8.c",
+    "src/q8-gemm/2x4c8-minmax-sse2.c",
+    "src/q8-gemm/4x4c2-minmax-sse2.c",
+    "src/q8-vadd/minmax-sse2.c",
     "src/u8-clamp/sse2-x64.c",
-    "src/u8-maxpool/9p8x-sse2-c16.c",
+    "src/u8-maxpool/9p8x-minmax-sse2-c16.c",
     "src/u8-rmax/sse2.c",
     "src/x32-pad/x2-sse2.c",
     "src/x32-zip/x2-sse2.c",
@@ -903,58 +932,58 @@
 AVX_UKERNELS = [
     "src/f32-clamp/gen/avx-x8.c",
     "src/f32-clamp/gen/avx-x16.c",
-    "src/f32-dwconv/gen/up16x4-avx-acc2.c",
-    "src/f32-dwconv/gen/up16x4-avx.c",
-    "src/f32-dwconv/gen/up8x4-avx-acc2.c",
-    "src/f32-dwconv/gen/up8x4-avx.c",
-    "src/f32-dwconv/gen/up16x9-avx-acc2.c",
-    "src/f32-dwconv/gen/up16x9-avx.c",
-    "src/f32-dwconv/gen/up8x9-avx-acc2.c",
-    "src/f32-dwconv/gen/up8x9-avx.c",
-    "src/f32-dwconv/gen/up16x25-avx-acc2.c",
-    "src/f32-dwconv/gen/up16x25-avx.c",
-    "src/f32-dwconv/gen/up8x25-avx-acc2.c",
-    "src/f32-dwconv/gen/up8x25-avx.c",
-    "src/f32-gemm/gen/1x8-avx-broadcast.c",
-    "src/f32-gemm/gen/4x8-avx-broadcast.c",
-    "src/f32-gemm/gen/5x8-avx-broadcast.c",
-    "src/f32-gemm/gen/6x8-avx-broadcast.c",
-    "src/f32-gemm/gen/7x8-avx-broadcast.c",
-    "src/f32-gemm/gen/1x16-avx-broadcast.c",
-    "src/f32-gemm/gen/3x16-avx-broadcast.c",
-    "src/f32-gemm/gen/4x16-avx-broadcast.c",
-    "src/f32-gemm/gen/5x16-avx-broadcast.c",
-    "src/f32-gemm/gen-inc/1x8-avx-broadcast.c",
-    "src/f32-gemm/gen-inc/4x8-avx-broadcast.c",
-    "src/f32-gemm/gen-inc/5x8-avx-broadcast.c",
-    "src/f32-gemm/gen-inc/6x8-avx-broadcast.c",
-    "src/f32-gemm/gen-inc/7x8-avx-broadcast.c",
-    "src/f32-gemm/gen-inc/1x16-avx-broadcast.c",
-    "src/f32-gemm/gen-inc/3x16-avx-broadcast.c",
-    "src/f32-gemm/gen-inc/4x16-avx-broadcast.c",
-    "src/f32-gemm/gen-inc/5x16-avx-broadcast.c",
+    "src/f32-dwconv/gen/up16x4-minmax-avx-acc2.c",
+    "src/f32-dwconv/gen/up16x4-minmax-avx.c",
+    "src/f32-dwconv/gen/up8x4-minmax-avx-acc2.c",
+    "src/f32-dwconv/gen/up8x4-minmax-avx.c",
+    "src/f32-dwconv/gen/up16x9-minmax-avx-acc2.c",
+    "src/f32-dwconv/gen/up16x9-minmax-avx.c",
+    "src/f32-dwconv/gen/up8x9-minmax-avx-acc2.c",
+    "src/f32-dwconv/gen/up8x9-minmax-avx.c",
+    "src/f32-dwconv/gen/up16x25-minmax-avx-acc2.c",
+    "src/f32-dwconv/gen/up16x25-minmax-avx.c",
+    "src/f32-dwconv/gen/up8x25-minmax-avx-acc2.c",
+    "src/f32-dwconv/gen/up8x25-minmax-avx.c",
+    "src/f32-gemm/gen/1x8-minmax-avx-broadcast.c",
+    "src/f32-gemm/gen/4x8-minmax-avx-broadcast.c",
+    "src/f32-gemm/gen/5x8-minmax-avx-broadcast.c",
+    "src/f32-gemm/gen/6x8-minmax-avx-broadcast.c",
+    "src/f32-gemm/gen/7x8-minmax-avx-broadcast.c",
+    "src/f32-gemm/gen/1x16-minmax-avx-broadcast.c",
+    "src/f32-gemm/gen/3x16-minmax-avx-broadcast.c",
+    "src/f32-gemm/gen/4x16-minmax-avx-broadcast.c",
+    "src/f32-gemm/gen/5x16-minmax-avx-broadcast.c",
+    "src/f32-gemm/gen-inc/1x8inc-minmax-avx-broadcast.c",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-avx-broadcast.c",
+    "src/f32-gemm/gen-inc/5x8inc-minmax-avx-broadcast.c",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-avx-broadcast.c",
+    "src/f32-gemm/gen-inc/7x8inc-minmax-avx-broadcast.c",
+    "src/f32-gemm/gen-inc/1x16inc-minmax-avx-broadcast.c",
+    "src/f32-gemm/gen-inc/3x16inc-minmax-avx-broadcast.c",
+    "src/f32-gemm/gen-inc/4x16inc-minmax-avx-broadcast.c",
+    "src/f32-gemm/gen-inc/5x16inc-minmax-avx-broadcast.c",
     "src/f32-hswish/gen/avx-x8.c",
     "src/f32-hswish/gen/avx-x16.c",
-    "src/f32-igemm/gen/1x8-avx-broadcast.c",
