sharp/docs/api-utility.md

Table of Contents

• cache
• concurrency
• counters
• simd

cache

Gets, or when options are provided sets, the limits of libvips' operation cache. Existing entries in the cache will be trimmed after any change in limits. This method always returns cache statistics, useful for determining how much working memory is required for a particular task.

Parameters

• options (Object | Boolean) Object with the following attributes, or Boolean where true uses default cache settings and false removes all caching.
  • options.memory Number? is the maximum memory in MB to use for this cache (optional, default 50)
  • options.files Number? is the maximum number of files to hold open (optional, default 20)
  • options.items Number? is the maximum number of operations to cache (optional, default 100)

Examples

const stats = sharp.cache();

sharp.cache( { items: 200 } );
sharp.cache( { files: 0 } );
sharp.cache(false);

Returns Object

concurrency

Gets, or when a concurrency is provided sets, the number of threads libvips' should create to process each image. The default value is the number of CPU cores. A value of 0 will reset to this default.

The maximum number of images that can be processed in parallel is limited by libuv's UV_THREADPOOL_SIZE environment variable.

This method always returns the current concurrency.

Parameters

• concurrency Number?

Examples

const threads = sharp.concurrency(); // 4
sharp.concurrency(2); // 2
sharp.concurrency(0); // 4

Returns Number concurrency

counters

Provides access to internal task counters.

• queue is the number of tasks this module has queued waiting for libuv to provide a worker thread from its pool.
• process is the number of resize tasks currently being processed.

Examples

const counters = sharp.counters(); // { queue: 2, process: 4 }

Returns Object

simd

Get and set use of SIMD vector unit instructions. Requires libvips to have been compiled with liborc support.

Improves the performance of resize, blur and sharpen operations by taking advantage of the SIMD vector unit of the CPU, e.g. Intel SSE and ARM NEON.

This feature is currently off by default but future versions may reverse this. Versions of liborc prior to 0.4.25 are known to segfault under heavy load.

Parameters

• simd Boolean? (optional, default false)

Examples

const simd = sharp.simd();
// simd is `true` if SIMD is currently enabled

const simd = sharp.simd(true);
// attempts to enable the use of SIMD, returning true if available

Returns Boolean