Tech News
← Home  ·  All topics

Cache

22 GoKawiil briefs on this topic

RTX 4090 teardown traces how STG.E writes results back to memory

A technical deep-dive follows the STG.E instruction—the global memory store operation in a simple vector-add CUDA kernel—through an RTX 4090's pipeline, from register read through the load/store unit, coalescer, L1 cache, crossbar, and L2 slice, down to DRAM. The author benchmarks how frequently a single warp can issue store instructions, finding a new STG.E can be pushed through the register file, LSU, and coalescer roughly every 6.1 cycles (about 2.3 nanoseconds at 2.6 GHz).

Cloudflare tests Zstandard compression in Pingora to shrink cache storage

Cloudflare built a prototype called Cache Transcoding, developed during an intern project, that compresses eligible cached assets with Zstandard before writing them to disk inside its Pingora proxy. Assets stay compressed while stored and while moving between data centers via Tiered Cache, then get decoded only when served to the client. Early tests show this shrinks eligible assets to about a third of their original size with only a small CPU overhead.

uv Python tool adds deduplication for wheel cache files

A pull request submitted to the uv project introduces a mechanism to deduplicate files stored in its wheel cache, reducing redundant copies of package files. The change was merged into the open-source Python packaging tool's codebase after review.

Smart TVs face storage limits as pre-installed apps and bloatware pile up

Most smart TVs ship with limited internal storage that quickly fills up due to app caches, temporary files, and manufacturer-installed bloatware that can be difficult or impossible to remove. Users can clear app cache through settings for temporary relief or uninstall unused apps, though bloatware removal is often restricted by manufacturers. An alternative workaround is attaching an external streaming device like a Fire TV Stick, Roku, or Google TV Streamer, which brings its own storage and app ecosystem independent of the TV's built-in memory.

Report explains why deleting app and browser caches often backfires

The report clarifies that caches are temporary storage used by browsers and apps to avoid re-downloading or recreating data like images, playlists and thumbnails. It notes there is no single universal cache on a device, since browsers, individual apps and the system each maintain separate caches for different purposes.

Cloudflare halves DNS cache memory, freeing 100TB across 1.1.1.1 network

Cloudflare reworked the Rust codebase powering its Big Pineapple DNS caching system, shrinking the average cache entry from 953 bytes to 420 bytes without adding any hardware. The changes—including replacing growable Vec and String types with fixed-size boxed slices and consolidating record storage into a single buffer—cut per-instance memory use from 9.3GB to 5.3GB while also boosting insert speed and lowering lookup latency.

Cloudflare shrinks 1.1.1.1 DNS cache entries to reclaim 100TB of RAM

Cloudflare redesigned how DNS cache entries are stored in memory for its 1.1.1.1 resolver platform, Big Pineapple, cutting each entry's size from 953 bytes to 420 bytes. The changes, detailed by systems engineer Sebastiaan Neuteboom, involved five Rust-level optimizations including fixed-size data structures, merged record lists, and raw wire-format storage instead of padded fields.

Artificial Analysis and Liquid AI launch benchmark for phone-based small AI models

Artificial Analysis has begun benchmarking small AI models that run directly on mobile phones, focusing on models that fit within 8 GB of memory after quantization, including KV cache at 8K context. The effort combines intelligence benchmarks with real on-device inference data gathered in partnership with Liquid AI, whose measurement methodology Artificial Analysis says it has independently validated.

Cloudflare cuts DNS cache memory use by 50%, freeing 100TB across 1.1.1.1 fleet

Cloudflare engineers made five successive changes to how DNS cache entries are stored in memory for Big Pineapple, the platform underlying 1.1.1.1 and related DNS services that hold over 250 billion cache entries at once. The changes shrank the per-entry memory footprint by more than half, freeing roughly 100 terabytes across the company's servers while also boosting insert throughput by 43% and cutting lookup latency by 19%.

Android troubleshooting guide targets recurring app crashes

A troubleshooting guide outlines steps Android users can take when an app repeatedly crashes, starting with checking the Google Play Store for pending updates via the 'Manage apps and device' menu. If no updates are available, users are advised to visit the app's info page to clear its cache and force stop it before relaunching.