--- title: "Local Performance Report" --- > Documentation Index > Fetch the complete documentation index at: https://openarkstudio.github.io/arknet/llms.txt > Use this file to discover all available pages before exploring further. # Local Performance Report The report contains the final measurements for each protocol: UDP on 2026-10-12, and other protocols and native coroutine TCP on 2026-10-10. Each protocol retains its own source hash and raw records. **1,620 local IPv4 measurements** compare throughput, p99 latency and IO models across the Asio dependency combinations. ## Test Conditions | Configuration | Test machine | | --- | --- | | Machine model | Mac16,11 | | Processor | Apple M4 Pro | | Architecture | arm64 | | Physical / logical CPUs | 14 / 14 | | Memory | 48 GiB | | Operating system | macOS 27.0.1 (26A434) | | Kernel release | 27.0.0 | Release build; measurement=1.0 s; warmup=0.25 s; repetitions=3. Main charts: **1024 bytes, 16 clients, no added computation**. Inflight messages per connection: UDP 1; other protocols 16. Both peers share the process and IO workers over local loopback. Average process CPU uses 100% for one logical CPU and can exceed 100% with multiple threads; estimated values are labeled, and incomplete sets of three CPU samples show N/A. On this 14-logical-CPU host, machine share = process CPU% / 14; 100% of one CPU is 7.14% of the machine. ## Key Results - **TCP / Standalone Asio / No added computation**: 4 contexts / 4 threads has this group's highest throughput median (1.98× one thread); its p99 is 1,346.00 us versus 1,491.79 us with one thread. - **UDP / Standalone Asio / No added computation**: 1 context / 1 thread has this group's highest throughput median (92,515 round trips/s), with p99 RTT of 207.79 us and process CPU utilization of 99.4%. - **Coroutine TCP / Standalone Asio / No added computation**: 1 context / 4 threads has this group's highest throughput median (1.84× one thread); its p99 is 870.46 us versus 1,028.71 us with one thread. - **Coroutine TCP / Standalone Asio / 10,000 CPU iterations per echo**: 4 contexts / 4 threads has this group's highest throughput median (3.54× one thread); its p99 is 1,749.29 us versus 5,687.25 us with one thread. The highest median applies only to the stated load; complete results appear below. ## Throughput ![Local throughput overview](/arknet/performance/assets/overview-throughput.svg) ## p99 Round-Trip Latency 99% of sampled round trips finish within this time; 1000 microseconds = 1 millisecond. ![Local p99 latency overview](/arknet/performance/assets/overview-latency.svg) ## Choosing an IO Model | Model | Workload | Tradeoff | | --- | --- | --- | | 1 context / 1 thread | Low concurrency, control services, light state handling | Long handlers delay every connection | | 1 context / several threads | Uneven connection activity, shared available workers | Use per-connection strands; shared scheduling has overhead | | Several contexts / 1 thread each | Partitioned sessions, tenants or stable assignment | A busy shard cannot borrow other shards' workers | [Coroutine IO and CPU workload comparisons](/arknet/concepts/io-models). ## Detailed Results by Protocol - [TCP](/arknet/performance/tcp) - [UDP](/arknet/performance/udp) - [WebSocket](/arknet/performance/websocket) - [TCP+TLS](/arknet/performance/tcps) - [WSS](/arknet/performance/wss) - [HTTP](/arknet/performance/http) - [HTTPS](/arknet/performance/https) - [Coroutine TCP](/arknet/performance/coroutine) Detailed pages retain all 64/1024/16384-byte and 1/16-client profiles. Bars show repeated-run medians; error bars show min/max, not confidence intervals. [All measured values (JSON)](/arknet/performance/summary.json ':ignore') | [Payload, batch and execution comparisons](/arknet/performance/dimensions) Coroutine window=16 measures batch RTT; the callback program measures message RTT. These results do not establish coroutine speedups. ## Toolchain and Build Apple Clang 21.0.0; CMake 4.3.3; Ninja; Release -O3 -DNDEBUG -std=gnu++20 -arch arm64; TLS ON; sanitizers OFF. Standalone Asio 1.38.2 with BHO Beast based on Boost 1.84 (API 351); Boost 1.90.0 with Asio 1.38.0 and Beast API 359; OpenSSL 3.6.4. Mac mini on AC power, low power mode disabled, no CPU affinity. The dependency combinations use different Beast versions, so differences between them also include dependency-version effects and cannot be attributed solely to Asio. ## Metric Definitions [Throughput, RTT, CPU, RSS and error definitions](/arknet/project/testing). ## Raw Data and Reproduction - TCP, TLS, HTTP, WebSocket and native coroutine source SHA-256: `127a500ad7e2fbe4c831cc3ebacc63acac96013e2cb8bf3893d7997ca6ea7f56` - UDP source SHA-256: `0771574af8ca56f1914deb544b956079e0a2f24021953ded5f5d524e3df334b1`; [conditions and raw records](/arknet/performance/udp-update). - [Completed