-    "src/f32-igemm/gen/4x8-avx-broadcast.c",
-    "src/f32-igemm/gen/5x8-avx-broadcast.c",
-    "src/f32-igemm/gen/6x8-avx-broadcast.c",
-    "src/f32-igemm/gen/7x8-avx-broadcast.c",
-    "src/f32-igemm/gen/1x16-avx-broadcast.c",
-    "src/f32-igemm/gen/3x16-avx-broadcast.c",
-    "src/f32-igemm/gen/4x16-avx-broadcast.c",
-    "src/f32-igemm/gen/5x16-avx-broadcast.c",
+    "src/f32-igemm/gen/1x8-minmax-avx-broadcast.c",
+    "src/f32-igemm/gen/4x8-minmax-avx-broadcast.c",
+    "src/f32-igemm/gen/5x8-minmax-avx-broadcast.c",
+    "src/f32-igemm/gen/6x8-minmax-avx-broadcast.c",
+    "src/f32-igemm/gen/7x8-minmax-avx-broadcast.c",
+    "src/f32-igemm/gen/1x16-minmax-avx-broadcast.c",
+    "src/f32-igemm/gen/3x16-minmax-avx-broadcast.c",
+    "src/f32-igemm/gen/4x16-minmax-avx-broadcast.c",
+    "src/f32-igemm/gen/5x16-minmax-avx-broadcast.c",
     "src/f32-prelu/gen/avx-2x8.c",
     "src/f32-prelu/gen/avx-2x16.c",
     "src/f32-rmax/avx.c",
-    "src/f32-vbinary/gen/vadd-avx-x8.c",
-    "src/f32-vbinary/gen/vadd-avx-x16.c",
-    "src/f32-vbinary/gen/vaddc-avx-x8.c",
-    "src/f32-vbinary/gen/vaddc-avx-x16.c",
-    "src/f32-vbinary/gen/vdiv-avx-x8.c",
-    "src/f32-vbinary/gen/vdiv-avx-x16.c",
-    "src/f32-vbinary/gen/vdivc-avx-x8.c",
-    "src/f32-vbinary/gen/vdivc-avx-x16.c",
+    "src/f32-vbinary/gen/vadd-minmax-avx-x8.c",
+    "src/f32-vbinary/gen/vadd-minmax-avx-x16.c",
+    "src/f32-vbinary/gen/vaddc-minmax-avx-x8.c",
+    "src/f32-vbinary/gen/vaddc-minmax-avx-x16.c",
+    "src/f32-vbinary/gen/vdiv-minmax-avx-x8.c",
+    "src/f32-vbinary/gen/vdiv-minmax-avx-x16.c",
+    "src/f32-vbinary/gen/vdivc-minmax-avx-x8.c",
+    "src/f32-vbinary/gen/vdivc-minmax-avx-x16.c",
     "src/f32-vbinary/gen/vmax-avx-x8.c",
     "src/f32-vbinary/gen/vmax-avx-x16.c",
     "src/f32-vbinary/gen/vmaxc-avx-x8.c",
@@ -963,78 +992,78 @@
     "src/f32-vbinary/gen/vmin-avx-x16.c",
     "src/f32-vbinary/gen/vminc-avx-x8.c",
     "src/f32-vbinary/gen/vminc-avx-x16.c",
-    "src/f32-vbinary/gen/vmul-avx-x8.c",
-    "src/f32-vbinary/gen/vmul-avx-x16.c",
-    "src/f32-vbinary/gen/vmulc-avx-x8.c",
-    "src/f32-vbinary/gen/vmulc-avx-x16.c",
-    "src/f32-vbinary/gen/vrdivc-avx-x8.c",
-    "src/f32-vbinary/gen/vrdivc-avx-x16.c",
-    "src/f32-vbinary/gen/vrsubc-avx-x8.c",
-    "src/f32-vbinary/gen/vrsubc-avx-x16.c",
-    "src/f32-vbinary/gen/vsub-avx-x8.c",
-    "src/f32-vbinary/gen/vsub-avx-x16.c",
-    "src/f32-vbinary/gen/vsubc-avx-x8.c",
-    "src/f32-vbinary/gen/vsubc-avx-x16.c",
+    "src/f32-vbinary/gen/vmul-minmax-avx-x8.c",
+    "src/f32-vbinary/gen/vmul-minmax-avx-x16.c",
+    "src/f32-vbinary/gen/vmulc-minmax-avx-x8.c",
+    "src/f32-vbinary/gen/vmulc-minmax-avx-x16.c",
+    "src/f32-vbinary/gen/vrdivc-minmax-avx-x8.c",
+    "src/f32-vbinary/gen/vrdivc-minmax-avx-x16.c",
+    "src/f32-vbinary/gen/vrsubc-minmax-avx-x8.c",
+    "src/f32-vbinary/gen/vrsubc-minmax-avx-x16.c",
+    "src/f32-vbinary/gen/vsub-minmax-avx-x8.c",
+    "src/f32-vbinary/gen/vsub-minmax-avx-x16.c",
+    "src/f32-vbinary/gen/vsubc-minmax-avx-x8.c",
+    "src/f32-vbinary/gen/vsubc-minmax-avx-x16.c",
     "src/f32-vscale/avx-unroll32.c",
 ]
 
 FMA3_UKERNELS = [
-    "src/f32-dwconv/gen/up16x4-fma3-acc2.c",
-    "src/f32-dwconv/gen/up16x4-fma3.c",
-    "src/f32-dwconv/gen/up8x4-fma3-acc2.c",
-    "src/f32-dwconv/gen/up8x4-fma3.c",
-    "src/f32-dwconv/gen/up16x9-fma3-acc2.c",
-    "src/f32-dwconv/gen/up16x9-fma3.c",
-    "src/f32-dwconv/gen/up8x9-fma3-acc2.c",
-    "src/f32-dwconv/gen/up8x9-fma3.c",
-    "src/f32-dwconv/gen/up16x25-fma3-acc2.c",
