AMD Instinct MI100 vs NVIDIA GeForce RTX 5070 Full Specs

Summary

7,680 Shaders 1.5GHz	6,144 Shaders 2.51GHz
32GB HBM21.23TB/s	12GB GDDR7672GB/s
Peak·FP32·FP16 23.07 TFLOPS	Peak·FP32·FP16 30.87 TFLOPS
Form Factor PCIe Card	Form Factor PCIe Card
TDP 300W	TDP 250W
Power Connectors - 2x 8-Pin	Power Connectors 1x 16-Pin 12VHPWR -

Highlights

Peak OPS

Peak OPS FP32 FP16 Tensor FP4 FP4 Tensor Sparse Tensor FP8-16 FP8-16 Tensor Sparse Tensor FP8-32 FP8-32 Tensor Sparse Tensor FP16-16 FP16-16 Tensor Sparse Tensor FP16-32 FP16-32 Tensor Sparse Tensor FP32 FP64 BF16 Tensor BF16 BF16 Tensor Sparse Tensor TF32 Tensor INT8 Tensor INT4 Memory Bandwidth TDP

Instinct MI100184.6 TFLOPSFP16-16 Tensor

x1

GeForce RTX 5070987.8 TFLOPSFP4 Tensor Sparse

x5.35

Show Top Score

Theoretical Performance

Clock Speed1502MHz UC·Base·Boost·OC	Clock Speed2512MHz UC·Base·Boost·OC
Peak OPS 184.6 TFLOPSFP16-16 Tensor	Peak OPS 987.8 TFLOPSFP4 Tensor Sparse
-	Tensor FP4 493.9 TFLOPS FP4 Tensor Sparse 987.8 TFLOPS
-	Tensor FP8-16 246.9 TFLOPS FP8-16 Tensor Sparse 493.9 TFLOPS Tensor FP8-32 123.5 TFLOPS FP8-32 Tensor Sparse 246.9 TFLOPS
Tensor FP16-16 184.6 TFLOPS FP16-16 Tensor Sparse - Tensor FP16-32 184.6 TFLOPS FP16-32 Tensor Sparse -	Tensor FP16-16 123.5 TFLOPS FP16-16 Tensor Sparse 246.9 TFLOPS Tensor FP16-32 61.73 TFLOPS FP16-32 Tensor Sparse 123.5 TFLOPS
BF16 - Tensor BF16 92.28 TFLOPS BF16 Tensor Sparse -	BF16 30.87 TFLOPS Tensor BF16 61.73 TFLOPS BF16 Tensor Sparse 123.5 TFLOPS
Tensor TF32 -	Tensor TF32 30.87 TFLOPS
FP32 23.07 TFLOPS Tensor FP32 46.14 TFLOPS	FP32 30.87 TFLOPS Tensor FP32 -
FP64 11.54 TFLOPS	FP64 482.3 GFLOPS
Tensor INT4 92.28 TOPS	Tensor INT4 -
Tensor INT8 92.28 TOPS	Tensor INT8 246.9 TOPS
Ray -	Ray 93.62 TOPS
Pixel Rate -	Pixel Rate 201 GPixel/s
Texture Rate 721 GTexel/s	Texture Rate 482.3 GTexel/s

Core Config

Shaders 7,680 Shaders	Shaders 6,144 Shaders
TMUs 480 TMUs	TMUs 192 TMUs
ROPs -	ROPs 80 ROPs
Tensor Cores 480 T-Cores	Tensor Cores 192 T-Cores
RT Cores -	RT Cores 48 RT-Cores
CUs 120 CUs	SMs 48 SMs

Clocks

Base Clock 1GHz	Base Clock 2.16GHz
Boost Clock 1.5GHz	Boost Clock 2.51GHz

Cache

L2 Cache 8.2MB shared

L2 Cache 41MB shared

Memory

32GB HBM2	12GB GDDR7
Memory Bus 4096-bit	Memory Bus 192-bit
Memory Speed 2.4GT/s	Memory Speed 28GT/s
Memory Bandwidth 1.23TB/s	Memory Bandwidth 672GB/s
ECC Yes	ECC No

Power & Thermals

TDP 300W	TDP 250W
Max Temp -	Max Temp 85°C Max

Video Output

Multi-Monitor -	Multi-Monitor 4
HDCP -	HDCP HDCP 2.3

Ports

-	3x DisplayPort 2.1
-	1x HDMI 2.1

Encoder

Encoder Model VCN 2.5

Encoder Model NVENC 9

Decoder

Decoder Model VCN 2.5

Decoder Model NVDEC 6

Card

Form Factor PCIe Card	Form Factor PCIe Card
PCIe 2-Slots	PCIe 2.1-Slots
Height 111 mm (4.37") Width 267 mm (10.51") Depth 37 mm (1.46")	Height 112 mm (4.41") Width 242 mm (9.53") Depth 42 mm (1.65")
Cooling Passive -	Cooling Open-Air 2x Fans
Power Connectors 2x 8-Pin	Power Connectors 1x 16-Pin 12VHPWR

Chip

Manufacturer AMD	Manufacturer NVIDIA
Chip Designer AMD	Chip Designer NVIDIA
Architecture CDNA 1	Architecture Blackwell
Family Instinct	Family GeForce 50
Branding	Branding
Codename Arcturus	Codename NV195
Chip Variant Arcturus XL	Chip Variant GB205-300-A1
Market Segment Server	Market Segment Desktop
Release Date Nov 16, 2020	Release Date Jan 6, 2025

Fabrication

Foundry TSMC	Foundry TSMC
Fabrication Node N7	Fabrication Node 4NP
Die Size 750mm²	Die Size 263mm²
Transistor Count 25.6 Billion	Transistor Count 31.1 Billion
Transistor Density 34.13 MTr/mm²	Transistor Density 118.3 MTr/mm²

Images

No images available

No images available

Competitors

AMDInstinct MI100 vs GPU NVIDIAGeForce RTX 5070 Ti AMDInstinct MI100 vs GPU NVIDIARTX PRO 4000 Blackwell SFF AMDInstinct MI100 vs GPU NVIDIARTX PRO 4500 Blackwell Show 12 more

NVIDIAGeForce RTX 5070 vs GPU AMDRadeon RX 9070 NVIDIAGeForce RTX 5070 vs GPU AMDRadeon RX 9070 XT NVIDIAGeForce RTX 5070 vs GPU NVIDIAGeForce RTX 5070 Ti Show 12 more