Results for “accelerator”

Statically analyze a Simulink model for Rapid Accelerator (raccel) mode incompatibilities. Identifies unsupported blocks (Interpreted MATLAB Function, Level-2 S-Functions without TLC), configuration issues, and data type restrictions that prevent code generation. Classifies each issue as BLOCKER (build error) or WARNING (build warning). Use BEFORE calling sim() in rapid-accelerator mode, when buildRapidAcceleratorTarget fails, or when the user asks "why won't my model run in raccel?" TRIGGER: any request explicitly involving rapid accelerator mode, raccel, raccel build errors, or buildRapidAcceleratorTarget failures.

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

Create and revise pitch decks, investor summaries, financial models, and accelerator applications using a shared set of facts. Cross-check figures, assumptions, funding requests, and milestones so fundraising documents remain consistent and their claims are easy to verify.

• Creating or revising an investor pitch deck
• Writing an investor summary or memo
• Building financial projections and use-of-funds tables

When the user wants to apply to startup accelerators, incubators, or fellowship programs. Also use when the user mentions "YC application", "Techstars", "accelerator", or "apply to programs".

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

Think and work like an expert Accelerator Physicist. Use when a task calls for Accelerator Physicist judgment. Reasons from beam optics, RF cavities, emittance budgets, and loss maps while treating halo and impedance-driven instabilities as first-class failure modes.

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

Write accelerator applications. TRIGGERS - Use when user needs help with accelerator-application related tasks.

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

Accelerator Application Specialist. Use to shortlist startup programs, draft applications, prepare founder videos, and rehearse accelerator interviews with a disciplined founder narrative.

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

Write accelerator applications. TRIGGERS - Use when user needs help with accelerator-application related tasks.

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

Build an HTML fundraising presentation for accelerator partners and angel investors. Organize the opening hook, traction, competitive advantage, and funding request into slides using a dedicated template, with speaker notes and presentation controls.

• Prepare a Demo Day fundraising presentation
• Explain traction and competitive advantage to investors
• Build an HTML deck with speaker notes

Review Zoom meetings, calls, recordings, and transcripts to catch up on discussions and extract decisions, action items, and owners. Use Zoom's official connections to manage meetings and access supported chat, document, task, and whiteboard features.

• Catch up on a missed meeting from recordings or transcripts
• Extract decisions, action items, and owners
• Create, update, or delete Zoom meetings

Inspect CPU, memory, disk, and accelerator information alongside container and scheduler limits. Produce a JSON snapshot with identifying details removed and conservative worker and memory estimates for planning local workloads, without running stress tests.

• Choosing worker counts before a large local job
• Checking CPU and memory limits inside a container
• Comparing resource snapshots before and after changes

Finds the base image needed to create a custom GKE node image. It translates requirements such as GKE version, operating system, processor architecture, accelerators, and sandbox settings into a matching image name.

• Finding a base image for custom GKE nodes
• Looking up images for GPU or TPU configurations
• Matching images to COS, Ubuntu, ARM64, or X86_64

Use this skill when the user is doing hands-on DOCA DPA host-side work on a BlueField — creating the `doca_dpa` Core context, loading a DPACC-compiled DPA app image (`doca_dpa_app`), creating DPA threads, launching kernels via `doca_dpa_kernel_launch_update_*`, draining `doca_dpa_completion`, running `doca_dpa_cap_*` discovery, choosing between the DPA comm component (inter-DPA messaging) and the DPA verbs component (in-kernel RDMA), or debugging `DOCA_ERROR_*` from `doca_dpa_*`. Trigger even without "DOCA DPA" or "Data-Path Accelerator": "run compute on the DPA from my host", "DPA kernel hangs, no completion", "DOCA_ERROR_DRIVER on launch", "DOCA/DPACC version skew", or "does this BlueField expose a DPA". Route elsewhere for DPA-side kernel programming itself, DPACC compiler internals, host↔DPU messaging (doca-comch), host-side RDMA (doca-rdma), and GPU-initiated networking (doca-gpunetio).

