The hottest Performance Substack posts right now

And their main takeaways
Category
Top Finance Topics

InferenceX v2: NVIDIA Blackwell Vs AMD vs Hopper - Formerly InferenceMAX

SemiAnalysis • 10506 implied HN points • 16 Feb 26
🕹 Technology Performance
  1. Nvidia’s Blackwell family (B200/B300/GB200/GB300) and NVL72 rack-scale systems deliver much higher inference throughput and far better tokens-per-dollar than prior Hopper GPUs, especially when paired with TensorRT-LLM, disaggregated prefill, and wide expert parallelism.
  2. AMD’s MI355X can be competitive on single-node FP8 SGLang setups, but its software stack struggles to compose FP4, disaggregated prefill, and wide EP together; AMD needs stronger upstream contributions, CI resources, and focus on composability to close the gap.
  3. Disaggregated prefill, wide expert parallelism, and multi-token prediction (MTP) are the key inference optimizations today, and when tuned against the throughput-vs-latency tradeoff they can massively lower cost per token while requiring accuracy checks to avoid silent regressions.

Is SwiftUI finally as fast as UIKit in iOS 26?

Jacob’s Tech Tavern • 3717 implied HN points • 09 Mar 26
🕹 Technology Performance
  1. iOS 26 brings big SwiftUI improvements focused on List updates and scroll performance that Apple emphasized at WWDC.
  2. A brutal stress test was built—a chaotic scrolling feed with high‑res GIFs, complex layouts, autoplaying animations, variable cell sizes, and multi-gesture interactions—to force 120fps and compare SwiftUI vs UIKit.
  3. Early real-world results show noticeable drops in scroll hitches on the iOS 26 SDK, suggesting SwiftUI may be nearing UIKit parity for demanding feeds, though some edge-case features still require falling back to UIKit.

My Warning About AI Music from 1988

The Honest Broker • 14029 implied HN points • 30 Jan 26
🎶 Music Performance
  1. Human imperfections are central to artistic expression; mistakes and rough edges can make music more expressive and emotionally powerful.
  2. Art is an expressive human activity, so works produced by machines—even if technically flawless—are categorically different from human-created masterpieces and lack the same expressive meaning.
  3. AI that closely imitates human creativity can feel unsettling, and this gap between human artistry and machine imitation can’t be bridged merely by better algorithms.

One Man In His Time Plays Many Parts

Gideon's Substack • 9 implied HN points • 24 Mar 26
🎭️ Culture Performance
  1. Writing for a quarterly gives critics time to think, but those pieces often appear after productions have closed, so theater criticism frequently becomes the only lasting record of ephemeral shows.
  2. Rising costs for live performers make large-cast classical plays harder to stage, so theaters respond with higher prices, star casting, inventive doubling, or by reimagining works as one-person shows.
  3. One-person shows can be theatrically inventive—using technology or a virtuoso performer to create many roles and worlds—but they can also feel lonely or mournful, reflecting social isolation as much as artistic choice.

My true love hath my heart

The Common Reader • 2445 implied HN points • 24 Feb 26
📚 Literature Performance
  1. They each give their heart to the other in a balanced, mutual exchange, showing deep love and trust.
  2. Their hearts unite them so that each guides the other's thoughts and feelings, making them feel like one.
  3. They share equal wounds and sorrows, and that shared pain becomes part of the bond that brings them happiness.
Get a weekly roundup of the best Substack posts, by hacker news affinity:

Death by 29 Cuts

Knicks Film School • 773 implied HN points • 24 Oct 24
🎾 Sports Performance
  1. The Knicks struggled in their opening game due to multiple defensive lapses, allowing the Celtics to score threes in several ways. These defensive issues were often caused by miscommunication or mistakes in execution.
  2. A lack of situational awareness cost the Knicks, leading to open shots for the Celtics after the Knicks failed to switch or cover properly. This shows the need for better communication on the court.
  3. Despite the disappointing loss, not all mistakes are worrying. Some missed plays will help the younger players learn, and there is hope for improvement as the season progresses.

