Performance measurement report for text-translation
Hardware
Deployed on kubernetes cluster in Amazon Elastic Kubernetes (EKS)
CPU tests
AWS instance type: c7a.2xlarge
CPU: AMD EPYC 9R14
GPU tests
AWS instance type (GPU): g5.2xlarge
CPU: AMD EPYC 7R32
GPU: NVIDIA A10G
Software
Image version (CPU tests): 1.0.0
Image version (GPU tests): 1.0.0
Dataset
Cornell Movie-Dialogs Corpus
Test methodology
One instance of the Text Translation microservice is deployed on the Kubernetes cluster in EKS.
The client sends randomly selected files to the microservice that translates text. Each file contains several sentences
(segments). Each tests is performed for 30 minutes, and the results are discarded.
⚠️ Warning
Due to a Kubernetes cluster limitation, only 7.7 CPUs are available for the microservice.
CPU
Settings
| Test |
PHX_NUM_THREADS_PER_INSTANCE |
PHX_NUM_INSTANCES_PER_DEVICE |
| 1 instances, 1 threads |
1 |
1 |
| 1 instances, 8 threads |
8 |
1 |
| 8 instances, 1 threads |
1 |
8 |
1 instances, 1 threads
| Model |
Version |
Processing time [s] |
Number of processed files |
Processed files per second |
Processed segments |
Processed segments per second |
Memory peak [MiB] |
Graphs |
| en_ar |
1.0.0+argos-1.0.0 |
1,800.71 (0h 30m 0.71s) |
2,222 |
1.2 |
81,411 |
45.21 |
510 |
CPU RAM |
| en_bn |
1.0.0+argos-1.9.0 |
1,801.18 (0h 30m 1.18s) |
1,414 |
0.8 |
51,710 |
28.71 |
545 |
CPU RAM |
| en_cs |
1.0.0+argos-1.9.0 |
1,802.50 (0h 30m 2.50s) |
1,461 |
0.8 |
53,600 |
29.74 |
638 |
CPU RAM |
| en_de |
1.0.0+argos-1.0.0 |
1,800.68 (0h 30m 0.68s) |
2,065 |
1.1 |
75,889 |
42.14 |
521 |
CPU RAM |
| en_es |
1.0.0+argos-1.0.0 |
1,802.20 (0h 30m 2.20s) |
2,005 |
1.1 |
73,639 |
40.86 |
762 |
CPU RAM |
| en_fa |
1.0.0+argos-1.5.0 |
1,801.90 (0h 30m 1.90s) |
1,641 |
0.9 |
59,538 |
33.04 |
686 |
CPU RAM |
| en_fr |
1.0.0+argos-1.9.0 |
1,802.44 (0h 30m 2.44s) |
1,444 |
0.8 |
52,106 |
28.91 |
765 |
CPU RAM |
| en_he |
1.0.0+argos-1.5.0 |
1,801.35 (0h 30m 1.35s) |
1,849 |
1.0 |
67,628 |
37.54 |
812 |
CPU RAM |
| en_hi |
1.0.0+argos-1.1.0 |
1,801.76 (0h 30m 1.76s) |
1,531 |
0.8 |
55,881 |
31.01 |
560 |
CPU RAM |
| en_ms |
1.0.0+argos-1.9.0 |
1,800.80 (0h 30m 0.80s) |
1,843 |
1.0 |
67,711 |
37.60 |
548 |
CPU RAM |
| en_pt-BR |
1.0.0+argos-1.9.0 |
1,801.90 (0h 30m 1.90s) |
1,624 |
0.9 |
59,699 |
33.13 |
762 |
CPU RAM |
| en_pt-PT |
1.0.0+argos-1.9.0 |
1,801.31 (0h 30m 1.31s) |
