Iluvatar TianGai 100 (BI-V100)
Product Overview
TianGai 100 (model BI-V100) is Iluvatar's first domestic fully self-developed general-purpose GPU training accelerator officially released in March 2021, using 7nm process, equipped with 32GB HBM2 memory, board-level power consumption 250W, supporting FP32/FP16/INT8 mixed-precision training, compatible with international mainstream GPU general computing models, supporting PyTorch/TensorFlow and other mainstream AI frameworks, is the founding work of Iluvatar's "TianGai" training product series.
Strategic Significance: Achieved a major breakthrough from 0 to 1 for domestic general-purpose GPU products, filling the gap in domestic cloud training GPUs.
Core Specifications
| Item | Parameter |
|---|
| Architecture | Iluvatar self-developed general-purpose GPU architecture |
| Process | 7nm (estimated TSMC) |
| FP16 | 128 TFLOPS |
| INT8 | 256 TOPS |
| INT32 | 32 TFLOPS |
| Memory Capacity | 32 GB HBM2 |
| Memory Bandwidth | Not disclosed (estimated ~1 TB/s) |
| TDP | 250 W (board-level power consumption) |
| Interface | PCIe Gen4.0 x16 |
| Inter-chip Interconnect | 64 GB/s bidirectional bandwidth |
| Cooling | Passive cooling |
| Dimensions | Full-length full-height dual-slot PCIe card |
| Release | March 2021 |
| Mass Production | Since 2021 |
| Software Stack | Iluvatar computing software stack (PyTorch/TensorFlow compatible) |
⚠️ Specification Note: Memory bandwidth not fully disclosed by official sources, subject to Iluvatar's subsequent official data sheet.
TianGai Series Product Line
| Product | Release | FP16 TFLOPS | Memory | Status |
|---|
| TianGai 100 (BI-V100) | 2021 | 128 TFLOPS | 32GB HBM2 | On sale |
| TianGai 150 (BI-V150) | 2023 | Not disclosed (estimated higher) | Not disclosed | On sale |
| TongYang TY1000 | 2024+ | Not disclosed | Not disclosed | Next generation |
Software Ecosystem
| Layer | Tool | Description |
|---|
| AI Framework | PyTorch / TensorFlow | Native compatibility |
| Programming Language | CUDA C++ / OpenCL | Supports mainstream programming models |
| Operator Library | Iluvatar computing software stack | Native operators + custom operators |
| Cluster Training | Supports distributed training | Multi-card interconnect |
Application Scenarios
- ✅ Domestic large model training (below 100 billion parameters)
- ✅ AI framework migration (CUDA programming model compatible)
- ✅ Government/state-owned enterprise AI projects (supply chain security)
- ✅ Scientific computing (FP32/INT32 support)
- ❌ Ultra-high compute requirements (FP16 128 TFLOPS lower than H100)
- ❌ Emerging FP8 precision (FP8 not supported)
Comparison with ZhiKai 100 (MR100)
| Metric | TianGai 100 (BI-V100) | ZhiKai 100 (MR100) | Difference |
|---|
| Positioning | Training (Training) | Inference (Inference) | Different scenarios |
| FP16 | 128 TFLOPS | 96 TFLOPS | TianGai stronger |
| INT8 | 256 TOPS | 192 TOPS | TianGai stronger |
| Video Decoding | Not supported | 128-channel 1080P | ZhiKai exclusive |
| TDP | 250W | Estimated 250-300W | Similar |
References