Do you use a laptop or a desk computer for stacking and processing?

As you know, PixInsight is a powerful tool, but also a very demanding one. Here you are the minimum request that PixInsight Team recommend (I'm under it, and it works fine for me, but I need a huge amount of time to stacking with WBPP): https://pixinsight.com/sysreq/
Of course you can stack with other programs like DSS but if I have a PixInsight license, why not to use all its potential. And I find WBPP 2.4.5 a great script to stack.

Me first. I use two laptops. A 2012 MacBook Pro retina with an i5 CPU and 16GB RAM and a 2021 DELL Windows 11 i7 CPU with 16GB RAM (both 4 core). Of course, the Dell is quicker than the MacBookPro. Both of them have integrated graphics in the CPU. I'm wondering if the best option is not to have a desktop powerful computer like a gaming pc (that I can use for other purposes too, like work and play).

Greetings & clear skies!

Daniel.

Hello Daniel,
I use an 8 year-old HP laptop running Windows 10 with 8GB of RAM to run the mount, cameras and accessories in my observatory. I capture all the images to a 2TB external hard drive, which I bring in at the end of the evening.

For processing, I use an HP desktop, running Windows 10, with a 2TB hard drive and 32GB of RAM. My processing programs are PixInsight and an older version of Photoshop. Every once in awhile, Pixinsight will crash while running 'Blink', but I assume this is because the developers of Pixinsight don't like Windows, so there must be some sort of hiccup. It's not a problem, though, as re-booting Pixinsight fixes the issue for the remainder of the session.

Hope this helps add info.

- - Steve

Hi Daniel,

Use a desktop specs:
AMD Ryzen 7 3700X 8-Core Processor
64GB RAM installed of which 16GB is dedicated to a RAM swap disk only used by PI.
NVMe M.2 1TB for system games etc
NVMe M.2 2TB for Astro data processing.
And 2 NAS systems for cold storage and redudancy of astro data

CS Sean

While there are very powerful laptops and while there are very good arguments these days for most people to work only on a laptop and just forget about desktops, desktops still have an advantage with respect to processing power. Due to the size limitations of a laptop it's more difficult to deal with the heat of powerful cpus and graphic cards. Of course current laptops are amazing and well suited for most tasks. I still prefer to run cpu, gpu and RAM intensive applications on a desktop.

PI is indeed quite slow with image integration, on the other hand there is no reason to sit at the machine and wait for the result. Just start WBPP and do what you like until it's finished. Most other PI processes are reasonably fast in my opinion with a reasonably strong machine.

Clear skies
Wolfgang

I clicked on Laptop but I also use my main desktop PC. Both are pretty powerful but for very large projects (like the Bat and Squid nebula), I prefer the desktop because it is still more powerful than my Laptop. The desktop has got 64GB RAM and the Laptop 32GB as example…

However, even with my desktop, when I processed in total about 1500 subs, it took almost a day to get all done (with darks, flats, bias).

CS

Georg

Hi Daniel,

I just upgraded for the Mac Studio which is doing quite a good the trick ! I use Pixinsight, and as aforementioned, you better have a powerful desktop or just be very patient 😂.

Best regards,
Aygen

While there are very powerful laptops and while there are very good arguments these days for most people to work only on a laptop and just forget about desktops, desktops still have an advantage with respect to processing power. Due to the size limitations of a laptop it's more difficult to deal with the heat of powerful cpus and graphic cards. Of course current laptops are amazing and well suited for most tasks. I still prefer to run cpu, gpu and RAM intensive applications on a desktop.

PI is indeed quite slow with image integration, on the other hand there is no reason to sit at the machine and wait for the result. Just start WBPP and do what you like until it's finished. Most other PI processes are reasonably fast in my opinion with a reasonably strong machine.

