Huawei Atlas · Accelerator Card
Atlas 150
The Atlas 150 accelerator card is powered by the Ascend 950PR. Through chip-architecture upgrades it delivers strong AI inference with high compute and large memory, supports SIMT and SIMD programming, and improves both performance and usability. It offers 1.56 PFLOPS @mxFP4 and flexible deployment for LLM inference and multimodal generation.
Key Highlights
- Strong compute: industry-leading 1561 TFLOPS @mxFP4
- SIMT support improves logic/memory-disjoint operator performance and developer ease
- HiF8/mxFP8/mxFP4 formats boost quantization. High bandwidth: 1.4TB/s memory bandwidth, PCIe 5.0, and up to 424GB/s card-to-card Lingqu interconnect.
Technical Specifications
| NPU Processor | Ascend 950PR |
|---|---|
| AI Compute | 1561 TFLOPS mxFP4, 804 TFLOPS mxFP8/HiF8/mxFP6, 804 TOPS INT8, 425 TFLOPS FP16/BF16 |
| Memory | 84GB HBM, 1.4TB/s, ECC |
| Interconnect | PCIe 5.0 x16 (128GBps bidirectional), Lingqu 2/4-card interconnect up to 424GB/s bidirectional |
| Max Power | <=600W |
| Cooling | passive |
| Form Factor | OH FL DW PCIe standard card |
| Dimensions | 295x39.04x137mm |
| Weight | 3.0kg |
| Operating Temp | 5C~35C |
Typical Use Cases
Multimodal generation, LLM inference, internet recommendation.
英文站hiascend.com无Atlas 150独立英文产品页（/en路径均重定向en-sg首页），英文文案为翻译稿；官网无独立PDF彩页/技术白皮书链接；img1: 视角1单机正面图；卡间灵衢互联分2卡/4卡两种模式，带宽见规格。
Related Products
Huawei Atlas
Atlas 200I DK A2 Developer Kit
Easy-to-use AI development kit for fast innovation View details →
Huawei Atlas
Atlas 200I A2 Acceleration Module
Edge AI accelerator module embedded in smart devices, robots and drones View details →
Huawei Atlas
Atlas 300I A2 Inference Card
High-performance AI inference card integrated into servers View details →
Huawei Atlas