---
title: "GPU"
description: "107 posts about GPU, summarised, each linking to the original."
---

# GPU
> 107 posts about GPU, summarised, each linking to the original.

## Articles

### [GPU virtualization at scale: AMD MI300X SR-IOV on Red Hat OpenStack Services on OpenShift](https://yomu.fyi/post/gpu-virtualization-at-scale-amd-mi300x-sr-iov-on-red-hat-openstack-ser.md)
- Company: [Red Hat](https://yomu.fyi/company/red-hat.md)
- Author: Sudhakar Molli, Bohdan Dobrelia, Miguel Carpio, Petr Kubica, Venkataramana Venugopal
- Published: Oct 1, 2026
- Type: Benchmark

Enterprise AI environments are moving from dedicated GPU servers to shared accelerator infrastructure, requiring higher GPU utilization with strong isolation. AMD Instinct MI300X GPUs support hardware-assisted SR-IOV in CPX mode, each exposing eight virtual functions with separate PCI and VRAM allocation. When integrated with Red Hat OpenStack Services on OpenShift, the control plane runs on OpenShift while RHEL compute nodes use EDPM, with Nova scheduling VFs via VFIO PCI passthrough. Validation on an 8-GPU system created 64 VFs, demonstrated 64 VMs with one VF each, and tested up to 32 VFs per VM. BabelStream benchmarks showed approximately 96-99% of bare-metal memory bandwidth, while the testing also identified limitations in GPU peer-to-peer DMA and runtime VF lifecycle.


### [Modal Clusters are generally available](https://yomu.fyi/post/modal-clusters-are-generally-available.md)
- Company: [Modal](https://yomu.fyi/company/modal.md)
- Author: Peyton Walters
- Published: Oct 1, 2026
- Type: Announcement

Modal announced general availability of Modal Clusters, a serverless multi-node compute primitive for AI workloads that provides up to 6.4 Tbps InfiniBand RDMA per node with PyTorch and NCCL support. A new gang scheduler places N nodes simultaneously by observing the fleet, assigning all pending clusters, and provisioning capacity when needed. RDMA support is fully automated behind a single rdma=True flag, with Modal contributing RDMA support to the gVisor container runtime and upstreaming the changes. Customers like Decagon, 1x, and Runway use Clusters for trillion-parameter fine-tuning, pre-training home robot world models, and multi-node video inference respectively. Clusters are available to all workspaces today through the @modal.clustered decorator, with per-second billing and no reservation required.


### [Distributed training on OpenShift AI 3.4 with Kubeflow Trainer v2](https://yomu.fyi/post/distributed-training-on-openshift-ai-3-4-with-kubeflow-trainer-v2.md)
- Company: [Red Hat](https://yomu.fyi/company/red-hat.md)
- Author: Linoy Bitan
- Published: Sep 29, 2026
- Type: Explainer

Red Hat OpenShift AI 3.4 introduces Kubeflow Trainer v2, a unified training API that replaces framework-specific CRDs like PyTorchJob and TFJob with a single TrainJob resource. Under the hood, each TrainJob is translated into a JobSet, a Kubernetes-native API for managing groups of Jobs as a single unit, which handles headless services for pod discovery and automatic failure recovery. The post validates the stack through two phases: manual validation with a minimal NCCL all-reduce test across 2 nodes, and real-world training of an embedding-based neural network on a fraud detection dataset using 1, 2, and 4 NVIDIA A10G GPUs. Benchmarking shows near-linear Distributed Data Parallel scaling at 95% efficiency, successful checkpoint recovery after pod kills, and GPU utilization confirmed via DCGM Exporter. The TrainJob API simplifies distributed training by automatically configuring torchrun, environment variables, and pod coordination, while raw JobSets offer more control for multi-stage pipelines and differentiated failure handling.


