diff --git a/csrc/float_fix.cc b/csrc/float_fix.cc
new file mode 100644
index 00000000..4a106bb7
--- /dev/null
+++ b/csrc/float_fix.cc
@@ -0,0 +1,95 @@
+#include <assert.h>
+#include <cinttypes>
+#include <cstdio>
+#include <iostream>
+#include <fstream>
+#include <sstream>
+
+
+// returns the bits in x[high:low] in the lowest positions 
+uint64_t BitRange(uint64_t x, int high, int low) {
+  int high_gap = 64 - high;
+  return x << high_gap >> (low + high_gap);
+}
+
+
+// uses magic numbers since can only handle float
+// logic from berkeley-hardfloat/src/main/scala/recodedFloatNToFloatN.scala
+uint64_t UnrecodeFloatFromDouble(uint64_t raw_input) {
+  uint64_t recoded_float = raw_input & 0x1ffffffff;  // lower 33 bits
+  uint64_t sign = BitRange(recoded_float, 31, 31);
+  uint64_t exp_in = BitRange(recoded_float, 30, 22);
+  uint64_t sig_in = BitRange(recoded_float, 21, 0);
+
+  bool is_high_subnormal_in = BitRange(exp_in, 6, 0) < 2;
+  bool is_subnormal = (BitRange(exp_in, 8, 6) == 1) ||
+                     ((BitRange(exp_in, 8, 7) == 1) && is_high_subnormal_in);
+  bool is_normal = ((BitRange(exp_in, 8, 7) == 1) && !is_high_subnormal_in) ||
+                    (BitRange(exp_in, 8, 7) == 2);
+  bool is_special = BitRange(exp_in, 8, 7) == 3;
+  bool is_NaN = is_special && BitRange(exp_in, 6, 6);
+
+  uint64_t denorm_shift_dist = 2 - BitRange(exp_in, 4, 0);
+  uint64_t subnormal_sig_out = (0x400000 | sig_in) >> denorm_shift_dist;
+  uint64_t normal_exp_out = BitRange(exp_in, 7, 0) - 129;
+
+  uint64_t exp_out = is_normal ? normal_exp_out : (is_special ? 255 : 0);
+  uint64_t sig_out = is_normal || is_NaN ? sig_in :
+                      is_subnormal ? subnormal_sig_out : 0;
+
+  uint64_t raw_output64 = (sign << 63) | (exp_out << 23) | sig_out;
+  // make sure nothing leaked over the top
+  assert((raw_output64 & 0xffffffff00000000) == uint64_t(0));
+  return raw_output64;
+}
+
+
+// checks if possibly recoded float inside double if upper 31 bits set
+bool NestedFloatPossible(uint64_t raw_input) {
+  const uint64_t mask = 0xfffffffe00000000;
+  return (raw_input & mask) == mask;
+}
+
+
+// best effort at replacing floats with unrecoded versions
+void DiffAndFix(std::string rocket_filename, std::string lspike_filename) {
+  std::ifstream rocket_log(rocket_filename);
+  if (!rocket_log.is_open()) {
+    std::cout << "Couldn't open file " << rocket_filename << std::endl;
+    std::exit(-2);
+  }
+  std::ifstream lspike_log(lspike_filename);
+  if (!lspike_log.is_open()) {
+    std::cout << "Couldn't open file " << lspike_filename << std::endl;
+    std::exit(-2);
+  }
+  std::stringstream ss;
+  ss << std::hex;
+  std::string rocket_line, lspike_line;
+  while(getline(rocket_log,rocket_line) && getline(lspike_log,lspike_line)) {
+    if ((rocket_line != lspike_line) &&  // mismatch and includes fp inst
+        (rocket_line.find(" f") != std::string::npos)) {
+      std::string fp_wb_str = rocket_line.substr(40, 16);
+      uint64_t raw_fp;
+      ss << fp_wb_str;
+      ss >> raw_fp;
+      if (NestedFloatPossible(raw_fp)) {
+        snprintf(const_cast<char*>(rocket_line.data()) + 40, 17,
+                 "%016" PRIx64, UnrecodeFloatFromDouble(raw_fp));
+      }
+    }
+    printf("%s\n", rocket_line.c_str());
+  }
+  rocket_log.close();
+  lspike_log.close();
+}
+
+
+int main(int argc, char** argv) {
+  if (argc < 3) {
+    std::cout << "Usage: float_fix rocket_output lspike_output" << std::endl;
+    return -1;
+  }
+  DiffAndFix(std::string(argv[1]), std::string(argv[2]));
+  return 0;
+}