GPU 2017 Benchmarks
BENCH
Bench gives you access to our internal benchmark data so that you can compare the products without searching for an older review. Make reliable comparisons between products by clicking
on product categories below!
Bench Results
NVIDIA Titan V
NVIDIA Titan Xp
NVIDIA Titan Xp
Product Ratings & Comparisons
Compute: General Matrix Multiply Single Precision (SGEMM)
GFLOPS - Higher is Better
GFLOPS - Higher is Better
13429
10879
Compute: General Matrix Multiply Double Precision (DGEMM)
GFLOPS - Higher is Better
GFLOPS - Higher is Better
6777
418
Compute: General Matrix Multiply Half Precision (HGEMM)
GFLOPS - Higher is Better
GFLOPS - Higher is Better
26463
218
Compute: Folding @ Home Double Precision
Nanoseconds per Day - Higher is Better
Nanoseconds per Day - Higher is Better
75.4
14.6
Compute: SPECviewperf 12.1 - 3dsmax-05 Composite
Frames per Second - Higher is Better
Frames per Second - Higher is Better
176.3
100.3
Synthetic: Beyond3D Suite - ALU Latency via Multiply-Adds (MADD)
Nanoseconds per MADD - Lower is Better
Nanoseconds per MADD - Lower is Better
2.35
3.3
Synthetic: 3DMark Fire Strike Ultra (4K) - Graphics Score
Score - Higher is Better
Score - Higher is Better
7679
7132
Synthetic: 3DMark Fire Strike Ultra (4K) - Graphics Test 1
Frames per Second - Higher is Better
Frames per Second - Higher is Better
46.84
39.3
Synthetic: 3DMark Fire Strike Ultra (4K) - Graphics Test 2
Frames per Second - Higher is Better
Frames per Second - Higher is Better
25.94
25.6
Compute: SiSoftware Sandra 2017 - GP Processing Double Precision (OpenCL)
GigaPixels per Second - Higher is Better
GigaPixels per Second - Higher is Better
10.69
0.72
Compute: SiSoftware Sandra 2017 - GP Processing Single Precision (OpenCL)
GigaPixels per Second - Higher is Better
GigaPixels per Second - Higher is Better
21
17.28
Compute: SiSoftware Sandra 2017 - GP Processing Half Precision (OpenCL)
GigaPixels per Second - Higher is Better
GigaPixels per Second - Higher is Better
21
17
Compute: SiSoftware Sandra 2017 - GP Processing Single Precision (Compute Shader)
GigaPixels per Second - Higher is Better
GigaPixels per Second - Higher is Better
25.25
18.43
Compute: SiSoftware Sandra 2017 - GP Processing Half Precision (Compute Shader)
GigaPixels per Second - Higher is Better
GigaPixels per Second - Higher is Better
24.22
16.12
Compute: SiSoftware Sandra 2017 - GP Financial Analysis Double Precision (OpenCL)
MOPT per Second - Higher is Better
MOPT per Second - Higher is Better
2.81
0.21
Compute: SiSoftware Sandra 2017 - GP Financial Analysis Single Precision (OpenCL)
MOPT per Second - Higher is Better
MOPT per Second - Higher is Better
7
4.12
Compute: SiSoftware Sandra 2017 - GP Financial Analysis Single Precision (Compute Shader)
MOPT per Second - Higher is Better
MOPT per Second - Higher is Better
4.3
2.49
Compute: SiSoftware Sandra 2017 - Video Shader Compute Double Precision (DX11)
GigaPixels per Second - Higher is Better
GigaPixels per Second - Higher is Better
6.08
0.45
Compute: SiSoftware Sandra 2017 - Video Shader Compute Single Precision (DX11)
GigaPixels per Second - Higher is Better
GigaPixels per Second - Higher is Better
12.7
11.61
Compute: SiSoftware Sandra 2017 - Video Shader Compute Half Precision (DX11)
GigaPixels per Second - Higher is Better
GigaPixels per Second - Higher is Better
12.9
11.7
Compute: SiSoftware Sandra 2017 - GP Processing Double Precision (Compute Shader)
GigaPixels per Second - Higher is Better
GigaPixels per Second - Higher is Better
7.74
0.22
Crysis: Warhead (DX10) - 3840x2160 - Enthusiast Quality, 4xSSAA
Frames per Second - Higher is Better
Frames per Second - Higher is Better
62
50.1
DL Inference: DeepBench - GEMM (DeepSpeech)
TFLOPS or TOPS - Higher is Better
TFLOPS or TOPS - Higher is Better
