Arthur Pastel
@apas.tel
Co-founder @codspeed.io • Building trustable and consistent performance tooling • Python, Rust, TypeScript • Creator of ODMantic • apas.tel
Excited to have spent some time building this inlining detection. Will come soon with an example use case showing how to leverage this new feature 🤓
Inlining Information is now available in @codspeed.io flame graphs.
You can now track how the compiler inlines your code directly in the UI.
codspeed.io/changelog/20...
You can now track how the compiler inlines your code directly in the UI.
codspeed.io/changelog/20...
Better granularity with Inlining Information - Changelog - CodSpeed
Stop guessing why performance changed between builds. See inlined frames directly in your flamegraphs for complete visibility into compiler optimizations.
codspeed.io
October 29, 2025 at 2:55 PM
Excited to have spent some time building this inlining detection. Will come soon with an example use case showing how to leverage this new feature 🤓
Excited to have it out! 🎉 This is just the first step toward adaptive instrumentation: get an overview with wall-time profiling, then zoom in with CPU simulation or other instruments to reach instruction-level detail.
This also edges us nearer to real-world sampling in production 👀
This also edges us nearer to real-world sampling in production 👀
Walltime profiling is now available across Rust, C/C++, Python, and Go! 🚀⚡
Get detailed flamegraphs powered by perf to pinpoint bottlenecks in real-world scenarios!
codspeed.io/changelog/20...
Get detailed flamegraphs powered by perf to pinpoint bottlenecks in real-world scenarios!
codspeed.io/changelog/20...
Introducing Walltime Profiling Across Languages - Changelog - CodSpeed
Walltime now includes profiling support powered by perf, available for Rust, C/C++, Python, Go, and Node.js to pinpoint bottlenecks in real-world scenarios.
codspeed.io
October 14, 2025 at 2:19 PM
Excited to have it out! 🎉 This is just the first step toward adaptive instrumentation: get an overview with wall-time profiling, then zoom in with CPU simulation or other instruments to reach instruction-level detail.
This also edges us nearer to real-world sampling in production 👀
This also edges us nearer to real-world sampling in production 👀
Reposted by Arthur Pastel
Want to write faster programs?
Start with the backbone of performance: the memory allocator
Last year, @apas.tel took us deep into an often overlooked topic and made it impossible to forget. Watch it again👇🏼
Start with the backbone of performance: the memory allocator
Last year, @apas.tel took us deep into an often overlooked topic and made it impossible to forget. Watch it again👇🏼
August 5, 2025 at 11:49 AM
Want to write faster programs?
Start with the backbone of performance: the memory allocator
Last year, @apas.tel took us deep into an often overlooked topic and made it impossible to forget. Watch it again👇🏼
Start with the backbone of performance: the memory allocator
Last year, @apas.tel took us deep into an often overlooked topic and made it impossible to forget. Watch it again👇🏼
Reposted by Arthur Pastel
What’s faster in Rust? 🦀 Your instinct says one thing. The benchmark says something else.
@apas.tel and @adriencaccia.bsky.social are bringing a live Rust performance quiz to #EuroRust25. You don’t want to miss it!
🧵1/3
@apas.tel and @adriencaccia.bsky.social are bringing a live Rust performance quiz to #EuroRust25. You don’t want to miss it!
🧵1/3
July 1, 2025 at 12:21 PM
What’s faster in Rust? 🦀 Your instinct says one thing. The benchmark says something else.
@apas.tel and @adriencaccia.bsky.social are bringing a live Rust performance quiz to #EuroRust25. You don’t want to miss it!
🧵1/3
@apas.tel and @adriencaccia.bsky.social are bringing a live Rust performance quiz to #EuroRust25. You don’t want to miss it!
🧵1/3
Announcing p99.chat, the assistant for software performance optimization 🚀
Describe your performance challenge, share your code, and p99 will analyze, instrument, and visualize the results, cutting the traditional setup overhead down to seconds!
Describe your performance challenge, share your code, and p99 will analyze, instrument, and visualize the results, cutting the traditional setup overhead down to seconds!
June 5, 2025 at 12:48 PM
Announcing p99.chat, the assistant for software performance optimization 🚀
Describe your performance challenge, share your code, and p99 will analyze, instrument, and visualize the results, cutting the traditional setup overhead down to seconds!
Describe your performance challenge, share your code, and p99 will analyze, instrument, and visualize the results, cutting the traditional setup overhead down to seconds!
We’re at @pyconit.bsky.social with @adriencaccia.bsky.social!
We’ll quizz you on Python performance again tomorrow! See you in room lasagna at 11:45!
We’ll quizz you on Python performance again tomorrow! See you in room lasagna at 11:45!
May 30, 2025 at 8:56 PM
We’re at @pyconit.bsky.social with @adriencaccia.bsky.social!