-    "src/f32-dwconv/gen/up16x25-fma3.c",
-    "src/f32-dwconv/gen/up8x25-fma3-acc2.c",
-    "src/f32-dwconv/gen/up8x25-fma3.c",
-    "src/f32-gemm/gen/1x8-fma3-broadcast.c",
-    "src/f32-gemm/gen/4x8-fma3-broadcast.c",
-    "src/f32-gemm/gen/5x8-fma3-broadcast.c",
-    "src/f32-gemm/gen/6x8-fma3-broadcast.c",
-    "src/f32-gemm/gen/7x8-fma3-broadcast.c",
-    "src/f32-gemm/gen/8x8-fma3-broadcast.c",
-    "src/f32-gemm/gen/1x16-fma3-broadcast.c",
-    "src/f32-gemm/gen/3x16-fma3-broadcast.c",
-    "src/f32-gemm/gen/4x16-fma3-broadcast.c",
-    "src/f32-gemm/gen/5x16-fma3-broadcast.c",
-    "src/f32-gemm/gen/1x16s4-fma3-broadcast.c",
-    "src/f32-gemm/gen/3x16s4-fma3-broadcast.c",
-    "src/f32-gemm/gen/4x16s4-fma3-broadcast.c",
-    "src/f32-gemm/gen/5x16s4-fma3-broadcast.c",
-    "src/f32-gemm/gen-inc/1x8-fma3-broadcast.c",
-    "src/f32-gemm/gen-inc/4x8-fma3-broadcast.c",
-    "src/f32-gemm/gen-inc/5x8-fma3-broadcast.c",
-    "src/f32-gemm/gen-inc/6x8-fma3-broadcast.c",
-    "src/f32-gemm/gen-inc/7x8-fma3-broadcast.c",
-    "src/f32-gemm/gen-inc/8x8-fma3-broadcast.c",
-    "src/f32-gemm/gen-inc/1x16-fma3-broadcast.c",
-    "src/f32-gemm/gen-inc/3x16-fma3-broadcast.c",
-    "src/f32-gemm/gen-inc/4x16-fma3-broadcast.c",
-    "src/f32-gemm/gen-inc/5x16-fma3-broadcast.c",
-    "src/f32-gemm/gen-inc/1x16s4-fma3-broadcast.c",
-    "src/f32-gemm/gen-inc/3x16s4-fma3-broadcast.c",
-    "src/f32-gemm/gen-inc/4x16s4-fma3-broadcast.c",
-    "src/f32-gemm/gen-inc/5x16s4-fma3-broadcast.c",
+    "src/f32-dwconv/gen/up16x4-minmax-fma3-acc2.c",
+    "src/f32-dwconv/gen/up16x4-minmax-fma3.c",
+    "src/f32-dwconv/gen/up8x4-minmax-fma3-acc2.c",
+    "src/f32-dwconv/gen/up8x4-minmax-fma3.c",
+    "src/f32-dwconv/gen/up16x9-minmax-fma3-acc2.c",
+    "src/f32-dwconv/gen/up16x9-minmax-fma3.c",
+    "src/f32-dwconv/gen/up8x9-minmax-fma3-acc2.c",
+    "src/f32-dwconv/gen/up8x9-minmax-fma3.c",
+    "src/f32-dwconv/gen/up16x25-minmax-fma3-acc2.c",
+    "src/f32-dwconv/gen/up16x25-minmax-fma3.c",
+    "src/f32-dwconv/gen/up8x25-minmax-fma3-acc2.c",
+    "src/f32-dwconv/gen/up8x25-minmax-fma3.c",
+    "src/f32-gemm/gen/1x8-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen/4x8-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen/5x8-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen/6x8-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen/7x8-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen/8x8-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen/1x16-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen/3x16-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen/4x16-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen/5x16-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen/1x16s4-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen/3x16s4-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen/4x16s4-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen/5x16s4-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen-inc/1x8inc-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen-inc/5x8inc-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen-inc/7x8inc-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen-inc/8x8inc-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen-inc/1x16inc-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen-inc/3x16inc-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen-inc/4x16inc-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen-inc/5x16inc-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen-inc/1x16s4inc-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen-inc/3x16s4inc-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen-inc/4x16s4inc-minmax-fma3-broadcast.c",
