--- title: "#benchmarking articles | Hashnode" description: "Explore 60 articles tagged #benchmarking on Hashnode. Tutorials, guides, and how-tos from the community." url: "https://hashnode.com/tag/benchmarking" --- Tag # #benchmarking 60 posts · 1 followers [JJozzieTech](https://hashnode.com/@jjozzietech) · [6d ago](https://hashnode.com/posts/2gbps-wan-validation/6aecd754c7f8995ea28e0e9e) ### [Why I tested from inside the network](https://jjozzietech.hashnode.dev/2gbps-wan-validation?utm_source=hashnode&utm_medium=feed) A 2 Gbps WAN validated from inside the production network — nine Ookla runs across three Sydney ISPs, with the measurement discipline shown. 6 min read · [jjozzietech.hashnode.dev](https://jjozzietech.hashnode.dev/2gbps-wan-validation?utm_source=hashnode&utm_medium=feed) [techpotions](https://hashnode.com/@techpotions) · [Oct 29](https://hashnode.com/posts/build-your-own-ai-coding-benchmarks/6ae385235e8fbbc22225b102) ### [Build Your Own AI Coding Benchmarks](https://techpotions.hashnode.dev/build-your-own-ai-coding-benchmarks?utm_source=hashnode&utm_medium=feed) Stop trusting public leaderboards built on generic repos. Build your own AI coding benchmarks tuned to your codebase: real tickets, precise scoring, and a refusal safety net. Run it cheap, run it often, and re-evaluate monthly. 8 min read · [techpotions.hashnode.dev](https://techpotions.hashnode.dev/build-your-own-ai-coding-benchmarks?utm_source=hashnode&utm_medium=feed) [Brock Claussen](https://hashnode.com/@claus) · [Oct 21](https://hashnode.com/posts/matched-recall/6ad8b7601fd070640089c8e6) ### [My unpublished benchmark post said 10x. The honest number is 1.2x to 1.8x.](https://blog.narcolepticfox.com/matched-recall?utm_source=hashnode&utm_medium=feed) An old Foxstash draft claimed it beat hnswlib by 2x. Its own table showed it losing 3x on one thread. Speed only means something at the same recall, on the same threads, on the dataset the label names. 5 min read · [blog.narcolepticfox.com](https://blog.narcolepticfox.com/matched-recall?utm_source=hashnode&utm_medium=feed) [2](https://hashnode.com/posts/matched-recall/6ad8b7601fd070640089c8e6) [Kuldeep Singh](https://hashnode.com/@ar5en1c) · [Oct 15](https://hashnode.com/posts/desktop-control-benchmark/6ad1401cc95ead9bd1583d6e) ### [I built desktop control so an agent could test my Chrome extension. Then I benchmarked it against four free tools.](https://ar5en1c.hashnode.dev/desktop-control-benchmark?utm_source=hashnode&utm_medium=feed) Same model, same Mac, 153 runs. One tool passed one more run than mine. Mine took half the time. 6 min read · [ar5en1c.hashnode.dev](https://ar5en1c.hashnode.dev/desktop-control-benchmark?utm_source=hashnode&utm_medium=feed) [6](https://hashnode.com/posts/desktop-control-benchmark/6ad1401cc95ead9bd1583d6e) [nRouter.ai](https://hashnode.com/@nrouterai) · [Oct 14](https://hashnode.com/posts/reading-a-benchmark-table-honestly-ttft-percentiles-from-a-warm-pool-gateway-test/6acf5de7b330e9f3b306e3c2) ### [Reading a Benchmark Table Honestly: TTFT Percentiles From a Warm-Pool Gateway Test](https://nrouter.hashnode.dev/reading-a-benchmark-table-honestly-ttft-percentiles-from-a-warm-pool-gateway-test?utm_source=hashnode&utm_medium=feed) Vendor benchmarks deserve suspicion. Most hide the workload, cherry-pick the tier, or quote a p50 and let you assume the tail behaves. So let's read one the way it should be read — method first, numbe 3 min read · [nrouter.hashnode.dev](https://nrouter.hashnode.dev/reading-a-benchmark-table-honestly-ttft-percentiles-from-a-warm-pool-gateway-test?utm_source=hashnode&utm_medium=feed) [samuel