1.5x speedup - that's great!
2x speedup - amazing!!!
5x speedup - somebody fucked up
2x speedup - amazing!!!
5x speedup - somebody fucked up
You know I’m senior because my immediate response to seeing a 20-fold speedup after a code change is being suspicious
February 27, 2025 at 1:23 PM
1.5x speedup - that's great!
2x speedup - amazing!!!
5x speedup - somebody fucked up
2x speedup - amazing!!!
5x speedup - somebody fucked up
TypeScript announces rewrite to Go, with a 10x editor and CLI speedup.
It's the biggest TS announcement I can remember.
www.totaltypescript.com/typescript-a...
It's the biggest TS announcement I can remember.
www.totaltypescript.com/typescript-a...
TypeScript Announces Go Rewrite, Achieves 10x Speedup
TypeScript announced a full rewrite of TypeScript in Go. In testing, this rewrite has achieved a 10x speedup in some repositories - and up to 15x in others.
www.totaltypescript.com
March 11, 2025 at 3:21 PM
TypeScript announces rewrite to Go, with a 10x editor and CLI speedup.
It's the biggest TS announcement I can remember.
www.totaltypescript.com/typescript-a...
It's the biggest TS announcement I can remember.
www.totaltypescript.com/typescript-a...
iyileştim. yarın eseceğim. sılayla muhteşem yüzyıl speedup izliyorum. bi yandan tiktok hikayelere ekleme işini yapıyordum şimdi klansky'dayım
July 16, 2026 at 6:28 PM
iyileştim. yarın eseceğim. sılayla muhteşem yüzyıl speedup izliyorum. bi yandan tiktok hikayelere ekleme işini yapıyordum şimdi klansky'dayım
~6.5x speedup on `bluesky-social/social-app` 😎
May 22, 2025 at 3:49 PM
~6.5x speedup on `bluesky-social/social-app` 😎
The speedup depends on the types of parameters you pass to initialize, but the "worst case" scenario is still a ~70% speedup
April 25, 2025 at 6:34 PM
The speedup depends on the types of parameters you pass to initialize, but the "worst case" scenario is still a ~70% speedup
My VM can now pin handles and do load/stores off the resulting pointer. It is 15% faster in the interpreter, which is less of a speedup than I hoped but it *is* a clear speedup. In transpiled C I expect more speedup cause the compiler can optimize it
I guess the lesson is that hashtables are *fast*
I guess the lesson is that hashtables are *fast*
July 27, 2025 at 8:44 AM
My VM can now pin handles and do load/stores off the resulting pointer. It is 15% faster in the interpreter, which is less of a speedup than I hoped but it *is* a clear speedup. In transpiled C I expect more speedup cause the compiler can optimize it
I guess the lesson is that hashtables are *fast*
I guess the lesson is that hashtables are *fast*
Here's a Claude Code transcript showing how I used the Opus 4.5 preview to help implement quite a complex new feature that shipped in the sqlite-utils 4.0a1 release.
gistpreview.github.io?f40971b69302...
gistpreview.github.io?f40971b69302...
November 24, 2025 at 9:17 PM
Here's a Claude Code transcript showing how I used the Opus 4.5 preview to help implement quite a complex new feature that shipped in the sqlite-utils 4.0a1 release.
gistpreview.github.io?f40971b69302...
gistpreview.github.io?f40971b69302...
Our leading theory is that it was an accident of increased horizontal provisioning on a service. But without knowing for sure the speedup could end at any time, so get those link clicks in while you can I guess.
September 30, 2025 at 1:21 AM
Our leading theory is that it was an accident of increased horizontal provisioning on a service. But without knowing for sure the speedup could end at any time, so get those link clicks in while you can I guess.
ICYMI: GitLab saw their memory consumption reduced by 100x by using the @rolldown.rs powered @vite.dev version (in addition to a 2.6x speedup) 🤯
April 30, 2025 at 11:53 AM
ICYMI: GitLab saw their memory consumption reduced by 100x by using the @rolldown.rs powered @vite.dev version (in addition to a 2.6x speedup) 🤯
My GopherCon talk was just posted!
GopherCon 2025: Porting the TypeScript Compiler to Go for a 10x Speedup V2 - Jake Bailey
YouTube video by Gopher Academy
youtu.be
January 7, 2026 at 2:50 AM
My GopherCon talk was just posted!
Wow, the speedup just delivers on this one. 😆
March 16, 2026 at 7:24 PM
Wow, the speedup just delivers on this one. 😆
Improvements
* Significant speedup after fixing bug where everything got measured twice.
Issues
* Cuts are broken.
* Significant speedup after fixing bug where everything got measured twice.
Issues
* Cuts are broken.
October 5, 2025 at 12:15 AM
Improvements
* Significant speedup after fixing bug where everything got measured twice.
Issues
* Cuts are broken.
* Significant speedup after fixing bug where everything got measured twice.
Issues
* Cuts are broken.
this also introduces incremental compilation which can speedup builds by up to 30x versus last release!
September 25, 2026 at 12:09 AM
this also introduces incremental compilation which can speedup builds by up to 30x versus last release!
