Advice needed: hackintosh build for extreme lightroom performance?

[iLLucionist]

From my experience, Lightroom is more disk intensive rather than cpu intensive.

Rapidly going through your photoshoot pictures, ranking photos, color-coding, and doing quick adjustments on RAW files, with file sizes that are tens of megabytes in size each, you want at least a 7200rpm drive, or better yet a fast SSD drive when working on your pictures.

Also best to have a separate hard drive to store your photos from your OSX system drive. It helps.

Lots of memory, 24GB at least. 32GB will be better. (when you edit your RAW from LR to Photoshop, it's a really huge TIFF file).

Thanks! I won't go for less than 32GB. Do think 64GB is even better? Or is 32GB the sweetspot?

For storage I'll try 950 Pro m.2 512 GB for OS and 850 Pro 1 TB sata 600 for photos and other data.

 

[pastrychef]

Thanks! Quick question.. you said you overclocked your CPU to 4.7Ghz. Do you need to overclock memory as well?

Why did you settle on G.Skill Trident Z Series? Did you consider Kingston Hyper Fury X / Savage or Crucial Ballistix LT?

I really want to know what memory is best to go with, especially if I'm overclocking the CPU a bit.

Thanks!

No, you can leave memory at stock speeds if you like.

Yes, I went with the G.Skill Trident Z because they were on the Asus verified memory list. No, I didn't consider anything else.

If you use M.2 SSD, remember that NVMe is not natively supported by macOS and you will need to install additional drivers before macOS will recognize it. AHCI M.2 SSDs are natively supported and no additional drivers are needed.

 

[iLLucionist]

No, you can leave memory at stock speeds if you like.

Yes, I went with the G.Skill Trident Z because they were on the Asus verified memory list. No, I didn't consider anything else.

If you use M.2 SSD, remember that NVMe is not natively supported by macOS and you will need to install additional drivers before macOS will recognize it. AHCI M.2 SSDs are natively supported and no additional drivers are needed.

Thanks for the quick reply, again! I've read somewhere that 950 pro comes in two versions... nvme and ahci, which can be seen on the serial number (V for nvme, H for ahci), and should be supported. If I have to get the nvme-version, is it going to work? Or is it not worth the effort and too much fiddling around with files / settings?

 

[pastrychef]

I have never used NVMe M.2 SSDs. I've always made sure to get AHCI versions. From what I've read from others' experiences, it can be a PITA. I prefer to keep things as simple as possible.

 

[powerpcg5]

Thanks! I won't go for less than 32GB. Do think 64GB is even better? Or is 32GB the sweetspot?

I originally had 16GB on my machine, then increased to 24GB after finding some extra RAM sticks. In my case, going from 16GB to 24GB already reduced my disk "Page Outs" to 0.6%.

And this is even with Windows 10 Parallels (+ Visual Studio) virtual machine + Photoshop + Lightroom running all at same time...
and still OSX reports 14GB free space (40% Pressure)

I was planning on 32GB but seeing the numbers, I can see 24GB is enough for me, even with my permanent VM always running in the background.

 

[iLLucionist]

I originally had 16GB on my machine, then increased to 24GB after finding some extra RAM sticks. In my case, going from 16GB to 24GB already reduced my disk "Page Outs" to 0.6%.

And this is even with Windows 10 Parallels (+ Visual Studio) virtual machine + Photoshop + Lightroom running all at same time...
and still OSX reports 14GB free space (40% Pressure)

I was planning on 32GB but seeing the numbers, I can see 24GB is enough for me, even with my permanent VM always running in the background.

View attachment 216267

Thanks! This is very informative. I guess I'll stick to 32GB for now then and save me some money.

 

[sentineljones]

While I've built many PCs long ago, I'm currently finalizing my parts list for my first hackintosh. As noted above, LRs biggest problem for me has been disk i/o, and I suspect that is (or was?) your pain point as well.

I run a 2010 MacPro, with 2x3TB 7200RPM drives set in RAID1 (mirror) for all my project files, including my LR catalogs. It's painfully, agonizingly slow. LR's problem doesn't seem to be memory (32GB) or CPU (2x2.4GHz quad core xeon). While upgrading the video card in this old beast from a stock whatever it was to an nvidia 6xx w/ 6GB of VRAM helped a little, the disk noise makes it apparent that's where biggest my issue is.

I don't think it's just the old magnetic drives, I think the older system hardware/architecture has a lot to do with it as well.

 

[normski321]

So am jumping in on this because am wondering if what I am seeing makes sense.
I recently built my Hackintosh. Think profile has details but basically the CustomMac Pro with 32GB ram, Nvidia 1060 6GB graphics card and I6700K CPU. Have boot SSD but have my Lightroom catalog and photos that I am working on on a separate 500GB NVMe M.2 SSD. Goal was a quick machine for photography purposes.

When I benchmark my machine, CPU, Heaven and disk speeds are all where they should be i.e. fast and all components work correctly within OS.

When using Lightroom however, I notice photo 'loading' is noticeably quicker when I turn the graphics card 'off' in the Lightroom preferences rather than on. When 'on', I see the 'loading' message for about 0.5sec before the image is sharp as I move between photos in the catalog. When the card is 'off', loading is pretty much instantaneous. However, when editing photos in the 'Develop' module, the changes are smoother when sliding adjustments with the graphics card turned on than off. Editing 1:1 previews I see the same trend although everything is quicker for both graphics card being 'on' and 'off' within Lightroom preferences.

Am wondering if there is an issue with Lightroom / Graphics card particularly related to the speed photos are 'loaded' and then displayed on the screen? Hopefully this makes sense...


UPDATE: Have, I think, answered my own question:
Turns out certain processes are optimized in LR for graphics cards and some are not. Full link here
" GPUs are marvelous at high-speed computation, but there's some overhead. For example, it takes time to get data from the main processor (CPU) over to the GPU. In the case of high-res images and big screens, that can take a LOT of time. This means that some operations may actually take longer when using the GPU, such as the time to load the full-resolution image, and the time to switch from one image to another."