line use and diff

src/kyupy/sim.py

@@ -187,7 +187,6 @@ class SimOps:
         levels = []
 
         ppio2idx = dict((n, i) for i, n in enumerate(circuit.s_nodes))
-        pis = set([n for n in circuit.s_nodes if len(n.ins) == 0])
         ppos = set([n for n in circuit.s_nodes if len(n.ins) > 0])
         readers = np.array([1 if l.reader in ppos else len(l.reader.outs) for l in circuit.lines], dtype=np.int32)  # for ref-counting forks
 
@@ -311,6 +310,15 @@ class SimOps:
             if len(n.ins) > 0:
                 self.c_locs[self.ppo_offset + i], self.c_caps[self.ppo_offset + i] = self.c_locs[n.ins[0]], self.c_caps[n.ins[0]]
 
+        # line use information
+        self.line_use_start = np.full(self.c_locs_len, -1, dtype=np.int32)
+        self.line_use_stop = np.full(self.c_locs_len, len(self.levels), dtype=np.int32)
+        for i, lv in enumerate(self.levels):
+            for op in lv:
+                self.line_use_start[op[1]] = i
+                for x in [2, 3, 4, 5]:
+                    self.line_use_stop[op[x]] = i
+
         self.c_len = h.max_size
 
         d = defaultdict(int)

src/kyupy/wave_sim.py

@@ -13,10 +13,11 @@ Two simulators are available: :py:class:`WaveSim` runs on the CPU, and the deriv
 """
 
 import math
+from collections import defaultdict
 
 import numpy as np
 
-from . import numba, cuda, sim, cdiv, eng
+from . import log, numba, cuda, sim, cdiv, eng
 
 
 TMAX = np.float32(2 ** 127)
@@ -59,8 +60,8 @@ class WaveSim(sim.SimOps):
         self.delays = np.zeros((len(delays), self.c_locs_len, 2, 2), dtype=delays.dtype)
         self.delays[:, :delays.shape[1]] = delays
 
-        self.c = np.zeros((self.c_len, sims), dtype=np.float32) + TMAX
+        self.c = np.full((self.c_len, self.sims), TMAX, dtype=np.float32)
-        self.s = np.zeros((11, self.s_len, sims), dtype=np.float32)
+        self.s = np.zeros((11, self.s_len, self.sims), dtype=np.float32)
         """Information about the logic values and transitions around the sequential elements (flip-flops) and ports.
 
         The first 3 values are read by :py:func:`s_to_c`.
@@ -98,7 +99,26 @@ class WaveSim(sim.SimOps):
         self.simctl_int[0] = range(sims)  # unique seed for each sim by default, zero this to pick same delays for all sims.
         self.simctl_int[1] = 2  # random picking by default.
 
-        self.nbytes = sum([a.nbytes for a in (self.c, self.s, self.c_locs, self.c_caps, self.ops, self.simctl_int)])
+        # flat array for line use information
+        line_use = defaultdict(list)
+        for lidx in range(len(self.circuit.lines)):
+            if self.line_use_start[lidx] < 0: continue
+            if self.line_use_stop[lidx] < 0:
+                log.warn(f'line {lidx} never read?')
+            for i in range(self.line_use_start[lidx], self.line_use_stop[lidx]):
+                line_use[i].append(lidx)
+
+        self.line_use_counts = np.array([len(line_use[i]) for i in range(len(self.levels))], dtype=np.int32)
+        self.line_use_offsets = np.zeros_like(self.line_use_counts)
+        self.line_use_offsets[1:] = self.line_use_counts.cumsum()[:-1]
+        self.line_use = np.hstack([line_use[i] for i in range(len(self.levels))])
+
+        self.h = np.zeros((self.c_locs_len, sims), dtype=np.float32)  # hashes of generated waveforms
+        self.h_base = np.zeros_like(self.h)  # base hashes to compare to
+
+        self.error_counts = np.zeros(self.s_len, dtype=np.uint32)  # number of capture errors by PPO
+
+        self.nbytes = sum([a.nbytes for a in (self.c, self.s, self.h, self.c_locs, self.c_caps, self.ops, self.simctl_int)])
 
     def __repr__(self):
         dev = 'GPU' if hasattr(self.c, 'copy_to_host') else 'CPU'
@@ -124,7 +144,7 @@ class WaveSim(sim.SimOps):
         """
         sims = min(sims or self.sims, self.sims)
         for op_start, op_stop in zip(self.level_starts, self.level_stops):
-            level_eval_cpu(self.ops, op_start, op_stop, self.c, self.c_locs, self.c_caps, self.abuf, 0, sims, self.delays, self.simctl_int, seed)
+            level_eval_cpu(self.ops, op_start, op_stop, self.c, self.c_locs, self.c_caps, self.h, self.abuf, 0, sims, self.delays, self.simctl_int, seed)
 
     def c_to_s(self, time=TMAX, sd=0.0, seed=1):
         """Simulates a capture operation at all sequential elements and primary outputs.
@@ -152,7 +172,7 @@ class WaveSim(sim.SimOps):
         self.s[2, self.ppio_s_locs] = self.s[8, self.ppio_s_locs]
 