We’ll quizz you on Python performance again tomorrow! See you in room lasagna at 11:45!
We’ll quizz you on Python performance again tomorrow! See you in room lasagna at 11:45!
Reposted by Arthur Pastel
🦀 @apas.tel - The Impact of Memory Allocators on Performance
Memory allocators are often overlooked. But Arthur makes the case that they’re a key lever for performance in Rust. Learn how they work and when to care.
🧵 5/11
📺 Watch: youtu.be/pJ-FRRB5E84
Memory allocators are often overlooked. But Arthur makes the case that they’re a key lever for performance in Rust. Learn how they work and when to care.
🧵 5/11
📺 Watch: youtu.be/pJ-FRRB5E84
The Impact of Memory Allocators on Performance: A Deep Dive - Arthur Pastel | EuroRust 2024
YouTube video by EuroRust
youtu.be
May 15, 2025 at 8:21 AM
🦀 @apas.tel - The Impact of Memory Allocators on Performance
Memory allocators are often overlooked. But Arthur makes the case that they’re a key lever for performance in Rust. Learn how they work and when to care.
🧵 5/11
📺 Watch: youtu.be/pJ-FRRB5E84
Memory allocators are often overlooked. But Arthur makes the case that they’re a key lever for performance in Rust. Learn how they work and when to care.
🧵 5/11
📺 Watch: youtu.be/pJ-FRRB5E84
We just improved our C++ integration. Now with first-class Bazel support for larger projects!
codspeed.io/changelog/20...
codspeed.io/changelog/20...
First Class Bazel Support for C++ - Changelog - CodSpeed
New updates and improvements released to Codspeed.
codspeed.io
May 7, 2025 at 3:30 PM
We just improved our C++ integration. Now with first-class Bazel support for larger projects!
codspeed.io/changelog/20...
codspeed.io/changelog/20...
We just released support for C++ benchmarks, enabling the use of our google benchmark compatibility layer to measure the performance of C++ codebases!
Learn more: codspeed.io/changelog/20...
Learn more: codspeed.io/changelog/20...
March 27, 2025 at 2:54 PM
We just released support for C++ benchmarks, enabling the use of our google benchmark compatibility layer to measure the performance of C++ codebases!
Learn more: codspeed.io/changelog/20...
Learn more: codspeed.io/changelog/20...
Super excited about our latest changes to make performance testing more accessible! Two significant changes:
✅ The Free plan now includes private repositories (up to 5 users)
✅ Macro Runners are now globally available (bare-metal ARM64, 16 cores, 32GB RAM) with 120 min/month included for free
✅ The Free plan now includes private repositories (up to 5 users)
✅ Macro Runners are now globally available (bare-metal ARM64, 16 cores, 32GB RAM) with 120 min/month included for free
March 4, 2025 at 3:56 PM
Super excited about our latest changes to make performance testing more accessible! Two significant changes:
✅ The Free plan now includes private repositories (up to 5 users)
✅ Macro Runners are now globally available (bare-metal ARM64, 16 cores, 32GB RAM) with 120 min/month included for free
✅ The Free plan now includes private repositories (up to 5 users)
✅ Macro Runners are now globally available (bare-metal ARM64, 16 cores, 32GB RAM) with 120 min/month included for free
We now support sharded benchmark execution to provide even faster performance feedback!
February 28, 2025 at 3:24 PM
We now support sharded benchmark execution to provide even faster performance feedback!
Single-core performance gains have largely plateaued. That means no more relying on “free” hardware speed boosts. CPUs haven't stopped improving, but we must actively tap into new architectures. Concurrency, SIMD, and thoughtful optimizations can still unlock impressive speedups.
February 19, 2025 at 9:48 AM
Single-core performance gains have largely plateaued. That means no more relying on “free” hardware speed boosts. CPUs haven't stopped improving, but we must actively tap into new architectures. Concurrency, SIMD, and thoughtful optimizations can still unlock impressive speedups.
`x * x` generates exactly the same assembly as `x.pow(2)` in Rust and this goes even beyond powers of 2, thanks to LLVM loop unrolling and exponentiation by squaring.
February 17, 2025 at 2:48 PM
`x * x` generates exactly the same assembly as `x.pow(2)` in Rust and this goes even beyond powers of 2, thanks to LLVM loop unrolling and exponentiation by squaring.
Reposted by Arthur Pastel
https://codspeed.io is pretty cool and class creation will be 38–52% faster in attrs 25.2.0
https://github.com/python-attrs/attrs/pull/1407
https://github.com/python-attrs/attrs/pull/1407
February 16, 2025 at 9:50 AM
https://codspeed.io is pretty cool and class creation will be 38–52% faster in attrs 25.2.0
https://github.com/python-attrs/attrs/pull/1407
https://github.com/python-attrs/attrs/pull/1407
The big issue with such an approach is that the instrumentation is limited to the interpreter itself while for native binary instrumentation the boundary is system calls and only the user space can be instrumented.