Clear skies
Wolfgang

I run, professionally, very high end (meaning costing tens of thousands £/$/Euro for a single year-long license) programs that use gobbles of memory (from 32 GB to 256GB) for days on end and I do it on high end laptops. So I don't see why I shouldn't run PI on laptops. Given the choice I'd probably run it on a WS if the noise isn't ovewhelming, as the price point is more convenient, but that would restrict it do handle just that and I can hardly think of more of a waste of money and energy. At the end of the day I don't get paid for running PI faster, at all.

PI is indeed quite slow with image integration, on the other hand there is no reason to sit at the machine and wait for the result. Just start WBPP and do what you like until it's finished. Most other PI processes are reasonably fast in my opinion with a reasonably strong machine.

Clear skies
Wolfgang

Yes! That’s what I do.

44 lights, 24 bias, 24 darks, 24 flats, 24 dark-flats lasts 1h 48 min in the DELL laptop and more than 2h 30 min in the MacBook.

Greetings!

Daniel.

There is no real performance differences between laptops and desktops anymore these days. There are however advantages and disadvantages to weigh up against each other:

Desktops:

+ Cheaper to build and upgrade
+ Flexibility to upgrade - more storage ports, memory slots and easy cpu/gpu swapping
+ Better cooling capacity, both air and water options
+ Customisation, if you are into that, you can add all kinds of mods like rgb lighting, cool fans ect to personalise your system

- mobility, moving it around is not really practical.
- bulky system
- can be noisy and more power hungry

Laptops:

+ mobility
+ small footprint

- high end systems are very expensive
- limited upgradability

General system advise for an astro-processing system:

Get the most RAM you can afford
Get the highest core count CPU you can afford, put your money into these.

An expensive GPU, associated with gaming systems, has very little application in our software (for now).

If possible, isolate a dedicated high speed hard drive for your processing, and transfer everything once your done to cheaper high capacity storage for back-up.

Hope it helps!

Hi Daniel,

I have a linux workstation because my 2019 macbook pro was too slow. I just received a M1 Max Macbook Pro 16 that completely changed the game, the benchmark CPU is at 17000, TOTAL at 12300 ! I guess the AMD Ryzen 9 will be on ebay soon.
Integration testing, mure denoise testing, deconvolution testing,… are so fast that I can reach way better results.

There is no real performance differences between laptops and desktops anymore these days. There are however advantages and disadvantages to weigh up against each other:

Desktops:

...
+ Better cooling capacity, both air and water options
...

An expensive GPU, associated with gaming systems, has very little application in our software (for now).
...

I agree with most of your statements. However, the better cooling capabilities will always give the desktops a performance edge.
For most users and applications this is not much of an issue because modern laptops are way good enough. So I would phrase it like this: the performance difference is real but for most users not relevant.

Concerning the GPU I would like to stress the "for now" part. Neural networks play very well together with GPUs and with more and more algorithms using trained neural networks this technology may become more important.

Clear skies
Wolfgang

I found with PixInsight, it's all about CPU core counts, RAM, and swap performance. A desktop will be hard to beat in terms of price/performance. Also, laptops have heat management that will throttle performance during long running processes and sustained high load. I run a desktop with 16 core CPU, 64GB of RAM and NVMe drives. To give you an idea how much storage speed matters, just upgrading the drive improved my swap rate by 50%.

If you want faster Tensorflow-based inference (StarNet, etc…) your GPU will have an impact as well. Just about any RTX-class NVIDIA GPU beats CPUs hands-down.

No brainer for me … bang for the buck a desktop/Workstation wins every time.

* Laptops typically use slower processors that are optimised for battery usage and because of their limited cooling capabilities. A water-cooled desktop/workstation with the right CPU will blow any laptop out of the water.
* Desktop GPUs are much more capable and can be used to optimise processes such as Starnet ++. Eventually, PixInsight will start to use GPU acceleration and then it will make a big difference.
* Desktops usually have more expansion capabilities for NVme drives
* Desktops are cheaper
* I can design and build my own desktop to be optimal for what I need … that's almost impossible with a laptop

I rambled on in the "Storage and Computers" thread about the systems I use. The short answer is laptop, but expansion to desktop is in the works.

