From: Michael Nolan Date: Fri, 7 Feb 2020 14:29:32 +0000 (-0500) Subject: modify reorder_bits to copy the MSB of the partition to each bit X-Git-Tag: ls180-24jan2020~244 X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=d07c689dc09ada1f3b408dcff9858b514a42dc80;p=ieee754fpu.git modify reorder_bits to copy the MSB of the partition to each bit Previously, it would move the MSB to the LSB, and set the rest of the bits to 0. --- diff --git a/src/ieee754/part_cmp/eq_gt_ge.py b/src/ieee754/part_cmp/eq_gt_ge.py index 7207c0f4..9772522a 100644 --- a/src/ieee754/part_cmp/eq_gt_ge.py +++ b/src/ieee754/part_cmp/eq_gt_ge.py @@ -53,7 +53,6 @@ class PartitionedEqGtGe(Elaboratable): m.submodules.gtc = gtc = GTCombiner(self.mwidth) m.submodules.reorder = reorder = ReorderResults(self.mwidth) - m.submodules.ripple = ripple = RippleLSB(self.mwidth) # make a series of "eqs" and "gts", splitting a and b into # partition chunks @@ -99,10 +98,8 @@ class PartitionedEqGtGe(Elaboratable): comb += reorder.results_in.eq(results) comb += reorder.gates.eq(self.partition_points.as_sig()) - comb += ripple.results_in.eq(reorder.output) - comb += ripple.gates.eq(self.partition_points.as_sig()) - comb += self.output.eq(ripple.output) + comb += self.output.eq(reorder.output) return m diff --git a/src/ieee754/part_cmp/formal/proof_eq_gt_ge.py b/src/ieee754/part_cmp/formal/proof_eq_gt_ge.py index 8d223746..13c42343 100644 --- a/src/ieee754/part_cmp/formal/proof_eq_gt_ge.py +++ b/src/ieee754/part_cmp/formal/proof_eq_gt_ge.py @@ -58,21 +58,21 @@ class EqualsDriver(Elaboratable): with m.Switch(gates): with m.Case(0b00): comb += Assert(out[0] == (a == b)) - comb += Assert(out[1] == 0) - comb += Assert(out[2] == 0) + comb += Assert(out[1] == out[0]) + comb += Assert(out[2] == out[1]) with m.Case(0b01): comb += Assert(out[0] == (a_intervals[0] == b_intervals[0])) comb += Assert(out[1] == ((a_intervals[1] == \ b_intervals[1]) & (a_intervals[2] == \ b_intervals[2]))) - comb += Assert(out[2] == 0) + comb += Assert(out[2] == out[1]) with m.Case(0b10): comb += Assert(out[0] == ((a_intervals[0] == \ b_intervals[0]) & (a_intervals[1] == \ b_intervals[1]))) - comb += Assert(out[1] == 0) + comb += Assert(out[1] == out[0]) comb += Assert(out[2] == (a_intervals[2] == b_intervals[2])) with m.Case(0b11): for i in range(mwidth-1): @@ -81,17 +81,19 @@ class EqualsDriver(Elaboratable): with m.If(opcode == 0b01): with m.Switch(gates): with m.Case(0b00): - comb += Assert(out == (a > b)) + comb += Assert(out[0] == (a > b)) + comb += Assert(out[1] == out[0]) + comb += Assert(out[2] == out[1]) with m.Case(0b01): comb += Assert(out[0] == (a_intervals[0] > b_intervals[0])) comb += Assert(out[1] == (Cat(*a_intervals[1:3]) > \ Cat(*b_intervals[1:3]))) - comb += Assert(out[2] == 0) + comb += Assert(out[2] == out[1]) with m.Case(0b10): comb += Assert(out[0] == (Cat(*a_intervals[0:2]) > \ Cat(*b_intervals[0:2]))) - comb += Assert(out[1] == 0) + comb += Assert(out[1] == out[0]) comb += Assert(out[2] == (a_intervals[2] > b_intervals[2])) with m.Case(0b11): for i in range(mwidth-1): @@ -100,17 +102,19 @@ class EqualsDriver(Elaboratable): with m.If(opcode == 0b10): with m.Switch(gates): with m.Case(0b00): - comb += Assert(out == (a >= b)) + comb += Assert(out[0] == (a >= b)) + comb += Assert(out[1] == out[0]) + comb += Assert(out[2] == out[1]) with m.Case(0b01): comb += Assert(out[0] == (a_intervals[0] >= b_intervals[0])) comb += Assert(out[1] == (Cat(*a_intervals[1:3]) >= \ Cat(*b_intervals[1:3]))) - comb += Assert(out[2] == 0) + comb += Assert(out[2] == out[1]) with m.Case(0b10): comb += Assert(out[0] == (Cat(*a_intervals[0:2]) >= \ Cat(*b_intervals[0:2]))) - comb += Assert(out[1] == 0) + comb += Assert(out[1] == out[0]) comb += Assert(out[2] == (a_intervals[2] >= b_intervals[2])) with m.Case(0b11): for i in range(mwidth-1): diff --git a/src/ieee754/part_cmp/reorder_results.py b/src/ieee754/part_cmp/reorder_results.py index 577f1702..41c64652 100644 --- a/src/ieee754/part_cmp/reorder_results.py +++ b/src/ieee754/part_cmp/reorder_results.py @@ -23,7 +23,7 @@ class ReorderResults(Elaboratable): for i in range(width-2, -1, -1): # counts down from width-1 to 0 cur = Signal() comb += cur.eq(current_result) - comb += self.output[i+1].eq(cur & self.gates[i]) + comb += self.output[i+1].eq(cur) current_result = Mux(self.gates[i], self.results_in[i], cur) comb += self.output[0].eq(current_result)