NVIDIA BioNeMo kernels hit 7.1x speedup on triangle attention in Terray's drug discovery benchmarks
AllThingsApx · x · 2026-09-11
Terray Therapeutics benchmarked NVIDIA's newly released BioNeMo Inference Runtime kernels (CuTeDSL) on its production architecture for TerraBind, its binding affinity and structure prediction model.
Key results:
- TerraBind's dominant cost is the O(n³) triangle attention operation central to modern structure-based drug discovery
- On H100 GPUs, the new kernels are up to 7.1x faster than standard PyTorch attention and up to 1.8x faster than Terray's current optimized cuEquivariance kernel stack
- The advantage grows at larger crop sizes, which matter most for protein–ligand complexes
- Integration was straightforward: kernels running and benchmarked within a day of access, module-level swaps only, no pipeline rewrites required
More from Infra
- AI chip startup Tenstorrent raises $1.5B at $6.2B valuation, more than doubling — Beth_Kindig · 2026-10-27
- Anthropic discloses up to $120.5B SpaceX compute deal, ~$3.1B monthly at full capacity — XFreeze · 2026-10-27
- Satire: swapping Raft for an LLM-style decision model tanks write throughput 99.98% — irombie · 2026-10-27
- Cadence joins the domain-specific AI model camp for chip design — karlfreund · 2026-10-27
- Qdrant builds Universal I/O to run vector search mmap code on object storage — qdrant_engine · 2026-10-27
- Nebius secures $3.5B Citi financing facility for GPU purchases and AI data centers — Beth_Kindig · 2026-10-26