{ "as_of": "2026-11-14", "scope": "Bounded read-only review of existing xOneRS source, docs, audit and issue evidence for website performance content. No benchmark executed; no new issues; no Site files changed.", "source_repo": "XDCIndia/xOneRS", "source_commit": "5af188dbeb6199a859e9823720a9c8df689fcb87", "measurement_attribution": "Repository-reported results, not independent reproduction.", "headline_recommendation": "Use the merged replay CPU self-comparison as the featured result, with workload and host context. Keep RPC/memory observations in details and root parallelism under In review.", "facts": [ { "id": "PERF-01", "area": "execution", "kind": "reported_measured_self_comparison", "website_use": "featured_with_context", "headline": "52% less CPU in a 20,000-block replay", "claim": "The merged libsecp256k1 recovery change reports roughly 873 to 417 CPU seconds for sequential net2040 replay from genesis through block 20,000.", "measurements": { "metric": "process user + system CPU seconds", "before_interleaved_runs": [ 874.7, 873.0 ], "after_interleaved_runs": [ 416.0, 417.3 ], "reported_reduction_percent": 52, "blocks": 20000, "mismatches_per_reported_run": 0, "archive_hash_checks_per_run": 200, "v2_epoch_switches_per_run": 11 }, "context": { "date": "2026-11-14", "cpu": "Intel i5-13500", "logical_cpus": 20, "host": "shared build host", "load1_mean_per_interleaved_run": [ 62, 50, 76, 53 ], "workload": "sequential replay from genesis, release, nice -n 19, checkpoint-every 0", "before_reported_commit": "b055158", "after_commit": "5839f946e600e5af141d9f43b30e4b0a91b41c8d" }, "sources": [ { "path": "docs/perf/ethrex-storage-comparison.md", "url": "https://github.com/XDCIndia/xOneRS/blob/5af188dbeb6199a859e9823720a9c8df689fcb87/docs/perf/ethrex-storage-comparison.md", "commit": "5af188dbeb6199a859e9823720a9c8df689fcb87", "section": "5. Implemented" }, { "path": "crates/xone-primitives/src/seal.rs", "url": "https://github.com/XDCIndia/xOneRS/blob/5af188dbeb6199a859e9823720a9c8df689fcb87/crates/xone-primitives/src/seal.rs", "commit": "5af188dbeb6199a859e9823720a9c8df689fcb87" } ], "links": [ "https://github.com/XDCIndia/xOneRS/issues/243", "https://github.com/XDCIndia/xOneRS/pull/319" ], "implementation_status": "PR #319 merged; source at the inspected pin imports secp256k1 and uses a verification context.", "limits": [ "Repository-reported measurements; not independently reproduced in this pass.", "Self-comparison against an earlier xOneRS build, not a competitor ranking.", "CPU time is not wall-clock latency, network finality or node follower throughput.", "PR explicitly says node follower throughput was not measured.", "Shared-host load varied; do not turn this into a generic 2x faster claim." ], "suggested_copy": "A merged signer-recovery improvement cut reported CPU use by 52% in a 20,000-block net2040 replay. Shared-host test; zero mismatches reported in each run." }, { "id": "PERF-02", "area": "execution", "kind": "reported_experiment_open_pr", "website_use": "roadmap_only", "headline": "Parallel root computation is under review", "claim": "PR #334 computes transaction and receipt roots on scoped threads beside the state root for blocks with at least 64 transactions.", "measurements": { "metric": "root-phase seconds on blocks with at least 64 transactions", "parallel_runs_0_to_14000": [ { "before": 101.1, "after": 79.6, "reported_reduction_percent": 21, "load1_mean": 107 }, { "before": 120.5, "after": 100.5, "reported_reduction_percent": 17, "load1_mean": 119 } ], "sequential_pair_0_to_20000": { "before": 91.7, "after": 89.0, "reported_reduction_percent": 3, "load1_mean": [ 109, 108 ] }, "mismatches_per_reported_run": 0 }, "sources": [ { "path": "docs/perf/ethrex-storage-comparison.md", "url": "https://github.com/XDCIndia/xOneRS/blob/5af188dbeb6199a859e9823720a9c8df689fcb87/docs/perf/ethrex-storage-comparison.md", "commit": "5af188dbeb6199a859e9823720a9c8df689fcb87", "section": "5. Implemented" }, { "path": "crates/xone-executor/src/lib.rs", "url": "https://github.com/XDCIndia/xOneRS/blob/5af188dbeb6199a859e9823720a9c8df689fcb87/crates/xone-executor/src/lib.rs", "commit": "5af188dbeb6199a859e9823720a9c8df689fcb87" } ], "links": [ "https://github.com/XDCIndia/xOneRS/issues/244", "https://github.com/XDCIndia/xOneRS/pull/334" ], "implementation_status": "GitHub check: PR #334 open, not merged; inspected main source still computes roots sequentially. PR head 35169fb475e828f0459706cfdbb326389d66dce7.", "limits": [ "The documentation calls this implemented work, but the PR remains open; label it an experiment.", "These numbers concern one phase, not end-to-end execution.", "The sequential-pair difference was 3% and described as within noise.", "CPU work did not fall; benefits depend on free cores.", "Do not headline a general 21% speed improvement." ], "suggested_copy": "In review: parallel root computation for large blocks, with replay equivalence checks and performance evidence attached to the PR." }, { "id": "PERF-03", "area": "rpc", "kind": "historical_narrow_observation", "website_use": "detail_only", "headline": "RPC timings need method-level context", "claim": "A historical live-node sample of 300 eth_blockNumber calls reported p50 0.5 ms, p99 8 ms and maximum 64 ms.", "measurements": { "method": "eth_blockNumber", "calls": 300, "p50_ms": 0.5, "p99_ms": 8, "max_ms": 64 }, "context": { "date_context": "2026-11-12 lag investigation", "network": "net2040", "build_context": "46e8769 live-node investigation", "load_model": "not a documented saturation or open-loop benchmark" }, "sources": [ { "path": "docs/ops/net2040-follow.md", "url": "https://github.com/XDCIndia/xOneRS/blob/5af188dbeb6199a859e9823720a9c8df689fcb87/docs/ops/net2040-follow.md", "commit": "5af188dbeb6199a859e9823720a9c8df689fcb87", "section": "Lag / Causes" } ], "links": [ "https://github.com/XDCIndia/xOneRS/issues/197", "https://github.com/XDCIndia/xOneRS/issues/13" ], "limits": [ "A single cheap method and small sample do not establish broad RPC latency.", "No comparable sustained RPC benchmark found in this scoped pass.", "No cold/warm or concurrency control is supplied for this sample." ], "suggested_copy": "A 300-call eth_blockNumber sample recorded 0.5 ms p50 and 8 ms p99. Full RPC load testing is tracked separately." }, { "id": "PERF-04", "area": "sync", "kind": "historical_operational_run", "website_use": "detail_only", "headline": "Every sync result carries its workload", "claim": "A historical net2040 run reached block 766,981 from genesis in 108 minutes on a shared host while workspace tests also ran.", "measurements": { "start_height": 0, "end_height": 766981, "elapsed_minutes": 108, "host_load_average_range": [ 17, 27 ] }, "context": { "date": "2026-11-12", "reported_binary": "xOne/v0.1.0-af707ce7-dirty-12112026", "network": "net2040", "role": "follower", "note": "This measurement is in the net2040 proof subsection of the Apothem follow document." }, "sources": [ { "path": "docs/ops/apothem-follow.md", "url": "https://github.com/XDCIndia/xOneRS/blob/5af188dbeb6199a859e9823720a9c8df689fcb87/docs/ops/apothem-follow.md", "commit": "5af188dbeb6199a859e9823720a9c8df689fcb87", "section": "Proof on net2040" } ], "links": [ "https://github.com/XDCIndia/xOneRS/issues/139", "https://github.com/XDCIndia/xOneRS/pull/141" ], "limits": [ "Historical run on a dirty development build; not a current release SLA.", "Not Apothem or Ethereum mainnet sync time.", "No same-host, same-retention peer-client comparison is established." ], "suggested_copy": "Published sync notes record the exact network, build, block range and host load, so progress can be checked against evidence." }, { "id": "PERF-05", "area": "memory", "kind": "historical_operational_observation", "website_use": "detail_only", "headline": "Track heap and mapped files separately", "claim": "In the same bounded-history net2040 run, RssAnon was 1.41 GB at blocks 200,794, 502,077 and 766,981; total RSS rose from 3.85 GB to 4.21 GB as mapped MDBX pages became resident.", "measurements": { "samples": [ { "height": 200794, "rss_gb": 3.85, "rss_anon_gb": 1.41, "rss_file_gb": 2.44 }, { "height": 502077, "rss_gb": 4.19, "rss_anon_gb": 