1,598 |
0.9 |
58,148 |
32.28 |
552 |
CPU RAM |
| en_ru |
1.0.0+argos-1.9.0 |
1,803.96 (0h 30m 3.96s) |
560 |
0.3 |
20,821 |
11.54 |
1,202 |
CPU RAM |
| en_sk |
1.0.0+argos-1.9.0 |
1,801.89 (0h 30m 1.89s) |
1,473 |
0.8 |
54,296 |
30.13 |
798 |
CPU RAM |
| en_sq |
1.0.0+argos-1.9.0 |
1,800.84 (0h 30m 0.84s) |
1,814 |
1.0 |
66,275 |
36.80 |
465 |
CPU RAM |
| en_zh |
1.0.0+argos-1.9.0 |
1,802.52 (0h 30m 2.52s) |
1,619 |
0.9 |
60,260 |
33.43 |
488 |
CPU RAM |
1 instances, 8 threads
| Model |
Version |
Processing time [s] |
Number of processed files |
Processed files per second |
Processed segments |
Processed segments per second |
Memory peak [MiB] |
Graphs |
| en_ar |
1.0.0+argos-1.0.0 |
1,800.88 (0h 30m 0.88s) |
4,623 |
2.6 |
169,413 |
94.07 |
560 |
CPU RAM |
| en_bn |
1.0.0+argos-1.9.0 |
1,801.15 (0h 30m 1.15s) |
3,105 |
1.7 |
115,365 |
64.05 |
549 |
CPU RAM |
| en_cs |
1.0.0+argos-1.9.0 |
1,800.83 (0h 30m 0.83s) |
3,536 |
2.0 |
128,981 |
71.62 |
640 |
CPU RAM |
| en_de |
1.0.0+argos-1.0.0 |
1,800.51 (0h 30m 0.51s) |
4,192 |
2.3 |
153,406 |
85.20 |
549 |
CPU RAM |
| en_es |
1.0.0+argos-1.0.0 |
1,800.52 (0h 30m 0.52s) |
4,095 |
2.3 |
150,560 |
83.62 |
557 |
CPU RAM |
| en_fa |
1.0.0+argos-1.5.0 |
1,800.57 (0h 30m 0.57s) |
3,614 |
2.0 |
132,799 |
73.75 |
620 |
CPU RAM |
| en_fr |
1.0.0+argos-1.9.0 |
1,800.86 (0h 30m 0.86s) |
3,301 |
1.8 |
121,218 |
67.31 |
514 |
CPU RAM |
| en_he |
1.0.0+argos-1.5.0 |
1,801.19 (0h 30m 1.19s) |
4,127 |
2.3 |
151,590 |
84.16 |
640 |
CPU RAM |
| en_hi |
1.0.0+argos-1.1.0 |
1,801.07 (0h 30m 1.07s) |
3,322 |
1.8 |
121,786 |
67.62 |
590 |
CPU RAM |
| en_ms |
1.0.0+argos-1.9.0 |
1,801.75 (0h 30m 1.75s) |
3,959 |
2.2 |
146,583 |
81.36 |
659 |
CPU RAM |
| en_pt-BR |
1.0.0+argos-1.9.0 |
1,801.41 (0h 30m 1.41s) |
3,768 |
2.1 |
138,820 |
77.06 |
564 |
CPU RAM |
| en_pt-PT |
1.0.0+argos-1.9.0 |
1,800.53 (0h 30m 0.53s) |
3,531 |
2.0 |
129,719 |
72.04 |
585 |
CPU RAM |
| en_ru |
1.0.0+argos-1.9.0 |
1,801.70 (0h 30m 1.70s) |
1,599 |
0.9 |
58,208 |
32.31 |
2,234 |
CPU RAM |
| en_sk |
1.0.0+argos-1.9.0 |
1,800.52 (0h 30m 0.52s) |
3,520 |
2.0 |
129,534 |
71.94 |
684 |
CPU RAM |
| en_sq |
1.0.0+argos-1.9.0 |
1,800.94 (0h 30m 0.94s) |
4,105 |
2.3 |
151,395 |
84.06 |
745 |
CPU RAM |
| en_zh |
1.0.0+argos-1.9.0 |
1,801.82 (0h 30m 1.82s) |
3,850 |
2.1 |
141,178 |
78.35 |
488 |
CPU RAM |
8 instances, 1 threads
| Model |
