@Soupy Do you plan to build (or at least assemble) it yourself?
Thanks! So about 18% faster for those two modules.
Its a small sample size, but if we were to use the average of those two modules as representative of all modules, and we used 20 modules per image, then 8 cores would take ~26 seconds per image, and 6 cores ~32. If we had 1000 images, that is ~7.2 hours compared to ~8.9. So for individual images the difference between 6 and 8 cores is not that significant, but for batching large amounts it is.
Is the moral of the story to put the cache folder on the shared drive available to both OS? Or to keep it separate? Most of my RT work will be done on the Linux side, but it may be convenient to install it on both.
Is the benefit of 32 over 16 only noticeable when you have multiple programs opened at once, or is it noticeable within just a single program?
No, that is unfortunately beyond me at this stage.
Both, depending on the situation and/or program(s) used.
Here are some of my experiences:
I noticed that you seem to use the same programs that I do (or at least have an overlap): GIMP, RawTherapee and darktable.
GIMP is rather memory hungry on its own. The average camera nowadays produces 18-24 MP shots and if you are using just a few layers to get the result you want/need (final touches after the RAW edit) you are fine, but a larger workflow with many layers and/or layer groups will benefit greatly from more memory. I had the pleasure of working with a D810 the other day and those extra GB’s are very welcome when working on 36MP shots!
If using darktable or RawTherapee on their own 16GB is enough. As mentioned I often use multiple versions of one program (dev vs stable) or have multiple editors open at the same time. In that case 16GB is probably not enough.
A while back there was a topic here that dealt with recreating the style of a Russian photographer. I started my own little project and one of the steps was averaging 297 24.4MP shots. I wasn’t able to do that with HDRMerge or Hugin due to memory limits. ImageMagick’s convert was able to do it though, although slowly.
So, in the end it all comes down to what it is that you (want to) do with your newly build machine. Only you can actually answer that. Do hope that I gave you some footholds.
PS: I can confirm from experience what Ingo has mentioned: RawTherapee definitely leans more towards CPU usage compared to darktable (opencl and all that). GIMP is a bit of a mixed bag: It does have (experimental) opencl, but I turned it off: Too many crashes when doing B&W stuff.
EDIT: This is the topic I was talking about: How to emulate Titarenko's long exposure stacking with digital?
1 Like
Exactly.
1 Like
It’s helped a great deal, thanks very much. I initially had 16gb pencilled in as it seemed enough for both RT or DT on their own, which is how I use them. But I hadn’t taken into consideration higher MP cameras, or multi-stacking projects - which I haven’t done a lot of, but will do more of in the future. So now I think 32gb is the way to go.
Based on the way I use each program, and pricing considerations at the store, I think I will prioritise GPU and darktable for very large batches, but continue to use both for single images or small batches. My new system will end up quite similar to yours. Just need to determine if the price increase from gtx 1660 super to rtx 2060 is worth it. gtx is about 73% cost of rtx. And like you, taking Ryzen 5 over 7 might make it worth it.
Hm. Permit me to have a different opinion.
Nowadays, it is not difficult to assemble one’s own computer.
I have just built my third real thing. And if I can do
it, so can you…
Have fun!
Claes in Lund, Sweden
1 Like
+1. I’m about to assemble my umpteenth one. Just takes a screwdriver, and a bit of hand-holding. Here’s a decent hand: How to Build a PC (2024): Hardware Suggestions, Instructions, and More | WIRED
Yeah, just make sure you don’t wear wool socks 
Jokes aside, I agree - it’s not difficult at all, just make sure to watch some brief tutorial video if you’re doing it for the first time. The whole process takes about 30 to 60 minutes. Hope I didn’t sound like bragging, I just wanted to motivate to do it yourself if you hesitate @Soupy .
1 Like
I believe you. I am more worried about selecting incompatible parts, because I don’t know about all the hardware. To solve that I could perhaps get a quote from the store to see what they come up with, then just get it myself. But its a time/cost thing. Time it takes to learn how to do it all myself vs cost of getting store to do it. In my life right now, time is a little more precious than cost, so I don’t think I will learn just yet. Maybe my next machine, in another ten years time (hoping this one will last that long, Haha).
PCPartpicker can help you there. It gives for each component selection criteria, and when you select a part for your list, it will tell you about the compatibility. Start with a CPU, and it’ll offer compatible alternatives for the other parts…
3 Likes
plenty of forums about the place full of people who are more than happy to not just check over your build but throw in their 2c to improve it.
it does still take time of course, and depending on where you’re located the part that takes the longest can actually be finding the retailers that stock all of the parts you want for a decent price and to minimise shipping costs.
either way have fun with it, not much can beat a new computer 
Yes, I do understand your reluctance to put together the Super Build yourself.
Yes, you will make a faulty decision here and there…
Just to make this thread more valuable for future forumers: these are my worst two goofs
in putting together my Own Computer:
- Power supply was powerful enough – but it did not have a sufficient amount of connectors for the motherboard I selected.
- Picked a Linux distro that could not handle a modern Nvidia graphics card (TU116).
My main goal was to build a silent computer having RawTherapee, darktable, and The Gimp as main tasks. Size was of no importance (I picked a tower).
Step 1: select CPU (in my case a Ryzen 3900X).
Step 2: pick a suitable motherboard.
Step 3: … follow the recommendations on sites like https://pcpartpicker.com or similar, like the Swedish Datorbyggare - Inet.se If I had followed them I would have been able to eliminate Goof #1, above.
When you have decided, watch unboxing/assembly videos on YT, dealing with exactly the same components as you picked…
Have fun!
Claes in Lund, Sweden
1 Like
And +1more
It does, even you have a high powered machine like today’s Ryzen or core i9.
For long time the 1660 had the best price performance ratio.
Today I’d start with something cheap, maybe even 2nd hand and wait for the newly announced RTX 3000 series. Like @Claes said (indirectly 
) , just make sure you use nvidia-drivers instead of Nouveau
I agree. With my Ryzen 9 and nVidia RTX 2060, I can tell a difference with or without openCL.
Ive researched more and think I’ve been convinced to upgrade to the rtx. Not sure I’ll wait for the 3000 series, prices are sky high at local stores while 2060 prices are good. Maybe the 3000 is helping bring the 2060 down. Ryzen 5 is also at a good price while ryzen 7 is not, so I’m leaning towards the 5. Either way, this is going to be a significant update on my current machine.
1 Like
If that is the case, you can just follow the spec from @Jade_NL. I saw several performance outputs and they are marvelous.
Just make sure you get a 2060super or 2070super (both where similar priced here).
For the RTX: I was a fanboy of Gigabyte, because their winddorce with 0% Fan speed where cool. Since RTX they are (still on 2070super) struggling with the oscillating fan issue. Very annoying. So stay away from that 
1 Like