GPU

AMD Radeon AI PRO R9700

Edit@17 days ago

Intergrated Memory(VRAM)
Capacity

32 GB

(GDDR6 256-bit)

Bandwidth

644 GB/s

92 Token/s

Vector Compute
FP64
1.50 T
FP32
47.84 T
FP16
95.68 T
BF16
X
INT32
X
INT8
X

AMD Radeon AI PRO R9700 General-Purpose Float-Point performance (Vector Performance / Scalar Performance)

FP64: 1.50 TFLOPS

FP32: 47.84 TFLOPS

FP16: 95.68 TFLOPS

Matirx Compute
FP64
X
FP32
X
FP16
191.36 T
382.72 T
FP8
382.72 T
765.44 T
TF32
X
BF16
191.36 T
382.72 T
INT16
X
INT8
382.75 T
765.50 T
INT4
765.50 T
1531 T

AMD Radeon AI PRO R9700 AI performance (Tensor Performance / Matrix Performance)

FP16: 191.36 TFLOPS, with sparsity: 382.72 TFLOPS

FP8: 382.72 TFLOPS, with sparsity: 765.44 TFLOPS

BF16: 191.36 TFLOPS, with sparsity: 382.72 TFLOPS

INT8: 382.75 TOPS, with sparsity: 765.50 TOPS

INT4: 765.50 TOPS, with sparsity: 1531 TOPS

Hardware Specs
AMD Radeon AI PRO R9700 is a 4nm chip, has 53900 million transistors, launched by AMD at 2025. It has 32 GB built-in(On-Board/On-Chip) memory with bandwidth up to 644 GB/s. It has 4096 general-purpose ALUs(CUDA cores/Shader cores) and 128 matrix cores(Tensor cores) .
Process Node
4 nm
Launch Year
2025
Vector(CUDA) Cores
4096
Matrix(Tensor) Cores
128
Core Frequency
1660 ~ 2920 MHz
Cache
64MB