Guardado en:
| Autores principales: | , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | https://arxiv.org/abs/2510.15893 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866914100479524864 |
|---|---|
| author | Bernadskiy, Mikhail Carson, Peter Graham, Thomas Groves, Taylor Lee, Ho John Yeh, Eric |
| author_facet | Bernadskiy, Mikhail Carson, Peter Graham, Thomas Groves, Taylor Lee, Ho John Yeh, Eric |
| contents | The unabated growth in AI workload demands is driving the need for concerted advances in compute, memory, and interconnect performance. As traditional semiconductor scaling slows, high-speed interconnects have emerged as the new scaling engine, enabling the creation of larger logical GPUs by linking many GPUs into a single, low-latency, high-bandwidth compute domain. While initial scale-up fabrics leveraged copper interconnects for their power and cost advantages, the maximum reach of passive electrical interconnects (approximately 1 meter) effectively limits the scale-up domain to within a single rack. The advent of 3D-stacked optics and logic offers a transformative, power-efficient scale-up solution for connecting hundreds of GPU packages (thousands of GPUs) across multiple data center racks. This work explores the design tradeoffs of scale-up technologies and demonstrates how frontier LLMs necessitate novel photonic solutions to achieve aggressive power and performance targets. We model the benefits of 3D CPO (Passage) enabled GPUs and switches within the scale-up domain when training Frontier Mixture of Experts (MoE) models exceeding one trillion parameters. Our results show that the substantial increases in bandwidth and radix enabled by 3D CPO allow for an 8X increase in scale-up capability. This affords new opportunities for multi-dimensional parallelism within the scale-up domain and results in a 2.7X reduction in time-to-train, unlocking unprecedented model scaling. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_15893 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Accelerating Frontier MoE Training with 3D Integrated Optics Bernadskiy, Mikhail Carson, Peter Graham, Thomas Groves, Taylor Lee, Ho John Yeh, Eric Hardware Architecture Artificial Intelligence Distributed, Parallel, and Cluster Computing Machine Learning 68M10, 68M14 B.4.3; C.2.4; C.4; I.2 The unabated growth in AI workload demands is driving the need for concerted advances in compute, memory, and interconnect performance. As traditional semiconductor scaling slows, high-speed interconnects have emerged as the new scaling engine, enabling the creation of larger logical GPUs by linking many GPUs into a single, low-latency, high-bandwidth compute domain. While initial scale-up fabrics leveraged copper interconnects for their power and cost advantages, the maximum reach of passive electrical interconnects (approximately 1 meter) effectively limits the scale-up domain to within a single rack. The advent of 3D-stacked optics and logic offers a transformative, power-efficient scale-up solution for connecting hundreds of GPU packages (thousands of GPUs) across multiple data center racks. This work explores the design tradeoffs of scale-up technologies and demonstrates how frontier LLMs necessitate novel photonic solutions to achieve aggressive power and performance targets. We model the benefits of 3D CPO (Passage) enabled GPUs and switches within the scale-up domain when training Frontier Mixture of Experts (MoE) models exceeding one trillion parameters. Our results show that the substantial increases in bandwidth and radix enabled by 3D CPO allow for an 8X increase in scale-up capability. This affords new opportunities for multi-dimensional parallelism within the scale-up domain and results in a 2.7X reduction in time-to-train, unlocking unprecedented model scaling. |
| title | Accelerating Frontier MoE Training with 3D Integrated Optics |
| topic | Hardware Architecture Artificial Intelligence Distributed, Parallel, and Cluster Computing Machine Learning 68M10, 68M14 B.4.3; C.2.4; C.4; I.2 |
| url | https://arxiv.org/abs/2510.15893 |