Results for “throughput”

Diagnoses and improves measured latency, throughput, memory, CPU, bundle, rendering, query, or resource performance by defining a target, profiling a representative workload, locating the bottleneck, making one change at a time, and comparing before/after evidence. Use for performance work or regressions. Not for guessing optimizations without a reproducible workload or for generic refactoring.

The English summary is being prepared. Showing the original description.

Go concurrency patterns for high-throughput web applications including worker pools, rate limiting, race detection, and safe shared state management. Use when implementing background task processing, rate limiters, or concurrent request handling.

The English summary is being prepared. Showing the original description.

Reviews Elixir code for performance issues including GenServer bottlenecks, memory usage, and concurrency patterns. Use when reviewing high-throughput code or investigating performance issues.

The English summary is being prepared. Showing the original description.

Design a performance benchmark for an API — test scenarios, metrics, and tooling. Use when asked to "benchmark this API", "design a load test", or "measure throughput".

The English summary is being prepared. Showing the original description.

Measure encoding/decoding throughput across Syrup implementations. This skill provides bandwidth benchmarks for comparing serialization performance.

The English summary is being prepared. Showing the original description.

Expert task distributor specializing in intelligent work allocation, load balancing, and queue management. Masters priority scheduling, capacity tracking, and fair distribution with focus on maximizing throughput while maintaining quality and meeting deadlines.

The English summary is being prepared. Showing the original description.

Use when benchmarking, profiling, or A/B-testing CUDA kernels or end-to-end perf in the imp inference engine on RTX 5090 (sm_120), including refreshing tests/perf_baseline.json or publishing numbers to docs/BENCHMARKS.md and the README. Triggers on "benchmark kernel", "profile cuda", "ncu", "nsys", "kernel timing", "kernel sum", "occupancy", "bandwidth bound", "compute bound", "roofline", "perf baseline", "is this regression real", "decode dropped", "aggregate throughput", "two-image A/B", "prefill kernel A/B". Do NOT use for writing/optimizing kernel code (sm120-cuda-expert) or output-quality checks (check-degeneration).

The English summary is being prepared. Showing the original description.

Evaluates midstream assets with throughput analysis, fee-based vs commodity-exposed revenue, and contract structure assessment. Use when analyzing midstream investments, evaluating pipeline assets, or assessing gathering systems.

The English summary is being prepared. Showing the original description.

Generalised autonomous optimisation loop — soft RLVR for any artifact a user can measure. Use this skill whenever a user wants to iteratively improve an artifact — code, prompts, documents, configs, designs, content — by running structured experiments, evaluating results against a multi-dimensional rubric, and learning from each attempt. Triggers include: "optimise this", "keep improving until it's good", "run experiments on", "autoresearch", "iterate on this overnight", "try different approaches and pick the best", or any request implying repeated evaluate-and-improve cycles. Also use when the user wants to improve a system prompt, a data pipeline, a writing style, or any artifact where quality can be decomposed into measurable tracks. For inference optimisation tasks (model latency, throughput, quantization, GPU deployment), a* delegates the low-level tuning to AITune while maintaining quality tracking and learning.

The English summary is being prepared. Showing the original description.

Interactively gather parameters for `python -m cli.inference.text_generate`, generate a concrete text generation simulation command, explain assumptions, ask for execution confirmation, then run the command and summarize performance metrics. Use when the user wants to validate a specific model, hardware profile, batch/query size, prefill or decode scenario, TP/DP/EP/MOE parallel strategy, profiling database, chrome trace, empirical metrics export, or a best row produced by throughput_optimizer.

The English summary is being prepared. Showing the original description.

Optimize AI system latency and throughput. TRIGGERS - Use when user needs help with ai-latency-optimizer related tasks.

The English summary is being prepared. Showing the original description.

Optimize multi-agent systems with coordinated profiling, workload distribution, and cost-aware orchestration. Use when improving agent performance, throughput, or reliability.

The English summary is being prepared. Showing the original description.