How reliable is your node?

Kerman Kohli • 99 implied HN points • 29 Oct 24
🔮 Crypto Performance
  1. RPC calls to blockchain nodes only succeed about 78.5% of the time on average. This means that sometimes you might have trouble getting the data you need.
  2. The performance of nodes varies depending on the blockchain you’re accessing, the RPC provider you choose, and even the time of day you make your requests.
  3. To ensure better reliability, it’s smart to use multiple node providers rather than depending on just one. This way, if one fails, you have a backup.

Size Does Matter: Why -Os Beat -O2 on My ESP32-S3

atomic14 • 346 implied HN points • 07 Mar 26
🕹 Technology Performance
  1. On the ESP32-S3, compiling with -Os (optimize for size) gave better results than using -O2 (optimize for speed).
  2. Binary size can matter more than you might expect on constrained microcontrollers, so smaller builds can be preferable.
  3. This challenges the common assumption that higher optimization levels focused on speed are always the best choice for embedded targets.

Could Liverpool Really Lose Mo Salah, TAA, *and* VVD After This Season?

No Grass in the Clouds • 99 implied HN points • 24 Oct 24
🎾 Sports Performance
  1. Liverpool's best players, Mo Salah, Trent Alexander-Arnold, and Virgil van Dijk, could leave after this season due to uncertain contract situations.
  2. The club has faced issues with its leadership and management, making it tough to handle player contracts effectively.
  3. Players may hesitate to sign long-term deals when they see instability in the club's management and direction.

Emulating future RISC-V Vector extensions

Fprox’s Substack • 145 implied HN points • 08 Mar 26
🕹 Technology Performance
  1. You can emulate proposed RISC‑V Vector extensions by translating them into RVV 1.0 intrinsics, so programs using new instructions can run on existing RVV1.0 hardware without compiler or hardware support for the new ops.
  2. The generated emulation is functional and easy to run but not optimal: the code is verbose and much slower than a dedicated hardware implementation, though it still lets you measure real performance and iterate on designs.
  3. The tool is Python‑driven and open source, already supports several draft extensions, and is useful for extension designers and early application developers to prototype and test features before toolchain or hardware support exists.

TPRR check-in: Who has made strides earning volume so far?

Stealing Signals • 599 implied HN points • 03 Oct 24
🎾 Sports Performance
  1. Routes data is really important for understanding how well players are performing. Different sources measure these routes in different ways, which can create confusion.
  2. The NFL has started providing its own routes data, which could help standardize how we analyze player performance. This might make comparisons easier and clearer moving forward.
  3. Stats like TPRR (Targets Per Route Run) help us understand player efficiency, but they need to be used alongside other context like player roles and QB performance for better insights.

A Tale of Two Halftime Shows

Common Sense with Bari Weiss • 1038 implied HN points • 09 Feb 26
🎭️ Culture Performance
  1. The Super Bowl halftime is one of the rare national events people watch live together, so whatever happens there carries outsized cultural weight.
  2. Bad Bunny’s halftime leaned into localism and community, recreating a small Puerto Rican town with colorful, multigenerational, human-scale moments.
  3. That joyful, local approach stood in sharp contrast to more sterile or grievance-driven presentations, like the grayscale Turning Point USA-style shows or industrial, cube-lit productions.

Why Do Children Hate Music Lessons?

The Honest Broker • 17221 implied HN points • 06 Aug 25
🎶 Music Performance
  1. Children often dislike music lessons because they feel boring and formal, unlike the fun of making music for joy. Switching the focus from 'lessons' to 'play' can change this experience.
  2. The pressure from parents and the educational system makes music feel like a chore, not a hobby. This can take away the excitement and fun of learning an instrument.
  3. Competitions and perfectionism in music lessons can ruin the enjoyment children get from playing music. It's important to create an environment where making music is fun and not just about being the best.

