Skip to main content
NVIDIA
Explore
Models
Skills
Blueprints
GPUs
Docs
Help Center
Getting Started
  1. Create and verify your account to unlock full access to NVIDIA NIM APIs.
ResourcesDeveloper ForumsContact Support
FAQs
  • View All Playbooks
    View All Playbooks

    onboarding

    • Set Up Local Network Access
    • Open WebUI with Ollama

    data science

    • Single-cell RNA Sequencing
    • Portfolio Optimization
    • CUDA-X Data Science
    • Build Knowledge Graphs with txt2kg
    • Optimized JAX

    tools

    • DGX Dashboard
    • RAG Application in AI Workbench
    • Set up Tailscale on Your Spark
    • VS Code
    • Connect Three DGX Spark in a Ring Topology
    • Connect Multiple DGX Spark through a Switch

    fine tuning

    • FLUX.1 Dreambooth LoRA Fine-tuning
    • LLaMA Factory
    • Fine-tune with NeMo
    • Fine-tune with Pytorch
    • Unsloth on DGX Spark

    use case

    • Run Hermes Agent with a Local LLM
    • cuTile Kernels
    • CLI Coding Agent
    • Run NemoClaw with a Local LLM
    • šŸ¦ž Set Up Example NemoClaw Agents šŸ¦ž
    • Live VLM WebUI
    • Install and Use Isaac Sim and Isaac Lab
    • Vibe Coding in VS Code
    • Build and Deploy a Multi-Agent Chatbot
    • Connect Two Sparks
    • NCCL for Multiple Sparks
    • Build a Video Search and Summarization (VSS) Agent
    • Spark & Reachy Photo Booth
    • Secure AI Agents with OpenShell
    • Run OpenClaw with a Local LLM

    inference

    • Generate Images and Videos with ComfyUI
    • Serve LLMs with vLLM
    • Speculative Decoding
    • Run models with llama.cpp on DGX Spark
    • Nemotron Model Family on DGX Spark
    • Serve LLMs with SGLang
    • TRT LLM for Inference
    • Quantize Models to NVFP4 with NVIDIA Model Optimizer
    • Multi-modal Inference
    • NIM on Spark
    • LM Studio on DGX Spark

    cuTile Kernels

    60 MIN

    Run cuTile kernel benchmarks, FMHA implementation, and LLM inference on DGX Spark and B300

    • Benchmarking
    • Cross-Platform
    • DeepSeek
    • Docker
    • FMHA
    • Flash Attention
    • GPU Development
    • LLM Inference
    • Qwen2
    • TileGym
    • cuTile
    View on GitHub
    OverviewOverviewKernel BenchmarksKernel BenchmarksEnd-to-End InferenceEnd-to-End InferenceFMHA ImplementationFMHA ImplementationPlatform ComparisonPlatform ComparisonTroubleshootingTroubleshooting

    DGX Spark vs B300 Performance Comparison

    This page summarizes performance scaling between DGX Spark (GB10) and B300 for both kernel benchmarks and end-to-end LLM inference.

    Kernel Benchmark Scaling

    Use the ratios below as a reference for how kernel performance scales from DGX Spark (GB10) to B300.

    KernelMetricB300 / GB10
    FMHA (causal, 8192)TFLOPS13.7x
    FMHA (non-causal, 8192)TFLOPS15.1x
    MatMul (8192)TFLOPS18.9x
    BMM (batch8, 4096)TFLOPS19.4x
    Group GEMM (4096)TFLOPS23.9x
    RMSNorm (4096)GB/s33.1x
    RoPE (16384)GB/s22.8x

    Key Observations:

    • Compute-heavy kernels typically scale 14-24x from GB10 to B300
    • Memory-bound kernels can scale 20-33x due to HBM bandwidth advantage

    Qwen2-7B Performance

    End-to-End Throughput

    ConfigurationDGX SparkB300Platform Speedup
    cuTile18.52 tok/s257.33 tok/s13.9x

    CUDA Kernel Time

    ConfigurationDGX SparkB300Platform Speedup
    cuTile43,080 ms2,954 ms14.6x

    cuTile Kernel Breakdown

    DGX Spark (GB10):

    KernelCUDA Time (ms)Calls
    fmha_kernel4,185.928
    swiglu_forward_kernel2,459.81,400
    attention_decode_kernel_grouped2,271.81,372
    rms_norm_kernel_static_persistent634.757
    rope_kernel355.61,400

    B300:

    KernelCUDA Time (ms)Speedup vs Spark
    fmha_kernel337.912.4x
    swiglu_forward_kernel226.310.9x
    attention_decode_kernel_grouped111.020.5x
    rms_norm_kernel_static_persistent29.721.4x
    rope_kernel16.721.3x

    Same code, different architectures - cuTile JIT compiles for sm_121 (Spark) and sm_103 (B300)

    Platform Specifications

    SpecificationDGX Spark (GB10)B300
    Compute Capabilitysm_121 (12.1)sm_103 (10.3)
    SMs48132
    Memory128 GB LPDDR5x192 GB HBM3e
    Memory Bandwidth273 GB/s8 TB/s

    Resources

    • TileGym Repository
    • cuTile Python Documentation
    • Tile IR Specification
    • DGX Spark Documentation
    • DGX Spark Forum
    • Qwen2 on HuggingFace
    • DeepSeek-V2-Lite on HuggingFace
    • NVIDIA Blog - Tuning Flash Attention in CUDA Tile
    • Flash Attention Paper
    Terms of Use
    Privacy Policy
    Your Privacy Choices
    Contact

    Copyright Ā© 2026 NVIDIA Corporation