### [Running open-Jev in SQL on Databricks](https://yomu.fyi/post/running-open-jev-in-sql-on-databricks.md)
- Company: [Databricks](https://yomu.fyi/company/databricks.md)
- Author: Cyrus Cowley, Tejas Sundaresan, Hanlin Tang, Anshul Gupta, Julia Powell
- Published: Sep 24, 2026
- Type: Tutorial

System One decision models such as Jev produce well-calibrated decisions from discrete options and are fast and cheap for large-scale applications, with potential use cases ranging from analyzing customer support transcripts to applying complex decisions at scale. Open-weight versions like SemIf-OpenJev are available, and Databricks enables serving them on governed data using serverless GPUs via AI Runtime, which provides on-demand GPU compute without managing infrastructure. The workflow is packaged in an importable notebook that downloads the model files, logs and registers the model using Express Deployments, and creates a GPU Model Serving endpoint automatically. The endpoint can be called from SQL via ai\_query, which supports arbitrary custom model APIs, allowing structured request formats directly from SQL to return selected classifications and option probabilities as structured columns. The example classifies hotel reviews as good or bad, and users can import the notebook, select Serverless GPU, and run all to get started.


### [Hitting a billion tokens per minute on one GPU by combining a query planner and an inference engine](https://yomu.fyi/post/hitting-a-billion-tokens-per-minute-on-one-gpu-by-combining-a-query-pl.md)
- Company: [Modal](https://yomu.fyi/company/modal.md)
- Published: Sep 24, 2026
- Type: Problem & solution

Modal and CMU built Quail, a query-aware inference layer for AI-SQL that combines a query planner with an inference engine to process structured LLM transformations of database entries. On a multi-join query, Quail achieves over a billion tokens per minute per H100 GPU, over 10x faster than vLLM, costing under 6 cents per billion tokens on Modal. The gain comes from using the query plan to order requests for better KV cache management, including a revised Hydragen-style cascade attention and batching small requests to reduce host overhead. On a new AI-SQL benchmark, Quail runs 1.84x faster than vLLM, but falls behind on agent trace processing. The authors see this as part of a trend toward custom inference engines for specific workloads, with a reusable core like databases post-DataFusion, and note that much work was done by prompting coding agents.


### [Validate GPU Cluster Readiness Before AI Workloads Land](https://yomu.fyi/post/validate-gpu-cluster-readiness-before-ai-workloads-land.md)
- Company: [NVIDIA](https://yomu.fyi/company/nvidia-developer-blog.md)
- Author: Michelle Horton
- Published: Sep 23, 2026
- Type: Explainer

GPU clusters can pass standard health checks yet fail under real AI training workloads due to a single degraded GPU, network link, or misrouted traffic. NVIDIA Cluster Readiness Engine (NVCRE) is an open source Kubernetes controller that runs real distributed workloads across topology-aware node groups before production jobs land, measuring results and reporting which nodes failed each test. Its layered API uses Custom Resource Definitions for Certification, Workflow, and Job resources, attributing each failure to a specific node and category such as NCCL bandwidth or training goodput. The WorkloadRun API handles platform detection, NCCL settings, shared-memory volumes, and gang scheduling, so operators do not need to write those manifests by hand. NVCRE integrates with NVSentinel for passive telemetry and is available under Apache 2.0, with a roadmap for inference and automated lifecycle validation.


### [How to Use NVIDIA Warp and MjWarp to Accelerate Robotics Simulation and Learning Workflows](https://yomu.fyi/post/how-to-use-nvidia-warp-and-mjwarp-to-accelerate-robotics-simulation-an.md)
- Company: [Hugging Face](https://yomu.fyi/company/hugging-face.md)
- Author: Johnny Nuñez Cano, Asier Arranz, Rishabh Chadha, Ben Oliveri
- Published: Sep 23, 2026
- Type: Tutorial

NVIDIA Warp is a Python framework for writing GPU-accelerated kernels that compiles statically typed Python to CUDA. MuJoCo Warp (MJWarp) implements MuJoCo physics on Warp, enabling batched simulation of hundreds to thousands of independent robot environments on NVIDIA GPUs. The core API transition is small: mjw.put\_model and mjw.put\_data replace mujoco.MjModel and mujoco.MjData, and mjw.step advances the entire batch. Proper throughput measurement requires GPU synchronization and warm-up to avoid timing queue latency instead of actual GPU work. The post demonstrates moving an SO-101 follower arm from a single CPU world to 2,048 parallel MJWarp environments, validating parity and measuring aggregate world-steps per second.


