"BETTER WBPP SPEED"

Faus Márquez:
Hola Ferran, básicamente cambiar de ordenador o comprar un portátil con la CPU de Intel o AMD de 7 ó más y con 32GB de memoria RAM. Si tu ordenador es de sobremesa y no quieres cambiarlo puedes mirar ampliar la memoria RAM todo lo que te permita tu placa base eso disminuirá los tiempos de espera drásticamente.
Una pregunta Ferran, vi un correo que me llegó firmado por ti comentando el tema de la calibración de color pero no tenía tiempo y ahora no lo encuentro me podrías poner el enlace, es un tema crucial para dar con el color adecuado en nebulosas muy complejas a veces cargadas de estrellas.  Gracias y un abrazo

Faus Márquez:
En castellano por favor, y ya que estamos me ha surgido un problema que paso a relatarte, no sé si esta aplicación sirve para banda estrecha porque en el momento que uno las diferentes imágenes en RGB, la imagen resultante pierde los datos astrométricos y si quiero hacerlo uno por uno la aplicación me dice que solo vale para una imagen en RGB, ¿como lo haces tú en un trabajo de banda estrecha? Muchas gracias

Some people may not like my answer, I give it anyway: Investing in more RAM and faster SSDs is certainly a great decission if you are a power user and use a bunch of software which profits from it. If WBPP is the only use case on your computer which bothers you, investing in another specialized stacking software may be a better (and smaller) investment of money. Personally I gave up on WBPP and stack my images with Astro Pixel Processor. I’ll never look back. I don’t want to turn this into a “WBPP vs. other software” thread though. Everyone should just use what he/she likes.

Clear skies

Wolfgang

It kinda depends, like some folks suggested. And there are different points for each bottleneck to become an issue.

For WBPP it is most commonly CPU cores.

You need to have enough memory to complete the task. This tends to spike for some specific operations like drizzle integration. Having sufficient memory available can also reduce the need to lean on slower storage for temporary files (including by the OS) which can offer other performance upsides. The amount of memory needed depends on factors like quantity of files, resolution of files (including upscaling with drizzle), and settings. Memory can also be affected quite a bit by OS/CPU architecture (e.g., an Apple M processor uses memory much more efficiently than x86-64 due to hardware design).

Storage speed can matter too. Generally it is best if the OS and PixInsight are running off a fast SSD. This helps a lot with temporary/swap files and other various I/O operations. Where any additional or alternative working directories are created (PixInsight preferences) matters as well. Having very fast and sufficient storage media here can help to ease the consequences of memory constraints and can make a big difference in performance if memory is stressed. What matters a bit less is the working drive for a stacking project. These I/O operations during stacking add up, but usually the bottleneck is decidedly in processing rather than file operations. However, an SSD helps here (doesn’t need to be exceptional) and using something like a NAS for the project working drive can be very problematic. A NAS or an external spinning platter hard drive is great, and much more economical, for archiving project files, however.

So as long as there are no major failures in those regards, the bottleneck ends up being CPU cores. WBPP, in most operations, assigns processing of one file to one core. If it has access to 8 cores, it will process 8 files at a time, and it will typically ramp those cores to 100% for that processing (100% CPU utilization is normal). You can process more files simultaneously with more cores. And faster cores help to process each file faster. ’Course, more cores can also ramp up some of the other resource demands.

GPU is not very relevant for stacking. A GPU which can accelerate machine learning processes (NVIDIA CUDA or the Apple M chips) helps distinctly with those related tools, like StarXTerminator.

Unfortunately, once you’re actually processing, the number of tasks which are multi-threaded starts to fall off, and you can run into some single-threaded tasks that want a fast processor core and may lean a fair bit on some other resources. So it may be worth weighing that a bit given you can walk away and do something else during WBPP processing, even if it takes a while, but some of those other tasks may be ones where we are staring at the monitor.