+    "src/f32-gemm/gen-inc/5x16s4inc-minmax-fma3-broadcast.c",
     "src/f32-hswish/gen/fma3-x8.c",
     "src/f32-hswish/gen/fma3-x16.c",
-    "src/f32-igemm/gen/1x8-fma3-broadcast.c",
-    "src/f32-igemm/gen/4x8-fma3-broadcast.c",
-    "src/f32-igemm/gen/5x8-fma3-broadcast.c",
-    "src/f32-igemm/gen/6x8-fma3-broadcast.c",
-    "src/f32-igemm/gen/7x8-fma3-broadcast.c",
-    "src/f32-igemm/gen/8x8-fma3-broadcast.c",
-    "src/f32-igemm/gen/1x16-fma3-broadcast.c",
-    "src/f32-igemm/gen/3x16-fma3-broadcast.c",
-    "src/f32-igemm/gen/4x16-fma3-broadcast.c",
-    "src/f32-igemm/gen/5x16-fma3-broadcast.c",
-    "src/f32-igemm/gen/1x16s4-fma3-broadcast.c",
-    "src/f32-igemm/gen/3x16s4-fma3-broadcast.c",
-    "src/f32-igemm/gen/4x16s4-fma3-broadcast.c",
-    "src/f32-igemm/gen/5x16s4-fma3-broadcast.c",
+    "src/f32-igemm/gen/1x8-minmax-fma3-broadcast.c",
+    "src/f32-igemm/gen/4x8-minmax-fma3-broadcast.c",
+    "src/f32-igemm/gen/5x8-minmax-fma3-broadcast.c",
+    "src/f32-igemm/gen/6x8-minmax-fma3-broadcast.c",
+    "src/f32-igemm/gen/7x8-minmax-fma3-broadcast.c",
+    "src/f32-igemm/gen/8x8-minmax-fma3-broadcast.c",
+    "src/f32-igemm/gen/1x16-minmax-fma3-broadcast.c",
+    "src/f32-igemm/gen/3x16-minmax-fma3-broadcast.c",
+    "src/f32-igemm/gen/4x16-minmax-fma3-broadcast.c",
+    "src/f32-igemm/gen/5x16-minmax-fma3-broadcast.c",
+    "src/f32-igemm/gen/1x16s4-minmax-fma3-broadcast.c",
+    "src/f32-igemm/gen/3x16s4-minmax-fma3-broadcast.c",
+    "src/f32-igemm/gen/4x16s4-minmax-fma3-broadcast.c",
+    "src/f32-igemm/gen/5x16s4-minmax-fma3-broadcast.c",
 ]
 
 AVX2_UKERNELS = [
@@ -1144,38 +1173,38 @@
 AVX512F_UKERNELS = [
     "src/f32-clamp/gen/avx512f-x16.c",
     "src/f32-clamp/gen/avx512f-x32.c",
-    "src/f32-dwconv/gen/up32x4-avx512f-acc2.c",
-    "src/f32-dwconv/gen/up32x4-avx512f.c",
-    "src/f32-dwconv/gen/up16x4-avx512f-acc2.c",
-    "src/f32-dwconv/gen/up16x4-avx512f.c",
-    "src/f32-dwconv/gen/up32x9-avx512f-acc2.c",
-    "src/f32-dwconv/gen/up32x9-avx512f.c",
-    "src/f32-dwconv/gen/up16x9-avx512f-acc2.c",
-    "src/f32-dwconv/gen/up16x9-avx512f.c",
-    "src/f32-dwconv/gen/up32x25-avx512f-acc2.c",
-    "src/f32-dwconv/gen/up32x25-avx512f.c",
-    "src/f32-dwconv/gen/up16x25-avx512f-acc2.c",
-    "src/f32-dwconv/gen/up16x25-avx512f.c",
-    "src/f32-gemm/gen/1x16-avx512f-broadcast.c",
-    "src/f32-gemm/gen/4x16-avx512f-broadcast.c",
-    "src/f32-gemm/gen/5x16-avx512f-broadcast.c",
-    "src/f32-gemm/gen/6x16-avx512f-broadcast.c",
-    "src/f32-gemm/gen/7x16-avx512f-broadcast.c",
-    "src/f32-gemm/gen/8x16-avx512f-broadcast.c",
-    "src/f32-gemm/gen-inc/1x16-avx512f-broadcast.c",
-    "src/f32-gemm/gen-inc/4x16-avx512f-broadcast.c",
-    "src/f32-gemm/gen-inc/5x16-avx512f-broadcast.c",
-    "src/f32-gemm/gen-inc/6x16-avx512f-broadcast.c",
-    "src/f32-gemm/gen-inc/7x16-avx512f-broadcast.c",
-    "src/f32-gemm/gen-inc/8x16-avx512f-broadcast.c",
+    "src/f32-dwconv/gen/up32x4-minmax-avx512f-acc2.c",
+    "src/f32-dwconv/gen/up32x4-minmax-avx512f.c",
+    "src/f32-dwconv/gen/up16x4-minmax-avx512f-acc2.c",
+    "src/f32-dwconv/gen/up16x4-minmax-avx512f.c",
+    "src/f32-dwconv/gen/up32x9-minmax-avx512f-acc2.c",
+    "src/f32-dwconv/gen/up32x9-minmax-avx512f.c",
+    "src/f32-dwconv/gen/up16x9-minmax-avx512f-acc2.c",
+    "src/f32-dwconv/gen/up16x9-minmax-avx512f.c",
+    "src/f32-dwconv/gen/up32x25-minmax-avx512f-acc2.c",
