Papers › Glinthawk: A Two-Tiered Architecture for Offline LLM Inference

Glinthawk: A Two-Tiered Architecture for Offline LLM Inference

20 Jan 2025arXiv:2501.11779archive 2025-07-28

Pouya Hamadanian, Sadjad Fouladi

We introduce Glinthawk, an architecture for offline Large Language Model (LLM) inference. By leveraging a two-tiered structure, Glinthawk optimizes the utilization of the high-end accelerators ("Tier 1") by offloading the attention mechanism to lower-end compute tier ("Tier 2"). This separation allows the memory demand of the attention, known as the key-value cache, to scale independently from the model weights, enabling larger batch sizes and more efficient accelerator usage. Prototyped with NVIDIA T4 GPUs and standard CPU VMs, Glinthawk improves throughput by 5.9× and reduces cost of generation by 2.8×, compared to paged attention baselines. For long sequence lengths, it achieves 16.3× throughput improvement at 2.4× less cost. Our evaluation shows that this architecture can tolerate moderate network latency with minimal performance degradation, making it highly effective for latency-tolerant, throughput-focused applications such as batch processing. The prototype is publicly available at https://github.com/microsoft/glinthawk.

PaperPDFCode

Code

microsoft/glinthawk officialmentioned in paper report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Language ModelingLanguage ModellingLarge Language Model

1 archive task tag without a task page not shown.

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

Absolute Position EncodingsAdamAttentionBPEDense ConnectionsDropoutLabel SmoothingLayer NormalizationLinear LayerMulti-Head AttentionPosition-Wise Feed-Forward LayerResidual ConnectionSoftmaxTransformer

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections