KAdot

Every new EC2 generation is typically slightly more expensive, but the gains in the compute performance are 15-25% or even higher depending on your workload. The same dollar buys you a lot more compute power than a decade ago.

I heavily A/B tested Opus vs Sol for two weeks, giving Claude Code and Codex the same tasks and comparing the results. In my experience Sol is much closer to Opus than to Fable, with Opus often beating Sol. The only area where Sol is better is code reviews, which these benchmarks confirm. I wish they included Fable.

The benchmark is setup to test 80 MiB dataset on a machine with 32 GiB RAM, which doesn't represent a typical database workload. How does the key-value store perform on larger than RAM datasets? MMAP is fast when the dataset fits in memory, but it can slow to a crawl when it doesn't, especially if the workload is mostly random point lookups.

Going back to 4.7 with `claude --model claude-opus-4-7` fixed it for me.

Is it closed source?

This is not my experience. I've been running RocksDB for 4 years on thousands of machines, each storing terabytes of data, and I haven't seen a single correctness issue caused by RocksDB.

> We are interested in making OTAP pipelines safely embeddable, through strict controls on memory and through support for thread-per-core runtimes.

I'm curious about the thread-per-core runtimes, are there even any mature thread-per-core runtimes in Rust around?

The benchmark doesn't accurately represent the real-world database performance because the dataset is too small (roughly half a gigabyte based on [1]?), meaning it fits into the page cache bypassing disk I/O.

[1]: https://github.com/voidDB/voidDB/blob/master/test/bench_test...

The benchmark is not comparing apples to apples.

prost is the most widely used Protobuf implementation in Rust, maintained by the Tokio organization. prost generates structs and serialization/deserialization code for you.

easyproto according to GitHib Search is used only by two projects. easyproto provides primitives for serializing and deserializing Protobuf, and requires hand writing code to do both.

A fair comparison would be prost vs google.golang.org/protobuf, or easyproto vs parts of quick-protobuf.

In most cases you can make Go as fast as Rust, but from my experience writing performance-sensitive code in Go requires significantly larger time investment and overall requires deeper language expertise. Pebble (RocksDB replacement in Go by CockroachDB) is a good example of this, the codebase is littered with hand-inlined[1] functions, hand-unrolled loops and it's not[2] even using Go memory management for performance critical parts, it's using the C memory allocator and manual memory management.

[prost]: https://github.com/tokio-rs/prost [easyproto]: https://github.com/VictoriaMetrics/easyproto [google.golang.org/protobuf]: https://github.com/protocolbuffers/protobuf-go [quick-protobuf]: https://github.com/tafia/quick-protobuf [1]: https://github.com/cockroachdb/pebble/blob/c34894c46703fd823... [2]: https://github.com/cockroachdb/pebble/blob/master/docs/memor...

The redis.io landing page says "from the ones who built it" just wow.

The project is gone from GitHub already https://github.com/yuzu-emu/yuzu

> After discovering Chrome are eating my old Mac's battery, I turned to using Edge

What? Have they tried using Safari instead of a Chromium based browser? From my experience Safari is by far the best browser on Mac in terms of the energy efficiency.

Is there a way to disable all "collaborative" features, including removing the "Sign in" button?

A similar idea implemented in Go https://github.com/akrylysov/pogreb (disclaimer: I'm the author of this package).

The performance is not that great, but you can already play Cyberpunk 2077 on a M2 Pro laptop using Game Porting Toolkit https://www.youtube.com/watch?v=sPJpkRmsceU Keep in mind, it has to do x64 -> ARM64 transition and translate Direct X into Metal. I would expect a native version to work better.

https://artem.krylysov.com/blog/ I'm writing mostly about databases and performance.

My last two posts:

- How RocksDB works https://artem.krylysov.com/blog/2023/04/19/how-rocksdb-works...

- Let's build a Full-Text Search engine https://artem.krylysov.com/blog/2020/07/28/lets-build-a-full...

Great article. If someone is looking for a more advanced example, some time ago, as an exercise, I created an interpreter for a Python-like language in Python https://github.com/akrylysov/abrvalg

I rephrased some sentences to avoid confusion. Thanks for the feedback!

Binary searching SST file blocks is a stretch, I agree. The key-value pairs would need to have a specific shape. Enabling compression makes it completely impossible. I'll remove this from the article, thanks for the feedback!

Author here.

> Well written article

Thanks!

> It is not Tao it is ZippyDb

I don't work for Meta, so might have made a mistake. There is an old blog post[1] about Tao and there is a recent paper[2] mentioning that the graph database is powered by MyRocks, which runs on RocksDB.

[1]: https://engineering.fb.com/2013/06/25/core-data/tao-the-powe...

[2]: https://www.vldb.org/pvldb/vol13/p3217-matsunobu.pdf

macOS 13 Ventura introduced[1] support of running amd64 binaries with Rosetta inside of arm64 Linux VMs when using Apple Virtualization framework.

Latest Docker Desktop and Lima VM can take advantage on the new feature. It makes running amd64 containers significantly faster.

Here is a very short manual how to setup Docker with Lima VM and Rosetta https://gist.github.com/akrylysov/7c1ea3bac409da2758e525f2f8...

[1]: https://developer.apple.com/documentation/virtualization/run...

I'll take a look, thanks for the links! As for apps, playing from the web browser doesn't require additional apps and works surprisingly well. Bsimp is integrated with the Media Session API which lets you control the playback from the notification bar or the lock screen.

It costs me $11/month. $6 for a small DO droplet instance[1] and $5 for DO spaces storage[2] which gets me 250GB of space. I use both the instance and the storage also for other purposes. E.g. I also host my personal blog and store backups there.

If you want to save a few dollars, you can get it close to $5/month if you go with cheaper S3 compatible storage from Backblaze[3] or Cloudflare[4] + the cheapest $2.5 instance from Vultr[5].

[1]: https://www.digitalocean.com/pricing/droplets

[2]: https://www.digitalocean.com/pricing/spaces-object-storage

[3]: https://help.backblaze.com/hc/en-us/articles/360037814594-B2...

[4]: https://developers.cloudflare.com/r2/platform/pricing/

[5]: https://www.vultr.com/pricing

I ended up uploading my music library to S3 compatible cloud storage (DigitalOcean Spaces) and creating[1] a minimalistic audio player that plays music straight from S3. S3 serves as a live audio library and as a backup at the same time. The web interface lets me listen the music from a desktop or a mobile web browser.

[1]: https://github.com/akrylysov/bsimp