NVIDIA GA10B 8SM vs Qualcomm Adreno 722 Full Specs
512 Shaders 625MHz | 512 Shaders 450MHz |
Shared Memory34.1GB/sLPDDR5 | Shared Memory33.6GB/sLPDDR5X |
·· 640 GFLOPS | ·· 460.8 GFLOPS |
Form Factor iGPU | Form Factor iGPU |
TDP Shared | TDP 5W |
GA10B 8SM40.96 TOPSINT4 Tensor Sparse
Adreno 722921.6 GFLOPSFP16Clock Speed ·· | Clock Speed ·· |
Peak OPS 40.96 TOPSINT4 Tensor Sparse | Peak OPS 921.6 GFLOPSFP16 |
Tensor FP16-16 5.12 TFLOPSFP16-16 Tensor Sparse 10.24 TFLOPSTensor FP16-32 5.12 TFLOPSFP16-32 Tensor Sparse 10.24 TFLOPS | - |
Tensor BF16 5.12 TFLOPSBF16 Tensor Sparse 10.24 TFLOPS | - |
Tensor TF32 2.56 TFLOPS | Tensor TF32 - |
FP32 640 GFLOPS | FP32 460.8 GFLOPS |
- | FP64 115.2 GFLOPS |
Tensor INT4 20.48 TOPS | Tensor INT4 - |
Tensor INT8 10.24 TOPS | - |
Pixel Rate 5 GPixel/s | Pixel Rate 7.2 GPixel/s |
Texture Rate 2.5 GTexel/s | Texture Rate 14.4 GTexel/s |
Shaders 512 Shaders | Shaders 512 Shaders |
TMUs 4 TMUs | TMUs 32 TMUs |
ROPs 8 ROPs | ROPs 16 ROPs |
Tensor Cores 16 T-Cores | Tensor Cores - |
SMs 4 SMs | EUs 4 EUs |
Boost Clock 625MHz | Boost Clock 450MHz |
L2 Cache 1MB shared | L2 Cache - |
Shared MemoryLPDDR5 | Shared MemoryLPDDR5X |
Memory Bus 64-bit | Memory Bus 32-bit |
Memory Speed 4.3GT/s | Memory Speed 8.4GT/s |
Memory Bandwidth 34.1GB/s | Memory Bandwidth 33.6GB/s |
TDP Shared | TDP 5W |
Multi-Monitor - | Multi-Monitor 1 |
Encoder Model NVENC 7 | Encoder Model Hexagon |
Decoder Model NVDEC 5 | Decoder Model Hexagon |
Form Factor iGPU | Form Factor iGPU |
Manufacturer | Manufacturer |
Chip Designer | Chip Designer |
Architecture | Architecture |
Family | Family |
Branding  | Branding  |
Codename Hercules | Codename - |
Market Segment Embedded | Market Segment Smartphone |
Release Date Jan 1, 2023 | Release Date May 15, 2025 |
Foundry Samsung | Foundry TSMC |
Fabrication Node 8N | Fabrication Node N6 |
Transistor Count 17 Billion | Transistor Count - |
No images available
No images available
CPUs | CPUs |