“cpu”的搜索结果

Add PyTorch FSDP2 to training scripts so model state can be split across GPUs. Covers distributed initialization, mixed precision, CPU offloading, and checkpoint saving and loading for models that exceed a single GPU's memory.

中文简介正在准备中。当前显示英语简介。

• Training a model that exceeds one GPU's memory
• Adding FSDP2 to an existing training script
• Configuring mixed precision and CPU offloading

Convert language models to GGUF and reduce their size through quantization, which stores model weights at lower precision. Guides cover compression choices, local inference on CPUs and Apple Silicon, and integration with tools such as Ollama.

中文简介正在准备中。当前显示英语简介。

• Running a language model on a CPU-only computer
• Choosing model compression for available memory
• Configuring Metal acceleration on Apple Silicon

Profile app launches and individual user flows in iOS Simulator with ETTrace. Collect traces with readable function names, identify CPU-heavy call stacks, and compare performance changes when recording conditions are consistent.

中文简介正在准备中。当前显示英语简介。

• Investigating slow app launches
• Finding CPU-heavy work during a specific interaction
• Comparing performance before and after a change

Investigate observed slowdowns or increased CPU, memory allocation, and IO costs in a process you control. Use controlled timing and runtime-specific profiling tools to test a cause, then check performance and behavior after a fix.

中文简介正在准备中。当前显示英语简介。

• Investigate a command that has become slower
• Locate CPU or memory allocation hotspots
• Separate transport waits from backend processing costs

Choose diagnostic tools and collect performance traces for production .NET problems such as high CPU usage, memory growth, or slow requests. Get environment-specific collection commands and guidance on which tools to use for later analysis.

中文简介正在准备中。当前显示英语简介。

• Collecting traces during high CPU usage or slow requests
• Choosing diagnostic tools for Windows or Linux
• Collecting diagnostics from containers or Kubernetes

Run and supervise DOCA applications directly on an x86 host or BlueField Arm hardware. Covers process launch, hardware resource assignment, isolation between workloads, and troubleshooting application failures or BlueField recovery problems.

中文简介正在准备中。当前显示英语简介。

• Launching a DOCA application directly on hardware
• Investigating systemd restart loops
• Assigning CPU and device resources to multiple workloads

Debug Node.js from the terminal by pausing execution, inspecting variables, and tracing function calls. Use interactive commands or scripts to investigate failing tests, collect runtime state, and capture CPU profiles or memory snapshots.

中文简介正在准备中。当前显示英语简介。

• Inspecting intermediate state in failing Node.js tests
• Tracing function calls and unexpected variable values
• Investigating stalled asynchronous processing

Optimize scientific Python workloads for NVIDIA GPUs by choosing suitable libraries and reducing bottlenecks. Compare outputs with the CPU version and measure complete processing time, including data transfers, to determine whether the GPU implementation provides a useful improvement.

中文简介正在准备中。当前显示英语简介。

• Assess whether slow Python processing would benefit from a GPU
• Move NumPy or pandas workloads to suitable GPU libraries
• Validate GPU results against CPU outputs

Measure scripts, running processes, or whole systems to find performance bottlenecks. Select suitable tools, add measurement code when needed, and report CPU, GPU, memory, and communication findings with prioritized recommendations and a record of file changes.

中文简介正在准备中。当前显示英语简介。

• Finding slow functions in a Python script
• Investigating CPU and GPU memory usage
• Measuring communication overhead across multiple GPUs

Improve MNN's Vulkan backend by separating CPU scheduling bottlenecks from GPU computation costs. It guides operator and shader changes, new feature integration with compatibility fallbacks, and device testing to verify both correctness and end-to-end performance.

中文简介正在准备中。当前显示英语简介。

• Finding CPU or GPU bottlenecks in Vulkan inference
• Optimizing a specific operator or shader
• Integrating cooperative matrix or subgroup features

Set up application monitoring with structured logs, performance metrics, and traces that follow requests across services. Build dashboards and alerts, run load tests, investigate CPU and memory bottlenecks, and estimate future infrastructure capacity.

中文简介正在准备中。当前显示英语简介。

• Set up application dashboards and alerts
• Investigate production issues using logs, metrics, and traces
• Run load tests before traffic increases

Diagnose slow systems, crashes, and resource problems on servers, PCs, virtual machines, and containers. Quick checks identify affected components, while deeper analysis investigates causes and provides optimization advice or repair plans.