To counter Andy Wray's points above, which are mostly true: there is a lot to be said for portability. Especially if you do any remote shooting and want to spend some field time processing. Also, GPU prices make the "cheaper" aspect of desktops less true. Yes, the NFT collapse has created a GPU surplus that has resulted in marginal graphic card price drops in the last month, but it still adds significant cost to a home built system. Having just built my first system in 10 years, I was blown away to see four figure costs for graphics cards. I just saw the Nvidia 4000 series flagship is supposed to run $2900?!?!?! Yikes.

It's not helpful to describe PI as slow when the speed is absolutely determined by the computer being used.
Try running the Benchmark script then have a look at the results table at https://www.pixinsight.com/benchmark/

My laptop has Ryzen 5800H processor, 16gb ram and a decently fast SSD AND WBPP chewed through 30 lights and 10 each dark, flats and bias in under 7 minutes. The images taken with a ASI183 MC Pro 20 Mpx camera.

Michael

Michael:
It's not helpful to describe PI as slow when the speed is absolutely determined by the computer being used.
Try running the Benchmark script then have a look at the results table at https://www.pixinsight.com/benchmark/

My laptop has Ryzen 5800H processor, 16gb ram and a decently fast SSD AND WBPP chewed through 30 lights and 10 each dark, flats and bias in under 7 minutes. The images taken with a ASI183 MC Pro 20 Mpx camera.

Michael

When I wrote that image integration of PI is slow I meant it's slow compared to other programs like Deep Sky Stacker or Astropixelprocessor. Of course it's comparing apples to oranges since PI offers more elaborate options. Still DSS and APP usually deliver good image integration results in significantly less time.
I would be willing to bet that DSS and APP would deal with your 30 lights in less than 1 minute on your machine.

I still use PI because DSS failed on me once and when I had to decide between APP license and PI license I preferred the more powerful PI.

Clear skies
Wolfgang

I'm lucky enough to have a high-end Threadripper/Quadro/128GB RAM workstation that I built for scientific data analysis and calculations. It seemed silly not to use it for processing my astro data as well

With a CCD I have fewer subs with longer exposures so not quite same as many folk but it chews through the data in no time. PI isn't as optimized for lots of threads (APP much better in that regard) but it not sluggish in the least for post-processing, keeping in mind that I'm using it under Windows, the faster but smaller Linux disk is reserved for work. Now I think about it I might do a PI install though just to run the benchmark for curiosity. Starnet and any neural network based software is pretty fast with the GPU acceleration but then it should be given how much the GPU cost!
CS
Pam

I use a Lenovo IdeaPad 3 laptop with a i7 2.8GHz core upgraded RAM to 20GB, 1TB HD and a 2TB external drive for backup, Windows 10 for my capture and processing. I purchased it last year and runs smoothly for all my processing software including PixInsight and Topaz. I think I got it for around $700 US + RAM and external HD. I try to stay off the internet as much as possible with it and use my older laptop for surfing.

Eddie

I used to do stacking and processing on Lenovo ThinkPad P40 with i7 processor, GPU and 16 gigs or ram. It was fast but would heat up significantly during the star alignment, image integration and star mask generation etc. Then I upgraded to Lenovo ThinkPad P1 with i9 CPU, GPU, 32 gb RAM and, most importantly, color calibrated UHD screen. It is not a cheap machine but it is lightning fast! I can integrate multisession set of LRGB subs within 10-15 minutes. Processing in PI is absolutely fluid. I also use M1 Macbook Pro and Lenovo is uncomparably faster and smoother, because PI is not adapted to M1 chips yet.

D

I use my M1 MacBook Pro (13") for everything: acquisition, stacking, processing. A desktop would be better suited, but I like the portability. The M1 system is pretty quick with stacking, it doesn't really take an obscene amount of time although that depends on how many frames you're integrating of course. What impresses me is that Pixinsight is not a native ARM application and yet never feels sluggish. I can only imagine what a native build would be like.

