Making SIMD Code Scale to Any Vector Width
Comments on “Making SIMD Code Scale to Any Vector Width”
Claim-checked. Scored by all three judges, and its checkable claims were compared with the sources it cites. How far stories are checked
Stream HPC shows how to make SIMD code scale with vector width by storing vector components as lanes, one object per lane, instead of one vec3/vec4 per object. In a frustum-culling example on a Ryzen 7 5800 with Clang, cost fell from 13.4 ns to 0.67 ns per box on AVX2 and 1.84 ns on SSE. Gains came from splatting planes once outside the loop and writing results a block at a time.
Why it ranked
No. 14 of 110 with a score of 0.63. The panel rated it highest on technical depth (0.95) and new information (0.93), lowest on audience breadth (0.41). Placed No. 14, No. 30 by score alone: the placement terms moved other stories past it.
- AI judgeJudge 2claude-haiku-5-5
- Detailed, benchmarked SIMD engineering write-up with reproducible code and clear lessons for graphics and systems programmers; narrow audience, not a landmark, and contains no stories from the recent list.
- AI judgeJudge 3gpt-5.6-luna
- A highly reproducible SIMD engineering write-up with code, assembly, benchmarks, and concrete optimization lessons. It is technically deep and actionable, though mainly relevant to performance-oriented systems and graphics programmers.
Each judge wrote its line after answering the rubric, with the site and the author still hidden. Every answer behind this is in the table under Score.
Score
0.63total under rubric v2.3
- Significance0.47
- Reader interest0.72
- Audience breadth0.41
- Practical relevance0.74
- Substantiated0.88
- Technical depth0.95
- New information0.93
- Counts against: Promotional0.03
- Counts against: Overclaim0.16
Each judge’s answer per question (0–1), their mean and spread, and what the question adds to the score (weight × mean).
| Question | Judge 1 | Judge 2 | Judge 3 | Mean | Spread | Weight | Contribution (weight times mean) |
|---|---|---|---|---|---|---|---|
| SignificanceHow much does this change what people in tech know or can do? | 0.50 | 0.30 | 0.62 | 0.47 | 0.32 | +0.24 | +0.114 |
| Reader interestWould a well-informed senior engineer or tech founder want to read this today, and likely pass it on? | 0.62 | 0.70 | 0.84 | 0.72 | 0.22 | +0.26 | +0.187 |
| Audience breadthHow many people who build technology does it concern? | 0.45 | 0.35 | 0.42 | 0.41 | 0.10 | +0.13 | +0.053 |
| Practical relevanceHow useful is it to someone who builds software or tech products? | 0.74 | 0.70 | 0.78 | 0.74 | 0.08 | +0.22 | +0.163 |
| SubstantiatedHow well is it backed? Consider whether it is the primary source, and whether its main claims come with data, code, a demo, a reproducible method or citations. | 0.99 | 0.75 | 0.90 | 0.88 | 0.24 | +0.05 | +0.044 |
| Technical depthHow technically substantive is it? | 1.00 | 0.90 | 0.95 | 0.95 | 0.10 | +0.05 | +0.048 |
| New informationDoes it contain information absent from the stories covered in the last 7 days (list provided)? | 0.90 | 0.90 | 1.00 | 0.93 | 0.10 | +0.05 | +0.047 |
| PromotionalIs it mainly marketing with little substance (a press release, ad or sponsored post with no real information)? A launch post that carries real news is not promotional. | 0.04 | 0.03 | 0.01 | 0.03 | 0.03 | −0.19 | −0.005 |
| OverclaimDo the headline or conclusions claim more than the body supports? | 0.21 | 0.20 | 0.08 | 0.16 | 0.13 | −0.11 | −0.018 |
| Sum of contributions | +0.632 | ||||||
| Claim-check penalty | 0.000 | ||||||
| Total, as published | 0.632 | ||||||
Spread is the highest minus the lowest judge answer. marks a spread above 0.40: the judges disagree, and the story is marked contested.
The badge for this ranking, for the author's page or anyone's: one line of HTML, and it links back here.
<a href="https://rubric.hrih.si/editions/2026-10-29/14/"><img src="https://rubric.hrih.si/badge/2026-10-29/14.svg" alt="Ranked #14 on Rubric, 29 Oct 2026" width="245" height="28"></a>Claim check
Claim check: 8 no source given. Each claim is compared with the source the article itself cites; a flag is published only when both checkers agree.
No source givenNumber
The naive frustum culler costs 13.4 ns per box on the author's Ryzen 7 5800 test setup.
The article gives no link or citation for it.
The article says “The cost is 13.4 ns per box. We can do better”
The source says No source to compare with.
No source givenNumber
The nearest-corner culling test costs 3.0 ns per box, about four times faster than the naive implementation.
The article gives no link or citation for it.
The article says “That’s 3.0 ns per box, so we’re already four times faster than the basic implementation”
The source says No source to compare with.
No source givenCausal conclusion
Zen 3's branch predictor can memorize the early outs of about 16,000 boxes, which made the nearest-corner test look 4.4× faster than it is on new data.
The article gives no link or citation for it.
The article says “Zen 3 can memorize the early outs of about 16,000 boxes”
The source says No source to compare with.
No source givenBenchmark result
The one-box-per-lane culler runs at 1.24 ns per box with AVX2 and 2.87 ns with SSE, 10× faster than the nearest-corner test on fresh data with AVX2.
The article gives no link or citation for it.
The article says “This runs at 1.24 ns per box with AVX2 and 2.87 ns with SSE. On AVX2 that’s 10× faster than the nearest-corner test on fresh data”
The source says No source to compare with.
No source givenCausal conclusion
The compiler rebuilds the frustum splats in every loop iteration because six planes need 42 splats and AVX2 has only 16 ymm registers.
The article gives no link or citation for it.
The article says “The compiler does this because six planes need 42 splats and AVX2 only has 16 ymm registers”
The source says No source to compare with.
No source givenBenchmark result
Splatting the planes once reduces the culler to 0.98 ns per box on AVX2 and 1.85 ns on SSE.
The article gives no link or citation for it.
The article says “That gets AVX2 down to 0.98 ns per box and SSE to 1.85 ns”
The source says No source to compare with.
No source givenBenchmark result
With splatting once and block writeback, AVX2 takes 0.67 ns per box and SSE 1.84 ns.
The article gives no link or citation for it.
The article says “With both changes in place, AVX2 takes 0.67 ns per box1 and SSE 1.84 ns”
The source says No source to compare with.
No source givenNumber
From start to finish the culler got about 20× faster, or 25× on data the branch predictor hadn't seen before.
The article gives no link or citation for it.
The article says “From start to finish, the culler got about 20× faster, or 25× on data the branch predictor hadn’t seen before”
The source says No source to compare with.
Also covered by
No other coverage was clustered with this story.
Elsewhere
Elsewhere: HN 2 pts, not on front page
Recorded for context only, checked 29 Oct 2026, 05:47 UTC. It played no part in the ranking.