Github: https://github.com/jinyus/related_post_gen
They are running the same code so there might be some issue worth investigating.
#Why is node 2.8x faster than bun in this data processing benchmark?
merry robin
honest maple
I noticed replacing the .forEach functions with for loops made the performance a very close tie between deno, bun and node on my laptop
honest maple
I'm getting roughly the same ~2-3x longer Processing Time on Bun compared to Node on an Mac M1 pro for the benchmark out of the box.
on Bun v1.0.3
Node v18.17.0
merry robin
I guess this is more of a test between v8 and jscore. Though node also beats deno
honest maple
I guess this is more of a test between v8 and jscore. Though node also beats de...
I guess the big bun plus is the quality of life, package manager speed, and io and http performance
I tried running this benchmark: https://github.com/SaltyAom/bun-http-framework-benchmark
| Framework | Average | Get (/) | Params, query & header | Post JSON |
|------------------------|-------------|------------|------------------------|-----------|
| bun (bun) | 101,340.427 | 116,602.03 | 97,748.64 | 89,670.61 |
| hyper-express (node) | 93,630.72 | 112,906.79 | 103,652.45 | 64,332.92 |
| bun-web-standard (bun) | 86,040.763 | 104,407.93 | 83,313.92 | 70,400.44 |
| uws (node) | 85,733.407 | 92,558.09 | 83,677.56 | 80,964.57 |
| stricjs (bun) | 85,311.623 | 93,271.22 | 87,136.48 | 75,527.17 |
| vixeny (bun) | 84,878.343 | 93,518.46 | 89,003.9 | 72,112.67 |
| elysia (bun) | 81,757.027 | 88,578.9 | 83,384.45 | 73,307.73 |
| hono (bun) | 79,800.303 | 88,887.74 | 78,380.41 | 72,132.76 |
| nhttp (bun) | 79,582.877 | 93,642.51 | 73,851.31 | 71,254.81 |
| baojs (bun) | 78,973.687 | 86,773.42 | 77,495.79 | 72,651.85 |
| hono (deno) | 76,447.127 | 85,885.06 | 82,731.47 | 60,724.85 |
| hyperbun (bun) | 68,473.023 | 76,708.06 | 66,223.61 | 62,487.4 |
| nbit (bun) | 67,542.2 | 73,676.06 | 67,236.89 | 61,713.65 |
| xirelta (bun) | 56,987.37 | 57,096.95 | 53,740.52 | 60,124.64 |
| fast (deno) | 48,523.847 | 52,256.66 | 48,434.12 | 44,880.76 |
| fastify (node) | 47,316.27 | 51,433.01 | 46,861.36 | 43,654.44 |
| h3 (node) | 43,279.003 | 49,154.67 | 42,275.9 | 38,406.44 |
| fastify (bun) | 42,089.757 | 51,962.22 | 49,631.44 | 24,675.61 |
| koa (node) | 36,929.583 | 40,510.99 | 37,555.71 | 32,722.05 |
| adonis (node) | 35,185.427 | 33,866.05 | 35,881.92 | 35,808.31 |
| abc (deno) | 33,599.1 | 40,404.07 | 36,339.01 | 24,054.22 |
| cheetah (deno) | 33,178.747 | 34,518.58 | 42,557.33 | 22,460.33 |
| oak (deno) | 31,013.76 | 34,215.38 | 32,901.24 | 25,924.66 |
| express (bun) | 29,080.373 | 35,570.24 | 33,889.22 | 17,781.66 |
| hapi (node) | 27,637.873 | 37,840.73 | 19,964 | 25,108.89 |
| hono (node) | 17,864.063 | 20,325.2 | 19,278.49 | 13,988.5 |
| express (node) | 15,953.73 | 17,646.83 | 16,430.56 | 13,783.8 |
| nest (node) | 14,937.14 | 16,244.13 | 15,297.12 | 13,270.17 |
| acorn (deno) | 5,631.123 | 10,952.12 | 3,320.73 | 2,620.52 |
both hyper-express (node) and bun gives you a very close ~6x req/s compared to express (node)
honest maple
I'm getting roughly the same ~2-3x longer Processing Time on Bun compared to Nod...
I'm seeing the same performance if I try to compile the bun benchmark js
honest maple
Github: https://github.com/jinyus/related_post_gen
They are running the same co...
Out of curiosity I set up a Rust FFI in Bun for this.
https://github.com/jinyus/related_post_gen/compare/main...Airtune:related_post_gen:main
Node: 154ms
Bun: 324ms
Rust: 34ms
Bun+Rust FFI: 46ms
Interesting to see FFI being a pretty good option for your bun project if you really need the extra performance
merry robin
Update: bun is now the fastest js runtime:
JS (Bun) 7.42 s
JS (Deno) 14.19 s
JS (Node) 14.60 s