Inferbase
Beta
AMD Instinct MI210
Datacenter
The MI210 brings CDNA 2 architecture to PCIe form factor with 64GB HBM2e memory. A practical choice for datacenter deployments without OAM infrastructure requirements.
VRAM
64 GB
Memory
HBM2e
Bandwidth
1638 GB/s
TDP
300W
Suitable Workloads
Medium Language Models
Inference for models up to 70B parameters
Enterprise Deployment
Designed for 24/7 datacenter operations
Key Highlights
- Built on CDNA 2 architecture
- 416 AI accelerator cores for tensor operations
- 6,656 compute cores for parallel processing
- PCIe form factor
Memory
VRAM64 GB
Memory TypeHBM2e
Memory Bandwidth1638 GB/s
Compute Performance
FP32 (Single Precision)22.6 TFLOPS
FP16 (Half Precision)181 TFLOPS
BF16181 TFLOPS
INT8181 TOPS
Architecture
ArchitectureCDNA 2
Compute Cores6,656
AI Accelerators416
Release Year2022
Power & Physical
TDP300 W
Max Power300 W
Form FactorPCIe
PCIe GenerationGen4 x16
Notes
Single-GCD PCIe card for mainstream datacenter deployment.
Models That May Fit on AMD Instinct MI210
Estimates based on INT8 quantization. Actual fit depends on framework and batch size.
Related Tools
Added Jan 27, 2026
Last updated: Jan 27, 2026
Build Your AI Stack with Confidence
Explore models, compare pricing and benchmarks, and right-size your infrastructure — all in one place.