+    "src/f32-dwconv/gen/up32x25-minmax-avx512f.c",
+    "src/f32-dwconv/gen/up16x25-minmax-avx512f-acc2.c",
+    "src/f32-dwconv/gen/up16x25-minmax-avx512f.c",
+    "src/f32-gemm/gen/1x16-minmax-avx512f-broadcast.c",
+    "src/f32-gemm/gen/4x16-minmax-avx512f-broadcast.c",
+    "src/f32-gemm/gen/5x16-minmax-avx512f-broadcast.c",
+    "src/f32-gemm/gen/6x16-minmax-avx512f-broadcast.c",
+    "src/f32-gemm/gen/7x16-minmax-avx512f-broadcast.c",
+    "src/f32-gemm/gen/8x16-minmax-avx512f-broadcast.c",
+    "src/f32-gemm/gen-inc/1x16inc-minmax-avx512f-broadcast.c",
+    "src/f32-gemm/gen-inc/4x16inc-minmax-avx512f-broadcast.c",
+    "src/f32-gemm/gen-inc/5x16inc-minmax-avx512f-broadcast.c",
+    "src/f32-gemm/gen-inc/6x16inc-minmax-avx512f-broadcast.c",
+    "src/f32-gemm/gen-inc/7x16inc-minmax-avx512f-broadcast.c",
+    "src/f32-gemm/gen-inc/8x16inc-minmax-avx512f-broadcast.c",
     "src/f32-hswish/gen/avx512f-x16.c",
     "src/f32-hswish/gen/avx512f-x32.c",
-    "src/f32-igemm/gen/1x16-avx512f-broadcast.c",
-    "src/f32-igemm/gen/4x16-avx512f-broadcast.c",
-    "src/f32-igemm/gen/5x16-avx512f-broadcast.c",
-    "src/f32-igemm/gen/6x16-avx512f-broadcast.c",
-    "src/f32-igemm/gen/7x16-avx512f-broadcast.c",
-    "src/f32-igemm/gen/8x16-avx512f-broadcast.c",
+    "src/f32-igemm/gen/1x16-minmax-avx512f-broadcast.c",
+    "src/f32-igemm/gen/4x16-minmax-avx512f-broadcast.c",
+    "src/f32-igemm/gen/5x16-minmax-avx512f-broadcast.c",
+    "src/f32-igemm/gen/6x16-minmax-avx512f-broadcast.c",
+    "src/f32-igemm/gen/7x16-minmax-avx512f-broadcast.c",
+    "src/f32-igemm/gen/8x16-minmax-avx512f-broadcast.c",
     "src/f32-prelu/gen/avx512f-2x16.c",
     "src/f32-prelu/gen/avx512f-2x32.c",
     "src/f32-raddexpminusmax/gen/avx512f-p5-scalef-x128.c",
@@ -1215,14 +1244,14 @@
     "src/f32-raddstoreexpminusmax/gen/avx512f-p5-scalef-x192-acc3.c",
     "src/f32-raddstoreexpminusmax/gen/avx512f-p5-scalef-x192-acc6.c",
     "src/f32-rmax/avx512f.c",
-    "src/f32-vbinary/gen/vadd-avx512f-x16.c",
-    "src/f32-vbinary/gen/vadd-avx512f-x32.c",
-    "src/f32-vbinary/gen/vaddc-avx512f-x16.c",
-    "src/f32-vbinary/gen/vaddc-avx512f-x32.c",
-    "src/f32-vbinary/gen/vdiv-avx512f-x16.c",
-    "src/f32-vbinary/gen/vdiv-avx512f-x32.c",
-    "src/f32-vbinary/gen/vdivc-avx512f-x16.c",
-    "src/f32-vbinary/gen/vdivc-avx512f-x32.c",
+    "src/f32-vbinary/gen/vadd-minmax-avx512f-x16.c",
+    "src/f32-vbinary/gen/vadd-minmax-avx512f-x32.c",
+    "src/f32-vbinary/gen/vaddc-minmax-avx512f-x16.c",
+    "src/f32-vbinary/gen/vaddc-minmax-avx512f-x32.c",
+    "src/f32-vbinary/gen/vdiv-minmax-avx512f-x16.c",
+    "src/f32-vbinary/gen/vdiv-minmax-avx512f-x32.c",
+    "src/f32-vbinary/gen/vdivc-minmax-avx512f-x16.c",
+    "src/f32-vbinary/gen/vdivc-minmax-avx512f-x32.c",
     "src/f32-vbinary/gen/vmax-avx512f-x16.c",
     "src/f32-vbinary/gen/vmax-avx512f-x32.c",
     "src/f32-vbinary/gen/vmaxc-avx512f-x16.c",
@@ -1231,18 +1260,18 @@
     "src/f32-vbinary/gen/vmin-avx512f-x32.c",
     "src/f32-vbinary/gen/vminc-avx512f-x16.c",
     "src/f32-vbinary/gen/vminc-avx512f-x32.c",
-    "src/f32-vbinary/gen/vmul-avx512f-x16.c",
-    "src/f32-vbinary/gen/vmul-avx512f-x32.c",
-    "src/f32-vbinary/gen/vmulc-avx512f-x16.c",
-    "src/f32-vbinary/gen/vmulc-avx512f-x32.c",
-    "src/f32-vbinary/gen/vrdivc-avx512f-x16.c",
-    "src/f32-vbinary/gen/vrdivc-avx512f-x32.c",
-    "src/f32-vbinary/gen/vrsubc-avx512f-x16.c",
-    "src/f32-vbinary/gen/vrsubc-avx512f-x32.c",
-    "src/f32-vbinary/gen/vsub-avx512f-x16.c",
-    "src/f32-vbinary/gen/vsub-avx512f-x32.c",