### [Manage Kubernetes Node Fleets with NodeWright](https://yomu.fyi/post/manage-kubernetes-node-fleets-with-nodewright.md)
- Company: [NVIDIA](https://yomu.fyi/company/nvidia-developer-blog.md)
- Author: Michelle Horton
- Published: Sep 23, 2026
- Type: Problem & solution

Managing Kubernetes node fleets, especially GPU nodes, is difficult because host-level changes can disrupt long-running training jobs, and traditional tools are not cluster-aware. NodeWright, an open-source Kubernetes-native package manager from NVIDIA, solves this by declaratively configuring and updating node operating systems while respecting PodDisruptionBudgets and non-interruptible workload labels. The operator orchestrates a sequence of cordon, wait, drain, apply, interrupt, and uncordon on each node, using container image packages that carry modifications such as kernel parameters, security agents, and verification scripts. It supports progressive rollouts across a fleet via DeploymentPolicy compartments with configurable disruption budgets and strategies like batch, canary, and rolling. NodeWright integrates with NVIDIA AI Cluster Runtime for validated configurations and is part of the DSX OS open-source portfolio, available under Apache 2.0.


### [How I massively improved my AI inference performance without buying new hardware](https://yomu.fyi/post/how-i-massively-improved-my-ai-inference-performance-without-buying-ne.md)
- Company: [Red Hat](https://yomu.fyi/company/red-hat.md)
- Author: Boaz Ben Shabat
- Published: Sep 23, 2026
- Type: Problem & solution

Default inference configurations for large language models often leave significant performance unused, even on expensive hardware like 16 NVIDIA H200 GPUs across two nodes. The author tested this setup with vLLM and Red Hat AI Inference 3.4 on a CoreWeave Kubernetes cluster, using enterprise RAG workloads with 2,000-token inputs and 100 concurrent users. By systematically comparing aggregated and prefill-decode (PD) disaggregated architectures, and tuning tensor parallelism (TP) and PD split, latency was dramatically reduced. For Qwen3-32B, time-to-first-token (TTFT) on the best aggregated configuration (2xTP8) dropped from 995ms to 304ms, and with optimized routing to 287ms — a 71% improvement without hardware changes. For Llama-70B, TTFT fell from 1,292ms to 463ms, a 64% gain. The author also built ServeIt Studio, an open-source tool that automates the full optimization pipeline on Kubernetes.


### [How to serve trillions of tokens for trillion-parameter coding agents](https://yomu.fyi/post/how-to-serve-trillions-of-tokens-for-trillion-parameter-coding-agents.md)
- Company: [Modal](https://yomu.fyi/company/modal.md)
- Published: Sep 23, 2026
- Type: Explainer

Serving inference for coding agents requires trillion-parameter sequence models like Moonshot AI’s Kimi K2.6 to process trillions of input and output tokens at extremely high relative and absolute performance. To be economically viable, these services must achieve large fractions of hardware peak performance, using contemporary matrix math accelerators like Tensor Cores at petaFLOP per second scale. Modal optimized by batching multiple sequences together, caching KV state, and patching SGLang with NVFP4 micro-scaling on Blackwell GPUs like the B200. Per-replica performance improved 2.8x per user and 5.6x across users, transforming ruinously expensive into a price-competitive service. One deployment processed hundreds of billions of tokens daily and trillions total, showing that careful engineering and benchmarking are key to scaling inference.