7.57
6.34
DL Training: DeepBench - GEMM (Language Modelling)
TFLOPS - Higher is Better
TFLOPS - Higher is Better
3.08
1.26
DL Inference: DeepBench - Sparse GEMM
TFLOPS - Higher is Better
TFLOPS - Higher is Better
2.94
2.94
DL Training: DeepBench - GEMM (DeepSpeech)
TFLOPS - Higher is Better
TFLOPS - Higher is Better
8.71
7.18
DL Inference: DeepBench - GEMM (DeepSpeech, GRU Affine)
TFLOPS or TOPS - Higher is Better
TFLOPS or TOPS - Higher is Better
12.1
10.6
DL Training: DeepBench - GEMM (Speaker ID)
TFLOPS - Higher is Better
TFLOPS - Higher is Better
7.74
8.91
DL Training: DeepBench - Convolutions (Speaker ID)
TFLOPS - Higher is Better
TFLOPS - Higher is Better
16.02
12.9
DL Training: DeepBench - Convolutions
TFLOPS - Higher is Better
TFLOPS - Higher is Better
8.89
7.74
DL Training: DeepBench - Convolutions (Vision)
TFLOPS - Higher is Better
TFLOPS - Higher is Better
13.22
11.2
DL Inference: DeepBench - GEMM (Language Modelling)
TFLOPS or TOPS - Higher is Better
TFLOPS or TOPS - Higher is Better
0.7
0.29
DL Training: DeepBench - Convolutions (Resnet)
TFLOPS - Higher is Better
TFLOPS - Higher is Better
6.52
5.64
DL Training: DeepBench - RNN (Vanilla)
TFLOPS - Higher is Better
TFLOPS - Higher is Better
4.36
3.82
DL Inference: DeepBench - GEMM (DeepSpeech, Output Layer)
TFLOPS or TOPS - Higher is Better
TFLOPS or TOPS - Higher is Better
7.43
5.65
DL Inference: DeepBench - GEMM
TFLOPS or TOPS - Higher is Better
TFLOPS or TOPS - Higher is Better
7
5.97
DL Inference: DeepBench - RNN (LSTM)
TFLOPS - Higher is Better
TFLOPS - Higher is Better
0.27
0.21
DL Inference: DeepBench - Convolutions (Resnet)
TFLOPS or TOPS - Higher is Better
TFLOPS or TOPS - Higher is Better
2.78
2.66
DL Inference: DeepBench - Convolutions (Vision)
TFLOPS or TOPS - Higher is Better
TFLOPS or TOPS - Higher is Better
3.92
4.76
DL Inference: DeepBench - Convolutions (DeepSpeech)
TFLOPS or TOPS - Higher is Better
TFLOPS or TOPS - Higher is Better
5.74
4.31
DL Inference: DeepBench - Convolutions
TFLOPS or TOPS - Higher is Better
TFLOPS or TOPS - Higher is Better
3.86
4.52
DL Inference: DeepBench - RNN (GRU)
TFLOPS - Higher is Better
TFLOPS - Higher is Better
0.55
0.41
DL Training: DeepBench - RNN (LSTM)
TFLOPS - Higher is Better
TFLOPS - Higher is Better
4.3
3.39
DL Training: DeepBench - RNN (GRU)
TFLOPS - Higher is Better
TFLOPS - Higher is Better
5.54
4.98
DL Training: DLBS Caffe2 - ResNet50 and ImageNet Throughput
Images per Second - Higher is Better
Images per Second - Higher is Better
236
213
DL Inference: DLBS Caffe2 - ResNet50 and ImageNet Throughput
Images per Second - Higher is Better
Images per Second - Higher is Better
841
718
DL Inference: DLBS TensorRT- ResNet50 and ImageNet Throughput
Images per Second - Higher is Better
Images per Second - Higher is Better
1465
1305
DL Training: DAWNBench CIFAR10 - KervResNet34 Training Time
Time to 94% Accuracy in Minutes - Lower is Better
Time to 94% Accuracy in Minutes - Lower is Better
25.37
31.4
DL Training: DAWNBench CIFAR10 - KervResNet34 Training Cost
kWh - Lower is Better
kWh - Lower is Better
0.12
0.19
DL Training: DAWNBench CIFAR10 - Basenet (ResNet18) Time to 94% Accuracy
Minutes:Seconds - Lower is Better
Minutes:Seconds - Lower is Better
5.98
11.78
DL Training: DAWNBench CIFAR10 - Basenet (ResNet18) Training Cost
kWh - Lower is Better
kWh - Lower is Better
0.02
0.07
DL Training: DeepBench - GEMM
TFLOPS - Higher is Better
TFLOPS - Higher is Better
8.42
6.91
Battlefield 1 - 3840x2160 - Ultra Quality
Frames per Second - Higher is Better
Frames per Second - Higher is Better
96
83.6
Battlefield 1 - 99th Percentile - 3840x2160 - Ultra Quality
Frames per Second at the 99th Percentile - Higher is Better
Frames per Second at the 99th Percentile - Higher is Better
69
69
