Fold-CP: A Context Parallelism Framework for Biomolecular Modeling

Understanding cellular machinery requires atomic-scale reconstruction of large biomolecular assemblies. However, predicting the structures of these systems has been constrained by hardware memory requirements of models like AlphaFold 3, imposing a practical ceiling of a few thousand residues that can be processed on a single GPU. Here we present NVIDIA BioNeMo Fold-CP, a context parallelism framework that overcomes this barrier by distributing the inference and training pipelines of co-folding models across multiple GPUs. We use the Boltz models as open source reference architectures and implement custom multidimensional primitives that efficiently parallelize both the dense triangular updates and the irregular, data-dependent pattern of window-batched local attention. Our approach achieves efficient memory scaling; for an N-token input distributed across P GPUs, per-device memory scales as $O(N^2/P)$, enabling the structure prediction of assemblies exceeding 30,000 residues on 64 NVIDIA B300 GPUs. We demonstrate the scientific utility of this approach through successful developer use cases: Fold-CP enabled the scoring of over 90% of Comprehensive Resource of Mammalian protein complexes (CORUM) database, as well as folding of disease-relevant PI4KA lipid kinase complex bound to an intrinsically disordered region without cropping. By providing a scalable pathway for modeling massive systems with full global context, Fold-CP represents a significant step toward the realization of a virtual cell.

Design of a Standard-Compliant Real-Time Neural Receiver for 5G NR

We detail the steps required to deploy a multi-user multiple-input multiple-output (MU-MIMO) neural receiver (NRX) in an actual cellular communication system. This raises several exciting research challenges, including the need for real-time inference and compatibility with the 5G NR standard. As the network configuration in a practical setup can change dynamically within milliseconds, we propose an adaptive NRX architecture capable of supporting dynamic modulation and coding scheme (MCS) configurations without the need for any re-training and without additional inference cost. We optimize the latency of the neural network (NN) architecture to achieve inference times of less than 1ms on an NVIDIA A100 GPU using the TensorRT inference library. These latency constraints effectively limit the size of the NN and we quantify the resulting signal-to-noise ratio (SNR) degradation as less than 0.7 dB when compared to a preliminary non-real-time NRX architecture. Finally, we explore the potential for site-specific adaptation of the receiver by investigating the required size of the training dataset and the number of fine-tuning iterations to optimize the NRX for specific radio environments using a ray tracing-based channel model. The resulting NRX is ready for deployment in a real-time 5G NR system and the source code including the TensorRT experiments is available online.