CPU upgrade for PL8 on Windows

Hi, long time lurker here…

(Have done as much research as I can, including searching the forum, but always happy to be pointed to another thread I have likely missed)

Is the consensus that for exporting, it’s beneficial to have a CPU with higher core count, and for general usage (moving between modules, adding effects, zooming…etc) it’s better to have single core speed?

I’ve recently upgraded my GPU to a 4060, which has significantly sped up denoising :ok_hand:, however I’m in the market for a new CPU and am looking to stay with AMD’s AM4 platform but with one of their more beefy CPUs (5900XT, 5950X, 5700/5800X3D).

For me personally ,waiting a little longer to export isn’t a problem if the interactions I have with the UI are super snappy. I like to edit quickly as I lose track of where I’m going with an edit…

I’d like to add my current Ryzen7 3700x runs fine, albeit after some careful CPU and memory overclocking.

Also discovered that disabling OpenCL speeds up the UI in general on my system.

I still can’t find any definitive answer as to core count/single core speed question, and would love to be educated.

I don’t have a definitive answer for you but while upgrading the Cpu will improve performance across the whole of your system, I’m not sure you’ll notice much improvement in export times from PL by doing that.

Adding more RAM will also improve performance across the whole of your system but again I’m not sure it will make much of a difference to the export times from PL.

It is the capabilities of the Gpu that determines the export times from PL but you’ve already got a high end GPU so there’s not much more you can do there.

RAM doesn’t speed up Photolab at all.
The CPU speed will help the editing speed. I still think there are inefficiencies in the software that will prevent “blazing” editing speed.

I have a 5900HS based laptop and sometimes put the CPU in a low power state (TDP <25W) and the editing speed is only a little bit slower than when the machine is at maximum performance (turbo).

The Ryzen 3700 is no slouch. The 5xxx series is definitely faster but I would expect a night and day difference in performance.

A tiny backstory, I came across the (not really an) issue when editing a photo of a car. When all the lens correction and sharpening within masks and everything else was finished, 100’s of tiny specs of dust suddenly appeared after export, which I’d missed, or disregarded as it didn’t seem to distract too much. So when I finally saw everything light up I went a little crazy with the healing tool.

In this particular case, loads of specs of dust allowed me to just zoom in and click away over the specs as if i was playing starcraft or some other ‘click heavy’ game for 10 minutes, and everything slowed down a whole lot, enough for me to investigate CPU usage and so on, which led me to overclocking it a bit, which definitely helped.

Re: the RAM, I have 32GB at 3600MHz so enough capacity and fast enough speed, and the overclocking/tuning part was to just lock the timings to those of the CPU for stability. I’ve never had DXO come close to using it all. It does however seem to increase as the edits go on, and never really goes down once usage has risen… There are a few threads on here with a similar theme.

Exporting takes an average of 11 - 13 seconds per 42MP Sony A7Rii snap, with Denoising and masks and most of the Customize modules enabled at 100% quality JPG (with no resizing). Which is plenty fast enough for me.

I’d be very interested to know if newer CPU’s can handle the same ‘abuse’ more efficiently, purely through newer architecture, and how it affects usage. Also would still love to answer the more cores/ better single core question, but I reckon only the devs can answer that one.

Also SUPER interested in PL9 (if/when it emerges) and if there will be some improvements from within.

I’m really not trying to moan, I’m having a great time using the software, just a stickler for squeezing out performance :slightly_smiling_face:

*just edited some awful spelling

That could be a consequence of the fact that PL does not show the fully sharpened image at zoom levels <75%. This is a very long standing source of constant gripes, e.g. this topic started in 2018:

There are many other topics about this feature of PL.

The new loupe in PL8 at least allows you to view part of the sharpened image at a high zoom level.

That’s another known issue, repeatedly using the retouch tool making everything grind, e.g. this post by @Joanna:

BlockquoteThat could be a consequence of the fact that PL does not show the fully sharpened image at zoom levels <75%.

It may have also been the result of JPG interpolation or something like that (really not sure). The loupe tool really did help after i’d spotted it, but again, not a huge issue, just a bit of a pain going back and forth when the specs looked so lit up in the jpg and not in the loupe. Maybe I wasn’t being meticulous enough…?

Again, all of my ‘problems’ are not really that bad. I feel like this software is built for out and out quality, which it delivers in droves. The speediness of other software I’ve used doesn’t outweigh. Just trying to save myself buying and returning multiple CPU’s.

Thank you for the links to the other threads, I’ll have a read after work