-    "src/f32-vbinary/gen/vsubc-avx512f-x16.c",
-    "src/f32-vbinary/gen/vsubc-avx512f-x32.c",
+    "src/f32-vbinary/gen/vmul-minmax-avx512f-x16.c",
+    "src/f32-vbinary/gen/vmul-minmax-avx512f-x32.c",
+    "src/f32-vbinary/gen/vmulc-minmax-avx512f-x16.c",
+    "src/f32-vbinary/gen/vmulc-minmax-avx512f-x32.c",
+    "src/f32-vbinary/gen/vrdivc-minmax-avx512f-x16.c",
+    "src/f32-vbinary/gen/vrdivc-minmax-avx512f-x32.c",
+    "src/f32-vbinary/gen/vrsubc-minmax-avx512f-x16.c",
+    "src/f32-vbinary/gen/vrsubc-minmax-avx512f-x32.c",
+    "src/f32-vbinary/gen/vsub-minmax-avx512f-x16.c",
+    "src/f32-vbinary/gen/vsub-minmax-avx512f-x32.c",
+    "src/f32-vbinary/gen/vsubc-minmax-avx512f-x16.c",
+    "src/f32-vbinary/gen/vsubc-minmax-avx512f-x32.c",
     "src/f32-vscale/avx512f-unroll64.c",
     "src/f32-vscaleexpminusmax/gen/avx512f-p5-scalef-x16.c",
     "src/f32-vscaleexpminusmax/gen/avx512f-p5-scalef-x32.c",
@@ -1276,75 +1305,89 @@
 ]
 
 AARCH32_ASM_UKERNELS = [
-    "src/q8-dwconv/up8x9-aarch32-neon.S",
-    "src/f32-gemm/4x8-aarch32-neon-cortex-a53.S",
-    "src/f32-gemm/4x8-aarch32-neon-cortex-a55.S",
-    "src/f32-gemm/gen/4x8-aarch32-neon-cortex-a75.S",
-    "src/f32-gemm/gen/4x8-aarch32-neon-pld-cortex-a75.S",
-    "src/f32-gemm/4x8-aarch32-neon-ld64.S",
-    "src/f32-igemm/4x8-aarch32-neon-ld64.S",
-    "src/f32-igemm/gen/4x8-aarch32-neon-cortex-a75.S",
-    "src/f32-igemm/gen/4x8-aarch32-neon-pld-cortex-a75.S",
-    "src/f32-igemm/4x8-aarch32-neon-cortex-a53.S",
-    "src/f32-igemm/4x8-aarch32-neon-cortex-a55.S",
+    "src/q8-dwconv/up8x9-minmax-aarch32-neon.S",
+    "src/f32-gemm/4x8-minmax-aarch32-neon-cortex-a53.S",
+    "src/f32-gemm/4x8-minmax-aarch32-neon-cortex-a55.S",
+    "src/f32-gemm/gen/4x8-minmax-aarch32-neon-cortex-a75.S",
+    "src/f32-gemm/gen/4x8-minmax-aarch32-neon-pld-cortex-a75.S",
+    "src/f32-gemm/4x8-minmax-aarch32-neon-ld64.S",
+    "src/f32-igemm/4x8-minmax-aarch32-neon-ld64.S",
+    "src/f32-igemm/gen/4x8-minmax-aarch32-neon-cortex-a75.S",
+    "src/f32-igemm/gen/4x8-minmax-aarch32-neon-pld-cortex-a75.S",
+    "src/f32-igemm/4x8-minmax-aarch32-neon-cortex-a53.S",
+    "src/f32-igemm/4x8-minmax-aarch32-neon-cortex-a55.S",
 ]
 
 AARCH64_ASM_UKERNELS = [
-    "src/f32-dwconv/up4x9-aarch64-neonfma-cortex-a55.S",
-    "src/f32-dwconv/up4x9-aarch64-neonfma.S",
-    "src/f32-gemm/gen/1x12-aarch64-neonfma-cortex-a53.S",
-    "src/f32-gemm/gen/1x8-aarch64-neonfma-cortex-a53.S",
-    "src/f32-gemm/gen/1x8-aarch64-neonfma-cortex-a57.S",
-    "src/f32-gemm/gen/1x8-aarch64-neonfma-cortex-a75.S",
-    "src/f32-gemm/gen/4x12-aarch64-neonfma-cortex-a53.S",
-    "src/f32-gemm/gen/4x8-aarch64-neonfma-cortex-a53.S",
-    "src/f32-gemm/gen/4x8-aarch64-neonfma-cortex-a57.S",
-    "src/f32-gemm/gen/4x8-aarch64-neonfma-cortex-a75.S",
-    "src/f32-gemm/gen/4x8-aarch64-neonfma-ld128.S",
-    "src/f32-gemm/gen/4x8-aarch64-neonfma-ld64.S",
-    "src/f32-gemm/gen/5x8-aarch64-neonfma-cortex-a57.S",
-    "src/f32-gemm/gen/5x8-aarch64-neonfma-cortex-a75.S",
-    "src/f32-gemm/gen/6x8-aarch64-neonfma-cortex-a53.S",
-    "src/f32-gemm/gen/6x8-aarch64-neonfma-cortex-a55.S",
-    "src/f32-gemm/gen/6x8-aarch64-neonfma-cortex-a73.S",
-    "src/f32-gemm/gen/6x8-aarch64-neonfma-cortex-a57.S",
-    "src/f32-gemm/gen/6x8-aarch64-neonfma-cortex-a75.S",
-    "src/f32-gemm/gen/6x8-aarch64-neonfma-ld128.S",
-    "src/f32-gemm/gen/6x8-aarch64-neonfma-ld64.S",
-    "src/f32-gemm/gen-inc/1x12-aarch64-neonfma-cortex-a53.S",
-    "src/f32-gemm/gen-inc/1x8-aarch64-neonfma-cortex-a53.S",