Version |
Processing time [s] |
Number of processed files |
Processed files per second |
Processed segments |
Processed segments per second |
Memory peak [MiB] |
Graphs |
| en_ar |
1.0.0+argos-1.0.0 |
1,801.88 (0h 30m 1.88s) |
13,056 |
7.2 (0.9 per instance) |
478,056 |
265.31 |
1,970 |
CPU RAM |
| en_bn |
1.0.0+argos-1.9.0 |
1,807.01 (0h 30m 7.01s) |
8,299 |
4.6 (0.6 per instance) |
305,117 |
168.85 |
2,011 |
CPU RAM |
| en_cs |
1.0.0+argos-1.9.0 |
1,803.74 (0h 30m 3.74s) |
8,689 |
4.8 (0.6 per instance) |
319,942 |
177.38 |
2,643 |
CPU RAM |
| en_de |
1.0.0+argos-1.0.0 |
1,802.63 (0h 30m 2.63s) |
11,639 |
6.5 (0.8 per instance) |
426,472 |
236.58 |
2,484 |
CPU RAM |
| en_es |
1.0.0+argos-1.0.0 |
1,803.46 (0h 30m 3.46s) |
11,507 |
6.4 (0.8 per instance) |
421,952 |
233.97 |
2,166 |
CPU RAM |
| en_fa |
1.0.0+argos-1.5.0 |
1,802.31 (0h 30m 2.31s) |
9,393 |
5.2 (0.7 per instance) |
344,726 |
191.27 |
2,063 |
CPU RAM |
| en_fr |
1.0.0+argos-1.9.0 |
1,803.26 (0h 30m 3.26s) |
8,336 |
4.6 (0.6 per instance) |
305,733 |
169.54 |
2,707 |
CPU RAM |
| en_he |
1.0.0+argos-1.5.0 |
1,804.22 (0h 30m 4.22s) |
10,924 |
6.1 (0.8 per instance) |
400,321 |
221.88 |
2,484 |
CPU RAM |
| en_hi |
1.0.0+argos-1.1.0 |
1,802.87 (0h 30m 2.87s) |
8,789 |
4.9 (0.6 per instance) |
320,760 |
177.92 |
1,912 |
CPU RAM |
| en_ms |
1.0.0+argos-1.9.0 |
1,802.03 (0h 30m 2.03s) |
10,826 |
6.0 (0.8 per instance) |
396,489 |
220.02 |
1,685 |
CPU RAM |
| en_pt-BR |
1.0.0+argos-1.9.0 |
1,802.41 (0h 30m 2.41s) |
9,831 |
5.5 (0.7 per instance) |
359,802 |
199.62 |
1,794 |
CPU RAM |
| en_pt-PT |
1.0.0+argos-1.9.0 |
1,804.10 (0h 30m 4.10s) |
9,490 |
5.3 (0.7 per instance) |
347,219 |
192.46 |
1,973 |
CPU RAM |
| en_ru |
1.0.0+argos-1.9.0 |
1,813.53 (0h 30m 13.53s) |
3,275 |
1.8 (0.2 per instance) |
119,881 |
66.10 |
4,498 |
CPU RAM |
| en_sk |
1.0.0+argos-1.9.0 |
1,804.84 (0h 30m 4.84s) |
8,913 |
4.9 (0.6 per instance) |
325,545 |
180.37 |
1,660 |
CPU RAM |
| en_sq |
1.0.0+argos-1.9.0 |
1,802.29 (0h 30m 2.29s) |
10,527 |
5.8 (0.7 per instance) |
385,547 |
213.92 |
1,939 |
CPU RAM |
| en_zh |
1.0.0+argos-1.9.0 |
1,803.48 (0h 30m 3.48s) |
9,749 |
5.4 (0.7 per instance) |
358,617 |
198.85 |
2,662 |
CPU RAM |
GPU
Settings
| Test |
PHX_NUM_THREADS_PER_INSTANCE |
PHX_NUM_INSTANCES_PER_DEVICE |
| 1 instances, 1 threads |
1 |
1 |
| 8 instances, 1 threads |
1 |
8 |
1 instances, 1 threads
