HeyJaredLow Latency Trading Insights
Fresh insights on low latency and high frequency trading. Mostly new relevant technology, C++, FPGA, Verilog, tutorials, short notes and of course, rants. Problem subscribing? Try https://lucisqr.lemonsqueezy.com/buy/c78e2dd6-81c0-44cd-bfa3-13db7158a050
- Indexed issues, last 90 days
- 23
- Latest publication
- Sep 29, 2026
- Audience
- Checking…
- Earliest in this view
- Aug 20, 2026
Latest issues
Weekly digest of what crossed the source scan, September 21-27. (opens the original)
Read excerpt
The top toolchain story is Fangrui Song's cold look at three linker I/O tricks that mold uses but lld declines, with numbers showing where the warm-cache wins collapse. The market structure half was unusually busy: IEX enters options, three exchanges filed for perpetual single-stock futures in the same week, and the 23/5 overnight trading infrastructure keeps filling in.https://hftuniversity.com/post/low-latency-radar-sep-21-27-2026<div class="prea
Your Sliding-Window Min Is Amortised O(1). That's the Problem. (opens the original)
Read excerpt
The rolling message-rate counter is a solved problem. Three hundred messages a second, a ring of time buckets, add to the current bucket, subtract the ones that aged out of the window, return the sum. That is the rolling-counter challenge on the platform, and the fast solutions all look the same because there is only one good shape. It is O(1) per update and it is boring, and it is boring for exactly one reason: a count has an inverse. When an event leaves the window you subtract it. Nothing to
You Can Beat the Compiler. llvm-mca Tells You When. (opens the original)
Read excerpt
Every few months someone insists a patient engineer can hand-write assembly that beats the compiler. They are right. OpenSSL, glibc, and ffmpeg all ship hand-tuned assembly that beats -O3, built the way the claim describes: take the compiler's output, change it by hand, measure, repeat. The sentence they leave out is that nobody has beaten a modern out-of-order core by staring at instructions since the Pentium Pro. The hand-tuning that wins is a feedback loop, and the feedback comes from an inst
LLVM's New Eytzinger Table Is Not the One in the Paper (opens the original)
Read excerpt
Two weeks ago LLVM grew a new container. Between July 13 and July 16, Kazu Hirata landed EytzingerTableSpan and EytzingerTable in the ADT library, a sorted array stored in breadth-first order so that binary search touches memory in a cache-friendly pattern. The motivation, per the commit message, is an RFC on faster sample profile loading: the binary sample profile format now stores its guid table in Eytzinger order, and the reader searches it in place without deserializing into a map.If this so
Stroustrup Says Throw Away the Clever Code. (opens the original)
Read excerpt
In a recent podcast appearance, Bjarne Stroustrup made a claim that should make every HFT engineer flinch:“My standard techniques for improving things actually is to first throw away the clever stuff and then see if you run faster or slower. Usually you run faster.”He’s been saying versions of this for years. The argument is that modern compilers see more, optimize harder, and that hand-tuned 1990s code is now pessimised on hardware whose branch predictors and cache hierarchies the cleverness wa
Publishing over time
Last 90 days. Choose a month to open its work.
Recurring subjects
Named in the text we hold. One piece can cover several.
Audience
No verified audience measurement yet.
About this data
Counts cover the work we have indexed. Tone needs enough text and a confident classification. Excerpts and episode notes are not full articles or transcripts.
Identity or attribution wrong? Suggest a correction.