3.91
#...
·
·
2
likes
|
|---|
|
Hi, This is meant to be an instructional post. As couple of month ago, after quite a while processing images in Pixinsight manually, i.e stacking darks, stacking dark for flats and and calibrating flats, etc... I took the plunge and started using the Weighted Batch Pre-Processing script in Pixinsight. WBPP has a learning curve, but after doing it a couple of times, you get the hang of it. Great script, I really enjoy it. Therefore, something that would take me 6 hours before to do manually was now down to 30 minutes to an hour. Great, I thought. I can dive into my archive and re-process some of my old stuff. Here is what I did wrong in using the process however: 1) I got excited and starting re-processing my archive of files dating back to 2016. Did lots of them. 2) I did not have the right computer for the task. 3) I did several runs one after another - that was great. I had all kinds of data to play with and produce new images, and none of this tedious pre-processing. Then disaster struck. In retrospect, a tell-tale sign would have been the computer shutting down on my in the middle of a batch....Hint - it was overheating ! As a matter of fact, it overheated quite a bit and worse - continuously. I was batch processing on a physical hard drive HDD, archiving on a second hard drive. Both these drives failed, including my archive, which was then lost... As far as my processing hard drive, it was completely seized...Could not retrieve anything from it. Computer could no longer access it. The repair shop told me it was toast. I also damaged the motherboard on my computer from the overheating. Two of the SATA ports are damaged - you guessed it - the ones connected to my HDD. Therefore, I was sadly left with no choice but to get myself a new computer. I also had lost my archive of images in the process. After reading the recommended requirements for Pixinsight, I believe I came as close as I can to the "ideal" computer to do the job: New I9-14900 processor. with 24 cores. (My old one had 8 cores) 128 GB RAM. 8TB storage (only) hard drive. 2TB SSD for system 2TB SSD for processing only. Now, with processing off a solid state drive (SSD), there is a lesser chance of overdoing it. It is faster too, as a side benefit. Since then, I have been running WBPP very gingerly and keeping a close eye on my processor temperature. So far, so good. Again, this is my fault really, for not being familiar with the minimum requirements and also being excited and using the process to excess. Hopefully, this can help others. PS, I got most most of my archive back, from different stashes I had around my place.... |
0.00
#...
·
|
|---|
|
Where did you get your new machine from? Have you got a link at all please? |
|
3.91
#...
·
|
|---|
|
Custom made locally from a computer dealer. Motherboard based on ASUS, if that helps. Added CPU, RAM and hard drive. Also added liquid cooling. |
0.90
#...
·
|
|---|
| Liquid cooling is a great idea for a powerful computer that will be running intensively. |
0.90
#...
·
·
1
like
|
|---|
|
What was your computer? 😳 CPUs are supposed to have thermal protection. They should, in theory, throttle down when overheating, shutting down as a last resort. I can't imagine software such as PixInsight can lead to such damage. Was your CPU overclocked, or did you have an insane amount of dust inside the case? |
0.90
#...
·
|
|---|
|
Actually I'm having a strange issue with WBPP which might be related to CPU load. I'm not doing many batches one after the other, but the batches I do are still quite a challenge: each sub is 205 Mb (it's a huge sensor chip) and, with master flats darks etc all pre-made, it still takes a long time to process. My last run with 63 frames took 3h 41m. And I get horizontal black lines of about 20 pixels height on a few of the registered subs (which don't appear if I do them individually). This is still under investigation on the PI forum, but may be related to processor load. My processor isn't exactly sluggish: its and Apple Silicon M1Max and scores well on the PI benchmark. |
15.85
#...
·
|
|---|
|
I have a very similar setup with the exception mind is the Ryzen 9, 128gb ram as well as 3060x GPU, I also have water cooling on mine. You should be very happy with your setup! I’m sorry you cooked your computer but in the end you got a very capable setup now that you don’t have to worry about! Congrats on your new computer! |
|
3.91
#...
·
|
|---|
|
Some of you seem to be doubtful that the computer cooked. Here is what happened: I had a batch of archives of about 150 subs for LRGB. That is 600 subs. Add to that 50 Darks, 50 Flats, 50 Dark for flats for each channel, you soon have over 1000 files to process. I got greedy and processed several imaging sessions - like 10 in an evening. Each session about 30-45 minutes to run. That is a lot of files to access on a HDD. I did this for several nights in a row. Sometimes the PC would shut down on overload. Sometimes it would not. I did not clue in. I did however, monitor the CPU temperature after a while and it was operating at about 91-95°C. That chip was designed for 80° maximum. From what I read, the thermal protection on the motherboard does not kick in until it hits 100°C for a short time. So the load was not severe enough to kick out in overload on a particular run, but he prolonged exposure to higher than normal (sustained) temperatures eventually fried the CPU. As for the HDD, I should have been using an SSD. The poor drive could not keep up with that many file read/write on a sustained basis. By the HDD failing, it also damaged the SATA port on the motherboard. The PC still boot up, but the temperature immediately climbs to high 80's°C. So not safe to run for any length of time. certainly not for processing images... I am not a computer guy, but I know what happened is certainly not normal. It was however subjected to a lot of sustained abuse for a while. It just decided that it had enough of this and quit ! |
|
0.90
#...
·
·
1
like
|
|---|
|
I was not doubtful; these things can obviously happen, especially with Intel-like architectures, as they are so bad at thermal efficiency. I was just surprised it could happen using PixInsight on a non-overclocked CPU without a defect of some kind. A CPU should start throttling down way before hitting 100°. In my experience, I used computers for years with repeated hours-long compilations without any issue, with the CPU hitting almost 100° every time. They were at about 75° with absolutely no load. This is usually not an issue; unfortunately, it can be pretty standard on some Intel chips. So, I was just wondering if your PC had some specific defect that could have led to the problem you experienced. If some ports on your motherboard are damaged, it's most likely scrapped. But if the PC still boots, the CPU should be OK. The higher-than-normal temperature probably indicates that the extensive usage caused some issues with the thermal paste. However, reapplying fresh paste should eliminate the issue. Anyway, congrats on your new PC; this is a good setup! |
1.81
#...
·
|
|---|
| i think in the PI forum it was discovered that some motherboards have default BIOS settings that are far too aggressive and set up the CPU to be able to overheat and be overclocked for too long under load. i can't remember the exact manufacturer or what generation CPU we're talking about here. but i guess the point is that a lot of the thermal throttling parameters are under control of the BIOS and as such could be misconfigured by the manufacturer that is trying to give you "extreme" performance by default. |