You know I’m senior because my immediate response to seeing a 20-fold speedup after a code change is being suspicious
February 27, 2025 at 1:05 PM
You know I’m senior because my immediate response to seeing a 20-fold speedup after a code change is being suspicious
What do we take away?
1. It seems likely that for some important settings, recent AI tooling has not increased productivity (and may in fact decrease it).
2. Self-reports of speedup are unreliable—to understand AI’s impact on productivity, we need experiments in the wild.
1. It seems likely that for some important settings, recent AI tooling has not increased productivity (and may in fact decrease it).
2. Self-reports of speedup are unreliable—to understand AI’s impact on productivity, we need experiments in the wild.
July 10, 2025 at 7:47 PM
What do we take away?
1. It seems likely that for some important settings, recent AI tooling has not increased productivity (and may in fact decrease it).
2. Self-reports of speedup are unreliable—to understand AI’s impact on productivity, we need experiments in the wild.
1. It seems likely that for some important settings, recent AI tooling has not increased productivity (and may in fact decrease it).
2. Self-reports of speedup are unreliable—to understand AI’s impact on productivity, we need experiments in the wild.
yaay i got a 15% prefill speedup in ds4 by rewriting one of the MoE kernels for WMMA
May 19, 2026 at 8:10 AM
yaay i got a 15% prefill speedup in ds4 by rewriting one of the MoE kernels for WMMA
Flash-KMeans
An IO-aware implementation of exact k-means that rethinks the algorithm around modern GPU bottlenecks. Flash-KMeans achieves 30x speedup over cuML and 200x speedup over FAISS.
Paper: arxiv.org/abs/2603.09229
Code: github.com/svg-project/...
An IO-aware implementation of exact k-means that rethinks the algorithm around modern GPU bottlenecks. Flash-KMeans achieves 30x speedup over cuML and 200x speedup over FAISS.
Paper: arxiv.org/abs/2603.09229
Code: github.com/svg-project/...
March 17, 2026 at 8:11 PM
Flash-KMeans
An IO-aware implementation of exact k-means that rethinks the algorithm around modern GPU bottlenecks. Flash-KMeans achieves 30x speedup over cuML and 200x speedup over FAISS.
Paper: arxiv.org/abs/2603.09229
Code: github.com/svg-project/...
An IO-aware implementation of exact k-means that rethinks the algorithm around modern GPU bottlenecks. Flash-KMeans achieves 30x speedup over cuML and 200x speedup over FAISS.
Paper: arxiv.org/abs/2603.09229
Code: github.com/svg-project/...
The speedup is even better with the JIT turned on
December 13, 2025 at 8:11 PM
The speedup is even better with the JIT turned on
holy shit the "slowdown" and "speedup" are different plugins LMAOOOO
March 2, 2025 at 11:11 PM
holy shit the "slowdown" and "speedup" are different plugins LMAOOOO
* Sega MD 3D engine update *
Added partial/offscreen quad rendering - was painfull coding.
Massively unrolled the line filling code for quads/triangles , 15-20% speedup.
Implemented an algorithm to account for fluctuating framerate for Z forward movement - it partially works.
Getting there !
Added partial/offscreen quad rendering - was painfull coding.
Massively unrolled the line filling code for quads/triangles , 15-20% speedup.
Implemented an algorithm to account for fluctuating framerate for Z forward movement - it partially works.
Getting there !
May 8, 2026 at 8:10 AM
* Sega MD 3D engine update *
Added partial/offscreen quad rendering - was painfull coding.
Massively unrolled the line filling code for quads/triangles , 15-20% speedup.
Implemented an algorithm to account for fluctuating framerate for Z forward movement - it partially works.
Getting there !
Added partial/offscreen quad rendering - was painfull coding.
Massively unrolled the line filling code for quads/triangles , 15-20% speedup.
Implemented an algorithm to account for fluctuating framerate for Z forward movement - it partially works.
Getting there !
full odds no speedup!! ✨
September 17, 2026 at 10:20 PM
full odds no speedup!! ✨
YouTube essay reviews will be like "Does this Retro JRPG still hold up today? Here is the opinion I got by skipping half the dialogue and holding down speedup in every battle."
September 13, 2026 at 12:29 AM
YouTube essay reviews will be like "Does this Retro JRPG still hold up today? Here is the opinion I got by skipping half the dialogue and holding down speedup in every battle."
One thing I'm really excited about in Ruby 4.0 is that object allocation is going to get a nice speed boost
December 13, 2025 at 8:09 PM
One thing I'm really excited about in Ruby 4.0 is that object allocation is going to get a nice speed boost
Goodhart strikes again. It's so funny that the thing they want for speedup now does the opposite. The employees aren't stupid.
‘Amazon employees are using an internal AI tool to automate non-essential tasks in a bid to show managers they are using the technology more frequently.‘ www.ft.com/content/8ee0...
Amazon staff use AI tool for unnecessary tasks to inflate usage scores
In-house MeshClaw tool enables employees to delegate jobs to AI agents and climb company’s AI leaderboard
www.ft.com
May 12, 2026 at 9:01 PM
Goodhart strikes again. It's so funny that the thing they want for speedup now does the opposite. The employees aren't stupid.