Semiconductor & Computer Engineering
From WikiChip
Difference between revisions of "nervana/microarchitectures/lake crest"
(→Architecture) |
|||
| Line 16: | Line 16: | ||
== Architecture == | == Architecture == | ||
| + | Lake Crest was designed from the ground up for [[deep learning]] | ||
* Tensor-based architecture | * Tensor-based architecture | ||
** Nervana Engine | ** Nervana Engine | ||
* [[Flexpoint]] number format | * [[Flexpoint]] number format | ||
* HBM2 memory | * HBM2 memory | ||
| + | ** 32 GiB of in-package memory | ||
| + | ** 8 Tbit/s bandwidth | ||
| + | * 12 bi-directional high-bandwidth direct chip-to-chip interconnect | ||
{{expand list}} | {{expand list}} | ||
Revision as of 23:41, 5 April 2018
| Edit Values | |
| Lake Crest µarch | |
| General Info | |
| Arch Type | NPU |
| Designer | Nervana |
| Manufacturer | TSMC |
| Introduction | November 17, 2016 |
| Process | 28 nm |
| Succession | |
Lake Crest is a neural processor microarchitecture designed by Nervana.
Process Technology
Lake Crest is fabricated on TSMC's 28 nm process.
Architecture
Lake Crest was designed from the ground up for deep learning
- Tensor-based architecture
- Nervana Engine
- Flexpoint number format
- HBM2 memory
- 32 GiB of in-package memory
- 8 Tbit/s bandwidth
- 12 bi-directional high-bandwidth direct chip-to-chip interconnect
This list is incomplete; you can help by expanding it.
Block Diagram
 |
This section is empty; you can help add the missing info by editing this page. |
Memory Hierarchy
- 32 GiB on-package HBM2
- 1 TiB/s
Facts about "Lake Crest - Microarchitectures - Intel Nervana"
| codename | Lake Crest + |
| designer | Nervana + |
| first launched | November 17, 2016 + |
| full page name | nervana/microarchitectures/lake crest + |
| instance of | microarchitecture + |
| manufacturer | TSMC + |
| name | Lake Crest + |
| process | 28 nm (0.028 μm, 2.8e-5 mm) + |