"Apply Theory of Constraints (TOC) to identify and manage system bottlenecks. Use this skill when the user needs to find what limits throughput, optimize a constrained process, apply the Five Focusing Steps, or implement Drum-Buffer-Rope scheduling — even if they say 'our output is stuck', 'what's the bottleneck', or 'why can't we produce more'.".

The English summary is being prepared. Showing the original description.

High-throughput inference on FriendliAI — optimized serving engine for low-latency, cost-efficient LLMs.

The English summary is being prepared. Showing the original description.

Use when a task needs the judgment of a Dining Room Attendant or Bartender Helper — triaging which table to bus first during a rush, sequencing a table reset, setting bar-back par levels and restock timing, writing or triaging a side-work checklist, or diagnosing why table turn time or bar throughput has slipped.

The English summary is being prepared. Showing the original description.

Use when a task needs the judgment of a Coating/Painting/Spraying Machine Operator — calculating wet film thickness from a target dry film thickness and the coating's percent volume solids, deciding whether flash-off time can be shortened to increase throughput, verifying cure by actual part temperature rather than oven setpoint, or tracing an adhesion failure back to surface prep rather than the coating application.

The English summary is being prepared. Showing the original description.

Use when a task needs the judgment of an Amusement and Recreation Attendant — running ride dispatch and restraint checks, calling a weather or mechanical hold, managing queue throughput and wait-time complaints, operating a game booth or coin-operated attraction, or handling an on-ride evacuation.

The English summary is being prepared. Showing the original description.

Raise real concurrency in asyncio LLM batch scorers built on the OpenAI SDK (AsyncOpenAI, including OpenAI-compatible providers like DeepSeek). Use when: (1) raising an asyncio.Semaphore above ~100 produces no throughput gain, (2) a batch pipeline saturates near 100 in-flight requests despite a larger semaphore, (3) planning a high-concurrency campaign against a provider with no hard rate limit (DeepSeek v4-flash tolerates 2000+ in flight). Root cause: AsyncOpenAI's default httpx pool caps max_connections at 100, silently bottlenecking any larger semaphore — you must pass a custom http_client with httpx.Limits sized to the semaphore.

The English summary is being prepared. Showing the original description.

Use when working on the SAM3 plus Sapiens2 pose estimation pipeline, batch GPU processing, or VRAM and throughput issues.

The English summary is being prepared. Showing the original description.

Investigates Google Cloud networking issues by analyzing logs, metrics, and diagnostics. Use when investigating VPC Flow Logs (including cost estimation), NAT, firewall, or threat logs, querying latency and throughput metrics, or running Connectivity Tests for path diagnostics. Don't use for generic VM management or non-observability tasks.

The English summary is being prepared. Showing the original description.

Mounts Cloud Storage buckets as a POSIX file system with Cloud Storage FUSE (gcsfuse). Use when interacting with gcsfuse: decide whether FUSE, native gs:// reads, or Filestore/Managed Lustre fits a workload, deploy tuned mounts on GKE, Compute Engine, or Cloud Run, enable and size file, stat, and list caches, tune mount flags or config-file settings, apply workload profiles, keep ML checkpointing safe (rename atomicity, hierarchical namespace/HNS, close-time finalization, concurrent writers), or diagnose slow training, low throughput, or bill spikes with gcsfuse metrics. Covers mount semantics, gcsfuse CLI and config files, GKE gcsfuse CSI driver (Workload Identity principal:// bindings, profile StorageClasses, sidecar sizing), and Cloud Run volume mounts. Don't use for bucket administration or data management without a mount (google-cloud-storage-basics) or fully POSIX-compliant shared file systems (Filestore, Managed Lustre).

The English summary is being prepared. Showing the original description.

Optimize mining extraction operations by analyzing ore grade control, processing plant throughput, metallurgical recovery rates, energy consumption, and water balance.

The English summary is being prepared. Showing the original description.

Use when the user asks to make something faster, try many variants, run recursive optimization, benchmark latency/throughput/cost, or choose the best implementation by repeated measured tests.

The English summary is being prepared. Showing the original description.

Use when large data ingestion, backfill, export, ETL, warehouse loading, manifest catch-up, or table synchronization needs to become much faster while preserving data correctness.

The English summary is being prepared. Showing the original description.