AMD's Ryzen 7 9850X3D: A Review

More Than Moore • 186 implied HN points • 01 Mar 26
🕹 Technology Performance
  1. The Ryzen 7 9850X3D is basically a higher‑binned 9800X3D with faster clocks, but it only delivers tiny performance gains while drawing significantly more power and costing more.
  2. AMD’s 3D V‑Cache really helps CPU‑bound, cache‑hungry games and makes memory speed matter less, but it doesn’t improve compute‑heavy workloads and offers no advantage for AI paths that need an NPU.
  3. On value, the 9800X3D or cheaper Intel options give better performance‑per‑dollar, so most buyers should pick the cheaper chip and spend any savings on other parts like memory amid volatile DRAM prices.

The Met Turned a Masterpiece Into an Anti-ICE Screed

Common Sense with Bari Weiss • 1279 implied HN points • 16 Jan 26
🎭️ Culture Performance
  1. The Met’s new production reimagines Carmen as a contemporary story about ICE agents guarding a gun factory on the U.S.–Mexico border.
  2. While bold reinterpretations can breathe new life into classics, this staging is criticized for stripping away the opera’s original music, color, and sensual charm.
  3. Overlaying current political issues onto Carmen clashes with its traditional themes of passion and natural vitality, making the update feel forced and ineffective.

Copy-on-write teaches you EVERYTHING about Swift Internals 🐮

Jacob’s Tech Tavern • 2405 implied HN points • 15 Dec 25
🕹 Technology Performance
  1. isKnownUniquelyReferenced is the tiny runtime check Swift uses to tell if a heap-backed value has only one owner. It’s the key mechanism that makes copy-on-write work under the hood.
  2. Copy-on-write lets structs behave like independent value types while sharing heap storage until you mutate them, at which point a uniqueness check triggers a deep copy. This gives easy-to-reason-about value semantics with low memory overhead.
  3. Many core Swift types (Array, Set, Dictionary, String, Data) use copy-on-write, and you can implement it yourself by wrapping your value in a reference box and using isKnownUniquelyReferenced to decide when to copy.

The Precious Eyeblink

Software Design: Tidy First? • 1414 implied HN points • 29 Dec 25
🕹 Technology Performance
  1. Human attention slips if feedback takes longer than about 400 milliseconds, so tools should aim to give immediate responses to keep people in flow.
  2. There’s a tradeoff between completeness and speed: faster, partial feedback often helps more than slow, perfect answers because delays invite distraction.
  3. Tool designers should prioritize the most important feedback first, degrade gracefully with partial results, let users choose the completeness/speed tradeoff, and measure time-to-first-feedback so latency is kept low.

How Do You Know If You’re a Good Leader?

Fish Food for Thought • 42 implied HN points • 11 Mar 26
💼 Business Performance
  1. Doubt and introspection are part of good leadership, not proof you're failing. Managing uncertainty and reflecting privately helps you make clearer public decisions.
  2. You're judged differently by your boss, peers, and team, so evaluate yourself from all those angles. Combine those perspectives to get a more accurate picture of your leadership.
  3. Seek real feedback and take ownership of perceptions by doing a 360-style review and looking for patterns. If feedback is valid, acknowledge it and make a plan; if you disagree, still address the impact rather than arguing intent.

Welcome!

Ben’s Blog 🏉 🧠🧑‍💻 • 21 implied HN points • 04 Mar 26
🏥 Health & Wellness Performance
  1. Sustainable performance needs both effort and careful energy management. Effort builds growth, but energy is what keeps that growth going.
  2. Life after professional sport can become a meaningful second career through coaching, speaking, and leading teams, turning past experience into purposeful work.
  3. Community action and storytelling — like free events and sharing personal stories — help fight suicide stigma and build resilience.

