RTX 2060 Vs GTX 1080Ti Deep Learning Benchmarks: Cheapest RTX card Vs Most Expensive GTX card | by Eric Perbos-Brinck
Training time comparison for 2060 and 1080Ti using the CIFAR-10 and CIFAR-100 datasets with fast.ai and PyTorch libraries.
TLDR #1: despite half its VRAM, and half its retail price, the RTX 2060 can blast past the 1080Ti in Computer Vision, once its Tensor Cores are activated with ‘FP16’ code in PyTorch + Fastai.
Less than a year ago, with its GP102 chip + 3584 CUDA Cores + 11GB of VRAM, the GTX 1080Ti was the apex GPU of last-gen Nvidia Pascal range (bar the Titan editions).
The demand was so high that retail prices often exceeded $900, way above the official $699 MSRP.
In Fall 2018, Nvidia launched its newest Turing line-up, named “RTX”, with Raytracing Cores and Tensor Cores. Prices as a whole jumped significantly: for example, the RTX 2080Ti retails for $1,150 and more.
One key feature for Machine Learning in the Turing / RTX range is the Tensor Core: according to Nvidia, this enables computation running in “Floating Point 16”, instead of the regular “Floating Point 32″, and cut down the time for training a Deep Learning model by up to 50%.
About a month ago (Jan 7, 2019), Nvidia released the cheapest GPU of the Turing line-up: the RTX 2060.
Using Jupyter notebooks, I trained ResNet models 18 to 152 on each Cifar dataset with FP32 then FP16, to compare the time required for 30 epochs.
With Fastai, switching from FP32 to FP16 training is as simple as adding `.to_fp16()`at the end of your regular code.
- The regular FP32 version, with a pre-trained Resnet 18 model:
learn = create_cnn(data, models.resnet18, metrics = accuracy)
- The FP16 version:
learn = create_cnn(data, models.resnet18, metrics = accuracy). to_fp16()
That’s it and you now have access to the RTX Tensor Cores !
Note: for more information on training with “FP16”, also known as “Mixed-Precision-Training (MPT), check those excellent posts.
- Mixed-Precision-Training by Sylvain Gugger (Fastai)
- Mixed-Precision-Training benchmarks with RTX 2080Ti & 1080Ti by Sanyam Bhutani (aka @init_27)
Note: ‘bs * 2’ indicates a batch_size twice larger, as in 256 vs. 128.
Hardware: I use a “real-life” high-end gaming PC with the following specs
- AMD Ryzen 7 1700X 3.4GHz 8-cores
- MSI X370 Krait Gaming motherboard
- 32 GB DDR4–2400 RAM
- 1 TB Nvme Samsung 960 EVO
- Asus GTX 1080Ti-11GB Turbo ($800)
- Palit RTX 2060–6GB ($350)
These parts are from my personal usage, and are not paid nor sponsored by any company, publisher or vendor.
Software: I use Ubuntu and Windows 10 in dual-boot.
- Ubuntu 18.04 + Anaconda/Python 3.7
- CUDA 10
- PyTorch 1.0 + fastai 1.0
- Nvidia drivers 415.xx
Note: before each training session, the cards were switched to a secondary PCie slot (x8) not handling the PC dual-monitors, thus assuring their computing power was 100% focused on the training.
I mention this as I’m using two monitors on a 2007 Ergotron LX Dual-Stand (my best long-term PC purchase ever !): one 24″ in 1080p (vertical) and one 27″ in 1440p (landscape) both connected to the same GPU, thus one can assume it would steal “some” computing power away from training.
A screenshot of my dual monitors: the 24″ vertical on left, the 27″ landscape on right.
BTW, if you want to check the impact of dual-monitor display on training performance, scroll down the bottom of the article for a comparison using the 1080Ti.
A quick summary of the two GPUs specs.
The 1080 Ti in the GTX 10 line-up (the last one):
The 2060 in the RTX 20 line-up (the first one):
Key points:
- The RTX 2060 has roughly HALF the number of CUDA cores of the 1080Ti (1920 vs. 3584)
- Its memory bandwith is about 70% of the 1080Ti (336 vs 484 GB/s)
- It has 240 Tensor Cores (source) for Deep Learning, the 1080Ti has none.
- It is rated for 160W of consumption, with a single 8-pin connector, while the 1080Ti is rated for 250W and needs a dual 8+6 pin connector.
