Huawei Atlas 200I A2 20 TOPS
Overview
The Huawei Atlas 200I A2 is an edge acceleration module built on the Ascend 310B, measuring just 82 by 60 by 7 millimetres and intended to be soldered down into a customer board. This is the higher of the two bins Huawei publishes, rated at 20 TOPS INT8 and 10 TFLOPS FP16 within 25 watts. Those figures reconcile almost exactly with Huawei's own hardware documentation: a single Da Vinci V300 AI core running at 1.224 GHz across 4,096 FP16 MACs gives 10.02 TFLOPS, and the same core across 8,192 INT8 MACs gives 20.05 TOPS, a 0.3 percent agreement between a 2019 conference deck and a 2023 driver guide. The module pairs the AI core with four TaiShan V200M CPU cores at 1.6 GHz and offers LPDDR4X in 12, 8 or 4 GB configurations at 51.2, 34.1 and 34.1 GB/s respectively; note that the bandwidth ladder is not monotonic in Huawei's own table, with 8 GB and 4 GB sharing a figure. Media support is 40 channels of 1080p30 decode.
Performance
Peak theoretical throughput by precision type
| Precision | Peak |
|---|---|
FP64 | No verified data available |
FP32 | No verified data available |
TF32 | No verified data available |
BF16 | No verified data available |
FP16 16-bit floating point | 10TFLOPS |
FP8 | No verified data available |
FP6 | No verified data available |
FP4 | No verified data available |
INT8 8-bit integer | 20TOPS |
Dense peak divided by accelerator TDP. Board power only: excludes host CPUs, networking, cooling and facility overhead. TDP for this part: 25 W.
The Atlas 200I A2 20 TOPS in the GPU landscape
Peak FP16 TFLOPS (dense) against TDP, single-GPU parts tracked by Flopper
Higher and further left is better: more half-precision throughput for less power.
Specifications
Architecture
Da Vinci
Form Factor
Module
Launch Year
2023
Process Node
No verified data available
Memory
12 GB LPDDR4X
Bandwidth
51.2 GB/s
TDP
25 W
Max power (Flopper estimate)
~29 W est. Flopper estimate: 25 W TDP x 1.15. The vendor publishes no maximum board power for this part.
Da Vinci AI Core
1
Spec Confidence
Official
Full Specifications
| Compute Engine | |
|---|---|
| Da Vinci AI Core | 1 |
| Boost Clock | 1.22 GHz |
| Memory | |
| Memory | 12 GB |
| Memory Type | LPDDR4X |
| Bandwidth | 51 GB/s |
| Interface Width | No verified data available |
| Power & Thermal | |
| TDP | 25 W |
| Max power (Flopper estimate) | ~29 W est. |
| Enterprise Features | |
| Compute APIs | CANN, MindSpore, MindX, ACL |
| Physical & Media | |
| Card Length | 82 mm |
| General | |
| Form Factor | Module |
| Architecture | Da Vinci |
| Process Node | No verified data available |
| Launch Year | 2023 |
Datasheet & Resources
Flopper Datasheet
Huawei Atlas 200I A2 20 TOPS specifications, generated from our database. Printable.
Vendor product page
Huawei documentation for the Atlas 200I A2 20 TOPS
Atlas 200I A2 加速模块 规格 (20/8 TOPS INT8, 10/4 TFLOPS FP16)
Huawei · Latest version
Huawei Ascend Developer Documentation
CANN guides, MindSpore, Da Vinci
Data Provenance
Every figure traced to a source
Primary Source
- Publisher
- Huawei
- Published
- No verified data available
Data Quality
- Spec confidence
- Official
- Clock basis
- Boost
- Core precisions with figures
- 2 of 9
- Normalization
- All values in TFLOPS
Compare cloud providers offering on-demand GPU instances for AI training, inference, and HPC workloads.
Browse GPU Cloud ProvidersSimilar GPUs
Frequently Asked Questions
How many TFLOPS does the Huawei Atlas 200I A2 have?
The Huawei Atlas 200I A2 delivers 10 TFLOPS FP16 at peak. Flopper does not currently have verified FP32 and FP8 throughput figures for it.
What is the power consumption of the Huawei Atlas 200I A2?
The Huawei Atlas 200I A2 has a TDP (Thermal Design Power) rating of 25 watts.
How much memory does the Huawei Atlas 200I A2 have?
The Huawei Atlas 200I A2 is equipped with 12 GB of memory with 51.2 GB/s of memory bandwidth.
What architecture is the Huawei Atlas 200I A2 based on?
The Huawei Atlas 200I A2 is based on the Da Vinci architecture, launched in 2023.
Stay Updated on GPU Releases
Get notified when new GPUs are added or specifications are updated.
No spam, unsubscribe anytime.