The show is over, say goodbye

Disaffected Newsletter • 4316 implied HN points • 21 Mar 24
🎭️ Culture Performance
  1. Madonna's recent performance highlighted her struggle with aging and the desire to cling to her past fame, which some people found sad and awkward.
  2. The concert's production quality was disappointing, with issues like poor sound and lip-syncing, leaving many fans feeling cheated.
  3. The author's views on Madonna shifted over time, reflecting a broader change in attitudes towards celebrity culture and its impacts on personal growth.

Huang's Law

The Chip Letter • 6115 implied HN points • 18 Jun 25
🕹 Technology Performance
  1. Huang's Law suggests that the performance of AI chips is improving much faster than what we used to call Moore's Law. It claims chips double their performance every year or so, which is a big leap forward.
  2. This new law emphasizes performance improvements related to AI, unlike Moore's Law, which was mostly about the number of transistors. It's all about how quickly these chips can process complex tasks.
  3. However, some experts think Huang's Law might not last as long as Moore's Law. While it's exciting now, it's still uncertain if this rapid improvement can continue in the future.

Going all in on becoming a Hyrox Athlete (Guest Post by Holly Gabrielle) | #323

Boundless by Paul Millerd • 66 implied HN points • 24 Feb 26
🎾 Sports Performance
  1. She quit a stable writing job to go all‑in on becoming an elite Hyrox athlete, choosing action over safety to chase a bold, concrete goal.
  2. Stepping away from work is about more than time — it’s about reclaiming mental energy for focused training, recovery, and better planning using an essentialist, "Hell yeah or no" approach.
  3. She’s rebuilding a fitness-focused brand with her partner and documenting the journey while living in cheaper cities to stretch savings; there’s little income now and clear financial risk, but she views the pursuit as worth the tradeoff.

The 4 types of men, quiet quitting, and impressions

The Rubesletter by Matt Ruby (of Vooza) | Sent every Tuesday • 570 implied HN points • 02 Jan 26
😂 Humor Performance
  1. A highlights reel of short comedy bits from the past year covers topics like the four types of men, quiet quitting, impressions, and other quick jokes.
  2. There’s a focus on standup clips and short-form comedy, with a top-10 YouTube Shorts list and ways to follow more on YouTube, Instagram, and TikTok.
  3. Subscribers are promised extra bonus content and a one-year 20% New Year’s discount, and readers are invited to give feedback or share the pieces they like.

Crushing on the singer or rock star

Rock 'n' Roll with Me • 419 implied HN points • 30 Jul 24
🎶 Music Performance
  1. It's normal to crush on musicians and rock stars. They create strong feelings in us through their art, and it feels like we connect with them on a deeper level.
  2. People often confuse their feelings for the artist with feelings for the person. We get swept up in the emotions their work brings out in us.
  3. Many of us don't pay enough attention to each other. When someone really listens, it makes us feel seen and loved, leading to these intense crushes.

🏰 Lessons on AI and art from the Magic Kingdom

Faster, Please! • 365 implied HN points • 14 Jan 26
🎭️ Culture Performance
  1. A theme park fired a popular character performer for breaking rules meant to protect the show's illusion and guest experience.
  2. The park would likely not replace her with a robot, which shows some creative roles rely on human presence, spontaneity, and authenticity.
  3. The episode highlights a broader lesson about AI: smart machines can help, but they often can't recreate the subtle human nuances and emotional authenticity that define many creative jobs.

Connection Pooling: Fundamentals, Challenges and Trade-offs

Engineering At Scale • 795 implied HN points • 29 Nov 25
🕹 Technology Performance
  1. Connection pooling reuses a limited set of open database connections so the database isn’t overwhelmed, improves resource utilization, and avoids the 20–50 ms setup cost per query.
  2. Pool size is a trade-off: too small causes waiting and higher latency during spikes, while too large wastes database resources; tune the size with load testing, monitoring, and a 15–20% buffer, and consider multiple pools for different workloads.
  3. Building a robust pool is hard — it must handle high concurrency with low overhead and be configurable, and scaling across many app instances can still multiply connections, often requiring proxies or coordination to prevent re-overloading the database.

