Enflame Yunsui i20
Overview
The Enflame (燧原科技) Yunsui i20 (云燧 i20) is an inference accelerator built on the DTU 2.5 (邃思2.5) GCU architecture on 12nm, launched December 2021 as a PCIe Gen4 x16 card. Vendor-stated dense peaks are 32 TFLOPS FP32, 128 TFLOPS TF32/FP16/BF16, and 256 TOPS INT8. Memory is 16 GB HBM2E at 819 GB/s in a 150W envelope; no scale-out interconnect fabric was published for the inference part, so interconnect fields are NULL.
Performance
Peak theoretical throughput by precision type
| Precision | Peak |
|---|---|
FP64 | No verified data available |
FP32 32-bit floating point | 32TFLOPS |
TF32 TensorFloat-32 | 128TFLOPS |
BF16 Brain Float 16 | 128TFLOPS |
FP16 16-bit floating point | 128TFLOPS |
FP8 | No verified data available |
FP6 | No verified data available |
FP4 | No verified data available |
INT8 8-bit integer | 256TOPS |
Dense peak divided by accelerator TDP. Board power only: excludes host CPUs, networking, cooling and facility overhead. TDP for this part: 150 W.
The Yunsui i20 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
GCU / DTU 2.5
Form Factor
PCIe
Launch Year
2021
Process Node
12nm
Memory
16 GB HBM2E
Bandwidth
819 GB/s
TDP
150 W
Max power (Flopper estimate)
~173 W est. Flopper estimate: 150 W TDP x 1.15. The vendor publishes no maximum board power for this part.
Spec Confidence
Vendor claimed
Full Specifications
| Memory | |
|---|---|
| Memory | 16 GB |
| Memory Type | HBM2E |
| Bandwidth | 819 GB/s |
| Interface Width | No verified data available |
| Power & Thermal | |
| TDP | 150 W |
| Max power (Flopper estimate) | ~173 W est. |
| General | |
| Form Factor | PCIe |
| Architecture | GCU / DTU 2.5 |
| Process Node | 12nm |
| Launch Year | 2021 |
Datasheet & Resources
Flopper Datasheet
Enflame Yunsui i20 specifications, generated from our database. Printable.
Vendor product page
Enflame documentation for the Yunsui i20
Enflame DTU 2.5 / Yunsui i20 inference accelerator
ChinaAET · 2021-12-20
Data Provenance
Every figure traced to a source
Primary Source
- Publisher
- ChinaAET
- Published
- 2021-12-20
Data Quality
- Spec confidence
- Vendor claimed
- Clock basis
- Boost
- Core precisions with figures
- 5 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 ProvidersSystems Using This GPU
Pre-configured systems featuring the Enflame Yunsui i20
| System | GPU Count | Peak Performance | Total Power | |
|---|---|---|---|---|
Enflame Cloud Blazer POD P2110 Cloud Blazer POD · rack · | 64x Yunsui i20 | No verified data available | No verified data available | View System |
Frequently Asked Questions
How many TFLOPS does the Enflame Yunsui i20 have?
The Enflame Yunsui i20 delivers 32 TFLOPS FP32 and 128 TFLOPS FP16 at peak. Flopper does not currently have a verified FP8 throughput figure for it.
What is the power consumption of the Enflame Yunsui i20?
The Enflame Yunsui i20 has a TDP (Thermal Design Power) rating of 150 watts.
How much memory does the Enflame Yunsui i20 have?
The Enflame Yunsui i20 is equipped with 16 GB of memory with 819 GB/s of memory bandwidth.
What architecture is the Enflame Yunsui i20 based on?
The Enflame Yunsui i20 is based on the GCU / DTU 2.5 architecture, launched in 2021.
Stay Updated on GPU Releases
Get notified when new GPUs are added or specifications are updated.
No spam, unsubscribe anytime.