Benchmark: DataView or BitTwiddling, Aligned and Unaligned

Script Preparation code:

let buf = new ArrayBuffer(12);
u32s = new Uint32Array(buf);
u8s = new Uint8Array(buf);
dv = new DataView(buf);
offsetAligned = 0;
offsetUnaligned = 1;
value = 0x01020304;

Tests:

Bit Twiddling, Aligned
u8s[offsetAligned + 0] = value >> 24; u8s[offsetAligned + 1] = value >> 16; u8s[offsetAligned + 2] = value >> 8; u8s[offsetAligned + 3] = value;
DataView, Aligned
dv.setInt32(offsetAligned, value, false)
BitTwiddling, Unaligned
u8s[offsetUnaligned + 0] = value >> 24; u8s[offsetUnaligned + 1] = value >> 16; u8s[offsetUnaligned + 2] = value >> 8; u8s[offsetUnaligned + 3] = value;
DataView, Unaligned
dv.setInt32(offsetUnaligned, value, false)

Rendered benchmark preparation results:

Suite status: <idle, ready to run>

Previous results

Test case name	Result
Bit Twiddling, Aligned
DataView, Aligned
BitTwiddling, Unaligned
DataView, Unaligned

Fastest: N/A

Slowest: N/A

Autogenerated LLM Summary (model llama3.2:3b, generated one month ago):

LLMs can make mistakes. Check important info.

I'll break down the benchmark and explain what's being tested.

**Benchmark Overview**

The test aims to compare the performance of two approaches: `DataView` (using the JavaScript `DataView` API) and bit twiddling (manually manipulating bits). The tests are designed to extract 32-bit integers from a byte array in different ways, with varying alignment requirements (`aligned` or `unaligned`). This allows for an apples-to-apples comparison of performance.

**Options Compared**

There are four test cases:

1. **DataView, Aligned**: Uses the `DataView` API to set 32 bits at a specified offset.
2. **Bit Twiddling, Aligned**: Manually performs bit twiddling operations (shifts and assignments) on the byte array, targeting the same aligned offset.
3. **DataView, Unaligned**: Similar to the above, but with an unaligned offset.
4. **Bit Twiddling, Unaligned**: Similar to the above, but with an unaligned offset.

**Pros and Cons**

* **DataView**:
	+ Pros: Simplifies bit manipulation, easier to maintain, and less prone to errors.
	+ Cons: May have additional overhead due to the API itself, and some browsers might not support it well (e.g., older versions).
* **Bit Twiddling**:
	+ Pros: Highly optimized, low-level control, and no external dependencies.
	+ Cons: Requires manual bit manipulation expertise, more prone to errors, and may be slower in modern browsers.

**Library (DataView)**

The `DataView` API provides a way to read and write binary data from ArrayBuffer or Int8Array views. It's designed for efficient and safe access to native buffer data types, like Uint32Array or Int32Array. In this benchmark, it's used to set 32-bit integers at specific offsets.

**Special JS Feature (No)**

There are no special JavaScript features being tested here. The focus is solely on the performance comparison between `DataView` and bit twiddling.

**Alternatives**

Other alternatives for efficient byte array manipulation in JavaScript include:

* Using `Buffer` and its methods (e.g., `writeUInt32LE`)
* Employing assembly language or native code to perform low-level operations
* Utilizing WebAssembly or other compiler-optimized languages

Keep in mind that these alternatives might have different performance characteristics, ease of use, and compatibility across browsers.

Latest run results:

Run details: (Test run date: one year ago)

User agent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/116.0.0.0 Safari/537.36

Browser/OS: Chrome 116 on Linux

View result in a separate tab

Test name	Executions per second
Bit Twiddling, Aligned	1213331.0 Ops/sec
DataView, Aligned	4665979.0 Ops/sec
BitTwiddling, Unaligned	1206346.1 Ops/sec
DataView, Unaligned	4622813.5 Ops/sec

Related benchmarks:

DataView or BitTwiddling, Aligned and Unaligned (version: 2)

Is DataView faster than directly bit twiddling?

Comparing performance of: Bit Twiddling, Aligned vs DataView, Aligned vs BitTwiddling, Unaligned vs DataView, Unaligned

Created: one year ago by: Registered User

Jump to the latest result

Bit Twiddling, Aligned

DataView, Aligned

BitTwiddling, Unaligned

DataView, Unaligned

Suite status: <idle, ready to run>

Experimental features:

Fastest: N/A

Slowest: N/A

Autogenerated LLM Summary (model llama3.2:3b, generated one month ago):