- It costs less than HALF the retail price of the 1080Ti (in Stockholm, Sweden).
Additional information:
- Methodology: to keep things comparable, I ran every benchmark in three versions.
— The version “FP32” and version “FP16” used the same batch_size for the 1080Ti and the 2060 (one could argue that the 1080Ti has approx twice the VRAM of the 2060, but I chose that approach. Feel free to run the tests yourself).
— The version “FP16 bs*2” used a batch_size twice larger, to benefit from the theory behind FP16 training (see the two linked posts above for details). - The Jupyter notebooks I used, including all duration for 30 epochs, are available in my GitHub repo.
You’ll need Fastai V1 to run them. - The spreadsheet I used for duration, time-scale and charts, is located the repo as well.
- Duration for “Time-To-Complete 30 epochs” in seconds:
- Time-scaled:
- Duration in seconds:
- Time-scaled:
- Duration in seconds:
- Time-scaled:
- Duration in seconds:
- Time-scaled:
- Duration in seconds:
- Time-scaled:
- Duration in seconds:
- Time-scaled:
- Duration in seconds:
- Time-scaled:
- Duration in seconds:
- Time-scaled:
- Duration in seconds:
- Time-scaled:
- Duration in seconds:
- Time-scaled:
I compared the performance of the 1080Ti as a stand-alone GPU (no display) in red color vs.
a main GPU (handling a dual-monitor, see earlier discussion) in blue color.
Note: I used Cifar-10.
- Duration in seconds:
- Time-scaled:
GeForce RTX 2060 vs GeForce GTX 1080
- Home
- VGA Benchmarks
- GeForce RTX 2060 vs GeForce GTX 1080
-
GeForce RTX 2060
100%
-
GeForce GTX 1080
106%
Relative performance
Reasons to consider GeForce RTX 2060 |
This is a much newer product, it might have better long term support. |
Supports PhysX |
Supports G-Sync |
Supports ShadowPlay (allows game streaming/recording with minimum performance penalty) |
Supports Direct3D 12 Async Compute |
Supports DirectX Raytracing (DXR) |
Supports Deep Learning Super-Sampling (DLSS) |
Reasons to consider GeForce GTX 1080 |
6% higher gaming performance. |
Supports PhysX |
Supports G-Sync |
Supports ShadowPlay (allows game streaming/recording with minimum performance penalty) |
Supports Direct3D 12 Async Compute |
Based on an outdated architecture (Nvidia Pascal), there may be no performance optimizations for current games and applications |
HWBench recommends GeForce GTX 1080
The GeForce GTX 1080 is the better performing card based on the game benchmark suite used (75 combinations of games and resolutions).
Core Configuration
GeForce RTX 2060 | GeForce GTX 1080 | |||
---|---|---|---|---|
GPU Name | TU106 (TU106-200A-KA-A1) | vs | GP104 (GP104-400-A1) | |
Fab Process | 12 nm | vs | 16 nm | |
Die Size | 445 mm² | vs | 314 mm² | |
Transistors | 10,800 million | vs | 7,200 million | |
Shaders | 1920 | vs | 2560 | |
Compute Units | 30 | vs | 20 | |
Core clock | 1365 MHz | vs | 1607 MHz | |
ROPs | 48 | vs | 64 | |
TMUs | 120 | vs | 160 |
Memory Configuration
GeForce RTX 2060 | GeForce GTX 1080 | |||
---|---|---|---|---|
Memory Type | GDDR6 | vs | GDDR5X | |
Bus Width | 192 bit | vs | 256 bit | |