Giving your AI a Job Interview

One Useful Thing • 1028 implied HN points • 12 Nov 25
🕹 Technology Performance
  1. Measuring AI performance is tricky because common tests can be flawed and sometimes don't really show how smart the AI is. We're often left uncertain about what these benchmarks actually mean.
  2. Using a more personal approach, like creating fun and unique tests, can help people understand how different AI models work. This way, you get a feel for the AI's strengths and weaknesses in a more relatable way.
  3. When companies choose AI tools, it's important to do thorough testing based on real tasks instead of just relying on average performance scores. Understanding specifically how well an AI can perform your unique tasks is key.

Epiphanies

Dada Drummer Almanach • 178 implied HN points • 01 Feb 26
🎶 Music Performance
  1. He finds freedom inside musical limits, stretching time and phrasing like a jazz musician to make his drumming and singing constantly surprising and expressive.
  2. He lives simply and generously in retirement, valuing small pleasures, leftist principles, and warm, kind interactions with fans and friends.
  3. His work turns ordinary details into deep feeling and has profoundly influenced other musicians, teaching new ways to play drums and sing.

Involutions: The License to Update in O(1)

Software Bits Newsletter • 206 implied HN points • 14 Jan 26
🕹 Technology Performance
  1. XOR is an involution: applying the same XOR twice cancels it out, so adding and removing an element use the same operation and let you update combined hashes in O(1).
  2. Zobrist hashing leverages XOR to update a chessboard hash with only a few XORs per move, enabling fast transposition-table lookups and huge search speedups; collisions are possible but usually acceptable or verifiable.
  3. More generally, pick the algebraic tool that matches your needs — use involutions like XOR for O(1) incremental updates when collisions are tolerable, rolling linear hashes for sliding windows, or Merkle trees when cryptographic integrity is required.

High Performance SwiftData Apps

Jacob’s Tech Tavern • 2842 implied HN points • 14 Jul 25
🕹 Technology Performance
  1. The app developed for Comic-Con was popular for its cool features but struggled with performance issues. As I used it, the app got slower, draining the battery and eventually crashing.
  2. I needed to improve the app's performance by optimizing how it used SwiftData without losing the cards I had already created. It was important to keep my collection safe while fixing the issues.
  3. This experience highlighted how vital it is to focus on app efficiency and data management to avoid frustration for users and devices alike.

Toohey’s Hot Start and Traore’s Champions League Debut Highlight Top International Prospect Performances

Chad Ford's NBA Big Board • 59 implied HN points • 03 Oct 24
🎾 Sports Performance
  1. Nolan Traore had an impressive debut in the Champions League, showing his talent despite earlier struggles. His performance helped his team win a big game.
  2. Alex Toohey is off to a strong start in Australia, making a name for himself early in the season. His skills are catching the attention of scouts.
  3. Ben Saraf and Noa Essengue continue to excel in Germany, proving they are top prospects. Their consistent play is making them standout players in their league.

The Freaks Are the Last to Go

Working Theorys • 276 implied HN points • 09 Jan 26
🎭️ Culture Performance
  1. Some creators—"freaks"—encode meaning so tightly that you have to work to understand it, and that effort is part of the pleasure.
  2. Modern culture prizes legibility and frictionless consumption, which flattens art and flattens us. Freaks deliberately keep mystery and timing so meaning arrives later and feels deeper.
  3. Freaks resist being named or sanitized by success and are rare, so find and support them early—they are the stubborn keepers of cultural depth.

Mrs Warren's Profession. A splendid play performed almost pefectly.

The Common Reader • 2090 implied HN points • 28 Jul 25
🎭️ Culture Performance
  1. Imelda Staunton gives an amazing performance in 'Mrs Warren's Profession.' She really captures the emotions and nuances of the play without just being loud or intense.
  2. The production is mostly well done, but the director added unnecessary elements like ghosts and emotional music that distract from Shaw's original message.
  3. Despite these flaws, the play is still worth seeing. Staunton's talent makes it a wonderful experience that shouldn't be missed.

