NVIDIA GA10B 16SM 765MHz vs NVIDIA GA10B 28SM Full Specs

Summary

1,024 Shaders 765MHz	1,792 Shaders 930MHz
Shared Memory102.4GB/sLPDDR5	Shared Memory204.8GB/sLPDDR5
Peak·FP32·FP16 1.57 TFLOPS	Peak·FP32·FP16 3.33 TFLOPS
Form Factor iGPU	Form Factor iGPU
TDP Shared	TDP Shared

Highlights

Peak OPS

Peak OPS FP32 FP16 Tensor FP16-16 FP16-16 Tensor Sparse Tensor FP16-32 FP16-32 Tensor Sparse Tensor BF16 BF16 Tensor Sparse Tensor TF32 Tensor INT8 Tensor INT4 Memory Bandwidth

GA10B 16SM 765MHz100.3 TOPSINT4 Tensor Sparse

x1

GA10B 28SM213.3 TOPSINT4 Tensor Sparse

x2.13

Show Top Score

Theoretical Performance

Clock Speed765MHz UC·Boost·OC	Clock Speed930MHz UC·Boost·OC
Peak OPS 100.3 TOPSINT4 Tensor Sparse	Peak OPS 213.3 TOPSINT4 Tensor Sparse
Tensor FP16-16 12.53 TFLOPS FP16-16 Tensor Sparse 25.07 TFLOPS Tensor FP16-32 12.53 TFLOPS FP16-32 Tensor Sparse 25.07 TFLOPS	Tensor FP16-16 26.66 TFLOPS FP16-16 Tensor Sparse 53.33 TFLOPS Tensor FP16-32 26.66 TFLOPS FP16-32 Tensor Sparse 53.33 TFLOPS
Tensor BF16 12.53 TFLOPS BF16 Tensor Sparse 25.07 TFLOPS	Tensor BF16 26.66 TFLOPS BF16 Tensor Sparse 53.33 TFLOPS
Tensor TF32 6.27 TFLOPS	Tensor TF32 13.33 TFLOPS
FP32 1.57 TFLOPS	FP32 3.33 TFLOPS
Tensor INT4 50.14 TOPS	Tensor INT4 106.7 TOPS
Tensor INT8 25.07 TOPS	Tensor INT8 53.33 TOPS
Pixel Rate 12.2 GPixel/s	Pixel Rate 22.3 GPixel/s
Texture Rate 6.1 GTexel/s	Texture Rate 13 GTexel/s

Core Config

Shaders 1,024 Shaders	Shaders 1,792 Shaders
TMUs 8 TMUs	TMUs 14 TMUs
ROPs 16 ROPs	ROPs 24 ROPs
Tensor Cores 32 T-Cores	Tensor Cores 56 T-Cores
SMs 8 SMs	SMs 14 SMs

Clocks

Boost Clock 765MHz

Boost Clock 930MHz

Cache

L2 Cache 2MB shared

L2 Cache 4.1MB shared

Memory

Shared MemoryLPDDR5	Shared MemoryLPDDR5
Memory Bus 128-bit	Memory Bus 256-bit
Memory Speed 6.4GT/s	Memory Speed 6.4GT/s
Memory Bandwidth 102.4GB/s	Memory Bandwidth 204.8GB/s

Encoder

Encoder Model NVENC 7

Encoder Model NVENC 7

Decoder

Decoder Model NVDEC 5

Decoder Model NVDEC 5

Card

Form Factor iGPU

Form Factor iGPU

Chip

Manufacturer NVIDIA	Manufacturer NVIDIA
Chip Designer NVIDIA	Chip Designer NVIDIA
Architecture Ampere	Architecture Ampere
Family Jetson	Family Jetson
Branding	Branding
Codename Hercules	Codename Hercules
Market Segment Embedded	Market Segment Embedded
Release Date Jan 1, 2023	Release Date Oct 1, 2022

Fabrication

Foundry Samsung	Foundry Samsung
Fabrication Node 8N	Fabrication Node 8N
Transistor Count 17 Billion	Transistor Count 17 Billion

Images

No images available

No images available

Used In

CPUs Orin NX 8GB

CPUs AGX Orin 32GB

Competitors

NVIDIAGA10B 16SM 765MHz vs GPU Apple5-Core A17 Pro NVIDIAGA10B 16SM 765MHz vs GPU Apple5-Core A18 Pro NVIDIAGA10B 16SM 765MHz vs GPU Apple5-Core A19 Pro Show 12 more

NVIDIAGA10B 28SM vs GPU QualcommAdreno 840 NVIDIAGA10B 28SM vs GPU QualcommAdreno 830 NVIDIAGA10B 28SM vs GPU ARMImmortalis-G925 MP12 1300MHz Show 12 more