-    "src/f32-gemm/gen-inc/1x8-aarch64-neonfma-cortex-a57.S",
-    "src/f32-gemm/gen-inc/1x8-aarch64-neonfma-cortex-a75.S",
-    "src/f32-gemm/gen-inc/4x12-aarch64-neonfma-cortex-a53.S",
-    "src/f32-gemm/gen-inc/4x8-aarch64-neonfma-cortex-a53.S",
-    "src/f32-gemm/gen-inc/4x8-aarch64-neonfma-cortex-a57.S",
-    "src/f32-gemm/gen-inc/4x8-aarch64-neonfma-cortex-a75.S",
-    "src/f32-gemm/gen-inc/4x8-aarch64-neonfma-ld128.S",
-    "src/f32-gemm/gen-inc/4x8-aarch64-neonfma-ld64.S",
-    "src/f32-gemm/gen-inc/5x8-aarch64-neonfma-cortex-a57.S",
-    "src/f32-gemm/gen-inc/5x8-aarch64-neonfma-cortex-a75.S",
-    "src/f32-gemm/gen-inc/6x8-aarch64-neonfma-cortex-a53.S",
-    "src/f32-gemm/gen-inc/6x8-aarch64-neonfma-cortex-a55.S",
-    "src/f32-gemm/gen-inc/6x8-aarch64-neonfma-cortex-a73.S",
-    "src/f32-gemm/gen-inc/6x8-aarch64-neonfma-cortex-a57.S",
-    "src/f32-gemm/gen-inc/6x8-aarch64-neonfma-cortex-a75.S",
-    "src/f32-gemm/gen-inc/6x8-aarch64-neonfma-ld128.S",
-    "src/f32-gemm/gen-inc/6x8-aarch64-neonfma-ld64.S",
-    "src/f32-igemm/1x12-aarch64-neonfma-cortex-a53.S",
-    "src/f32-igemm/1x8-aarch64-neonfma-cortex-a53.S",
-    "src/f32-igemm/gen/1x8-aarch64-neonfma-cortex-a57.S",
-    "src/f32-igemm/gen/1x8-aarch64-neonfma-cortex-a75.S",
-    "src/f32-igemm/4x12-aarch64-neonfma-cortex-a53.S",
-    "src/f32-igemm/4x8-aarch64-neonfma-cortex-a53.S",
-    "src/f32-igemm/gen/4x8-aarch64-neonfma-cortex-a57.S",
-    "src/f32-igemm/gen/4x8-aarch64-neonfma-cortex-a75.S",
-    "src/f32-igemm/gen/5x8-aarch64-neonfma-cortex-a57.S",
-    "src/f32-igemm/gen/5x8-aarch64-neonfma-cortex-a75.S",
-    "src/f32-igemm/6x8-aarch64-neonfma-cortex-a53.S",
-    "src/f32-igemm/6x8-aarch64-neonfma-cortex-a55.S",
-    "src/f32-igemm/6x8-aarch64-neonfma-cortex-a73.S",
-    "src/f32-igemm/gen/6x8-aarch64-neonfma-cortex-a57.S",
-    "src/f32-igemm/gen/6x8-aarch64-neonfma-cortex-a75.S",
+    "src/f16-gemm/gen/1x16-minmax-aarch64-neonfp16arith-ld32.S",
+    "src/f16-gemm/gen/4x16-minmax-aarch64-neonfp16arith-ld32.S",
+    "src/f16-gemm/gen/6x16-minmax-aarch64-neonfp16arith-ld32.S",
+    "src/f16-gemm/gen-inc/1x16inc-minmax-aarch64-neonfp16arith-ld32.S",
+    "src/f16-gemm/gen-inc/4x16inc-minmax-aarch64-neonfp16arith-ld32.S",
+    "src/f16-gemm/gen-inc/6x16inc-minmax-aarch64-neonfp16arith-ld32.S",
+    "src/f32-dwconv/up4x9-minmax-aarch64-neonfma-cortex-a55.S",
+    "src/f32-dwconv/up4x9-minmax-aarch64-neonfma.S",
+    "src/f32-gemm/gen/1x8-minmax-aarch64-neonfma-ld64.S",
+    "src/f32-gemm/gen/1x12-minmax-aarch64-neonfma-cortex-a53.S",
+    "src/f32-gemm/gen/1x8-minmax-aarch64-neonfma-cortex-a53.S",
+    "src/f32-gemm/gen/1x8-minmax-aarch64-neonfma-cortex-a57.S",
+    "src/f32-gemm/gen/1x8-minmax-aarch64-neonfma-cortex-a75.S",
+    "src/f32-gemm/gen/4x12-minmax-aarch64-neonfma-cortex-a53.S",
+    "src/f32-gemm/gen/4x8-minmax-aarch64-neonfma-cortex-a53.S",
+    "src/f32-gemm/gen/4x8-minmax-aarch64-neonfma-cortex-a55.S",
+    "src/f32-gemm/gen/4x8-minmax-aarch64-neonfma-cortex-a57.S",
+    "src/f32-gemm/gen/4x8-minmax-aarch64-neonfma-cortex-a75.S",
+    "src/f32-gemm/gen/4x8-minmax-aarch64-neonfma-ld128.S",
+    "src/f32-gemm/gen/4x8-minmax-aarch64-neonfma-ld64.S",
+    "src/f32-gemm/gen/5x8-minmax-aarch64-neonfma-cortex-a57.S",
+    "src/f32-gemm/gen/5x8-minmax-aarch64-neonfma-cortex-a75.S",
+    "src/f32-gemm/gen/6x8-minmax-aarch64-neonfma-cortex-a53.S",
+    "src/f32-gemm/gen/6x8-minmax-aarch64-neonfma-cortex-a55.S",
+    "src/f32-gemm/gen/6x8-minmax-aarch64-neonfma-cortex-a73.S",
+    "src/f32-gemm/gen/6x8-minmax-aarch64-neonfma-cortex-a57.S",