### [Topology-Aware Workload Scheduling with NVIDIA Topograph](https://yomu.fyi/post/topology-aware-workload-scheduling-with-nvidia-topograph.md)
- Company: [NVIDIA](https://yomu.fyi/company/nvidia-developer-blog.md)
- Author: Elizabeth Goodman
- Published: Sep 22, 2026
- Type: Explainer

AI factories are power-limited systems where GPU workload placement is a key optimization; poor placement fragments topology domains and forces traffic across shared links, reducing throughput and raising job costs. NVIDIA Topograph discovers cluster topology from cloud APIs or on-premises fabric systems, normalizes it into a common model, and publishes it in formats each workload manager expects, such as Kubernetes node labels, Slurm topology configuration, or Slinky ConfigMaps. It uses five components: API Server, Node Observer, Node Data Broker, Provider, and Engine to keep the topology view current as the cluster changes. The post covers deploying Topograph with Helm on Kubernetes, as a system package on Slurm, and as a Helm chart on Slinky, with support for cloud providers including Google Cloud, Lambda, Nebius, Nscale, and OCI. Topograph works alongside Dynamic Resource Allocation and KAI Scheduler to enable topology-aware gang scheduling, improving AI factory efficiency and tokens per watt.


### [Simplifying Model Serving Across Multiple GPUs with NVIDIA TensorRT Multi-Device Integration in NVIDIA Dynamo-Triton](https://yomu.fyi/post/simplifying-model-serving-across-multiple-gpus-with-nvidia-tensorrt-mu.md)
- Company: [NVIDIA](https://yomu.fyi/company/nvidia-developer-blog.md)
- Author: Tanya Lenz
- Published: Sep 21, 2026
- Type: Announcement

NVIDIA TensorRT multi-device inference enables a single TensorRT network to execute across multiple GPUs using NCCL-backed distributed collectives, supported from TensorRT 11.0. Integrated into Dynamo-Triton release 26.07, one model instance can own multiple GPUs and expose a single gRPC endpoint so the client does not coordinate GPU ranks. The post demonstrates this with Cosmos 3 Nano video generation, where the transformer uses Ulysses context parallelism to distribute 44,160 video tokens across up to eight GPUs. End-to-end latency drops from 156.6 seconds on one GPU to 34.2 seconds on eight GPUs, achieving a 6.09x transformer RPC speedup. Teams can trade additional GPU resources for shorter request latency while keeping the application interface and surrounding workflow stable.


### [Benchmarking LLM Inference at Scale with AIPerf](https://yomu.fyi/post/benchmarking-llm-inference-at-scale-with-aiperf.md)
- Company: [NVIDIA](https://yomu.fyi/company/nvidia-developer-blog.md)
- Author: Elizabeth Goodman
- Published: Sep 18, 2026
- Type: Benchmark

AIPerf is NVIDIA’s successor to GenAI-Perf, intended to benchmark large-scale LLM inference without letting the load client become the bottleneck. Its multiprocess architecture combines worker processes, separate record-processor services, and ZMQ coordination, while supporting 15+ endpoint types, public datasets, and trace-replay formats. The walkthrough serves Qwen3-0.6B through vLLM, using streaming synthetic requests with fixed 128-token inputs and outputs; AIPerf reports TTFT, ITL, request latency, output token throughput, percentile distributions, and optional GPU telemetry in CSV and JSON. A second profile uses Poisson arrivals averaging 10 requests per second, variable input and output lengths, and a fixed random seed, producing wider latency distributions than the static baseline as concurrency and workload variability increase.


### [How a global fintech scaled coding agent traffic with Dedicated Model Inference](https://yomu.fyi/post/how-a-global-fintech-scaled-coding-agent-traffic-with-dedicated-model.md)
- Company: [Together AI](https://yomu.fyi/company/together-ai.md)
- Published: Sep 18, 2026
- Type: Problem & solution

A global fintech runs its coding assistant on GLM 5.2 through Together’s Dedicated Model Inference (DMI), but spiky engineering-hours traffic outpaced static capacity planning. DMI gives the customer’s teams direct control over endpoint provisioning, scaling, configuration, model changes, and performance data through APIs, UI, and CLI. The workload is peak-load and relatively low-TPS, with ISL p50/p90/p95 of roughly 81K/163K/178K tokens and RPS p50/p90/p95 of 3/6/7, so concurrency headroom mattered more than raw throughput; the deployment used dozens of B200s, multiple replicas, and 256K context. When a reshaped endpoint later reached near-100% prefill capacity, requests queued for one to three minutes and decode throughput fell to roughly 5 tokens per second, the team restored cache-session-aware routing and widened max-inflight-per-worker, pushing the change same-day with zero downtime.