Memory Speed | 1750 MHz 14000 MHz effective |
vs | 1251 MHz 10008 MHz effective |
|
Memory Size | 6144 Mb | vs | 8192 Mb |
Additional details
GeForce RTX 2060 | GeForce GTX 1080 | |||
---|---|---|---|---|
TDP | 160 watts | vs | 180 watts | |
Release Date | 15 Jan 2019 | vs | 27 May 2016 |
-
GeForce RTX 2060
80. 64 GP/s
-
GeForce GTX 1080
102.80 GP/s
GigaPixels — higher is better
-
GeForce RTX 2060
201.60 GT/s
-
GeForce GTX 1080
257.10 GT/s
GigaTexels — higher is better
-
GeForce RTX 2060
336.00 GB/s
-
GeForce GTX 1080
320.00 GB/s
GB/s — higher is better
-
GeForce RTX 2060
6451.00 GFLOPs
-
GeForce GTX 1080
8228.00 GFLOPs
GFLOPs — higher is better
Ultra Quality, DirectX12, Windows 10×64
-
GeForce RTX 2060
50
-
GeForce GTX 1080
53
FPS (higher is better)
Ultra Quality, DirectX12, Windows 10×64
-
GeForce RTX 2060
107
-
GeForce GTX 1080
108
FPS (higher is better)
Ultra Quality, DirectX11, Windows 10×64
-
GeForce RTX 2060
62
-
GeForce GTX 1080
72
FPS (higher is better)
Ultra quality TTA DirectX11 Windows10 x64
-
GeForce RTX 2060
103
-
GeForce GTX 1080
111
FPS (higher is better)
Ultra High Quality, TAA, DirectX12, Windows 10 x64
-
GeForce RTX 2060
121
-
GeForce GTX 1080
125
FPS (higher is better)
Ultra High Quality TAA, Bahrain, DirectX12, Windows 10 x64
-
GeForce RTX 2060
145
-
GeForce GTX 1080
129
FPS (higher is better)
Very High Quality TAA, DirectX12, Windows 10 x64
-
GeForce RTX 2060
117
-
GeForce GTX 1080
121
FPS (higher is better)
Highest quality DirectX11 Windows10 x64
-
GeForce RTX 2060
123
-
GeForce GTX 1080
127
FPS (higher is better)
Ultra quality DirectX12 Windows10 x64
-
GeForce RTX 2060
96
-
GeForce GTX 1080
106
FPS (higher is better)
Ultra quality DirectX11 Windows10 x64
-
GeForce RTX 2060
103
-
GeForce GTX 1080
110
FPS (higher is better)
Ultra High Quality, DirectX12, Windows 10 x64
-
GeForce RTX 2060
81
-
GeForce GTX 1080
70
FPS (higher is better)
Ultra Quality, DirectX11, Windows 10×64
-
GeForce RTX 2060
106
-
GeForce GTX 1080
110
FPS (higher is better)
Ultra Quality, DirectX12, Windows 10 x64
-
GeForce RTX 2060
87
-
GeForce GTX 1080
92
FPS (higher is better)
DX11, Max Details, 16:1 AF, 2xMSAA
-
GeForce RTX 2060
123
-
GeForce GTX 1080
159
FPS (higher is better)
Very high quality DirectX11 Windows10 x64
-
GeForce RTX 2060
89
-
GeForce GTX 1080
91
FPS (higher is better)
Ultra Detail,16:1 AF, DirectX12, Windows 10×64
-
GeForce RTX 2060
55
-
GeForce GTX 1080
58
FPS (higher is better)
Ultra Quality, DirectX11, Windows 10×64
-
GeForce RTX 2060
40
-
GeForce GTX 1080
41
FPS (higher is better)
Ultra Quality, HR Textures, DirectX11, Windows 10 x64
-
GeForce RTX 2060
93
-
GeForce GTX 1080
99
FPS (higher is better)
UltraTX Max Quality, Vulkan, Windows 10 x64
-
GeForce RTX 2060
53
-
GeForce GTX 1080
50
FPS (higher is better)
Max Quality, DirectX12, Windows 10×64
-
GeForce RTX 2060
118
-
GeForce GTX 1080
128
FPS (higher is better)
Highest Details, Pure hair, HBAO+, DirectX12, Windows 10 x64
-
GeForce RTX 2060
86
-
GeForce GTX 1080
91
FPS (higher is better)
Ultra quality DirectX12 Windows10 x64
-
GeForce RTX 2060
131
-
GeForce GTX 1080
132
FPS (higher is better)