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

Expert-level particle physics knowledge. Use when working with the Standard Model, quarks, leptons, bosons, fundamental forces, Feynman diagrams, quantum field theory, particle accelerators, or particle detectors. Also use when the user mentions 'Standard Model', 'quark', 'lepton', 'boson', 'Higgs', 'gluon', 'neutrino', 'antimatter', 'Feynman diagram', 'QCD', 'QED', 'electroweak', 'LHC', or 'particle accelerator'.

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

Add a Keyboard Shortcuts help window (Ctrl+question) listing all app accelerators, plus the standard GNOME accelerator set. Use when wiring keyboard shortcuts or the Keyboard Shortcuts menu item.

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

Draft concise fundraising emails for investors, including first contact, introduction requests, follow-ups, and progress updates. Tailor each message to the investor’s focus, support the pitch with concrete evidence, and make the requested next step clear.

• Drafting a first email to an investor
• Requesting an introduction with a forwardable summary
• Planning follow-ups after no response

Find bottlenecks in large data transfers, imports, transformations, and historical data catch-up. Benchmark alternative processing methods, turn the fastest correct approach into a repeatable workflow, and verify file counts, row counts, and timestamps after execution.

• Speeding up large data imports and exports
• Reducing the time needed to load historical data
• Comparing batch sizes and worker counts

Adapt existing PyTorch training scripts to use multiple GPUs or machines with limited code changes. The skill guides configuration, launch commands, mixed precision, which uses lower-precision numbers, and DeepSpeed or FSDP settings for distributing training work and memory.

• Move a single-GPU training script to multiple GPUs
• Configure mixed precision training
• Set up DeepSpeed or FSDP to reduce memory use

Adapt PyTorch training scripts for multiple GPUs or machines with Accelerate. This skill guides code changes, configuration, and launch commands, including mixed precision, which combines numerical formats, and DeepSpeed or FSDP integration for distributed training.

• Moving single-GPU training to multiple GPUs
• Configuring mixed precision training
• Setting up DeepSpeed or FSDP

Reduce LLM weight usage through pruning, which removes selected model parameters. Explore Wanda and SparseGPT for compression without retraining, choose hardware-friendly patterns for faster inference, and compare the compressed model's accuracy with the original.

• Compressing an LLM without retraining
• Comparing pruning methods
• Exploring faster inference on supported hardware

Investigate slow TPU training on GKE by comparing diagnostic events with one-minute monitoring metrics. Separate memory or CPU bottlenecks from hardware and network issues, and identify the appropriate next steps.

• Investigating slower TPU training without pod crashes
• Checking PERFORMANCE_DEGRADATION events
• Separating resource bottlenecks from infrastructure issues

Configure GKE ComputeClasses, which define preferred computing resources for workloads. Set fallback options between Spot and on-demand VMs, target GPUs or machine families, control class usage, and investigate pods waiting for suitable nodes.

• Configuring fallback from Spot VMs to on-demand capacity
• Selecting GPUs, TPUs, or machine families
• Restricting ComputeClass editing and workload usage

Plan and troubleshoot large language and vision model training and inference. This guide uses hardware measurements to identify bottlenecks, address memory shortages and unstable training, and improve checkpointing, recovery, and performance across multiple GPUs and machines.

• Diagnosing slow training and GPU memory shortages
• Investigating loss spikes and invalid numerical values
• Debugging communication hangs across GPUs and machines

Guides LLM compression through pruning, which removes selected model weights. Covers Wanda and SparseGPT methods without retraining, accuracy comparisons, and structured weight patterns that can accelerate inference on compatible hardware.

• Trying model compression without retraining
• Comparing accuracy before and after pruning
• Choosing between Wanda and SparseGPT

为给定模型负载比较加速器的容量、有效计算、内存带宽、功耗与成本。用于硬件选型或解释标称峰值与实测差异。

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