Is this page helpful?

Release Notes#

Release 2.5.0#

Summary#

This release extends GPU support with NVIDIA B300 and GB300, and removes the per-GPU engine-build step that was required by earlier releases. OpenFold2 now starts and serves requests with a single, simplified runtime configuration.

Key Features#

  • NVIDIA B300: Added support for NVIDIA B300 GPU (288GB HBM3e)

  • NVIDIA GB300: Added support for NVIDIA GB300 Grace Blackwell Ultra superchip (288GB HBM3e)

  • Confidence outputs (pLDDT, PAE, pTM): Each predicted structure in structures_in_ranked_order now includes per-residue plddt, the predicted_aligned_error matrix and its max_predicted_aligned_error bin upper bound, aligned_confidence_probs, and the ptm_score confidence scalar. Refer to Response Fields in Example Requests and the API Reference for field-by-field details.

  • New optimized inference backend: The NIM now runs on a fully NVIDIA-tuned inference path, delivering faster and more consistent predictions out of the box.

Backend Optimizations#

The new inference backend introduces custom GPU kernels and graph-level optimizations:

  • New kernels: Fused Adaptive Layer Norm; DualGemm (sm80, two variants); gated sigmoid (sm80); FlashAttention v2/v3 for Triangle Attention and Attention Pair Bias.

  • Pre-computed masks and buffers: Pre-computed, left-aligned attention masks feed directly into the new dual-GEMM and attention kernels, with pre-computed buffers for the Pairformer and Diffusion Transformer (DiT) stacks.

Release 2.4.0#

Summary#

This release extends GPU support with the addition of NVIDIA GH200 superchips and the RTX PRO 6000 Blackwell Workstation Edition, adds Slurm support for HPC deployments, and adds support for NVIDIA H200.

Key Features#

  • NVIDIA H200: Added support for H200 with 141GB memory

  • NVIDIA GH200 144GB: Added support for GH200 with 144GB memory

  • NVIDIA RTX PRO 6000 Blackwell Workstation Edition: Added support for the professional workstation GPU with 96GB GDDR7 memory

  • Slurm support: Added support for running on Slurm-based HPC clusters (refer to Running on Slurm With Enroot in Prerequisites)

Release 2.3.0#

Summary#

This release extends GPU support with the addition of the GB10 architecture and updates cuEquivariance for enhanced performance.

Key Features#

  • GB10 DGX Spark support: Added support for GB10 DGX Spark SKUs with sequence lengths up to 1536 residues

  • cuEquivariance integration: Updated to cuEquivariance 0.8.1 for GB10 support

Release 2.2.0#

Summary#

This release extends GPU support with the addition of the GB200 architecture.

Key Features#

  • GB200 support: Added TensorRT-BioNeMo support for GB200 SKUs with sequence lengths up to 2048 residues

Release 2.1.0#

Summary#

This release focuses on security improvements, performance enhancements for newer GPU architectures, and extended sequence length support.

Key Features#

  • Security: Addressed all CVE (Common Vulnerabilities and Exposures) issues

  • cuEquivariance integration: Updated to cuEquivariance 0.7.0 for improved equivariant operations

  • Enhanced B200 support: Enabled trimul kernel optimization for B200 SKUs in PyTorch backend

  • Extended sequence length: Support for longer sequences up to 2048 residues on A100, H100, and B200 GPUs with TensorRT-BioNeMo

  • Telemetry control: NIM Telemetry helps NVIDIA deliver a faster, more reliable experience with greater compatibility across a wide range of environments, while maintaining strict privacy protections and giving users full control.

    Benefits:

    • Enhances performance and reliability: Provides anonymous system and NIM-level insights that help NVIDIA identify bottlenecks, tune performance across hardware configurations, and improve runtime stability.

    • Improves compatibility across deployments: Helps detect and resolve version, driver, and environment compatibility issues early, reducing friction across diverse infrastructure setups.

    • Accelerates troubleshooting and bug resolution: Allows NVIDIA to diagnose errors and regressions faster, leading to quicker support response times and higher overall availability.

    • Informs smarter optimizations and future releases: Real-world, aggregated telemetry data helps guide the optimization of NIM runtimes, model packaging, and deployment workflows, ensuring updates target the scenarios that matter most to users.

    • Protects user privacy and data security: Collects only minimal, anonymous metadata, such as hardware type and NIM version. No user data, input sequences, or prediction results are collected.

    • Fully optional and configurable: Telemetry collection is disabled by default. You can toggle telemetry at any time using environment variables.

    Configuration:

    • Set NIM_TELEMETRY_MODE=0 to disable telemetry (default)

    • Set NIM_TELEMETRY_MODE=1 to enable telemetry

    For more information about data privacy, what is collected, and how to configure telemetry, refer to:

Release 2.0.0#

Summary#

This release removes HHR-based template processing in favor of explicit mmCIF template support and adds TensorRT integration for new GPU architectures.

Key Changes#

  • Enhanced GPU support: TensorRT-BioNeMo integration for L40S, B200, and RTX 6000 Ada Generation

  • Removed: Removed the following functionalities:

    • The downloading and decompressing of a bundled database of structure files at NIM startup

    • The web API accepting input fields containing HHR content

    • The input data processing code that uses the HHR content to select template structures from the de-compressed database of structure files

  • Simplified: Template workflow now uses only mmCIF format using explicit_templates, where users can submit one or multiple mmCIF strings as part of their structure prediction requests.

  • Reduced footprint: Smaller container size and faster startup without bundled databases

  • Migration guide: Refer to the Migration Guide for upgrading from previous versions

Release 1.2.0#

Summary#

This release introduces support for user-supplied mmCIF input features.

Key Features#

  • Users can now submit one or multiple mmCIF strings as part of their structure prediction requests.

  • The structure prediction network can utilize these user-supplied mmCIF inputs for enhanced flexibility and custom workflows.

  • This feature enables new use cases where users provide their own structural templates or data in mmCIF format.

Release 1.1.0#

Summary#

This release introduces significant performance improvements through TensorRT (TRT) optimization.

Key Features#

  • Enhanced inference performance using TensorRT optimization

  • Improved model execution speed

  • Optimized memory usage during inference

Release 1.0.0#

Summary#

This is the first release of NVIDIA NIM for OpenFold2.