Ultra Quality, DirectX12, Async Compute ,Windows 10×64
-
GeForce RTX 2060
115
-
GeForce GTX 1080
127
FPS (higher is better)
DX11,Max Details, 16:1 HQ-AF, +AA
-
GeForce RTX 2060
106
-
GeForce GTX 1080
105
FPS (higher is better)
Ultra Quality, DirectX12, Windows 10×64
-
GeForce RTX 2060
51
-
GeForce GTX 1080
54
FPS (higher is better)
Ultra Quality, DirectX12, Windows 10×64
-
GeForce RTX 2060
39
-
GeForce GTX 1080
42
FPS (higher is better)
Ultra Quality, DirectX12, Windows 10×64
-
GeForce RTX 2060
81
-
GeForce GTX 1080
84
FPS (higher is better)
Ultra Quality, DirectX11, Windows 10×64
-
GeForce RTX 2060
46
-
GeForce GTX 1080
51
FPS (higher is better)
Ultra quality TTA DirectX11 Windows10 x64
-
GeForce RTX 2060
79
-
GeForce GTX 1080
84
FPS (higher is better)
Ultra High Quality, TAA, DirectX12, Windows 10 x64
-
GeForce RTX 2060
88
-
GeForce GTX 1080
95
FPS (higher is better)
Ultra High Quality TAA, Bahrain, DirectX12, Windows 10 x64
-
GeForce RTX 2060
85
-
GeForce GTX 1080
96
FPS (higher is better)
Very High Quality TAA, DirectX12, Windows 10 x64
-
GeForce RTX 2060
82
-
GeForce GTX 1080
86
FPS (higher is better)
Highest quality DirectX11 Windows10 x64
-
GeForce RTX 2060
90
-
GeForce GTX 1080
94
FPS (higher is better)
Ultra quality DirectX12 Windows10 x64
-
GeForce RTX 2060
64
-
GeForce GTX 1080
69
FPS (higher is better)
Ultra quality DirectX11 Windows10 x64
-
GeForce RTX 2060
76
-
GeForce GTX 1080
81
FPS (higher is better)
Ultra High Quality, DirectX12, Windows 10 x64
-
GeForce RTX 2060
57
-
GeForce GTX 1080
52
FPS (higher is better)
Ultra Quality, DirectX11, Windows 10×64
-
GeForce RTX 2060
80
-
GeForce GTX 1080
84
FPS (higher is better)
Ultra Quality, DirectX12, Windows 10 x64
-
GeForce RTX 2060
57
-
GeForce GTX 1080
65
FPS (higher is better)
DX11, Max Details, 16:1 AF, 2xMSAA
-
GeForce RTX 2060
105
-
GeForce GTX 1080
114
FPS (higher is better)
Very high quality DirectX11 Windows10 x64
-
GeForce RTX 2060
57
-
GeForce GTX 1080
58
FPS (higher is better)
Ultra Detail,16:1 AF, DirectX12, Windows 10×64
-
GeForce RTX 2060
42
-
GeForce GTX 1080
45
FPS (higher is better)
Ultra Quality, DirectX11, Windows 10×64
-
GeForce RTX 2060
31
-
GeForce GTX 1080
30
FPS (higher is better)
Ultra Quality, HR Textures, DirectX11, Windows 10 x64
-
GeForce RTX 2060
65
-
GeForce GTX 1080
68
FPS (higher is better)
UltraTX Max Quality, Vulkan, Windows 10 x64
-
GeForce RTX 2060
42
-
GeForce GTX 1080
39
FPS (higher is better)
Max Quality, DirectX12, Windows 10×64
-
GeForce RTX 2060
77
-
GeForce GTX 1080
83
FPS (higher is better)
Highest Details, Pure hair, HBAO+, DirectX12, Windows 10 x64
-
GeForce RTX 2060
57
-
GeForce GTX 1080
62
FPS (higher is better)
Ultra quality DirectX12 Windows10 x64
-
GeForce RTX 2060
84
-
GeForce GTX 1080
95
FPS (higher is better)
Ultra Quality, DirectX12, Async Compute ,Windows 10×64
-
GeForce RTX 2060
83
-
GeForce GTX 1080
91
FPS (higher is better)
DX11,Max Details, 16:1 HQ-AF, +AA
-
GeForce RTX 2060
76
-
GeForce GTX 1080
82
FPS (higher is better)
Ultra Quality, DirectX12, Windows 10×64
-
GeForce RTX 2060
37
-
GeForce GTX 1080
40
FPS (higher is better)
Ultra Quality, DirectX12, Windows 10×64
