Sample
Dorn/cuda/57
Human rating uses the original dataset scale.
Scores
| Method | Score | Scale | Dataset rank (0–100) |
|---|---|---|---|
| Posnett | 0.32 | 0–1; higher is more readable | 78.3Δ 17.8 |
| Scalabrino | 0.595 | 0–1; higher is more readable | 58.9Δ 1.5 |
| Dorn | 0.653 | 0–1; higher is more readable | 59.1Δ 1.4 |
| Mi | 0.529 | 0–1; higher is more readable | 56.8Δ 3.6 |
| LLOC | 8 | Logical lines; fewer is more readable | 63.9Δ 3.5 |
| DeepSeek V4 Pro | 18 | 0–20; mean of three runs | 98.3Δ 37.9 |
| GPT-6.1 Sol | 19 | 0–20; mean of three runs | 98.9Δ 38.4 |
| BTTF | 0.571 | 0–1; higher is more readable | 66.3Δ 5.8 |
| Human | 3.46 | 1–5; original rating | 60.4 |
Views
Select the downloaded ZIP or its extracted folder. Files stay in your browser.
Loading feature diagnosis…
Source regions
Feature contributions
Feature contributions
| Feature | Value | Contribution (log-odds) |
|---|---|---|
| Byte entropyHow varied the bytes in the source text are. Repeated bytes give a lower value; a more even mixture gives a higher value. | 4.757 | −7.135 |
| Halstead volumeHow much symbolic information the code contains. It combines the total number of operators and values with the number of distinct ones. | 184.477 | −6.088 |
| Source linesThe number of non-empty source lines. | 9 | +3.6 |
Intercept 8.87 + feature contributions = -0.753 (log-odds).
Score = sigmoid(-0.753) = 0.32.
| Score | 0.595 |
|---|
Feature contributions
| Feature | Value | Contribution (log-odds) |
|---|---|---|
| Identifier layout (horizontal)How rapidly the visual pattern of identifier characters changes from left to right. It measures spatial frequency, not the meaning of the names. | 60 | +1.124 |
| Comment layout (horizontal)How rapidly the visual pattern of comment characters changes from left to right. | 0 | −0.804 |
| Comment layout frequencyHow frequently the layout switches between lines containing only comments or whitespace and lines containing code. | 10 | +0.682 |
| Indentation frequencyHow rapidly indentation changes from line to line. | 9 | −0.426 |
| Number layout frequencyHow rapidly the number of numeric values per line changes through the code. | 1 | +0.398 |
| Identifier layout (vertical)How rapidly the visual pattern of identifier characters changes from top to bottom. | 8.951 | −0.285 |
| Operator visual areaThe share of the code’s visual area occupied by operators such as +, =, and *. | 0.11 | −0.128 |
Intercept 0.07 + feature contributions = 0.631 (log-odds).
Score = sigmoid(0.631) = 0.653.
| Score | 0.529 |
|---|
| Score | 8 |
|---|
Run 1
The code is highly readable. The function is concise, with a clear logical structure: prepare the volume, prepare the ray caster, generate the image, and clean up. Each action is on its own line, enhancing action granularity. The flow is natural: early returns for failure cases keep the main logic flat and avoid deep nesting. The function and method names clearly convey intent, though the prefix 'CUDAIsosurfaceVRScene::' adds some verbosity. Overall, the code is straightforward, well-organized, and easy to follow.
Run 2
The code is highly readable due to its clear, flat structure and concise length. Each function call represents a single, well-defined action, making the intent easy to follow: prepare data, prepare the renderer, generate an image, and clean up. There is minimal nesting and no complex control flow beyond a simple success check, which promotes clarity. The natural reading flow follows the logical sequence of operations. While the function names are self-explanatory, the code could be slightly improved by adding brief comments to explain the purpose of each step, but this is not necessary for understanding.
Run 3
The code is well-structured and highly readable. It follows a clear sequential flow: prepare data, prepare ray caster, generate the image, then clean up. The function has a single responsibility and each line represents one action, aiding readability. It uses early returns for failure, avoiding deep nesting. Names are descriptive, conveying intent clearly. The function is concise without being overly terse. Minor deductions could be argued because the failure paths are implicit (simply returning false without logging or error context), but that is a design choice rather than a readability issue. Overall, the code is nearly ideal for understanding its purpose at a glance.
Run 1
Code Structure: A clear sequence of preparation, rendering, cleanup, and success reporting. Nesting: Flat control flow with early returns avoids nested conditionals. Clarity of intent: Descriptive function names make the rendering workflow easy to understand, though cleanup requirements on failure are not evident. Code length: Short and focused, with no unnecessary statements. Action granularity: Each line performs one main operation; the inline checks and returns remain simple to read. Reading flow: The operations follow a natural order, and the final return clearly signals successful completion.
Run 2
Code Structure: The function is logically organized into preparation, image generation, cleanup, and success reporting. Nesting: Guard clauses keep the code flat and easy to follow. Clarity of intent: Descriptive function names make the rendering workflow clear, although cleanup behavior after the second preparation fails is not evident. Code length: Short and concise, with no unnecessary statements. Action granularity: Each line performs one primary operation; the compact conditional returns remain readable. Reading flow: Operations follow a natural sequence, and early returns clearly communicate failure paths.
Run 3
Code Structure: A logically organized sequence of preparation, rendering, and cleanup. Nesting: Flat control flow with early returns and no nested blocks. Clarity of intent: Descriptive function names make the rendering pipeline clear, though cleanup behavior after preparation failure is not explicit. Code length: Concise, with no unnecessary statements. Action granularity: Each line performs a single main operation; the preparation checks combine a call and conditional return but remain easy to read. Reading flow: Natural top-to-bottom progression, ending with an explicit success return.