speed up epoch processing by caching shuffling results (#153)

* speed up epoch processing by caching shuffling results

beacon_chain/spec/beaconstate.nim

@@ -151,6 +151,26 @@ func slash_validator*(state: var BeaconState, index: ValidatorIndex) =
   validator.withdrawable_epoch = get_current_epoch(state) +
     LATEST_SLASHED_EXIT_LENGTH
 
+func update_shuffling_cache*(state: var BeaconState) =
+  let
+    list_size = state.validator_registry.len.uint64
+    shuffling_seq = mapIt(
+      get_shuffled_seq(state.current_shuffling_seed, list_size),
+      # No intrinsic reason for this conversion; SSZ requirement artifact.
+      it.int)
+
+  doAssert state.shuffling_cache.index in [0, 1]
+
+  # Do a dance to keep everything JSON-encodable.
+  state.shuffling_cache.seeds[state.shuffling_cache.index] =
+    state.current_shuffling_seed
+  state.shuffling_cache.list_sizes[state.shuffling_cache.index] = list_size
+  if state.shuffling_cache.index == 0:
+    state.shuffling_cache.shuffling_0 = shuffling_seq
+  else:
+    state.shuffling_cache.shuffling_1 = shuffling_seq
+  state.shuffling_cache.index = 1 - state.shuffling_cache.index
+
 # https://github.com/ethereum/eth2.0-specs/blob/v0.3.0/specs/core/0_beacon-chain.md#on-genesis
 func get_genesis_beacon_state*(
     initial_validator_deposits: openArray[Deposit],
@@ -238,6 +258,8 @@ func get_genesis_beacon_state*(
     state.latest_active_index_roots[index] = genesis_active_index_root
   state.current_shuffling_seed = generate_seed(state, GENESIS_EPOCH)
 
+  update_shuffling_cache(state)
+
   state
 
 # TODO candidate for spec?

beacon_chain/spec/datatypes.nim

@@ -417,6 +417,9 @@ type
     eth1_data_votes*: seq[Eth1DataVote]
     deposit_index*: uint64
 
+    # Not in spec. TODO: don't serialize or deserialize this.
+    shuffling_cache*: ShufflingCache
+
   # https://github.com/ethereum/eth2.0-specs/blob/v0.3.0/specs/core/0_beacon-chain.md#validator
   Validator* = object
     pubkey*: ValidatorPubKey ##\
@@ -493,6 +496,21 @@ type
 
   # TODO: not in spec
   CrosslinkCommittee* = tuple[committee: seq[ValidatorIndex], shard: uint64]
+  ShufflingCache* = object
+    ## https://github.com/ethereum/eth2.0-specs/blob/v0.3.0/specs/core/0_beacon-chain.md#get_shuffling
+    ## Note: this definition and the next few definitions make heavy use of
+    ## repetitive computing. Production implementations are expected to
+    ## appropriately use caching/memoization to avoid redoing work.
+    ##
+    ## TODO use native ValidatorIndex, once this doesn't need to be serialized.
+    ## `seed` and `list_size` determine the shuffle. For now, only need two, at
+    ## any given time. If the next_epoch variations of shuffling get called, it
+    ## might increase to three at once.
+    seeds*: array[2, Eth2Digest]
+    list_sizes*: array[2, uint64]
+    index*: int
+    shuffling_0*: seq[int]
+    shuffling_1*: seq[int]
 
 func shortValidatorKey*(state: BeaconState, validatorIdx: int): string =
     ($state.validator_registry[validatorIdx].pubkey)[0..7]