measurements]: /arknet/performance/1620 - [Commands and validation scope](/arknet/performance/methodology) - [macmini-m4pro-20261010-boost-coroutine.json](https://github.com/OpenArkStudio/arknet/blob/main/benchmarks/results/macmini-m4pro-20261010-boost-coroutine.json): 180 measurements; 2026-10-10T06:22:06.443606+00:00; executable SHA-256 `14b22c5832f3d7731294a735d97ab5464cbf64c4efeb439f8f5f4b7576c70df4` - [macmini-m4pro-20261010-boost-http.json](https://github.com/OpenArkStudio/arknet/blob/main/benchmarks/results/macmini-m4pro-20261010-boost-http.json): 90 measurements; 2026-10-10T06:08:56.779694+00:00; executable SHA-256 `7b2989fd23e15a68a662899d1906f3b094138018b8ada5904309b4a90d96ea34` - [macmini-m4pro-20261010-boost-https.json](https://github.com/OpenArkStudio/arknet/blob/main/benchmarks/results/macmini-m4pro-20261010-boost-https.json): 90 measurements; 2026-10-10T06:16:12.072913+00:00; executable SHA-256 `7b2989fd23e15a68a662899d1906f3b094138018b8ada5904309b4a90d96ea34` - [macmini-m4pro-20261010-boost-tcp.json](https://github.com/OpenArkStudio/arknet/blob/main/benchmarks/results/macmini-m4pro-20261010-boost-tcp.json): 90 measurements; 2026-10-10T05:31:57.122216+00:00; executable SHA-256 `7b2989fd23e15a68a662899d1906f3b094138018b8ada5904309b4a90d96ea34` - [macmini-m4pro-20261010-boost-tcps.json](https://github.com/OpenArkStudio/arknet/blob/main/benchmarks/results/macmini-m4pro-20261010-boost-tcps.json): 90 measurements; 2026-10-10T05:56:20.110867+00:00; executable SHA-256 `7b2989fd23e15a68a662899d1906f3b094138018b8ada5904309b4a90d96ea34` - [boost-io.json](https://github.com/OpenArkStudio/arknet/blob/main/benchmarks/results/macmini-m4pro-20261012-udp/boost-io.json): 90 UDP measurements. - [macmini-m4pro-20261010-boost-websocket.json](https://github.com/OpenArkStudio/arknet/blob/main/benchmarks/results/macmini-m4pro-20261010-boost-websocket.json): 90 measurements; 2026-10-10T05:44:16.816290+00:00; executable SHA-256 `7b2989fd23e15a68a662899d1906f3b094138018b8ada5904309b4a90d96ea34` - [macmini-m4pro-20261010-boost-wss.json](https://github.com/OpenArkStudio/arknet/blob/main/benchmarks/results/macmini-m4pro-20261010-boost-wss.json): 90 measurements; 2026-10-10T06:02:39.046046+00:00; executable SHA-256 `7b2989fd23e15a68a662899d1906f3b094138018b8ada5904309b4a90d96ea34` - [macmini-m4pro-20261010-standalone-coroutine.json](https://github.com/OpenArkStudio/arknet/blob/main/benchmarks/results/macmini-m4pro-20261010-standalone-coroutine.json): 180 measurements; 2026-10-10T06:18:15.349061+00:00; executable SHA-256 `f2878cefe4ec4259114a9492fa7ac2235deecf797be63735ae51668d7a25b73f` - [macmini-m4pro-20261010-standalone-http.json](https://github.com/OpenArkStudio/arknet/blob/main/benchmarks/results/macmini-m4pro-20261010-standalone-http.json): 90 measurements; 2026-10-10T06:06:57.385745+00:00; executable SHA-256 `480122707b4b3ed592d1f580f5cf7092b8008ce52331f55f0e666e89c568be6b` - [macmini-m4pro-20261010-standalone-https.json](https://github.com/OpenArkStudio/arknet/blob/main/benchmarks/results/macmini-m4pro-20261010-standalone-https.json): 90 measurements; 2026-10-10T06:14:08.452400+00:00; executable SHA-256 `480122707b4b3ed592d1f580f5cf7092b8008ce52331f55f0e666e89c568be6b` - [macmini-m4pro-20261010-standalone-tcp.json](https://github.com/OpenArkStudio/arknet/blob/main/benchmarks/results/macmini-m4pro-20261010-standalone-tcp.json): 90 measurements; 2026-10-10T05:29:56.838785+00:00; executable SHA-256 `480122707b4b3ed592d1f580f5cf7092b8008ce52331f55f0e666e89c568be6b` - [macmini-m4pro-20261010-standalone-tcps.json](https://github.com/OpenArkStudio/arknet/blob/main/benchmarks/results/macmini-m4pro-20261010-standalone-tcps.json): 90 measurements; 2026-10-10T05:54:16.794729+00:00; executable SHA-256 `480122707b4b3ed592d1f580f5cf7092b8008ce52331f55f0e666e89c568be6b` - [standalone-io.json](https://github.com/OpenArkStudio/arknet/blob/main/benchmarks/results/macmini-m4pro-20261012-udp/standalone-io.json): 90 UDP measurements. - [macmini-m4pro-20261010-standalone-websocket.json](https://github.com/OpenArkStudio/arknet/blob/main/benchmarks/results/macmini-m4pro-20261010-standalone-websocket.json): 90 measurements; 2026-10-10T05:42:15.113877+00:00; executable SHA-256 `480122707b4b3ed592d1f580f5cf7092b8008ce52331f55f0e666e89c568be6b` - [macmini-m4pro-20261010-standalone-wss.json](https://github.com/OpenArkStudio/arknet/blob/main/benchmarks/results/macmini-m4pro-20261010-standalone-wss.json): 90 measurements; 2026-10-10T06:00:34.943293+00:00; executable SHA-256 `480122707b4b3ed592d1f580f5cf7092b8008ce52331f55f0e666e89c568be6b` ## Scope Local loopback results do not establish cross-host or production capacity. Short runs, scheduling and thermals introduce variance; repeat with the deployment workload. Source: https://openarkstudio.github.io/arknet/performance/overview/index.mdx