1.41, "rss_file_gb": 2.78 }, { "height": 766981, "rss_gb": 4.21, "rss_anon_gb": 1.41, "rss_file_gb": 2.8 } ], "peak_rss_anon_gb": 1.5, "sample_interval_seconds": 10 }, "sources": [ { "path": "docs/ops/apothem-follow.md", "url": "https://github.com/XDCIndia/xOneRS/blob/5af188dbeb6199a859e9823720a9c8df689fcb87/docs/ops/apothem-follow.md", "commit": "5af188dbeb6199a859e9823720a9c8df689fcb87", "section": "Proof on net2040" }, { "path": "crates/xone-node/src/services/hist.rs", "url": "https://github.com/XDCIndia/xOneRS/blob/5af188dbeb6199a859e9823720a9c8df689fcb87/crates/xone-node/src/services/hist.rs", "commit": "5af188dbeb6199a859e9823720a9c8df689fcb87" } ], "links": [ "https://github.com/XDCIndia/xOneRS/issues/145", "https://github.com/XDCIndia/xOneRS/issues/200" ], "limits": [ "Anonymous memory is not total RSS or cgroup memory.", "Historic net2040 state and workload; not a full-chain mainnet resource guarantee.", "The observed cap does not prove a permanent resource ceiling." ], "suggested_copy": "Bounded history separates the working set from retained chain data. Published measurements distinguish anonymous memory from file-backed RSS." }, { "id": "PERF-06", "area": "memory", "kind": "synthetic_store_measurement", "website_use": "detail_only", "headline": "Disk-backed state has a measured synthetic baseline", "claim": "A real-store synthetic workload reported anonymous memory of 1,329 MB for memory state and 1,157 MB for disk state, a reported 13% reduction for the flat-row stage.", "measurements": { "accounts": 500000, "storage_slots": 2000000, "memory_state_anon_mb": 1329, "disk_state_anon_mb": 1157, "reported_reduction_percent": 13, "retention_blocks": 8 }, "context": { "generator": "xone-state-bench", "per_block_accounts": 5000, "per_block_slots": 20000, "state": "trie still in memory for this stage" }, "sources": [ { "path": "docs/ops/disk-state.md", "url": "https://github.com/XDCIndia/xOneRS/blob/5af188dbeb6199a859e9823720a9c8df689fcb87/docs/ops/disk-state.md", "commit": "5af188dbeb6199a859e9823720a9c8df689fcb87", "section": "Numbers (part 1)" } ], "links": [ "https://github.com/XDCIndia/xOneRS/issues/145" ], "limits": [ "Synthetic measurement of one implementation stage.", "Not total RSS, not all memory savings from disk state, and not a mainnet result.", "Startup timings in the same doc use different block heights and must not be turned into a speed ratio." ], "suggested_copy": "Disk-backed flat state reduced anonymous memory by 13% in one published synthetic store test; persisted tries and streamed snap installation address the remaining working set." }, { "id": "PERF-07", "area": "storage", "kind": "source_architecture_and_unmeasured_work", "website_use": "architecture_and_roadmap", "headline": "Keep hot state separate from retained history", "claim": "The Rust store combines MDBX state/index tables with append-only XSEG history, disk-backed flat-state caches, persisted path-keyed trie nodes and streamed snap installation. Whole-segment pruning primitives exist; configurable retention, compression and bounded cleanup remain tracked work.", "sources": [ { "path": "docs/ops/disk-state.md", "url": "https://github.com/XDCIndia/xOneRS/blob/5af188dbeb6199a859e9823720a9c8df689fcb87/docs/ops/disk-state.md", "commit": "5af188dbeb6199a859e9823720a9c8df689fcb87" }, { "path": "crates/xone-sync/src/mdbx/mod.rs", "url": "https://github.com/XDCIndia/xOneRS/blob/5af188dbeb6199a859e9823720a9c8df689fcb87/crates/xone-sync/src/mdbx/mod.rs", "commit": "5af188dbeb6199a859e9823720a9c8df689fcb87" }, { "path": "docs/adr/ADR-005-storage-and-publication.md", "url": "https://github.com/XDCIndia/xOneRS/blob/5af188dbeb6199a859e9823720a9c8df689fcb87/docs/adr/ADR-005-storage-and-publication.md", "commit": "5af188dbeb6199a859e9823720a9c8df689fcb87" } ], "links": [ "https://github.com/XDCIndia/xOneRS/issues/135", "https://github.com/XDCIndia/xOneRS/issues/177", "https://github.com/XDCIndia/xOneRS/issues/194", "https://github.com/XDCIndia/xOneRS/issues/326", "https://github.com/XDCIndia/xOneRS/issues/329" ], "limits": [ "No controlled before/after