### [Translating CUDA Tile Operations from Python to Rust Using Agentic AI](https://yomu.fyi/post/translating-cuda-tile-operations-from-python-to-rust-using-agentic-ai.md)
- Company: [NVIDIA](https://yomu.fyi/company/nvidia-developer-blog.md)
- Author: Tanya Lenz
- Published: Sep 16, 2026
- Type: Tutorial

cuTile Rust (cutile-rs) extends Rust’s ownership model to tile-based GPU kernels, and this post describes an agentic workflow for translating cuTile Python and Triton-TileIR implementations. Because all three front ends target CUDA Tile IR and share the tileiras compiler, the workflow compares reference and translated IR while making Python’s implicit specializations explicit through Rust const generics, fixed dtype symbols, and FFI checks. A bounded multi-agent pipeline covers analysis, device kernels, host and FFI code, correctness, and benchmarking, with machine-checkable verdicts controlling progress. It ported all 24 public TileGym operators, roughly 40 kernels, and reported 99.5% average cuTile Python performance across 347 paired configurations on NVIDIA DGX B200. The skill ships with TileGym and uses acceptance checks requiring correctness, backend execution, and a geomean speedup of at least 0.95.


### [How NVIDIA Groq 3 LPX Deterministic Execution Drives Power-Efficient High-Interactivity Inference on NVIDIA Vera Rubin](https://yomu.fyi/post/how-nvidia-groq-3-lpx-deterministic-execution-drives-power-efficient-h.md)
- Company: [NVIDIA](https://yomu.fyi/company/nvidia-developer-blog.md)
- Author: Tanya Lenz
- Published: Sep 15, 2026
- Type: Explainer

AI factories are constrained by site power, making performance per watt especially important for high-interactivity inference. The NVIDIA Groq 3 LPX rack uses deterministic execution: its compiler schedules computation, data movement, and clock-cycle timing across 256 LPU chips, then derives a cycle-level current-demand curve. Preemptive Power (PEP) prepares the power-delivery network for upcoming spikes, while Clock Period Synthesis (CPS) lengthens or shortens selected cycles to reduce di/dt, voltage droop, and the voltage guardband. Internal testing reported more than 60% less voltage drop, while the source estimates a high-single-digit baseline-voltage reduction and potentially low-double-digit workload-power reduction; paired with Vera Rubin NVL72, the platform is described as delivering up to 35x higher throughput per megawatt than GB200 NVL72 for 2T+ parameter models at long context and high interactivity.


### [How NVIDIA NVLink 6 Delivers Multi-Layer Resiliency for AI Factories](https://yomu.fyi/post/how-nvidia-nvlink-6-delivers-multi-layer-resiliency-for-ai-factories.md)
- Company: [NVIDIA](https://yomu.fyi/company/nvidia-developer-blog.md)
- Author: Elizabeth Goodman
- Published: Sep 15, 2026
- Type: Explainer

NVIDIA NVLink 6 is presented as a scale-up networking fabric for large AI factories, where dropped packets, link degradation, and node interruptions can reduce training goodput or inference capacity. It connects 72 Rubin GPUs in the Vera Rubin NVL72 into a single scale-up domain and uses a multi-layer resiliency stack spanning physical, link, application, software, and system layers. Physical Layer techniques include lightweight FEC, Physical Layer Retry (PLR), UPHY recovery, and hardware replay buffers, while credit-based flow control (CBFC) prevents injection when the next hop lacks buffer capacity. Redundant switch trays and management paths, local fault containment, NMX high availability, checkpointing, topology-aware recovery, and targeted switch maintenance are described as mechanisms for containing failures and preserving workloads; the post concludes that this integrated design targets continuous operation and maximum cluster productivity.