+    "src/f32-gemm/gen/6x8-minmax-aarch64-neonfma-cortex-a75.S",
+    "src/f32-gemm/gen/6x8-minmax-aarch64-neonfma-ios.S",
+    "src/f32-gemm/gen/6x8-minmax-aarch64-neonfma-ld128.S",
+    "src/f32-gemm/gen/6x8-minmax-aarch64-neonfma-ld64.S",
+    "src/f32-gemm/gen-inc/1x8inc-minmax-aarch64-neonfma-ld64.S",
+    "src/f32-gemm/gen-inc/1x12inc-minmax-aarch64-neonfma-cortex-a53.S",
+    "src/f32-gemm/gen-inc/1x8inc-minmax-aarch64-neonfma-cortex-a53.S",
+    "src/f32-gemm/gen-inc/1x8inc-minmax-aarch64-neonfma-cortex-a57.S",
+    "src/f32-gemm/gen-inc/1x8inc-minmax-aarch64-neonfma-cortex-a75.S",
+    "src/f32-gemm/gen-inc/4x12inc-minmax-aarch64-neonfma-cortex-a53.S",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-aarch64-neonfma-cortex-a53.S",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-aarch64-neonfma-cortex-a55.S",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-aarch64-neonfma-cortex-a57.S",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-aarch64-neonfma-cortex-a75.S",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-aarch64-neonfma-ld128.S",
+    "src/f32-gemm/gen-inc/4x8inc-minmax-aarch64-neonfma-ld64.S",
+    "src/f32-gemm/gen-inc/5x8inc-minmax-aarch64-neonfma-cortex-a57.S",
+    "src/f32-gemm/gen-inc/5x8inc-minmax-aarch64-neonfma-cortex-a75.S",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-aarch64-neonfma-cortex-a53.S",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-aarch64-neonfma-cortex-a55.S",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-aarch64-neonfma-cortex-a73.S",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-aarch64-neonfma-cortex-a57.S",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-aarch64-neonfma-cortex-a75.S",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-aarch64-neonfma-ios.S",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-aarch64-neonfma-ld128.S",
+    "src/f32-gemm/gen-inc/6x8inc-minmax-aarch64-neonfma-ld64.S",
+    "src/f32-igemm/1x12-minmax-aarch64-neonfma-cortex-a53.S",
+    "src/f32-igemm/1x8-minmax-aarch64-neonfma-cortex-a53.S",
+    "src/f32-igemm/gen/1x8-minmax-aarch64-neonfma-cortex-a57.S",
+    "src/f32-igemm/gen/1x8-minmax-aarch64-neonfma-cortex-a75.S",
+    "src/f32-igemm/4x12-minmax-aarch64-neonfma-cortex-a53.S",
+    "src/f32-igemm/4x8-minmax-aarch64-neonfma-cortex-a53.S",
+    "src/f32-igemm/4x8-minmax-aarch64-neonfma-cortex-a55.S",
+    "src/f32-igemm/gen/4x8-minmax-aarch64-neonfma-cortex-a57.S",
+    "src/f32-igemm/gen/4x8-minmax-aarch64-neonfma-cortex-a75.S",
+    "src/f32-igemm/gen/5x8-minmax-aarch64-neonfma-cortex-a57.S",
+    "src/f32-igemm/gen/5x8-minmax-aarch64-neonfma-cortex-a75.S",
+    "src/f32-igemm/6x8-minmax-aarch64-neonfma-cortex-a53.S",
+    "src/f32-igemm/6x8-minmax-aarch64-neonfma-cortex-a55.S",
+    "src/f32-igemm/6x8-minmax-aarch64-neonfma-cortex-a73.S",
+    "src/f32-igemm/gen/6x8-minmax-aarch64-neonfma-cortex-a57.S",
+    "src/f32-igemm/gen/6x8-minmax-aarch64-neonfma-cortex-a75.S",
+    "src/f32-igemm/gen/6x8-minmax-aarch64-neonfma-ios.S",
 ]
 
 cc_defaults {
@@ -1582,6 +1625,9 @@
         },
         arm64: {
             srcs: AARCH64_ASM_UKERNELS,
+            clang_asflags: [
+                "-march=armv8.2-a+fp16",
+            ],
         },
         x86: { enabled: false, },
         x86_64: { enabled: false, },
@@ -1874,7 +1920,8 @@
         "src",
     ],
     cflags: [
-        "-Wno-unused-function"
+        "-Wno-unused-function",
+        "-Wno-unused-parameter",
     ],
     header_libs: [
         "fp16_headers",