Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

feat: gnovm benchmarking tool #2241

Open
wants to merge 17 commits into
base: master
Choose a base branch
from
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
97 changes: 97 additions & 0 deletions examples/gno.land/r/x/benchmark/storage/boards.gno
Original file line number Diff line number Diff line change
@@ -0,0 +1,97 @@
package storage

import (
"strconv"

"gno.land/p/demo/avl"
)

var boards avl.Tree

type Board interface {
AddPost(title, content string)
GetPost(id int) (Post, bool)
Size() int
}

// posts are persisted in an avl tree
type TreeBoard struct {
id int
posts *avl.Tree
}

func (b *TreeBoard) AddPost(title, content string) {
n := b.posts.Size()
p := Post{n, title, content}
b.posts.Set(strconv.Itoa(n), p)
}

func (b *TreeBoard) GetPost(id int) (Post, bool) {
p, ok := b.posts.Get(strconv.Itoa(id))
if ok {
return p.(Post), ok
} else {
return Post{}, ok
}
}

func (b *TreeBoard) Size() int {
return b.posts.Size()
}

// posts are persisted in a map
type MapBoard struct {
id int
posts map[int]Post
}

func (b *MapBoard) AddPost(title, content string) {
n := len(b.posts)
p := Post{n, title, content}
b.posts[n] = p
}

func (b *MapBoard) GetPost(id int) (Post, bool) {
p, ok := b.posts[id]
if ok {
return p, ok
} else {
return Post{}, ok
}
}

func (b *MapBoard) Size() int {
return len(b.posts)
}

// posts are persisted in a slice
type SliceBoard struct {
id int
posts []Post
}

func (b *SliceBoard) AddPost(title, content string) {
n := len(b.posts)
p := Post{n, title, content}
b.posts = append(b.posts, p)
}

func (b *SliceBoard) GetPost(id int) (Post, bool) {
if id < len(b.posts) {
p := b.posts[id]

return p, true
} else {
return Post{}, false
}
}

func (b *SliceBoard) Size() int {
return len(b.posts)
}

type Post struct {
id int
title string
content string
}
64 changes: 64 additions & 0 deletions examples/gno.land/r/x/benchmark/storage/forum.gno
Original file line number Diff line number Diff line change
@@ -0,0 +1,64 @@
package storage

import (
"strconv"

"gno.land/p/demo/avl"
)

func init() {
// we write to three common data structure for persistence
// avl.Tree, map and slice.
posts0 := avl.NewTree()
b0 := &TreeBoard{0, posts0}
boards.Set(strconv.Itoa(0), b0)

posts1 := make(map[int]Post)
b1 := &MapBoard{1, posts1}
boards.Set(strconv.Itoa(1), b1)

posts2 := []Post{}
b2 := &SliceBoard{2, posts2}
boards.Set(strconv.Itoa(2), b2)
}

// post to all boards.
func AddPost(title, content string) {
for i := 0; i < boards.Size(); i++ {
boardId := strconv.Itoa(i)
b, ok := boards.Get(boardId)
if ok {
b.(Board).AddPost(title, content)
}
}
}

func GetPost(boardId, postId int) string {
b, ok := boards.Get(strconv.Itoa(boardId))
var res string

if ok {
p, ok := b.(Board).GetPost(postId)
if ok {
res = p.title + "," + p.content
}
}
return res
}

func GetPostSize(boardId int) int {
b, ok := boards.Get(strconv.Itoa(boardId))
var res int

if ok {
res = b.(Board).Size()
} else {
res = -1
}

return res
}

func GetBoardSize() int {
return boards.Size()
}
3 changes: 3 additions & 0 deletions examples/gno.land/r/x/benchmark/storage/gno.mod
Original file line number Diff line number Diff line change
@@ -0,0 +1,3 @@
module gno.land/r/x/benchmark/storage

require gno.land/p/demo/avl v0.0.0-latest
30 changes: 30 additions & 0 deletions gnovm/pkg/benchops/Makefile
Original file line number Diff line number Diff line change
@@ -0,0 +1,30 @@
.PHONY: opcode storage build_opcode build_storage