beacon_chain/spec/validator.nim

@@ -15,7 +15,7 @@ import
 # https://github.com/ethereum/eth2.0-specs/blob/v0.3.0/specs/core/0_beacon-chain.md#get_shuffling
 # https://github.com/ethereum/eth2.0-specs/blob/v0.3.0/specs/core/0_beacon-chain.md#get_permuted_index
 func get_shuffled_seq*(seed: Eth2Digest,
-                       validators: openArray[Validator],
+                       list_size: uint64,
                        ): seq[ValidatorIndex] =
   ## Via https://github.com/protolambda/eth2-shuffle/blob/master/shuffle.go
   ## Shuffles ``validators`` into crosslink committees seeded by ``seed`` and
@@ -25,7 +25,6 @@ func get_shuffled_seq*(seed: Eth2Digest,
   ##
   ## Invert the inner/outer loops from the spec, essentially. Most useful
   ## hash result re-use occurs within a round.
-  let list_size = validators.len.uint64
   var
     # Share these buffers.
     pivot_buffer: array[(32+1), byte]
@@ -75,21 +74,31 @@ func get_shuffled_seq*(seed: Eth2Digest,
 # https://github.com/ethereum/eth2.0-specs/blob/v0.3.0/specs/core/0_beacon-chain.md#get_shuffling
 func get_shuffling*(seed: Eth2Digest,
                     validators: openArray[Validator],
-                    epoch: Epoch
+                    epoch: Epoch,
+                    shuffling_cache: ShufflingCache
                     ): seq[seq[ValidatorIndex]] =
   ## This function is factored to facilitate testing with
   ## https://github.com/ethereum/eth2.0-test-generators/tree/master/permutated_index
   ## test vectors, which the split of get_shuffling obfuscates.
+  let list_size = validators.len.uint64
 
   let
     active_validator_indices = get_active_validator_indices(validators, epoch)
     committees_per_epoch = get_epoch_committee_count(
       len(active_validator_indices)).int
+    # Both mapIt-type-conversions are an SSZ artifact. TODO remove.
+    shuffled_seq =
+      if shuffling_cache.seeds[0] == seed and
+         shuffling_cache.list_sizes[0] == list_size:
+        mapIt(shuffling_cache.shuffling_0, it.ValidatorIndex)
+      elif shuffling_cache.seeds[1] == seed and
+           shuffling_cache.list_sizes[1] == list_size:
+        mapIt(shuffling_cache.shuffling_1, it.ValidatorIndex)
+      else:
+        get_shuffled_seq(seed, list_size)
 
   # Split the shuffled list into committees_per_epoch pieces
-  result = split(
+  result = split(shuffled_seq, committees_per_epoch)
-    get_shuffled_seq(seed, validators),
-    committees_per_epoch)
   assert result.len() == committees_per_epoch # what split should do..
 
 # https://github.com/ethereum/eth2.0-specs/blob/v0.3.0/specs/core/0_beacon-chain.md#get_previous_epoch_committee_count
@@ -196,6 +205,7 @@ func get_crosslink_committees_at_slot*(state: BeaconState, slot: Slot,
       seed,
       state.validator_registry,
       shuffling_epoch,
+      state.shuffling_cache
     )
     offset = slot mod SLOTS_PER_EPOCH
     committees_per_slot = committees_per_epoch div SLOTS_PER_EPOCH

beacon_chain/state_transition.nim

@@ -916,6 +916,8 @@ func processEpoch(state: var BeaconState) =
         state.current_shuffling_seed = generate_seed(state, state.current_shuffling_epoch)
         # /Note/ that state.current_shuffling_start_shard is left unchanged
 
+  updateShufflingCache(state)
+
   ## Regardless of whether or not a validator set change happens run
   ## process_slashings(state) and process_exit_queue(state)
   process_slashings(state)

tests/test_validator.nim

@@ -19,9 +19,7 @@ func sumCommittees(v: openArray[seq[ValidatorIndex]], reqCommitteeLen: int): int
 func checkPermutation(index: int, list_size: uint64,
                       permuted_index: int, seed: Eth2Digest): bool =
   let
-    validators = repeat(
+    s = get_shuffled_seq(seed, list_size)
-      Validator(exit_epoch: FAR_FUTURE_EPOCH), list_size)
-    s = get_shuffled_seq(seed, validators)
   s[index] == permuted_index.ValidatorIndex
 
 suite "Validators":
@@ -35,7 +33,7 @@ suite "Validators":
         Validator(
           exit_epoch: FAR_FUTURE_EPOCH
         ), num_validators)
-      s = get_shuffling(Eth2Digest(), validators, 0)
+      s = get_shuffling(Eth2Digest(), validators, 0, ShufflingCache())
       committees = get_epoch_committee_count(len(validators)).int
     check:
       # def b(s): return "Eth2Digest(data: [0x" + "'u8, 0x".join((s[i:i+2] for i in range(0, 64, 2))) + "'u8])"