 
-def _wave_eval(op, cbuf, c_locs, c_caps, sim, delays, simctl_int, seed=0):
+def _wave_eval(op, cbuf, c_locs, c_caps, hbuf, sim, delays, simctl_int, seed):
     overflows = int(0)
 
     lut = op[0]
@@ -182,6 +202,8 @@ def _wave_eval(op, cbuf, c_locs, c_caps, sim, delays, simctl_int, seed=0):
     z_mem = c_locs[z_idx]
     z_cap = c_caps[z_idx]
 
+    h = np.float32(0)
+
     a_cur = int(0)
     b_cur = int(0)
     c_cur = int(0)
@@ -229,6 +251,7 @@ def _wave_eval(op, cbuf, c_locs, c_caps, sim, delays, simctl_int, seed=0):
             next_t = cbuf[d_mem + d_cur, sim] + delays[d_idx, (d_cur & 1) ^ 1, z_val ^ 1]
 
         if (z_cur & 1) != ((lut >> inputs) & 1):
+            h += h*3 + max(current_t, -10)  # hash based on generated transitions before filtering
             # we generate an edge in z_mem, if ...
             if (z_cur == 0                            # it is the first edge in z_mem ...
                 or next_t < current_t                 # -OR- the next edge on SAME input is EARLIER (need current edge to filter BOTH in next iteration) ...
@@ -240,8 +263,8 @@ def _wave_eval(op, cbuf, c_locs, c_caps, sim, delays, simctl_int, seed=0):
                     z_cur += 1
                 else:
                     overflows += 1
-                    previous_t = cbuf[z_mem + z_cur - 1, sim]
                     z_cur -= 1
+                    previous_t = cbuf[z_mem + z_cur, sim]
             else:
                 z_cur -= 1
                 previous_t = cbuf[z_mem + z_cur - 1, sim] if z_cur > 0 else TMIN
@@ -258,6 +281,8 @@ def _wave_eval(op, cbuf, c_locs, c_caps, sim, delays, simctl_int, seed=0):
     # generate or propagate overflow flag
     cbuf[z_mem + z_cur, sim] = TMAX_OVL if overflows > 0 else max(a, b, c, d)
 
+    hbuf[z_idx, sim] = h
+
     nrise = max(0, (z_cur+1) // 2 - (cbuf[z_mem, sim] == TMIN))
     nfall = z_cur // 2
 
@@ -268,11 +293,11 @@ wave_eval_cpu = numba.njit(_wave_eval)
 
 
 @numba.njit
-def level_eval_cpu(ops, op_start, op_stop, c, c_locs, c_caps, abuf, sim_start, sim_stop, delays, simctl_int, seed):
+def level_eval_cpu(ops, op_start, op_stop, c, c_locs, c_caps, hbuf, abuf, sim_start, sim_stop, delays, simctl_int, seed):
     for op_idx in range(op_start, op_stop):
         op = ops[op_idx]
         for sim in range(sim_start, sim_stop):
-            nrise, nfall = wave_eval_cpu(op, c, c_locs, c_caps, sim, delays, simctl_int[:, sim], seed)
+            nrise, nfall = wave_eval_cpu(op, c, c_locs, c_caps, hbuf, sim, delays, simctl_int[:, sim], seed)
             a_loc = op[6]
             a_wr = op[7]
             a_wf = op[8]
@@ -345,12 +370,18 @@ class WaveSimCuda(WaveSim):
         self.delays = cuda.to_device(self.delays)
         self.simctl_int = cuda.to_device(self.simctl_int)
         self.abuf = cuda.to_device(self.abuf)
+        self.h = cuda.to_device(self.h)
+        self.h_base = cuda.to_device(self.h_base)
+        self.line_use = cuda.to_device(self.line_use)
+        self.error_counts = cuda.to_device(self.error_counts)
+
+        self.retval_int = cuda.to_device(np.array([0], dtype=np.int32))
 
         self._block_dim = (32, 16)
 
     def __getstate__(self):
         state = self.__dict__.copy()
-        state['c'] = np.array(self.c)
+        del state['c']
         state['s'] = np.array(self.s)
         state['ops'] = np.array(self.ops)
         state['c_locs'] = np.array(self.c_locs)
@@ -358,11 +389,16 @@ class WaveSimCuda(WaveSim):
         state['delays'] = np.array(self.delays)
         state['simctl_int'] = np.array(self.simctl_int)
         state['abuf'] = np.array(self.abuf)
+        state['h'] = np.array(self.h)
+        state['h_base'] = np.array(self.h_base)
+        state['line_use'] = np.array(self.line_use)
+        state['error_counts'] = np.array(self.error_counts)
+        state['retval_int'] = np.array(self.retval_int)
         return state
 
     def __setstate__(self, state):
         self.__dict__.update(state)
-        self.c = cuda.to_device(self.c)
+        self.c = cuda.to_device(np.full((self.c_len, self.sims), TMAX, dtype=np.float32))
         self.s = cuda.to_device(self.s)
         self.ops = cuda.to_device(self.ops)
         self.c_locs = cuda.to_device(self.c_locs)
@@ -370,6 +406,11 @@ class WaveSimCuda(WaveSim):
         self.delays = cuda.to_device(self.delays)
         self.simctl_int = cuda.to_device(self.simctl_int)
         self.abuf = cuda.to_device(self.abuf)
+        self.h = cuda.to_device(self.h)
+        self.h_base = cuda.to_device(self.h_base)
+        self.line_use = cuda.to_device(self.line_use)
+        self.error_counts = cuda.to_device(self.error_counts)
+        self.retval_int = cuda.to_device(self.retval_int)
 
     def s_to_c(self):
         grid_dim = self._grid_dim(self.sims, self.s_len)
@@ -383,10 +424,18 @@ class WaveSimCuda(WaveSim):
             if op_from > op_start: continue
             if op_to is not None and op_to <= op_start: break
             grid_dim = self._grid_dim(sims, op_stop - op_start)
-            wave_eval_gpu[grid_dim, self._block_dim](self.ops, op_start, op_stop, self.c, self.c_locs, self.c_caps, self.abuf, int(0),
+            wave_eval_gpu[grid_dim, self._block_dim](self.ops, op_start, op_stop, self.c, self.c_locs, self.c_caps, self.h, self.abuf, int(0),
                 sims, self.delays, self.simctl_int, seed)
         cuda.synchronize()
 
+    def c_prop_level(self, level, sims=None, seed=1):
+        sims = min(sims or self.sims, self.sims)
+        op_start = self.level_starts[level]
+        op_stop = self.level_stops[level]
+        grid_dim = self._grid_dim(sims, op_stop - op_start)
+        wave_eval_gpu[grid_dim, self._block_dim](self.ops, op_start, op_stop, self.c, self.c_locs, self.c_caps, self.h, self.abuf, int(0),
+            sims, self.delays, self.simctl_int, seed)
+
     def c_to_s(self, time=TMAX, sd=0.0, seed=1):
         grid_dim = self._grid_dim(self.sims, self.s_len)
         wave_capture_gpu[grid_dim, self._block_dim](self.c, self.s, self.c_locs, self.c_caps, self.ppo_offset,
@@ -396,6 +445,52 @@ class WaveSimCuda(WaveSim):
         grid_dim = self._grid_dim(self.sims, self.s_len)
         ppo_to_ppi_gpu[grid_dim, self._block_dim](self.s, self.c_locs, time, self.ppi_offset, self.ppo_offset)
 
+    def set_base_hashes(self):
+        nitems = self.h_base.shape[0] * self.h_base.shape[1]
+        grid_dim = cdiv(nitems, 256)
+        memcpy_gpu[grid_dim, 256](self.h, self.h_base, nitems)
+
+    def compare_hashes_level(self, lv):
+        self.retval_int[0] = 0
+        grid_dim = self._grid_dim(self.sims, self.line_use_counts[lv])
+        diff_hash_gpu[grid_dim, self._block_dim](self.h, self.h_base, self.line_use, self.line_use_offsets[lv],
+                                                 self.line_use_counts[lv], self.retval_int)
+        return self.retval_int[0]
+
+    def calc_error_counts(self, sims=None):
+        sims = min(sims or self.sims, self.sims)
+        grid_dim = cdiv(self.s_len, 256)
+        calc_error_counts_gpu[grid_dim, 256](self.s, sims, self.error_counts)
+        return np.array(self.error_counts)
+
+
+@cuda.jit()
+def memcpy_gpu (src, dst, nitems):
+    tid = cuda.grid(1)
+    stride = cuda.gridDim.x * cuda.blockDim.x
+    for i in range(tid, nitems, stride):
+        dst.flat[i] = src.flat[i]
+
+
+@cuda.jit()
+def diff_hash_gpu(hbuf1, hbuf2, h_locs, h_locs_offset, h_locs_cnt, differs):
+    x, y = cuda.grid(2)
+    if x >= hbuf1.shape[1]: return
+    if y >= h_locs_cnt: return
+    h_loc = h_locs[h_locs_offset+y]
+    if hbuf1[h_loc, x] != hbuf2[h_loc, x]:
+        differs[0] = 1
+
+
+@cuda.jit()
+def calc_error_counts_gpu(s, sims, error_counts):
+    x = cuda.grid(1)
+    if x >= s.shape[1]: return
+    cnt = 0
+    for i in range(sims):
+        cnt += (s[6,x,i] != s[8,x,i])
+    error_counts[x] = cnt
+
 
