NVIDIA T4 vs NVIDIA V100 Full Specs

Summary

2,560 Shaders 1.59GHz	5,120 Shaders 1.38GHz
16GB GDDR6160GB/s	32GB HBM2896GB/s
Peak·FP32·FP16 8.14 TFLOPS	Peak·FP32·FP16 14.13 TFLOPS
Form Factor PCIe Card	Form Factor PCIe Card
TDP 70W	TDP 250W
Power Connectors -	Power Connectors 2x 8-Pin

Highlights

Peak OPS

Peak OPS FP32 FP16 Tensor FP16-16 Tensor FP16-32 FP64 Tensor BF16 Tensor TF32 Tensor INT8 Tensor INT4 Memory Bandwidth TDP

T4260.5 TOPSINT4 Tensor

x2.30

V100113 TFLOPSFP16-16 Tensor

x1

Show Top Score

Theoretical Performance

Clock Speed1590MHz UC·Base·Boost·OC	Clock Speed1380MHz UC·Base·Boost·OC
Peak OPS 260.5 TOPSINT4 Tensor	Peak OPS 113 TFLOPSFP16-16 Tensor
Tensor FP16-16 65.13 TFLOPS Tensor FP16-32 32.56 TFLOPS	Tensor FP16-16 113 TFLOPS Tensor FP16-32 113 TFLOPS
Tensor BF16 32.56 TFLOPS	-
Tensor TF32 32.56 TFLOPS	Tensor TF32 -
FP32 8.14 TFLOPS	FP32 14.13 TFLOPS
FP64 254.4 GFLOPS	FP64 7.07 TFLOPS
Tensor INT4 260.5 TOPS	Tensor INT4 -
Tensor INT8 130.3 TOPS	-
Ray 24.56 TOPS	Ray -
Pixel Rate 101.8 GPixel/s	Pixel Rate 176.6 GPixel/s
Texture Rate 254.4 GTexel/s	Texture Rate 441.6 GTexel/s

Core Config

Shaders 2,560 Shaders	Shaders 5,120 Shaders
TMUs 160 TMUs	TMUs 320 TMUs
ROPs 64 ROPs	ROPs 128 ROPs
Tensor Cores 320 T-Cores	Tensor Cores 640 T-Cores
RT Cores 40 RT-Cores	RT Cores -
SMs 40 SMs	SMs 80 SMs

Clocks

Base Clock 585MHz	Base Clock 1.23GHz
Boost Clock 1.59GHz	Boost Clock 1.38GHz

Cache

L2 Cache 4.1MB shared

L2 Cache 6.1MB shared

Memory

16GB GDDR6	32GB HBM2
Memory Bus 256-bit	Memory Bus 4096-bit
Memory Speed 5GT/s	Memory Speed 1.8GT/s
Memory Bandwidth 160GB/s	Memory Bandwidth 896GB/s

Power & Thermals

TDP 70W

TDP 250W

Encoder

Encoder Model NVENC 7

Encoder Model 3x NVENC 5

Decoder

Decoder Model NVDEC 4

Decoder Model NVDEC 3

Card

Form Factor PCIe Card	Form Factor PCIe Card
PCIe 1-Slots	PCIe 2-Slots
Height 69 mm (2.72") Width 168 mm (6.61") Depth 20 mm (0.79")	Height 111 mm (4.37") Width 267 mm (10.51") Depth 40 mm (1.57")
Cooling Passive	Cooling Passive
Power Connectors -	Power Connectors 2x 8-Pin

Chip

Manufacturer NVIDIA	Manufacturer NVIDIA
Chip Designer NVIDIA	Chip Designer NVIDIA
Architecture Turing	Architecture Volta
Family Server	Family Server
Branding	Branding
Codename NV164	Codename NV140
Chip Variant TU104-150-KC-A1	Chip Variant GV100-400-A1
Market Segment Server	Market Segment Server
Release Date Sep 13, 2018	Release Date Mar 27, 2018

Fabrication

Foundry TSMC	Foundry TSMC
Fabrication Node 12FFN	Fabrication Node 12FFN
Die Size 545mm²	Die Size 815mm²
Transistor Count 13.6 Billion	Transistor Count 21.1 Billion
Transistor Density 24.95 MTr/mm²	Transistor Density 25.89 MTr/mm²

Images

No images available

No images available

Competitors

NVIDIAT4 vs GPU AMDRadeon RX 7400 NVIDIAT4 vs GPU AMDRadeon PRO W7400 NVIDIAT4 vs GPU GIGABYTEGeForce RTX 3050 OC Low Profile 6G Show 12 more

NVIDIAV100 vs GPU AMDRadeon AI PRO R9700 NVIDIAV100 vs GPU AMDRadeon RX 9070 XT NVIDIAV100 vs GPU AMDRadeon RX 9070 Show 12 more