Tessera: Unlocking Heterogeneous GPUs through Kernel-Granularity Disaggregation
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866915932276785152 |
|---|---|
| author | Hu, Tiancheng Qin, Jin Wang, Zheng Hu, Junhao Wang, Yuzheng Chen, Lei Shan, Yizhou Zhang, Mingxing Cao, Ting Xia, Chunwei Cui, Huimin Xie, Tao Wang, Chenxi |
| author_facet | Hu, Tiancheng Qin, Jin Wang, Zheng Hu, Junhao Wang, Yuzheng Chen, Lei Shan, Yizhou Zhang, Mingxing Cao, Ting Xia, Chunwei Cui, Huimin Xie, Tao Wang, Chenxi |
| contents | Disaggregation maps parts of an AI workload to different types of GPUs, offering a path to utilize modern heterogeneous GPU clusters. However, existing solutions operate at a coarse granularity and are tightly coupled to specific model architectures, leaving much room for performance improvement. This paper presents Tessera, the first kernel disaggregation system to improve performance and cost efficiency on heterogeneous GPUs for large model inference. Our key insight is that kernels within a single application exhibit diverse resource demands, making them the most suitable granularity for aligning computation with hardware capabilities. Tessera integrates offline analysis with online adaptation by extracting precise inter-kernel dependencies from PTX to ensure correctness, overlapping communication with computation through a pipelined execution model, and employing workload-aware scheduling with lightweight runtime adaptation. Extensive evaluations across five heterogeneous GPUs and four model architectures, scaling up to 16 GPUs, show that Tessera improves serving throughput and cost efficiency by up to 2.3x and 1.6x, respectively, compared to existing disaggregation methods, while generalizing to model architectures where prior approaches do not apply. Surprisingly, a heterogeneous GPU pair under Tessera can even exceed the throughput of two homogeneous high-end GPUs at a lower cost. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_10180 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Tessera: Unlocking Heterogeneous GPUs through Kernel-Granularity Disaggregation Hu, Tiancheng Qin, Jin Wang, Zheng Hu, Junhao Wang, Yuzheng Chen, Lei Shan, Yizhou Zhang, Mingxing Cao, Ting Xia, Chunwei Cui, Huimin Xie, Tao Wang, Chenxi Distributed, Parallel, and Cluster Computing Machine Learning Disaggregation maps parts of an AI workload to different types of GPUs, offering a path to utilize modern heterogeneous GPU clusters. However, existing solutions operate at a coarse granularity and are tightly coupled to specific model architectures, leaving much room for performance improvement. This paper presents Tessera, the first kernel disaggregation system to improve performance and cost efficiency on heterogeneous GPUs for large model inference. Our key insight is that kernels within a single application exhibit diverse resource demands, making them the most suitable granularity for aligning computation with hardware capabilities. Tessera integrates offline analysis with online adaptation by extracting precise inter-kernel dependencies from PTX to ensure correctness, overlapping communication with computation through a pipelined execution model, and employing workload-aware scheduling with lightweight runtime adaptation. Extensive evaluations across five heterogeneous GPUs and four model architectures, scaling up to 16 GPUs, show that Tessera improves serving throughput and cost efficiency by up to 2.3x and 1.6x, respectively, compared to existing disaggregation methods, while generalizing to model architectures where prior approaches do not apply. Surprisingly, a heterogeneous GPU pair under Tessera can even exceed the throughput of two homogeneous high-end GPUs at a lower cost. |
| title | Tessera: Unlocking Heterogeneous GPUs through Kernel-Granularity Disaggregation |
| topic | Distributed, Parallel, and Cluster Computing Machine Learning |
| url | https://arxiv.org/abs/2604.10180 |