-
GeForce RTX 2060
24
-
GeForce GTX 1080
25
FPS (higher is better)
Ultra Quality, DirectX12, Windows 10×64
-
GeForce RTX 2060
44
-
GeForce GTX 1080
49
FPS (higher is better)
Ultra Quality, DirectX11, Windows 10×64
-
GeForce RTX 2060
19
-
GeForce GTX 1080
28
FPS (higher is better)
Ultra quality TTA DirectX11 Windows10 x64
-
GeForce RTX 2060
48
-
GeForce GTX 1080
51
FPS (higher is better)
Ultra High Quality, TAA, DirectX12, Windows 10 x64
-
GeForce RTX 2060
49
-
GeForce GTX 1080
55
FPS (higher is better)
Ultra High Quality TAA, Bahrain, DirectX12, Windows 10 x64
-
GeForce RTX 2060
51
-
GeForce GTX 1080
57
FPS (higher is better)
Very High Quality TAA, DirectX12, Windows 10 x64
-
GeForce RTX 2060
45
-
GeForce GTX 1080
47
FPS (higher is better)
Highest quality DirectX11 Windows10 x64
-
GeForce RTX 2060
50
-
GeForce GTX 1080
54
FPS (higher is better)
Ultra quality DirectX12 Windows10 x64
-
GeForce RTX 2060
33
-
GeForce GTX 1080
36
FPS (higher is better)
Ultra quality DirectX11 Windows10 x64
-
GeForce RTX 2060
39
-
GeForce GTX 1080
42
FPS (higher is better)
Ultra High Quality, DirectX12, Windows 10 x64
-
GeForce RTX 2060
34
-
GeForce GTX 1080
32
FPS (higher is better)
Ultra Quality, DirectX11, Windows 10×64
-
GeForce RTX 2060
43
-
GeForce GTX 1080
45
FPS (higher is better)
Ultra Quality, DirectX12, Windows 10 x64
-
GeForce RTX 2060
32
-
GeForce GTX 1080
36
FPS (higher is better)
DX11, Max Details, 16:1 AF, 2xMSAA
-
GeForce RTX 2060
56
-
GeForce GTX 1080
58
FPS (higher is better)
Very high quality DirectX11 Windows10 x64
-
GeForce RTX 2060
28
-
GeForce GTX 1080
28
FPS (higher is better)
Ultra Detail,16:1 AF, DirectX12, Windows 10×64
-
GeForce RTX 2060
26
-
GeForce GTX 1080
28
FPS (higher is better)
Ultra Quality, DirectX11, Windows 10×64
-
GeForce RTX 2060
17
-
GeForce GTX 1080
17
FPS (higher is better)
Ultra Quality, HR Textures, DirectX11, Windows 10 x64
-
GeForce RTX 2060
39
-
GeForce GTX 1080
38
FPS (higher is better)
UltraTX Max Quality, Vulkan, Windows 10 x64
-
GeForce RTX 2060
24
-
GeForce GTX 1080
24
FPS (higher is better)
Max Quality, DirectX12, Windows 10×64
-
GeForce RTX 2060
40
-
GeForce GTX 1080
41
FPS (higher is better)
Highest Details, Pure hair, HBAO+, DirectX12, Windows 10 x64
-
GeForce RTX 2060
29
-
GeForce GTX 1080
33
FPS (higher is better)
Ultra quality DirectX12 Windows10 x64
-
GeForce RTX 2060
48
-
GeForce GTX 1080
57
FPS (higher is better)
Ultra Quality, DirectX12, Async Compute ,Windows 10×64
-
GeForce RTX 2060
45
-
GeForce GTX 1080
53
FPS (higher is better)
DX11,Max Details, 16:1 HQ-AF, +AA
-
GeForce RTX 2060
41
-
GeForce GTX 1080
48
FPS (higher is better)
Ultra Quality, DirectX12, Windows 10×64
-
GeForce RTX 2060
22
-
GeForce GTX 1080
21
FPS (higher is better)
VS | ||
GeForce RTX 2060 | Radeon RX 6600 |
VS | ||
GeForce RTX 2060 | Radeon RX 6600 XT |
VS | ||
GeForce GTX 1080 | Radeon RX 6650 XT |
VS | ||
GeForce GTX 1080 | GeForce RTX 2060 Super |
VS | ||
GeForce RTX 2070 Super | Radeon RX 5700 XT |
VS | ||
Radeon RX 5700 XT | Radeon RX 5700 XT 50th Anniversary |
Please enable JavaScript to view the comments powered by Disqus.