Skip to content

Test Documentation Hub

Pantheon includes focused GPU diagnostics for compute, memory, interconnect, and AI execution paths. Choose a test for its target subsystem and failure mode.


Core & Compute

  • Omni Virus


    Asynchronously overlaps memory sweeps, FP16 tensor math, FP32 vector math, and SFU work to load multiple GPU pipelines together.

  • Voltage Virus


    Uses volatile math to force rapid ALU state switching and expose voltage-rail and VRM stability limits.

  • Pulse Virus


    Alternates heavy FMA load and short idle periods to create repeated transient power ramps.

  • Tensor Virus


    Saturates FP16 arithmetic with continuous fused multiply-add chains to expose half-precision datapath, power, and thermal issues.

  • MMA Virus


    Uses physical matrix multiply-accumulate instructions to push matrix cores toward sustained thermal and power limits.

  • Transformer Virus


    Uses platform-specific matrix instructions to exercise modern transformer-engine and matrix-core execution paths.

  • FP64 Chokehold


    Runs sustained FP64 fused multiply-add work to expose double-precision throughput, clocks, and power behavior.

  • Integer Virus


    Saturates INT32 ALUs with bit operations, rotations, and XOR cascades to isolate integer-specific datapaths.

  • SFU Virus


    Hammers high-latency SIN, COS, EXP, LOG, and reciprocal-square-root operations to exercise special-function units.

  • Incinerator


    Combines vector ALU work with local-memory bank conflicts for dense thermal and SRAM pressure.


Fixed-Function & Accelerators

  • RT Virus


    Floods dedicated ray-tracing hardware with non-coherent intersection work and BVH traversal.

  • Media Encoder Virus


    Feeds high-entropy input into the hardware video encoder to exercise fixed-function media logic.


Memory & Cache

  • Atomic Virus


    Forces concurrent wide-stride atomic read-modify-write operations to stress L2 arbitration and contention handling.

  • Cache Latency


    Defeats prefetching with dependent pointer-chasing random walks across the memory pool.

  • Memory Write, Aggressive


    Bypasses cache and heavily unrolls writes with alternating patterns to maximize physical write-path pressure.

  • Memory Write, Standard


    Measures standard sequential VRAM write bandwidth with non-temporal stores and rail-to-rail patterns.

  • Memory Read, Aggressive


    Uses heavily unrolled volatile pointer accesses to force direct memory fetches without relying on cache reuse.

  • Memory Read, Standard


    Measures standard sequential VRAM read bandwidth using wide, coalesced device reads.

  • Memory Bank Thrasher


    Strides reads across large page boundaries to drive row-buffer misses in the memory subsystem.

  • Memory Cache Fracturing


    Forces large numbers of uncoalesced reads to overload memory-controller queues and cache arbitration.

  • Memory Retention Bake


    Writes a known payload, heats the device with compute work, then checks whether memory retention errors occurred.

  • Memory Asymmetric Thermal


    Hammers an isolated memory region while drawing compute power to create a severe package thermal gradient.


Interconnect & Architecture

  • P2P Thrasher


    Saturates peer-to-peer DMA links across NVLink, Infinity Fabric, or PCIe where available.

  • All-Reduce


    Validates a two-GPU sum-and-broadcast collective with direct peer DMA where available and a clearly reported host-staged fallback otherwise.

  • TLB Avalanche


    Performs pseudo-random jumps across page boundaries to force translation-cache misses and page walks.

  • PCIe Thrasher


    Floods the bus with asynchronous host-to-device and device-to-host DMA transfers.

  • RAS Validator


    Continuously reads a pristine pattern to detect uncorrectable errors and observe active ECC-scrub behavior.

  • Memory TSV Thrasher


    Alternates high and low data patterns to maximize physical memory-bus toggle rate and TSV or PHY stress.

  • Memory PC Ping-Pong


    Reads from one memory region and writes another to exercise crossbar and pseudo-channel traffic.

  • Scheduler Virus


    Launches micro-kernels across many streams to pressure dispatcher multiplexing and scheduling behavior.

  • Baseline Metrics


    Initializes the GPU without compute or memory work so telemetry captures an idle baseline for comparison.


AI & ML

  • AI Workload Suites


    Lists the inference, training, runtime, RAG, and vision suites, including their commands, telemetry guidance, and diagnostic scope.

  • LLM Decode


    Uses dependent KV-cache gathers and projection-like math to isolate the cache, memory, and compute pressure of autoregressive generation.

  • LLM Prefill


    Combines causal context scans and projection-like math to isolate the throughput-oriented pressure of long prompt processing.

  • KV Cache Churn


    Performs sparse page reads and updates across a synthetic cache to resemble ragged requests and paged KV-cache maintenance.

  • Fused Attention


    Exercises the combined compute and memory-access pattern of causal fused attention, including context-window reuse.

  • RoPE Stress


    Applies paired rotary-position-style math to exercise the embedding transform path used by modern transformers.

  • Quantized GEMM


    Unpacks low-precision values and applies projection-style arithmetic to expose quantization and dequantization overhead.

  • Serving Mix


    Creates mixed execution pressure representative of requests with different prompt and generation demands, without claiming queue-level serving metrics.

  • Speculative Decode


    Exercises the alternating draft and verification style of speculative decoding to reveal its compute and cache interaction.

  • MoE Router


    Uses sparse Top-K-style routing and gated accumulation to exercise the local compute path behind mixture-of-experts inference.

  • Transformer Train Step


    Applies forward, backward, gradient, and optimizer-style arithmetic to create a controlled training-step pressure profile.

  • Allocation Fragmentation


    Repeatedly drives a working-set allocation pattern to help expose allocator pressure and memory-management instability.

  • Graph Replay


    Repeats a compact launch sequence to stress graph-capture and replay-adjacent runtime behavior without requiring a model framework.

  • RAG Embedding


    Applies embedding-vector projection work to stress the dense math and memory behavior relevant to retrieval and RAG pipelines.

  • Vision Encoder


    Applies image-tile projection work to stress the dense compute and memory behavior common in vision encoder front ends.