| Model |
Version |
Processing time [s] |
Number of processed files |
Processed files per second |
Processed segments |
Processed segments per second |
Memory peak [MiB] |
GPU memory peak [MiB] |
Graphs |
| en_ar |
1.0.0+argos-1.0.0 |
1,800.19 (0h 30m 0.19s) |
11,433 |
6.4 |
418,805 |
232.64 |
382 |
716 |
CPU RAM GPU GPU-RAM |
| en_bn |
1.0.0+argos-1.9.0 |
1,800.50 (0h 30m 0.50s) |
8,091 |
4.5 |
297,591 |
165.28 |
359 |
588 |
CPU RAM GPU GPU-RAM |
| en_cs |
1.0.0+argos-1.9.0 |
1,800.16 (0h 30m 0.16s) |
9,549 |
5.3 |
350,381 |
194.64 |
360 |
620 |
CPU RAM GPU GPU-RAM |
| en_de |
1.0.0+argos-1.0.0 |
1,800.37 (0h 30m 0.37s) |
10,622 |
5.9 |
390,650 |
216.98 |
377 |
716 |
CPU RAM GPU GPU-RAM |
| en_es |
1.0.0+argos-1.0.0 |
1,800.28 (0h 30m 0.28s) |
10,014 |
5.6 |
367,148 |
203.94 |
376 |
556 |
CPU RAM GPU GPU-RAM |
| en_fa |
1.0.0+argos-1.5.0 |
1,800.23 (0h 30m 0.23s) |
9,705 |
5.4 |
353,976 |
196.63 |
415 |
684 |
CPU RAM GPU GPU-RAM |
| en_fr |
1.0.0+argos-1.9.0 |
1,800.31 (0h 30m 0.31s) |
8,810 |
4.9 |
323,042 |
179.44 |
360 |
588 |
CPU RAM GPU GPU-RAM |
| en_he |
1.0.0+argos-1.5.0 |
1,800.34 (0h 30m 0.34s) |
10,481 |
5.8 |
384,201 |
213.40 |
416 |
972 |
CPU RAM GPU GPU-RAM |
| en_hi |
1.0.0+argos-1.1.0 |
1,800.51 (0h 30m 0.51s) |
8,527 |
4.7 |
311,633 |
173.08 |
393 |
652 |
CPU RAM GPU GPU-RAM |
| en_ms |
1.0.0+argos-1.9.0 |
1,800.26 (0h 30m 0.26s) |
10,518 |
5.8 |
383,238 |
212.88 |
354 |
556 |
CPU RAM GPU GPU-RAM |
| en_pt-BR |
1.0.0+argos-1.9.0 |
1,800.13 (0h 30m 0.13s) |
10,261 |
5.7 |
375,234 |
208.45 |
360 |
716 |
CPU RAM GPU GPU-RAM |
| en_pt-PT |
1.0.0+argos-1.9.0 |
1,800.23 (0h 30m 0.23s) |
9,462 |
5.3 |
347,015 |
192.76 |
360 |
652 |
CPU RAM GPU GPU-RAM |
| en_ru |
1.0.0+argos-1.9.0 |
1,800.30 (0h 30m 0.30s) |
6,445 |
3.6 |
236,080 |
131.13 |
581 |
1,132 |
CPU RAM GPU GPU-RAM |
| en_sk |
1.0.0+argos-1.9.0 |
1,800.29 (0h 30m 0.29s) |
9,286 |
5.2 |
341,311 |
189.59 |
359 |
684 |
CPU RAM GPU GPU-RAM |
| en_sq |
1.0.0+argos-1.9.0 |
1,800.27 (0h 30m 0.27s) |
10,569 |
5.9 |
385,420 |
214.09 |
366 |
556 |
CPU RAM GPU GPU-RAM |
| en_zh |
1.0.0+argos-1.9.0 |
1,800.32 (0h 30m 0.32s) |
9,774 |
5.4 |
357,984 |
198.85 |
363 |
652 |
CPU RAM GPU GPU-RAM |
8 instances, 1 threads
| Model |
Version |
Processing time [s] |
Number of processed files |
Processed files per second |
Processed segments |
Processed segments per second |