Ashes of the Singularity: Escalation - 3840x2160 - Extreme Quality
Frames per Second - Higher is Better
Frames per Second - Higher is Better
88.5
80.9
Ashes of the Singularity: Escalation - 99th Percentile - 3840x2160 - Extreme Quality
Frames per Second at the 99th Percentile - Higher is Better
Frames per Second at the 99th Percentile - Higher is Better
50
46
Doom - 3840x2160 - Ultra Quality
Frames per Second - Higher is Better
Frames per Second - Higher is Better
127.7
94.4
Doom - 99th Percentile - 3840x2160 - Ultra Quality
Frames per Second at the 99th Percentile - Higher is Better
Frames per Second at the 99th Percentile - Higher is Better
67.1
69.8
Ghost Recon Wildlands - 3840x2160 - Very High Quality
Frames per Second - Higher is Better
Frames per Second - Higher is Better
59.2
55.5
Deus Ex: Mankind Divided - 3840x2160 - Ultra Quality
Frames per Second - Higher is Better
Frames per Second - Higher is Better
49.7
41
Grand Theft Auto V - 3840x2160 - Very High Quality
Frames per Second - Higher is Better
Frames per Second - Higher is Better
62.8
58.4
Grand Theft Auto V - 99th Percentile - 3840x2160 - Very High Quality
Frames per Second at the 99th Percentile - Higher is Better
Frames per Second at the 99th Percentile - Higher is Better
22.7
33.3
Total War: Warhammer - 3840x2160 - Ultra Quality
Frames per Second - Higher is Better
Frames per Second - Higher is Better
79.3
70
Compute: CompuBench 2.0 - Level Set Segmentation 256
MegaVoxels per Second - Higher is Better
MegaVoxels per Second - Higher is Better
16974
16858
Compute: CompuBench 2.0 - N-Body Simulation 1024K
Iterations per Second - Higher is Better
Iterations per Second - Higher is Better
127.7
84.7
Compute: CompuBench 2.0 - Optical Flow
MegaPixels per Second - Higher is Better
MegaPixels per Second - Higher is Better
104.2
68.8
Compute: Folding @ Home Single Precision
Nanoseconds per Day - Higher is Better
Nanoseconds per Day - Higher is Better
122.1
125.8
Compute: Geekbench 4 - GPU Compute - Total Score
Points - Higher is Better
Points - Higher is Better
509384
243066
Synthetic: TessMark, Image Set 4, 64x Tessellation
Frames per Second - Higher is Better
Frames per Second - Higher is Better
703
604
Synthetic: Beyond3D Suite - Pixel Fillrate
GigaPixels per Second - Higher is Better
GigaPixels per Second - Higher is Better
129.8
109.9
Synthetic: Beyond3D Suite - Integer Texture Fillrate (INT8)
GigaTexels per Second - Higher is Better
GigaTexels per Second - Higher is Better
490.7
283.2
Synthetic: Beyond3D Suite - Floating Point Texture Fillrate (FP32)
GigaTexels per Second - Higher is Better
GigaTexels per Second - Higher is Better
138
111.5
Idle Power Consumption
Total System Power Consumption in Watts - Lower is Better
Total System Power Consumption in Watts - Lower is Better
96
82
Load Power Consumption - Battlefield 1
Total System Power Consumption in Watts - Lower is Better
Total System Power Consumption in Watts - Lower is Better
381
419
Load Power Consumption - FurMark
Total System Power Consumption in Watts - Lower is Better
Total System Power Consumption in Watts - Lower is Better
350
375
Idle GPU Temperature
Temperature in Celsius - Lower is Better
Temperature in Celsius - Lower is Better
36
30
Load GPU Temperature - Battlefield 1
Temperature in Celsius - Lower is Better
Temperature in Celsius - Lower is Better
83
82
Load GPU Temperature - FurMark
Temperature in Celsius - Lower is Better
Temperature in Celsius - Lower is Better
84
80
Idle Noise Levels
Noise In dB(A) - Lower is Better
Noise In dB(A) - Lower is Better
39.6
39.4
Load Noise Levels - Battlefield 1
Noise In dB(A) - Lower is Better
Noise In dB(A) - Lower is Better
49.5
49.8
Load Noise Levels - FurMark
Noise In dB(A) - Lower is Better
Noise In dB(A) - Lower is Better
50.5
50
Facebook
Twitter
RSS