Optimize memory buffer, simplify set32, use sha256-simd (#7060)

Hi,

I'm syncing Gnosis on a Celeron N5100 to get familiar with the codebase.
In the process I managed to optimize some things from profiling.
Since I'm not yet on the project Discord, I decided to open this PR as a
suggestion. This pass all tests here and gave me a nice boost for that
platform, although I didn't have time to benchmark it yet.

* reuse VM Memory objects with sync.Pool. It starts with 4k as `evmone`
[code
suggested](https://github.com/ethereum/evmone/blob/0897edb00179ac3d27f27c82749119003df851e7/lib/evmone/execution_state.hpp#L49)
as a good value.
* set32 simplification: mostly cosmetic
* sha256-simd: Celeron has SHA instructions. We should probably do the
same for torrent later, but this already helped as it is very CPU bound
on such a low end processor. Maybe that helps ARM as well.

cl/utils/crypto.go

@@ -14,7 +14,7 @@
 package utils
 
 import (
-	"crypto/sha256"
+	"github.com/minio/sha256-simd"
 	"hash"
 	"sync"
 )

cmd/erigon-cl/core/state/state.go

@@ -1,8 +1,8 @@
 package state
 
 import (
-	"crypto/sha256"
 	"encoding/binary"
+	"github.com/minio/sha256-simd"
 
 	lru2 "github.com/hashicorp/golang-lru/v2"
 	libcommon "github.com/ledgerwatch/erigon-lib/common"

core/vm/contracts.go

@@ -17,9 +17,9 @@
 package vm
 
 import (
-	"crypto/sha256"
 	"encoding/binary"
 	"errors"
+	"github.com/minio/sha256-simd"
 	"math/big"
 
 	"github.com/holiman/uint256"

core/vm/interpreter.go

@@ -18,6 +18,7 @@ package vm
 
 import (
 	"hash"
+	"sync"
 
 	"github.com/ledgerwatch/erigon-lib/chain"
 	libcommon "github.com/ledgerwatch/erigon-lib/common"
@@ -43,6 +44,12 @@ type Config struct {
 	ExtraEips []int // Additional EIPS that are to be enabled
 }
 
+var pool = sync.Pool{
+	New: func() any {
+		return NewMemory()
+	},
+}
+
 func (vmConfig *Config) HasEip3860(rules *chain.Rules) bool {
 	for _, eip := range vmConfig.ExtraEips {
 		if eip == 3860 {
@@ -192,8 +199,8 @@ func (in *EVMInterpreter) Run(contract *Contract, input []byte, readOnly bool) (
 	in.returnData = nil
 
 	var (
-		op          OpCode        // current opcode
-		mem         = NewMemory() // bound memory
+		op          OpCode // current opcode
+		mem         = pool.Get().(*Memory)
 		locStack    = stack.New()
 		callContext = &ScopeContext{
 			Memory:   mem,
@@ -215,6 +222,8 @@ func (in *EVMInterpreter) Run(contract *Contract, input []byte, readOnly bool) (
 	// Don't move this deferrred function, it's placed before the capturestate-deferred method,
 	// so that it get's executed _after_: the capturestate needs the stacks before
 	// they are returned to the pools
+	mem.Reset()
+	defer pool.Put(mem)
 	defer stack.ReturnNormalStack(locStack)
 	contract.Input = input
 
@@ -304,7 +313,8 @@ func (in *EVMInterpreter) Run(contract *Contract, input []byte, readOnly bool) (
 		err = nil // clear stop token error
 	}
 
-	return res, err
+	ret = append(ret, res...)
+	return
 }
 
 // Depth returns the current call stack depth.

core/vm/memory.go

@@ -30,7 +30,9 @@ type Memory struct {
 
 // NewMemory returns a new memory model.
 func NewMemory() *Memory {
-	return &Memory{}
+	return &Memory{
+		store: make([]byte, 0, 4*1024),
+	}
 }
 
 // Set sets offset + size to value
@@ -56,9 +58,8 @@ func (m *Memory) Set32(offset uint64, val *uint256.Int) {
 		panic("invalid memory: store empty")
 	}
 	// Zero the memory area
-	copy(m.store[offset:offset+32], []byte{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0})
-	// Fill in relevant bits
-	val.WriteToSlice(m.store[offset:])
+	copy(m.store[offset:offset+32], zeroes)
+	val.WriteToSlice(m.store[offset : offset+32])
 }
 
 // zeroes - pre-allocated zeroes for Resize()
@@ -77,6 +78,11 @@ func (m *Memory) Resize(size uint64) {
 	m.store = append(m.store, zeroes[:l]...)
 }
 
+func (m *Memory) Reset() {
+	m.lastGasCost = 0
+	m.store = m.store[:0]
+}
+
 // GetCopy returns offset + size as a new slice
 func (m *Memory) GetCopy(offset, size int64) (cpy []byte) {
 	if size == 0 {

crypto/ecies/ecies_test.go

@@ -33,10 +33,10 @@ import (
 	"bytes"
 	"crypto/elliptic"
 	"crypto/rand"
-	"crypto/sha256"
 	"encoding/hex"
 	"flag"
 	"fmt"
+	"github.com/minio/sha256-simd"
 	"math/big"
 	"os"
 	"testing"

crypto/ecies/params.go

@@ -37,9 +37,9 @@ import (
 	"crypto/aes"
 	"crypto/cipher"
 	"crypto/elliptic"
-	"crypto/sha256"
 	"crypto/sha512"
 	"fmt"
+	"github.com/minio/sha256-simd"
 	"hash"
 
 	ethcrypto "github.com/ledgerwatch/erigon/crypto"

p2p/discover/v5wire/crypto_test.go

@@ -20,7 +20,7 @@ import (
 	"bytes"
 	"crypto/ecdsa"
 	"crypto/elliptic"
-	"crypto/sha256"
+	"github.com/minio/sha256-simd"
 	"reflect"
 	"strings"
 	"testing"

p2p/discover/v5wire/encoding.go

@@ -22,10 +22,10 @@ import (
 	"crypto/cipher"
 	"crypto/ecdsa"
 	crand "crypto/rand"
-	"crypto/sha256"
 	"encoding/binary"
 	"errors"
 	"fmt"
+	"github.com/minio/sha256-simd"
 	"hash"
 
 	"github.com/ledgerwatch/erigon/common/mclock"

p2p/server_test.go

@@ -19,8 +19,8 @@ package p2p
 import (
 	"context"
 	"crypto/ecdsa"
-	"crypto/sha256"
 	"errors"
+	"github.com/minio/sha256-simd"
 	"io"
 	"math/rand"
 	"net"

turbo/trie/vtree/verkle_utils_test.go

@@ -1,7 +1,7 @@
 package vtree
 
 import (
-	"crypto/sha256"
+	"github.com/minio/sha256-simd"
 	"math/big"
 	"math/rand"
 	"testing"