Hu](https://hashnode.com/@sam-happy) · [Oct 10](https://hashnode.com/posts/how-many-cameras-can-an-edge-ai-box-really-handle/6aca0d69641202419e03cee6) ### [How Many Cameras Can an Edge AI Box Really Handle?](https://cosmoedge.hashnode.dev/how-many-cameras-can-an-edge-ai-box-really-handle?utm_source=hashnode&utm_medium=feed) Reading a real RK3576 load test beyond the headline FPS 8 min read · [cosmoedge.hashnode.dev](https://cosmoedge.hashnode.dev/how-many-cameras-can-an-edge-ai-box-really-handle?utm_source=hashnode&utm_medium=feed) [3](https://hashnode.com/posts/how-many-cameras-can-an-edge-ai-box-really-handle/6aca0d69641202419e03cee6) [Brock Claussen](https://hashnode.com/@claus) · [Oct 5](https://hashnode.com/posts/classify-the-errors-before-you-trust-the-score/6ac39f6da1a61e9cc915ac0e) ### [Classify the errors before you trust the score](https://blog.narcolepticfox.com/classify-the-errors-before-you-trust-the-score?utm_source=hashnode&utm_medium=feed) A forced model swap in Mousing, nine benchmark runs, and why the expensive model's lead mostly belonged to the answer key 5 min read · [blog.narcolepticfox.com](https://blog.narcolepticfox.com/classify-the-errors-before-you-trust-the-score?utm_source=hashnode&utm_medium=feed) [Levelbrook Consulting](https://hashnode.com/@levelbrook) · [Sep 21](https://hashnode.com/posts/how-to-read-a-coding-agent-benchmark-without-getting-sold/6ab14092d31db0960eeebe20) ### [How to read a coding-agent benchmark without getting sold](https://levelbrook.hashnode.dev/how-to-read-a-coding-agent-benchmark-without-getting-sold?utm_source=hashnode&utm_medium=feed) A nine-author study this week pulled a coding agent apart into its components and measured each one across 176 configurations. The findings are less e 9 min read · [levelbrook.hashnode.dev](https://levelbrook.hashnode.dev/how-to-read-a-coding-agent-benchmark-without-getting-sold?utm_source=hashnode&utm_medium=feed) [1](https://hashnode.com/posts/how-to-read-a-coding-agent-benchmark-without-getting-sold/6ab14092d31db0960eeebe20) [Debashish Ghosal](https://hashnode.com/@deghosal) · [Sep 16](https://hashnode.com/posts/one-hung-api-call-used-to-kill-my-1000-run-benchmark-heres-the-fix-mu3djf2n/6aa9e573df3f135ac1531d4f) ### [One Hung API Call Used to Kill My 1,000-Run Benchmark. Here's the Fix.](https://pragmatic-engineer.hashnode.dev/one-hung-api-call-used-to-kill-my-1000-run-benchmark-heres-the-fix-mu3djf2n?utm_source=hashnode&utm_medium=feed) A single hung call discarded hours of completed work. The timeout, checkpointing, and resume patterns that make long agent benchmarks survivable. 4 min read · [pragmatic-engineer.hashnode.dev](https://pragmatic-engineer.hashnode.dev/one-hung-api-call-used-to-kill-my-1000-run-benchmark-heres-the-fix-mu3djf2n?utm_source=hashnode&utm_medium=feed) [Oliver Zehentleitner](https://hashnode.com/@oliver-zehentleitner) · [Sep 13](https://hashnode.com/posts/picows-in-unicorn-binance-websocket-api-up-to-2-the-throughput-opt-in-for-now/6aa6bae3dc399e9dd58cbc15) ### [picows in UNICORN Binance WebSocket API: Up to 2× the Throughput, Opt-In for Now](https://blog.technopathy.club/picows-in-unicorn-binance-websocket-api-up-to-2-the-throughput-opt-in-for-now?utm_source=hashnode&utm_medium=feed) UBWA 2.16.0 can run on picows, a Cython WebSocket implementation, instead of websockets. Benchmarks, a 24 h soak against live Binance, three bugs found on the way, and why the default has not changed yet. 10 min read · [blog.technopathy.club](https://blog.technopathy.club/picows-in-unicorn-binance-websocket-api-up-to-2-the-throughput-opt-in-for-now?utm_source=hashnode&utm_medium=feed)