Literature in the Octagon

Life Since the Baby Boom • 1844 implied HN points • 17 Aug 25
📚 Literature Performance
  1. Poetry slams have turned poetry into a competitive sport, making it more popular but also more theatrical. Some think this takes away from the true meaning of poetry.
  2. There seems to be a growing trend of focusing on identity in both poetry and fiction, which some argue makes the writing less relatable or enjoyable.
  3. Literature slams could be a fun way to showcase fiction, but many writers may feel it’s too casual or worry about how the audience will react.

What Progress Has There Been in Industrial Robots?

Construction Physics • 18999 implied HN points • 10 Jan 24
🕹 Technology Performance
  1. Industrial robots have become more cost-effective over time, making them more accessible for various applications.
  2. Advances in industrial robots have led to significant improvements in precision and smooth, continuous motion capabilities.
  3. There has been a trend towards standard robotic architectures, with modern robots primarily consisting of robotic arms with electric drives and servo motors.

RISC-V and Post Quantum Cryptography

Fprox’s Substack • 186 implied HN points • 18 Jan 26
🕹 Technology Performance
  1. Quantum computers threaten today’s public-key cryptography, so governments and industry are already moving to post-quantum algorithms and rolling out standards and deployments now.
  2. Post-quantum schemes (e.g., Kyber, Dilithium, SPHINCS+, Falcon) rely on heavy math like NTT and Keccak, and they trade off key/signature sizes, signing speed, and verification cost differently.
  3. RISC-V can run PQC today using its vector extensions, but lacks dedicated PQC ISA support; targeted accelerations for NTT and Keccak (and vector crypto extensions) would greatly improve performance and are being explored by the community.

The One Property That Makes FlashAttention Possible

Software Bits Newsletter • 257 implied HN points • 29 Dec 25
🕹 Technology Performance
  1. Associativity is the key property that lets you split work, combine partial results, and safely parallelize or stream computations without changing the answer.
  2. Softmax has a hidden associative state — tracking a local max and a scaled sum lets you correct and merge chunked results, which is the math behind FlashAttention’s memory- and time-saving trick.
  3. When optimizing a global computation, look for a small combinable state and an associative combine rule; if it exists you can chunk and parallelize, and if it doesn’t (for example, median) you need a different algorithmic approach.

I Am Begging Sports Media to Stop Misusing Regression to the Mean

Freddie deBoer • 2165 implied HN points • 03 Jul 25
🎾 Sports Performance
  1. Regression to the mean means that extreme results are unlikely to happen again without some change in conditions. If a team's situation changes, it’s not just luck but a new factor affecting performance.
  2. Using regression to the mean incorrectly can lead to confusion. If someone thinks a team will do worse because they lost players, that’s not regression to the mean; it’s a different kind of prediction.
  3. There’s a risk of making mistakes by assuming past results will always influence future ones, like betting based on past game outcomes. Each situation should be treated by its own conditions.

Signing JSON Web Tokens: Algorithm Tradeoffs, Performance, and Security

ciamweekly • 62 implied HN points • 09 Feb 26
🕹 Technology Performance
  1. Pick your JWT algorithm based on tradeoffs: HMAC (HS) is very fast and simple but uses a shared secret and cannot provide non-repudiation, while asymmetric algorithms let you separate signing and verification.
  2. Prefer modern asymmetric schemes when possible: RSA-PSS is safer than old PKCS#1 v1.5, ECDSA gives small fast signatures but demands perfect nonce randomness, and EdDSA (Ed25519) is usually the best choice because it’s fast, secure, and uses deterministic nonces.
  3. Match algorithm to your environment and tooling: RSA has the widest compatibility but large signatures and slower signing, ECDSA risks come from RNG mistakes, and EdDSA may require checking HSM/KMS and library support before committing.