# Caching Performance > Explore the built-in LRU TokenCache in @humanspeak/svelte-markdown, measuring cold and warm render times plus repeat-render speedups in the demo. **Source:** [https://markdown.svelte.page/examples/caching-performance](https://markdown.svelte.page/examples/caching-performance) **Markdown mirror:** [https://markdown.svelte.page/examples/caching-performance.md](https://markdown.svelte.page/examples/caching-performance.md) --- This mirror preserves the prose, implementation notes, and runnable Svelte source behind the live example page. ## FIG-001: token cache benchmark. Render the same markdown once cold, once warm. The render log records every cycle; the stats card surfaces average time and cache size. Clear the cache to start the comparison over. **Metadata:** tag: `BENCHMARK` | cache: `lru 50d / 5m` ### Notes - First render parses markdown into tokens and stores them in an LRU cache. - Subsequent renders of the same content skip parsing and use cached tokens (50–200× faster). - LRU eviction and TTL expiration manage memory automatically. ### Source #### CachingBenchmark.svelte Source file: [src/lib/examples/caching-performance/demos/CachingBenchmark.svelte](https://github.com/humanspeak/svelte-markdown/blob/main/docs/src/lib/examples/caching-performance/demos/CachingBenchmark.svelte) ```svelte

Controls

Statistics

Render count: {renderCount}
Average time: {averageTime}ms
Cache size: {cache.size()}

Render Log

{#if renderTimes.length === 0}

Click Render to start tracking performance.

{:else}

#{entry.index} {entry.cached ? 'cached' : 'cold'} {entry.duration}ms

{/if}

Markdown Input

{input.length} characters

Rendered Output

{#if renderCount > 0} render #{renderCount} {/if}

{#if renderCount === 0}

Click Render to see the markdown output.

{:else}

{#key renderKey} {/key}

{/if}

```