aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorPavel V. Shatov (Meister) <meisterpaul1@yandex.ru>2019-03-23 11:06:42 +0300
committerPavel V. Shatov (Meister) <meisterpaul1@yandex.ru>2019-03-23 11:06:42 +0300
commit701e3f1670042c9f9c7c76c529eac43802b20c24 (patch)
treea4f2bfed057fc89b6729cc31872a3a5174977720
parentc30c0bd7c5f517a4acbf8e3840cd8cd5fb8a08c3 (diff)
Added optional output of intermediate quantities for debugging.
Reworked index rotation code for better readability.
-rw-r--r--modexpng_fpga_model.py61
1 files changed, 54 insertions, 7 deletions
diff --git a/modexpng_fpga_model.py b/modexpng_fpga_model.py
index 54db95f..d3b7841 100644
--- a/modexpng_fpga_model.py
+++ b/modexpng_fpga_model.py
@@ -71,6 +71,15 @@ _VECTOR_PATH = "/vector"
_VECTOR_CLASS = "Vector"
+# ------------------
+# Debugging Settings
+# ------------------
+DUMP_VECTORS = False
+DUMP_INDICES = False
+DUMP_MACS_CLEARING = False
+DUMP_MACS_ACCUMULATION = True
+
+
#
# Multi-Precision Integer
#
@@ -94,6 +103,15 @@ class ModExpNG_Operand():
self._init_from_words(words, length)
+ def format_verilog_concat(self, name):
+
+ for i in range(len(self.words)):
+ if i > 0:
+ if (i % 4) == 0: print("")
+ else: print(" ", end='')
+ print("%s[%2d] = 17'h%05x;" % (name, i, self.words[i]), end='')
+ print("")
+
def _init_from_words(self, words, count):
for i in range(count):
@@ -315,14 +333,17 @@ class ModExpNG_WordMultiplier():
def _rotate_indices(self, num_words):
for x in range(len(self._indices)):
- self._indices[x] -= 1
- if self._indices[x] < 0:
- self._indices[x] += num_words
+ if self._indices[x] > 0:
+ self._indices[x] -= 1
+ else:
+ self._indices[x] = num_words - 1
def _rotate_index_aux(self):
self._index_aux[0] -= 1
- def multiply_square(self, a_wide, b_narrow, ab_num_words):
+ def multiply_square(self, a_wide, b_narrow, ab_num_words, dump=False):
+
+ if dump: print("multiply_square()")
num_cols = ab_num_words // NUM_MULTS
@@ -335,8 +356,17 @@ class ModExpNG_WordMultiplier():
self._clear_all_macs()
self._preset_indices(col)
+ if dump and DUMP_MACS_CLEARING:
+ print("t= 0, col=%2d > clear > all" % (col))
+
for t in range(ab_num_words):
+ if dump and DUMP_INDICES:
+ print("t=%2d, col=%2d > indices:" % (t, col), end='')
+ for i in range(NUM_MULTS):
+ print(" %2d" % self._indices[i], end='')
+ print("")
+
# current b-word
bt = b_narrow.words[t]
@@ -348,6 +378,16 @@ class ModExpNG_WordMultiplier():
if t == (col * NUM_MULTS + x):
parts[t] = self._macs[x]
self._clear_one_mac(x)
+ if dump and DUMP_MACS_CLEARING:
+ print("t=%2d, col=%2d > clear > x=%d:" % (t, col, x))
+
+ if dump and DUMP_MACS_ACCUMULATION:
+ for i in range(NUM_MULTS):
+ if i > 0: print(" | ", end='')
+ print("[%d]: 0x%012x" % (i, self._macs[i]), end='')
+ print("")
+
+
# save the uppers part of product at end of column,
# for the last column don't save the very last part
@@ -570,13 +610,20 @@ class ModExpNG_Worker():
return ModExpNG_Operand(None, 2*ab_num_words, ab)
- def multiply(self, a, b, n, n_coeff, ab_num_words, reduce_only=False, multiply_only=False):
+ def multiply(self, a, b, n, n_coeff, ab_num_words, reduce_only=False, multiply_only=False, dump=False):
+
+ if dump and DUMP_VECTORS:
+ print("num_words = %d" % ab_num_words)
+ a.format_verilog_concat("A")
+ b.format_verilog_concat("B")
+ n.format_verilog_concat("N")
+ n_coeff.format_verilog_concat("N_COEFF")
# 1.
if reduce_only:
ab = a
else:
- ab_parts = self.multiplier.multiply_square(a, b, ab_num_words)
+ ab_parts = self.multiplier.multiply_square(a, b, ab_num_words, dump)
ab_words = self.recombinator.recombine_square(ab_parts, ab_num_words)
ab = ModExpNG_Operand(None, 2 * ab_num_words, ab_words)
@@ -678,7 +725,7 @@ if __name__ == "__main__":
mp_blind = worker.multiply(mp_blind_inverse_factor, vector.p_factor, vector.p, vector.p_coeff, pq_num_words)
mq_blind = worker.multiply(mq_blind_inverse_factor, vector.q_factor, vector.q, vector.q_coeff, pq_num_words)
- mp_blind_factor = worker.multiply(mp_blind, vector.p_factor, vector.p, vector.p_coeff, pq_num_words)
+ mp_blind_factor = worker.multiply(mp_blind, vector.p_factor, vector.p, vector.p_coeff, pq_num_words, dump=True)
mq_blind_factor = worker.multiply(mq_blind, vector.q_factor, vector.q, vector.q_coeff, pq_num_words)
sp_blind_factor = worker.exponentiate(ip_factor, mp_blind_factor, vector.dp, vector.p, vector.p_factor, vector.p_coeff, pq_num_words)