中文简介正在准备中。当前显示英语简介。

• Investigating a slow server or PC
• Finding causes of high CPU or low memory
• Checking disk space and network problems

Investigate Finelog logs and runtime measurements for Iris tasks, training, and vLLM. Discover available fields, write time-bounded SQL, summarize resource usage, compare central and regional records, and interpret counters correctly.

中文简介正在准备中。当前显示英语简介。

• Summarizing Iris task memory and CPU usage
• Comparing central and regional logs for forwarding delays
• Aggregating vLLM counters with reset handling

Measure how quickly a host or DPU CPU installs and deletes DOCA Flow rules. Choose a JSON workload and backend, run repeated benchmarks, and record hardware, software, and execution settings so results can be assessed.

中文简介正在准备中。当前显示英语简介。

• Measure BlueField rule installation rates
• Compare DPDK and DOCA backends
• Investigate differences from published benchmark results

Build and troubleshoot applications that move DEFLATE compression and decompression, or LZ4 decompression, to DOCA-supported hardware. Guidance covers device capabilities, buffer sizes, memory permissions, validation, and deciding when CPU processing is more suitable.

中文简介正在准备中。当前显示英语简介。

• Moving bulk DEFLATE compression to hardware
• Decompressing incoming network data
• Checking device support and buffer limits

Get guidance on installing CUDA-Q, writing quantum processing functions, and choosing CPU or GPU simulation targets. Also covers connecting to quantum processors, running a Bell-state example, and selecting approaches for execution across multiple devices.

中文简介正在准备中。当前显示英语简介。

• Installing CUDA-Q and checking a first example
• Choosing CPU or GPU simulation targets
• Writing and debugging Python quantum kernels

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.

中文简介正在准备中。当前显示英语简介。

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

Find relevant Google Cloud monitoring metrics for a service or resource. Retrieves current metric definitions through the API, filters them by the requested concepts, and presents selected metrics with their names, data types, units, and monitored resources.

中文简介正在准备中。当前显示英语简介。

• Find CPU metrics for Compute Engine
• Identify Cloud Storage write throughput and request metrics
• Check metric data types and units

Check CPU cores, GPUs, available memory, and disk space before demanding computing tasks. Save a JSON report with recommendations for parallel workers, GPU use, and processing data in chunks when memory is limited.

中文简介正在准备中。当前显示英语简介。

• Checking memory before analyzing large datasets
• Identifying available GPUs before model training
• Choosing worker counts for parallel file processing

Diagnose GPU out-of-memory errors during Megatron Bridge training and choose memory reduction settings. Covers fragmentation, LoRA input storage, recomputation, parallelism changes, and CPU offloading, explaining compatibility limits and potential tradeoffs between memory use and training throughput.

中文简介正在准备中。当前显示英语简介。

• Investigating GPU out-of-memory errors during training
• Reducing retained inputs in LoRA training
• Estimating memory needs before a large training job

Use to lock/cap Jetson CPU/GPU/EMC clocks, toggle EMC/CPU DVFS, or change cpufreq governors by editing BPMP DTB and nvpower.sh pre-flash. Do NOT use for live tuning or nvpmodel edits.

中文简介正在准备中。当前显示原文说明。

Use when diagnosing NPU + PyTorch or LLM Serving Host CPU affinity, NUMA locality, cgroup/cpuset constraints, CPU range conflicts, PyTorch/runtime threading, DataLoader, tokenizer, scheduler, vLLM-Ascend, SGLang, TTFT, TPOT, tokens/s, QPS, or multi-rank/multi-worker CPU binding issues.

中文简介正在准备中。当前显示原文说明。

测量并比较多台计算机达成一致结果的分布式共识方式，分析处理量、响应时间、扩展表现及资源使用情况，找出性能瓶颈并整理配置调整和改进建议。

• 比较 Byzantine、Raft、Gossip 性能
• 测量负载增加时的响应时间
• 查找处理器和内存限制

Create container-based hosted agents in Azure AI Foundry using Python and the Azure AI Projects SDK. Follow examples for resource settings, authentication, version management, and checks needed before treating a deployment as healthy.

中文简介正在准备中。当前显示英语简介。

• Creating a hosted agent from a custom container image
• Configuring CPU, memory, and environment variables
• Listing and deleting agent versions