Memory peak [MiB] |
GPU memory peak [MiB] |
Graphs |
| en_ar |
1.0.0+argos-1.0.0 |
1,801.36 (0h 30m 1.36s) |
18,317 |
10.2 (1.3 per instance) |
673,029 |
373.62 |
409 |
2,980 |
CPU RAM GPU GPU-RAM |
| en_bn |
1.0.0+argos-1.9.0 |
1,802.40 (0h 30m 2.40s) |
11,878 |
6.6 (0.8 per instance) |
436,126 |
241.97 |
370 |
2,148 |
CPU RAM GPU GPU-RAM |
| en_cs |
1.0.0+argos-1.9.0 |
1,801.05 (0h 30m 1.05s) |
15,401 |
8.6 (1.1 per instance) |
566,360 |
314.46 |
371 |
1,892 |
CPU RAM GPU GPU-RAM |
| en_de |
1.0.0+argos-1.0.0 |
1,802.10 (0h 30m 2.10s) |
15,841 |
8.8 (1.1 per instance) |
581,968 |
322.94 |
397 |
2,628 |
CPU RAM GPU GPU-RAM |
| en_es |
1.0.0+argos-1.0.0 |
1,801.29 (0h 30m 1.29s) |
15,032 |
8.3 (1.0 per instance) |
551,371 |
306.10 |
390 |
1,636 |
CPU RAM GPU GPU-RAM |
| en_fa |
1.0.0+argos-1.5.0 |
1,801.99 (0h 30m 1.99s) |
14,609 |
8.1 (1.0 per instance) |
534,165 |
296.43 |
429 |
2,116 |
CPU RAM GPU GPU-RAM |
| en_fr |
1.0.0+argos-1.9.0 |
1,801.58 (0h 30m 1.58s) |
13,688 |
7.6 (0.9 per instance) |
502,242 |
278.78 |
370 |
2,532 |
CPU RAM GPU GPU-RAM |
| en_he |
1.0.0+argos-1.5.0 |
1,801.50 (0h 30m 1.50s) |
16,426 |
9.1 (1.1 per instance) |
601,247 |
333.75 |
427 |
2,052 |
CPU RAM GPU GPU-RAM |
| en_hi |
1.0.0+argos-1.1.0 |
1,802.15 (0h 30m 2.15s) |
11,564 |
6.4 (0.8 per instance) |
422,289 |
234.33 |
403 |
2,020 |
CPU RAM GPU GPU-RAM |
| en_ms |
1.0.0+argos-1.9.0 |
1,801.18 (0h 30m 1.18s) |
16,173 |
9.0 (1.1 per instance) |
593,173 |
329.33 |
365 |
1,764 |
CPU RAM GPU GPU-RAM |
| en_pt-BR |
1.0.0+argos-1.9.0 |
1,801.53 (0h 30m 1.53s) |
16,210 |
9.0 (1.1 per instance) |
595,914 |
330.78 |
371 |
1,828 |
CPU RAM GPU GPU-RAM |
| en_pt-PT |
1.0.0+argos-1.9.0 |
1,801.81 (0h 30m 1.81s) |
14,995 |
8.3 (1.0 per instance) |
552,106 |
306.42 |
371 |
1,860 |
CPU RAM GPU GPU-RAM |
| en_ru |
1.0.0+argos-1.9.0 |
1,803.65 (0h 30m 3.65s) |
8,452 |
4.7 (0.6 per instance) |
307,743 |
170.62 |
591 |
3,844 |
CPU RAM GPU GPU-RAM |
| en_sk |
1.0.0+argos-1.9.0 |
1,801.97 (0h 30m 1.97s) |
14,721 |
8.2 (1.0 per instance) |
540,390 |
299.89 |
371 |
1,924 |
CPU RAM GPU GPU-RAM |
| en_sq |
1.0.0+argos-1.9.0 |
1,801.47 (0h 30m 1.47s) |
18,624 |
10.3 (1.3 per instance) |
682,890 |
379.07 |
370 |
1,988 |
CPU RAM GPU GPU-RAM |
| en_zh |
1.0.0+argos-1.9.0 |
1,801.78 (0h 30m 1.78s) |
16,756 |
9.3 (1.2 per instance) |
612,439 |
339.91 |
373 |
1,796 |
CPU RAM GPU GPU-RAM |