I agree with most of the comments here, but IMHO I still feel that a desktop is the best way to go for performance. I recently updated from an i5 6400, 16gigs ram and 4gigs Nivea GPU OS/ windows 10, and have been blown away at how much faster the new machine is. Its configuration is as follows: Ryzen 9 ADM 5950x 3.4mgz water cooled 16 cores, 3060 12gig GPU, 96gigs, DDR4 viper ram, and I use an 512 ssd drive and use the 1tb internal storage, with 6tb external storage. Windows 11 OS

While it does run incredibly fast, I do have PI crash on me after a long run in and out of programs. This happened with me with my old i5 which is the reason I decided to upgrade to the Ryzen 9. In talking with several people on PI I’ve found many others have this same problem and I’m lead to believe its a windows problem. I have communicated back and forth with tech support for PI and my personal experience is they will not admit there is issues with there software.

I’ve been told that there is a Linux environment that you can use on a windows OS that can then allow PI to run under its native OS. I don’t know if it will take the issue I have a way but I’m considering trying it. For now I just save my “projects” now so that I loose only a little should the computer crash on me.

Still though, its a bummer!

Dale

Greetings!

There is no alternative to a desktop computer - faster and you have big hardware, colour calibrated EIZO screen, that all you don't have on a laptop,
even with a standard screen you have the advantage of a 27 or 30inch screen, I don't like to process an astro-image on a 13-16inch screen.
:-)

Clear skies!

Dale Penkala:
I’ve been told that there is a Linux environment that you can use on a windows OS that can then allow PI to run under its native OS. I don’t know if it will take the issue I have a way but I’m considering trying it. For now I just save my “projects” now so that I loose only a little should the computer crash on me.

They keep singing from the same hymn sheet I see. They'll never admit their hardware coverage is dubious to say the least. Do not bother switching to Linux. I did the same and didn't fix a thing. I had to buy another machine (Windows) and it never crashes.

andrea tasselli:
Dale Penkala:
I’ve been told that there is a Linux environment that you can use on a windows OS that can then allow PI to run under its native OS. I don’t know if it will take the issue I have a way but I’m considering trying it. For now I just save my “projects” now so that I loose only a little should the computer crash on me.

They keep singing from the same hymn sheet I see. They'll never admit their hardware coverage is dubious to say the least. Do not bother switching to Linux. I did the same and didn't fix a thing. I had to buy another machine (Windows) and it never crashes.

Thanks for the response Andrea,
many have told me the same thing. I’ve made many different changes to my machines configuration and while it definitely helps quite a bit it still will crash, but this is mainly after 5-6hrs if continuous use and in and out of programs.

Dale

We should not turn this into a Windows vs. Linux/Unix thread. In my professional life I worked with a lot of operating systems and while there is a good reason that most of the large scale business applications run on the Unix OS family I would not be willing to join the Windows bashing community. In all cases a smooth running OS requires some maintenance, in particular keeping the hardware drivers up-to-date and having the latest patches installed. I have experienced very few crashes on my windows machines over the years and at least as many "hangups" on MacOS. While watching a colourful rotating wheel is nicer than a blue screen with a cryptic error message (remember the infamous: "Who is General Failure and why is he reading my hard disk?") both is not what you want to see. There is no such thing like bug-free software so the occasional crash is part of life - just like your mount not guiding properly. If it happens frequently or even regularly then something is wrong with your setup.

If you want to run PI under Linux the most straightforward way is to create a Linux USB boot stick. The next step would be to create a linux partition on your PC:
How to Dual Boot Any Linux Distribution With Windows – and Get Rid of It When You Need To (freecodecamp.org)

You could also run Linux in a Virtual box under Windows but this just introduces an additional layer which may fail. I would prefer the dual boot solution.

Clear skies
Wolfgang

For staking and process your astrophotography, do you use a laptop or a desk computer? Please feel free to explain which equipment do you have and how powerful is it (CPU, RAM, OS, Graphics).