# Build target
build_opcode:
go build -tags "benchmarkingops" -o build/gnobench ./cmd

# Build target
build_storage:
go build -tags "benchmarkingstorage" -o build/gnobench ./cmd

# Extract the latest commit hash
COMMIT_HASH := $(shell git rev-parse --short=7 HEAD)

# Run target
opcode: build_opcode
./build/gnobench -out opcode_results_$(COMMIT_HASH).csv

# Run target
storage: build_storage
./build/gnobench -out store_results_$(COMMIT_HASH).csv

# command to run dependency utilities, like goimports.
rundep=go run -modfile ../misc/devdeps/go.mod

# flags for `make fmt`. -w will write the result to the destination files.
GOFMT_FLAGS ?= -w

.PHONY: fmt
fmt:
$(rundep) mvdan.cc/gofumpt $(GOFMT_FLAGS) .
90 changes: 90 additions & 0 deletions gnovm/pkg/benchops/README.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,90 @@
# `gnobench` the time consumed for GnoVM OpCode execution and store access

`gnobench` benchmarks the time consumed for each VM CPU OpCode and persistent access to the store, including marshalling and unmarshalling of realm objects.

## Usage

### Simple mode

The benchmark only involves the GnoVM and the persistent store. It benchmarks the bare minimum components, and the results are isolated from other components. We use standardize gno contract to perform the benchmarking.

This mode is the best for benchmarking each major release and/or changes in GnoVM.

make opcode
make storage

### Production mode

It benchmarks the node in the production environment with minimum overhead.
We can not only benchmark with standardize the contract but also capture the live usage in production environment.
It gives us a complete picture of the node perform.


1. Build the production node with benchmarking flags:

`go build -tags "benchmarkingstorage benchmarkingops" gno.land/cmd/gnoland`

2. Run the node in the production environment. It will dump benchmark data to a benchmarks.bin file.

3. call the realm contracts at `gno.land/r/x/benchmark/opcodes` and `gno.land/r/x/benchmark/storage`

4. Stop the server after the benchmarking session is complete.

5. Run the following command to convert the binary dump:

`gnobench -bin path_to_benchmarks.bin`

it converts the binary dump to results.csv and results_stats.csv.


## Results

The benchmarking results are stored in two files:
1. The raw results are saved in results.csv.

| Operation | Elapsed Time | Disk IO Bytes |
|-----------------|--------------|---------------|
| OpEval | 40333 | 0 |
| OpPopBlock | 208 | 0 |
| OpHalt | 167 | 0 |
| OpEval | 500 | 0 |
| OpInterfaceType | 458 | 0 |
| OpPopBlock | 166 | 0 |
| OpHalt | 125 | 0 |
| OpInterfaceType | 21125 | 0 |
| OpEval | 541 | 0 |
| OpEval | 209 | 0 |
| OpInterfaceType | 334 | 0 |



2. The averages and standard deviations are summarized in results_stats.csv.

| Operation | Avg Time | Avg Size | Time Std Dev | Count |
|----------------|----------|----------|--------------|-------|
| OpAdd | 101 | 0 | 45 | 300 |
| OpAddAssign | 309 | 0 | 1620 | 100 |
| OpArrayLit | 242 | 0 | 170 | 700 |
| OpArrayType | 144 | 0 | 100 | 714 |
| OpAssign | 136 | 0 | 95 | 2900 |
| OpBand | 92 | 0 | 30 | 100 |
| OpBandAssign | 127 | 0 | 62 | 100 |
| OpBandn | 97 | 0 | 54 | 100 |
| OpBandnAssign | 125 | 0 | 113 | 100 |
| OpBinary1 | 128 | 0 | 767 | 502 |
| OpBody | 127 | 0 | 145 | 13700 |

## Design consideration

### Minimum Overhead and Footprint

- Constant build flags enable benchmarking.
- Encode operations and measurements in binary.
- Dump to a local file in binary.
- No logging, printout, or network access involved.

### Accurate

- Pause the timer for storage access while performing VM opcode benchmarking.
- Measure each OpCode execution in nanoseconds.
- Store access includes the duration for Amino marshalling and unmarshalling.
Loading