disk-pruning benchmark was found.", "Deleting records is not the same as shrinking an MDBX file.", "Full, archive and anchored/minimal sizes retain different data and cannot be ranked as equivalent.", "Do not publish a disk savings percentage or lower-resource promise from these sources." ], "suggested_copy": "Separate hot state from retained history. Track pruning, compressed segments and bounded cleanup through public implementation issues." }, { "id": "PERF-08", "area": "methodology", "kind": "benchmark_gap", "website_use": "required_context", "headline": "A controlled cross-client baseline remains to be published", "claim": "The checked-in performance contract and baseline procedure explicitly say their standardized baseline has not yet been run. The storage decision record says its cross-client spikes use different hosts, loads and runtimes and are not a ranking.", "sources": [ { "path": "docs/perf.md", "url": "https://github.com/XDCIndia/xOneRS/blob/5af188dbeb6199a859e9823720a9c8df689fcb87/docs/perf.md", "commit": "5af188dbeb6199a859e9823720a9c8df689fcb87" }, { "path": "bench/README.md", "url": "https://github.com/XDCIndia/xOneRS/blob/5af188dbeb6199a859e9823720a9c8df689fcb87/bench/README.md", "commit": "5af188dbeb6199a859e9823720a9c8df689fcb87" }, { "path": "docs/decisions/0135-storage-architecture.md", "url": "https://github.com/XDCIndia/xOneRS/blob/5af188dbeb6199a859e9823720a9c8df689fcb87/docs/decisions/0135-storage-architecture.md", "commit": "5af188dbeb6199a859e9823720a9c8df689fcb87", "section": "3 and 6" } ], "links": [ "https://github.com/XDCIndia/xOneRS/issues/13", "https://github.com/XDCIndia/xOneRS/issues/135", "https://github.com/XDCIndia/xOneRS/issues/197", "https://github.com/XDCIndia/xOneRS/issues/200" ], "limits": [ "Historical self-comparisons and operational observations exist; absence applies to a controlled comparable baseline.", "No supported claim that xOne is faster than Geth, Reth, Erigon, Nethermind or Besu.", "Do not use targets from docs/perf.md as achieved latency or throughput." ], "suggested_copy": "We publish workload-specific measurements and their limits. A controlled cross-client baseline is still pending." } ], "improvement_work": [ { "area": "execution", "title": "Remove repeated root computation", "issues": [ 245 ], "status": "tracked; not established as implemented in the inspected source", "benefit": "Reduce duplicate validation work while preserving every check.", "links": [ "https://github.com/XDCIndia/xOneRS/issues/245" ] }, { "area": "execution", "title": "Parallel storage roots and broader sender recovery", "issues": [ 246, 250 ], "status": "proposed", "benefit": "Explore additional parallelism; gains require measurement on the target workload.", "links": [ "https://github.com/XDCIndia/xOneRS/issues/246", "https://github.com/XDCIndia/xOneRS/issues/250" ] }, { "area": "execution", "title": "Pipeline execution with commit and cache analyzed bytecode", "issues": [ 247, 248, 249 ], "status": "proposed", "benefit": "Move work off the import critical path without changing publication safety.", "links": [ "https://github.com/XDCIndia/xOneRS/issues/247", "https://github.com/XDCIndia/xOneRS/issues/248", "https://github.com/XDCIndia/xOneRS/issues/249" ] }, { "area": "rpc", "title": "Measure and accelerate historical queries", "issues": [ 197, 258, 333 ], "status": "tracked", "benefit": "Open-loop method-level replay, bloom filtering and a bounded persistent log index.", "links": [ "https://github.com/XDCIndia/xOneRS/issues/197", "https://github.com/XDCIndia/xOneRS/issues/258", "https://github.com/XDCIndia/xOneRS/issues/333" ] }, { "area": "storage", "title": "Pruning, compression and bounded cleanup", "issues": [ 177, 194, 326, 329 ], "status": "tracked", "benefit": "Make retained data and cleanup costs explicit; no savings percentage measured here.", "links": [ "https://github.com/XDCIndia/xOneRS/issues/177", "https://github.com/XDCIndia/xOneRS/issues/194", "https://github.com/XDCIndia/xOneRS/issues/326", "https://github.com/XDCIndia/xOneRS/issues/329" ] }, { "area": "memory", "title": "Disk-backed