 @cuda.jit()
 def wave_assign_gpu(c, s, c_locs, ppi_offset):
@@ -425,7 +520,7 @@ _wave_eval_gpu = cuda.jit(_wave_eval, device=True)
 
 
 @cuda.jit()
-def wave_eval_gpu(ops, op_start, op_stop, cbuf, c_locs, c_caps, abuf, sim_start, sim_stop, delays, simctl_int, seed):
+def wave_eval_gpu(ops, op_start, op_stop, cbuf, c_locs, c_caps, hbuf, abuf, sim_start, sim_stop, delays, simctl_int, seed):
     x, y = cuda.grid(2)
     sim = sim_start + x
     op_idx = op_start + y
@@ -437,7 +532,7 @@ def wave_eval_gpu(ops, op_start, op_stop, cbuf, c_locs, c_caps, abuf, sim_start,
     a_wr = op[7]
     a_wf = op[8]
 
-    nrise, nfall = _wave_eval_gpu(op, cbuf, c_locs, c_caps, sim, delays, simctl_int[:, sim], seed)
+    nrise, nfall = _wave_eval_gpu(op, cbuf, c_locs, c_caps, hbuf, sim, delays, simctl_int[:, sim], seed)
 
     # accumulate WSA into abuf
     if a_loc >= 0:

tests/test_wave_sim.py

@@ -8,9 +8,10 @@ from kyupy.logic import mvarray
 def test_xnor2_delays():
     op = (sim.XNOR2, 2, 0, 1, 3, 3, -1, 0, 0)
     #op = (0b0111, 4, 0, 1)
-    c = np.full((4*16, 1), TMAX)  # 4 waveforms of capacity 16
+    c = np.full((4*16, 1), TMAX, dtype=np.float32)  # 4 waveforms of capacity 16
     c_locs = np.zeros((4,), dtype='int')
     c_caps = np.zeros((4,), dtype='int')
+    h = np.zeros((4, 1), dtype=np.float32)
 
     for i in range(4): c_locs[i], c_caps[i] = i*16, 16  # 1:1 mapping
     delays = np.zeros((1, 4, 2, 2))
@@ -27,7 +28,7 @@ def test_xnor2_delays():
 
     def wave_assert(inputs, output):
         for i, a in zip(inputs, c.reshape(-1,16)): a[:len(i)] = i
-        wave_eval_cpu(op, c, c_locs, c_caps, 0, delays, simctl_int)
+        wave_eval_cpu(op, c, c_locs, c_caps, h, 0, delays, simctl_int, 0)
         for i, v in enumerate(output): np.testing.assert_allclose(c.reshape(-1,16)[2,i], v)
 
     wave_assert([[TMIN,TMAX],[TMIN,TMAX]], [TMIN,TMAX])      # XNOR(1,1) => 1
@@ -40,9 +41,10 @@ def test_xnor2_delays():
 def test_nand_delays():
     op = (sim.NAND4, 4, 0, 1, 2, 3, -1, 0, 0)
     #op = (0b0111, 4, 0, 1)
-    c = np.full((5*16, 1), TMAX)  # 5 waveforms of capacity 16
+    c = np.full((5*16, 1), TMAX, dtype=np.float32)  # 5 waveforms of capacity 16
     c_locs = np.zeros((5,), dtype='int')
     c_caps = np.zeros((5,), dtype='int')
+    h = np.zeros((5, 1), dtype=np.float32)
 
     for i in range(5): c_locs[i], c_caps[i] = i*16, 16  # 1:1 mapping
 
@@ -64,7 +66,7 @@ def test_nand_delays():
 
     def wave_assert(inputs, output):
         for i, a in zip(inputs, c.reshape(-1,16)): a[:len(i)] = i
-        wave_eval_cpu(op, c, c_locs, c_caps, 0, delays, simctl_int)
+        wave_eval_cpu(op, c, c_locs, c_caps, h, 0, delays, simctl_int, 0)
         for i, v in enumerate(output): np.testing.assert_allclose(c.reshape(-1,16)[4,i], v)
 
     wave_assert([[TMAX,TMAX],[TMAX,TMAX],[TMIN,TMAX],[TMIN,TMAX]], [TMIN,TMAX]) # NAND(0,0,1,1) => 1