Version 5.3.3 is overheating my laptop
DesBw opened this issue · comments
Current Behavior
I am sorry, I don't have any reproducible issue. But, Lstmtraining in the latest version seems to consume more energy; triggering overheating and shutting down of my pc. I never had this issue before (version 5.2).
Expected Behavior
No response
Suggested Fix
No response
tesseract -v
5.3.3
Operating System
Ubuntu 22.04 Jammy
Other Operating System
Ubuntu 23.04
uname -a
No response
Compiler
No response
CPU
Intel i7-3610QM
Virtualization / Containers
No response
Other Information
I used to run the training for days & weeks; millions of iterations, with no issue. Now, I cannot run it more than 50,000 iterations. It shuts down due to thermal issues (overheating).
10 Power Consumers
| Usage | Events/s | Category | Description |
|---|---|---|---|
| 286% | 11.5 | Process | [PID 46987] lstmtraining --debug_interval 0 --traineddata data... |
| 0.2% | 258.0 | kWork | dbs_work_handler |
| 0.7% | 108.6 | Timer | tick_sched_timer |
| 0.2% | 24.3 | kWork | blk_mq_run_work_fn |
| 0.1% | 23.1 | Process | [PID 15] [rcu_preempt] |
| 5.5% | 0.10 | Process | [PID 44751] /usr/bin/gnome-system-monitor |
| 0.1% | 17.4 | Interrupt | [4] block(softirq) |
| 0.0% | 12.0 | kWork | psi_avgs_work |
| 0.0% | 9.2 | kWork | ieee80211_iface_work |
| 1.4% | 1.5 | Process | [PID 6464] /usr/lib/xorg/Xorg vt2 -displayfd 3 -auth /run/user/1000/gdm/Xauthority ... |
No response
Please run tesseract -v with 5.3.3. Which AVX variants does your laptop support? Depending on your CPU AVX instructions require more power than other instructions. It looks like your tesseract binary was built with multithreading support (usage is greater than 100%). Running single threaded might reduce the power consumption sufficiently to solve the overheating problem.
Thank you for answering dear Stefan. THis is what I get when I run tesseract -v:
tesseract 5.3.3 leptonica-1.82.0 libgif 5.2.1 : libjpeg 8d (libjpeg-turbo 2.1.2) : libpng 1.6.39 : libtiff 4.5.0 : zlib 1.2.13 : libwebp 1.2.4 : libopenjp2 2.5.0 Found AVX Found SSE4.1 Found OpenMP 201511 Found libarchive 3.6.2 zlib/1.2.13 liblzma/5.4.0 bz2lib/1.0.8 liblz4/1.9.4 libzstd/1.5.2 Found libcurl/7.88.1 OpenSSL/3.0.8 zlib/1.2.13 brotli/1.0.9 zstd/1.5.4 libidn2/2.3.3 libpsl/0.21.2 (+libidn2/2.3.3) libssh/0.10.4/openssl/zlib nghttp2/1.52.0 librtmp/2.3
My processor is Sandy Bridge from 2011 (i7). So, it appears to support AVX (256bit).
Yes, and therefore it will also use AVX for the training process. Try OMP_THREAD_LIMIT=1; export OMP_THREAD_LIMIT before starting the training. The CPU usage should then stay below 100%.
Thank you so much for the help. I will do that.