state and resource regression budgets", "issues": [ 145, 200 ], "status": "partial source implementation plus open qualification work", "benefit": "Bound the working set and verify RSS, CPU and disk limits by node role.", "links": [ "https://github.com/XDCIndia/xOneRS/issues/145", "https://github.com/XDCIndia/xOneRS/issues/200" ] }, { "area": "observability", "title": "Per-phase slow-block timing", "issues": [ 226 ], "status": "tracked", "benefit": "Replace temporary profiling timers with repeatable phase evidence.", "links": [ "https://github.com/XDCIndia/xOneRS/issues/226" ] } ], "competitor_comparison": { "controlled_like_for_like_results_found": false, "scope": "Audited xOne files only; no broad new competitor search performed.", "notes": [ "The docs mention geth-XDC dense-range observations, but these do not establish controlled same-host, same-mode comparability.", "Do not transpose ethrex, Reth or other upstream benchmark numbers into xOne claims.", "Existing competitor retention/operations references can inform design; no competitor performance numbers were needed for this website update." ] }, "website_guardrails": [ "Label measurements, experiments, targets and proposed work separately.", "A 52% CPU reduction is not 52% lower transaction latency.", "Do not imply network finality, TPS, mainnet sync time or production readiness from a replay.", "Keep RPC method/sample size and memory metric definitions visible.", "Archive/minimal/full nodes keep different data; avoid a common disk-size ranking.", "Use source pins and PR links; recheck status before changing an experimental item to shipped." ], "github_status_checked": [ { "number": 319, "url": "https://github.com/XDCIndia/xOneRS/pull/319", "title": "xone-primitives: recover signers with libsecp256k1", "state": "closed", "merged": true, "merged_at": "2026-11-14T07:15:52Z", "head_sha": "5839f946e600e5af141d9f43b30e4b0a91b41c8d", "merge_commit_sha": "dccecf890984fee2b51957a4ca78131b94814218", "updated_at": "2026-11-14T08:10:15Z" }, { "number": 334, "url": "https://github.com/XDCIndia/xOneRS/pull/334", "title": "xone-executor: compute the tx, receipts and state roots of a large block at the same time", "state": "open", "merged": false, "merged_at": null, "head_sha": "35169fb475e828f0459706cfdbb326389d66dce7", "merge_commit_sha": "21a244226f52083b646e328d56a84dfbd82ee690", "updated_at": "2026-11-14T08:13:07Z" }, { "number": 335, "url": "https://github.com/XDCIndia/xOneRS/pull/335", "title": "docs: ethrex and xOneRS storage, execution pipeline and speed, with ranked improvements", "state": "closed", "merged": true, "merged_at": "2026-11-14T08:10:43Z", "head_sha": "28002cb2b112b641b01291cd5915312b6fd96c60", "merge_commit_sha": "e57a4ec13cafb7c1e8e10afa07a15e10a87941c5", "updated_at": "2026-11-14T08:10:43Z" }, { "number": 243, "url": "https://github.com/XDCIndia/xOneRS/issues/243", "title": "perf: ecrecover on libsecp256k1 (recovery is 71% of replay time on k256)", "state": "closed", "merged": null, "merged_at": null, "head_sha": null, "merge_commit_sha": null, "updated_at": "2026-11-14T07:15:53Z" }, { "number": 244, "url": "https://github.com/XDCIndia/xOneRS/issues/244", "title": "perf: compute the tx root, the receipts and the state root of a large block at the same time", "state": "open", "merged": null, "merged_at": null, "head_sha": null, "merge_commit_sha": null, "updated_at": "2026-11-14T05:43:39Z" }, { "number": 145, "url": "https://github.com/XDCIndia/xOneRS/issues/145", "title": "state: disk-backed state (flat, trie, streamed snap install) for bounded memory on Apothem", "state": "open", "merged": null, "merged_at": null, "head_sha": null, "merge_commit_sha": null, "updated_at": "2026-11-14T02:36:10Z" }, { "number": 197, "url": "https://github.com/XDCIndia/xOneRS/issues/197", "title": "[GAP-0042] RPC traffic recorder and open-loop replay benchmark (record real requests, replay them at a set rate, report latency per method)", "state": "open", "merged": null, "merged_at": null, "head_sha": null, "merge_commit_sha": null, "updated_at": "2026-11-14T04:36:38Z" } ] }