For languages running on a VM (Python, Java, JS, etc.), performance instrumentation should happen at the VM level. Execution environments vary, but the interpreter is the single source of truth for performance behavior.
February 13, 2025 at 11:46 PM
The big issue with such an approach is that the instrumentation is limited to the interpreter itself while for native binary instrumentation the boundary is system calls and only the user space can be instrumented.
We just released a Divan 🛋️ integration. So far, this is the best developer experience you can hope for building performance tests in Rust 🦀!
codspeed.io/changelog/20...
codspeed.io/changelog/20...
Divan Support for Rust - Changelog - CodSpeed
New updates and improvements released to Codspeed.
codspeed.io
February 13, 2025 at 3:59 PM
We just released a Divan 🛋️ integration. So far, this is the best developer experience you can hope for building performance tests in Rust 🦀!
codspeed.io/changelog/20...
codspeed.io/changelog/20...
Started hacking on a small TUI for managing EC2 instances and SSH connections. Can’t wait to ditch AWS Console's clunkiness.
February 12, 2025 at 10:21 PM
Started hacking on a small TUI for managing EC2 instances and SSH connections. Can’t wait to ditch AWS Console's clunkiness.
For languages running on a VM (Python, Java, JS, etc.), performance instrumentation should happen at the VM level. Execution environments vary, but the interpreter is the single source of truth for performance behavior.
February 11, 2025 at 9:34 PM
For languages running on a VM (Python, Java, JS, etc.), performance instrumentation should happen at the VM level. Execution environments vary, but the interpreter is the single source of truth for performance behavior.
Lets go! I’ll be in Pittsburgh again in may for PyCon US 2025! Super excited to try a new interactive talk format this year!
February 10, 2025 at 10:44 PM
Lets go! I’ll be in Pittsburgh again in may for PyCon US 2025! Super excited to try a new interactive talk format this year!
We've just released some new changes, improving the UX for large-scale results display and sharing results when there are many benchmarks!
Check out the demos:
Check out the demos:
Improvements of the Benchmark and Branch pages - Changelog - CodSpeed
New updates and improvements released to Codspeed.
codspeed.io
December 17, 2024 at 9:48 AM
We've just released some new changes, improving the UX for large-scale results display and sharing results when there are many benchmarks!
Check out the demos:
Check out the demos:
Reposted by Arthur Pastel
Day 3 of the Advent of CodSpeed is done!
You can still be a part of it and test your code's performance against the other participants at codspeed.io/advent
You can still be a part of it and test your code's performance against the other participants at codspeed.io/advent
December 4, 2024 at 5:30 PM
Day 3 of the Advent of CodSpeed is done!
You can still be a part of it and test your code's performance against the other participants at codspeed.io/advent
You can still be a part of it and test your code's performance against the other participants at codspeed.io/advent
Reposted by Arthur Pastel
[Sponsored post]
This December, @codspeed.io is hosting an epic Rust Performance Leaderboard for the Advent of Code! 🦀 The fastest solutions win amazing prizes like a MacBook, Nintendo Switch, and more!
➡️ codspeed.io/advent
#rust #rustlang #eurorust24 #adventofcode #codspeed #programming
This December, @codspeed.io is hosting an epic Rust Performance Leaderboard for the Advent of Code! 🦀 The fastest solutions win amazing prizes like a MacBook, Nintendo Switch, and more!
➡️ codspeed.io/advent
#rust #rustlang #eurorust24 #adventofcode #codspeed #programming
Advent of CodSpeed - CodSpeed
A performance challenge based on the Advent of Code problems.
codspeed.io
December 4, 2024 at 1:48 PM
[Sponsored post]
This December, @codspeed.io is hosting an epic Rust Performance Leaderboard for the Advent of Code! 🦀 The fastest solutions win amazing prizes like a MacBook, Nintendo Switch, and more!
➡️ codspeed.io/advent
#rust #rustlang #eurorust24 #adventofcode #codspeed #programming
This December, @codspeed.io is hosting an epic Rust Performance Leaderboard for the Advent of Code! 🦀 The fastest solutions win amazing prizes like a MacBook, Nintendo Switch, and more!
➡️ codspeed.io/advent
#rust #rustlang #eurorust24 #adventofcode #codspeed #programming
Reposted by Arthur Pastel
It's a wrap for Day 1 of the Advent of CodSpeed. We saw a lot of interesting optimizations and ideas. Let's continue pushing through this advent month! 🎄
December 2, 2024 at 6:57 PM
It's a wrap for Day 1 of the Advent of CodSpeed. We saw a lot of interesting optimizations and ideas. Let's continue pushing through this advent month! 🎄