### [Accelerating Dropless MoE Training in JAX with NVIDIA Transformer Engine](https://yomu.fyi/post/accelerating-dropless-moe-training-in-jax-with-nvidia-transformer-engi.md)
- Company: [NVIDIA](https://yomu.fyi/company/nvidia-developer-blog.md)
- Author: Tanya Lenz
- Published: Sep 14, 2026
- Type: Explainer

Production-scale mixture-of-experts (MoE) training must handle uneven, data-dependent expert loads, creating ragged tensors and expensive dispatch, combine, and all-to-all communication. Dropless MoE processes every selected token, preserving model quality but requiring kernels that support variable token counts without padding or dropping. The post presents JAX and NVIDIA Transformer Engine optimizations centered on group-aware MXFP8 quantization, grouped GEMM, and fused expert-parallel (EP) dispatch and combine using NCCL EP. Grouped GEMM computes each expert’s actual workload in one kernel path, while EP reduces movement overhead and can deduplicate tokens crossing networks. For DeepSeek-V3 training on NVIDIA GB200, the reported throughput increased from 103 to 1,068 TFLOPS/GPU, a 10.4x improvement; the post also reports roughly 10x throughput improvement and 97% scaling efficiency at 1,024 GPUs.


### [Understanding W8A8 INT8 LLM quantization: Accuracy and performance results](https://yomu.fyi/post/understanding-w8a8-int8-llm-quantization-accuracy-and-performance-resu.md)
- Company: [Red Hat](https://yomu.fyi/company/red-hat.md)
- Author: Sana Fayyaz
- Published: Sep 14, 2026
- Type: Benchmark

The post evaluates whether W8A8 INT8 quantization can reduce the size of a Llama 3.1 8B Instruct model without sacrificing accuracy or inference performance. Using SmoothQuant to address activation outliers and GPTQ with error compensation for BF16 weights, the model was compressed from 14.9 GB to 8.0 GB after calibration on WikiText-2. Identical zero-shot lm-eval-harness tests across MMLU, ARC Easy, HellaSwag, and IFEval found MMLU, ARC Easy, and HellaSwag deltas smaller than their standard errors, while the IFEval change was described as likely within a similar noise range; the post concludes no measurable degradation. vLLM and GuideLLM tests on one NVIDIA L40S reported 29% more concurrent requests, 24% lower synchronous TTFT, and 44% faster token generation at maximum load, while ITL degraded more quickly under high concurrency because of dynamic activation quantization overhead.


### [Product updates: Sandbox Sidecars, new models, a refreshed dashboard, and more](https://yomu.fyi/post/product-updates-sandbox-sidecars-new-models-a-refreshed-dashboard-and.md)
- Company: [Modal](https://yomu.fyi/company/modal.md)
- Published: Sep 14, 2026
- Type: Announcement

Modal added day-zero support for Kimi K3, Qwen 3.8, GLM 5.3, and GLM 5.3 Flash on Auto Endpoints, with custom DFlash speculators boosting throughput to 460 tokens per second. New environment-level budgets give Team and Enterprise workspaces per-environment compute limits, while the pricing multiplier for broad region selection dropped from 1.5x to 1.15x. The dashboard got a refreshed navigation, color system, and beta light mode, and Private Environments now support a default No Access role. Sandbox Sidecars are in public alpha, running additional containers via low-latency internal networking, and VM Sandboxes are in public beta with a real Linux kernel for Docker and eBPF. Sandbox directory snapshots and a new filesystem API are now GA, and regional Proxies allow selecting a region for static outbound IPs, while SDK updates improved log APIs and autoscaler reporting.


[Newer posts](https://yomu.fyi/topic/gpu.md) · [Older posts](https://yomu.fyi/topic/gpu/page/3.md)