diff --git a/arch/inst/F/fadd.s.yaml b/arch/inst/F/fadd.s.yaml
index 5553eedffa..98a28185fe 100644
--- a/arch/inst/F/fadd.s.yaml
+++ b/arch/inst/F/fadd.s.yaml
@@ -27,7 +27,8 @@ access:
   vu: always
 data_independent_timing: true
 operation(): |
-  RoundingMode mode = rm_to_mode(X[rm], $encoding);
+  check_f_ok($encoding);
+  RoundingMode mode = rm_to_mode(rm, $encoding);
   X[fd] = f32_add(X[fs1], X[fs2], mode);
 
 sail(): |
diff --git a/arch/inst/F/fcvt.s.w.yaml b/arch/inst/F/fcvt.s.w.yaml
index 9e0d646c01..f79c946b02 100644
--- a/arch/inst/F/fcvt.s.w.yaml
+++ b/arch/inst/F/fcvt.s.w.yaml
@@ -34,20 +34,8 @@ access:
 data_independent_timing: false
 operation(): |
   check_f_ok($encoding);
-
-  Bits<32> int_value = X[rs1];
-
-  Bits<1> sign = int_value[31];
-
   RoundingMode rounding_mode = rm_to_mode(rm, $encoding);
-
-  if ((int_value & 32'h7fff_ffff) == 0) {
-    X[fd] = (sign == 1) ? packToF32UI(1, 0x9E, 0) : 0;
-  } else {
-    Bits<32> absA = (sign == 1) ? -int_value : int_value;
-    X[fd] = softfloat_normRoundPackToF32( sign, 0x9C, absA, rounding_mode );
-  }
-
+  X[fd] = i32_to_f32(X[rs1], rounding_mode);
   mark_f_state_dirty();
 
 sail(): |
diff --git a/arch/inst/F/fcvt.s.wu.yaml b/arch/inst/F/fcvt.s.wu.yaml
index 57fab4e22c..a0474330c6 100644
--- a/arch/inst/F/fcvt.s.wu.yaml
+++ b/arch/inst/F/fcvt.s.wu.yaml
@@ -3,9 +3,18 @@
 $schema: "inst_schema.json#"
 kind: instruction
 name: fcvt.s.wu
-long_name: No synopsis available.
+long_name: Convert unsigned 32-bit integer to single-precision float
 description: |
-  No description available.
+  Converts a 32-bit unsigned integer in integer register _rs1_ into a floating-point number in
+  floating-point register _fd_.
+
+  All floating-point to integer and integer to floating-point conversion instructions round
+  according to the _rm_ field.
+  A floating-point register can be initialized to floating-point positive zero using
+  `fcvt.s.w rd, x0`, which will never set any exception flags.
+
+  All floating-point conversion instructions set the Inexact exception flag if the rounded
+  result differs from the operand value and the Invalid exception flag is not set.
 definedBy: F
 assembly: fd, xs1, rm
 encoding:
@@ -24,7 +33,10 @@ access:
   vu: always
 data_independent_timing: true
 operation(): |
-
+  check_f_ok($encoding);
+  RoundingMode rounding_mode = rm_to_mode(rm, $encoding);
+  X[fd] = ui32_to_f32(X[rs1], rounding_mode);
+  mark_f_state_dirty();
 sail(): |
   {
     assert(sizeof(xlen) >= 64);
diff --git a/arch/inst/F/fsub.s.yaml b/arch/inst/F/fsub.s.yaml
index b0622f32ad..905c1a6b7e 100644
--- a/arch/inst/F/fsub.s.yaml
+++ b/arch/inst/F/fsub.s.yaml
@@ -27,7 +27,8 @@ access:
   vu: always
 data_independent_timing: true
 operation(): |
-  RoundingMode mode = rm_to_mode(X[rm], $encoding);
+  check_f_ok($encoding);
+  RoundingMode mode = rm_to_mode(rm, $encoding);
   X[fd] = f32_sub(X[fs1], X[fs2], mode);
 sail(): |
   {
diff --git a/arch/isa/fp.idl b/arch/isa/fp.idl
index c5386b455d..e49ce20585 100644
--- a/arch/isa/fp.idl
+++ b/arch/isa/fp.idl
@@ -882,3 +882,43 @@ function f32_sub {
     }
   }
 }
+
+function i32_to_f32 {
+  returns U32
+  arguments
+    U32 a,
+    RoundingMode mode
+  description {
+    Converts 32-bit signed integer to 32-bit floating point
+  }
+  body {
+    # sign of integer, it is 1 when negative
+    Bits<1> sign = a[31];
+    if ((a & 0x7FFFFFFF) == 0) {
+      return (sign == 1) ? packToF32UI(1, 0x9E, 0) : packToF32UI(0, 0, 0);
+    }
+    U32 magnitude_of_A = returnMag(a);
+    return softfloat_normRoundPackToF32(sign, 0x9C, magnitude_of_A, mode);
+  }
+}
+
+function ui32_to_f32 {
+  returns U32
+  arguments
+    U32 a,
+    RoundingMode mode
+  description {
+    Converts 32-bit unsigned integer to 32-bit floating point
+  }
+  body {
+    # sign of integer, it is 1 when negative
+    if (a == 0) {
+      return a;
+    }
+    if (a[31] == 1) {
+      return softfloat_roundPackToF32(0, 0x9D, a>>1 | (a & 1), mode);
+    } else {
+      return softfloat_normRoundPackToF32(0, 0x9C, a, mode);
+    }
+  }
+}