Test setup
Object recognition performance tests were done using a dedicated machine that only runs the Object Recognition service with the following hardware:
-
CPU: 12th Gen Intel(R) Core(TM) i9-12900KF @ 3.20 GHz.
-
NVIDIA GPU: NVIDIA GeForce RTX 3080 Ti.
-
Intel GPU: Intel UHD Graphics 750.
During the test, the following metrics were observed and collected:
-
Input frame rate
-
Output frame rate
-
Memory usage (RAM)
-
CPU usage
-
GPU usage for model interference
-
GPU usage for video decompression
-
GPU memory usage
Tests were performed by using H.264 compressed video clip with 1920x1080 resolution.
Test results
Performance Test Results for Vehicles
CPU: 12th Gen Intel(R) Core(TM) i9-12900KF @ 3.20 GHz
|
Streams |
Input FPS |
Output FPS |
Process CPU |
Process memory |
NVIDIA GPU |
NVIDIA decode |
GPU memory |
|---|---|---|---|---|---|---|---|
|
1 |
4 |
4 |
25% |
4.6 GB |
4% |
3% |
3.5 GB |
|
2 |
8 |
8 |
45 % |
4.8 GB |
8% |
8% |
3.8 GB |
|
3 |
10 |
10 |
88% |
5.6 GB |
11% |
11% |
4.1 GB |
NVIDIA GPU: NVIDIA GeForce RTX 3080 Ti
|
Streams |
Input FPS |
Output FPS |
Process CPU |
Process memory |
NVIDIA GPU |
NVIDIA decode |
GPU memory |
|---|---|---|---|---|---|---|---|
|
1 |
4 |
4 |
3% |
2.8 GB |
25% |
3% |
3.9 GB |
|
2 |
8 |
8 |
4 % |
3.5 GB |
30% |
5% |
4.2 GB |
|
3 |
12 |
12 |
5% |
3.7 GB |
30% |
5% |
4.6 GB |
|
4 |
16 |
16 |
5% |
3.8 GB |
35% |
5% |
4.9 GB |
|
5 |
20 |
20 |
6% |
4.0 GB |
35% |
4% |
5.2 GB |
|
6 |
24 |
24 |
6% |
4.1 GB |
35% |
5% |
5.5 GB |
|
7 |
28 |
28 |
7% |
4.3 GB |
43% |
6% |
5.9 GB |
|
8 |
32 |
32 |
8% |
4.4 GB |
50% |
6% |
6.2 GB |
|
9 |
36 |
36 |
10% |
4.6 GB |
95% |
6% |
6.6 GB |
|
10 |
40 |
40 |
10% |
4.8 GB |
95% |
6% |
7.0 GB |
|
11 |
44 |
44 |
13% |
4.9 GB |
98% |
6% |
7.4 GB |
|
12 |
48 |
48 |
13% |
5.1 GB |
98% |
6% |
7.8 GB |
|
13 |
52 |
52 |
17% |
5.2 GB |
98% |
6% |
8.1 GB |
|
14 |
56 |
56 |
17% |
5.7 GB |
99% |
7% |
8.5 GB |
|
15 |
60 |
60 |
20% |
6.5 GB |
99% |
7% |
8.9 GB |
|
16 |
55 |
55 |
21% |
5.5 GB |
98% |
7% |
9.2 GB |
Intel GPU: Intel UHD Graphics 750
|
Streams |
Input FPS |
Output FPS |
Process CPU |
Process memory |
NVIDIA GPU |
NVIDIA decode |
Intel GPU |
Intel GPU memory |
|---|---|---|---|---|---|---|---|---|
|
1 |
4 |
4 |
23% |
3.5 GB |
2% |
2% |
50% |
0.7 GB |
|
2 |
8 |
8 |
50% |
4.2 GB |
3% |
3% |
99% |
0.9 GB |
|
3 |
7 |
7 |
50% |
4.6 GB |
3% |
3% |
100% |
0.9 GB |
Performance test results for persons
CPU: 12th Gen Intel(R) Core(TM) i9-12900KF 3.20 GHz
|
Streams |
Input FPS |
Output FPS |
Process CPU |
Process memory |
NVIDIA GPU |
NVIDIA decode |
GPU memory |
|---|---|---|---|---|---|---|---|
|
1 |
4 |
4 |
60% |
4.9 GB |
3% |
3% |
3.5 GB |
|
2 |
6 |
6 |
86 % |
5.4 GB |
3% |
3% |
3.8 GB |
NVIDIA GPU: NVIDIA GeForce RTX 3080 Ti
|
Streams |
Input FPS |
Output FPS |
Process CPU |
Process memory |
NVIDIA GPU |
NVIDIA decode |
GPU memory |
|---|---|---|---|---|---|---|---|
|
1 |
4 |
4 |
4% |
3.3 GB |
31% |
2% |
4.6 GB |
|
2 |
8 |
8 |
6% |
3.6 GB |
35% |
2% |
5.0 GB |
|
3 |
12 |
12 |
9% |
3.7 GB |
90% |
2% |
5.3 GB |
|
4 |
16 |
16 |
12% |
3.9 GB |
95% |
2% |
5.6 GB |
|
5 |
20 |
20 |
14% |
4.3 GB |
98% |
2% |
6.1 GB |
|
6 |
20 |
20 |
17% |
5.0 GB |
99% |
2% |
6.4 GB |
Intel GPU: Intel UHD Graphics 750
|
Streams |
Input FPS |
Output FPS |
Process CPU |
Process memory |
NVIDIA GPU |
NVIDIA decode |
Intel GPU |
Intel GPU memory |
|---|---|---|---|---|---|---|---|---|
|
1 |
4 |
4 |
23% |
0.6 GB |
4% |
4% |
60% |
0.7 GB |
|
2 |
2 |
0 |
75% |
3.1 GB |
6% |
6% |
100% |
1.0 GB |
These tests are indicative and may not be directly applicable to production systems.