Handle CAST from fp-type to narrow integer types

Be explicit that the value is converted to int32 and then clamped to the
output type range.

Also remove use of apply_clip, replacing with either apply_clip_s or apply_clip_u

Signed-off-by: Eric Kunze <eric.kunze@arm.com>
Change-Id: I0efc7f4d653c1bef13c9989cfe2647208dad9617

5 files changed, 21 insertions, 6 deletions

diff --git a/chapters/introduction.adoc b/chapters/introduction.adoc
index 8b5be21..ae5c7b1 100644
--- a/chapters/introduction.adoc
+++ b/chapters/introduction.adoc
@@ -739,7 +739,7 @@ void generate_lookup_table(int16_t *table, int32_t (*reference)(int32_t))
 {
     for (int i = -256; i <= 256; i++) {
         int32_t value = (*reference)(i);
-        table[i + 256] = static_cast<int16_t>(apply_clip<int32_t>(value, -32768, +32767));
+        table[i + 256] = static_cast<int16_t>(apply_clip_s<int32_t>(value, -32768, +32767));
     }
 }
 ----
diff --git a/chapters/pseudocode.adoc b/chapters/pseudocode.adoc
index 0de7f39..9e3b7bd 100644
--- a/chapters/pseudocode.adoc
+++ b/chapters/pseudocode.adoc
@@ -262,6 +262,13 @@ in_t apply_clip_s<in_t>(in_t value, in_t min_val, in_t max_val) {
     return value;
 }
 
+in_t apply_clip_u<in_t>(in_t value, in_t min_val, in_t max_val) {
+    REQUIRE(zero_extend<int64_t>(min_val) <= zero_extend<int64_t>(max_val));
+    value = apply_max_u<in_t>(value, min_val);
+    value = apply_min_u<in_t>(value, max_val);
+    return value;
+}
+
 in_t apply_exp<in_t>(in_t input) {
     return e to the power input
 }
@@ -296,6 +303,10 @@ in_t apply_max_s<in_t>(in_t a, in_t b) {
     if (sign_extend<int64_t>(a) >= sign_extend<int64_t>(b)) return a; else return b;
 }
 
+in_t apply_max_u<in_t>(in_t a, in_t b) {
+    if (zero_extend<uint64_t>(a) >= zero_extend<int64_t>(b)) return a; else return b;
+}
+
 in_t apply_min_s<in_t>(in_t a, in_t b) {
     if (is_floating_point(in_t)) {
         if (isNaN(a) || isNaN(b)) {
@@ -307,6 +318,10 @@ in_t apply_min_s<in_t>(in_t a, in_t b) {
     if (sign_extend<int64_t>(a) < sign_extend<int64_t>(b)) return a; else return b;
 }
 
+in_t apply_min_u<in_t>(in_t a, in_t b) {
+    if (zero_extend<int64_t>(a) < zero_extend<int64_t>(b)) return a; else return b;
+}
+
 in_t apply_mul_s<in_t>(in_t a, in_t b) {
     if (is_floating_point(in_t)) return a * b;
     int64_t c = sign_extend<int64_t>(a) * sign_extend<int64_t>(b);
diff --git a/pseudocode/operators/CAST.tosac b/pseudocode/operators/CAST.tosac
index 8d816ca..fac73e3 100644
--- a/pseudocode/operators/CAST.tosac
+++ b/pseudocode/operators/CAST.tosac
@@ -17,11 +17,11 @@ for_each(index in shape) {
     } else if (out_t == fp16_t || out_t == bf16_t || out_t == fp32_t) {
         out = round_to_nearest_float(in);
     } else if (in_t == fp16_t || in_t == bf16_t || in_t == fp32_t) {
-        out = apply_clip<out_t>(round_to_nearest_int(in), minimum<out_t>, maximum<out_t>);
+        out = truncate<out_t>(apply_clip_s<i32_t>(round_to_nearest_int(in), minimum<out_t>, maximum<out_t>));
     } else if (sizeof(out_t) >= sizeof(in_t)) {
         out = sign_extend<out_t>(in);
     } else {
-        out = truncate(in);
+        out = truncate<out_t>(in);
     }
     tensor_write<out_t>(output, shape, index, out);
 }
diff --git a/pseudocode/operators/CLAMP.tosac b/pseudocode/operators/CLAMP.tosac
index 7a26d50..cfac632 100644
--- a/pseudocode/operators/CLAMP.tosac
+++ b/pseudocode/operators/CLAMP.tosac
@@ -10,6 +10,6 @@
 ERROR_IF(max_val < min_val);
 for_each(index in shape) {
     in_out_t value = tensor_read<in_out_t>(input, shape, index);
-    value = apply_clip<in_out_t>(value, min_val, max_val);
+    value = apply_clip_s<in_out_t>(value, min_val, max_val);
     tensor_write<in_out_t>(output, shape, index, value);
 }
diff --git a/pseudocode/operators/RESCALE.tosac b/pseudocode/operators/RESCALE.tosac
index c29bddd..3ebb8d6 100644
--- a/pseudocode/operators/RESCALE.tosac
+++ b/pseudocode/operators/RESCALE.tosac
@@ -43,14 +43,14 @@ for_each(index in shape) {
     if (output_unsigned) {
         int32_t extended_out_zp = zero_extend<int32_t>(output_zp);
         result = apply_add_s<int32_t>(result, extended_out_zp);
-        out_t out = static_cast<out_t>(apply_clip<int32_t>(result,
+        out_t out = static_cast<out_t>(apply_clip_u<i32_t>(result,
                                                            minimum_u<out_t>,
                                                            maximum_u<out_t>));
     }
     else {
         int32_t extended_out_zp = sign_extend<int32_t>(output_zp);
         result = apply_add_s<int32_t>(result, extended_out_zp);
-        out_t out = static_cast<out_t>(apply_clip<int32_t>(result,
+        out_t out = static_cast<out_t>(apply_clip_s<i32_t>(result,
                                                            minimum_s<out_t>,
                                                            maximum_s<out_t>));
     }