GPU USELESS For Lightroom Classic!

Started 3 months ago | Discussions thread
IanYorke Veteran Member • Posts: 4,325
Re: GPU USELESS For Lightroom Classic!

Batdude wrote:

skyglider wrote:

Batdude wrote:

skyglider wrote:


I'm wondering. When you say import/export process is a LOT faster than your previous i7-6700, are you just using the i7-6700 as a reference to that computer?

Correct. Just to give you an example, my previous system would not import 1900 RAW photos in one minute and fourty seconds.

Or are you attributing the lot faster import/export speed to the difference between your AMD CPU vs the Intel i7-6700?

Isn't that the same thing as above?

No, not the same thing. First case is like putting a name to a computer. If your i7 PC was named Sam, then saying that the 3900x PC (the whole system including fast SSDs) can import/export files a lot faster than your Sam PC (that could have slower SATA SSD or mechanical hard drive).

In the second case, it's like saying that the import/export speed difference is attributed to the 3900x CPU being a lot faster than an i7-6700 CPU (not taking into consideration the type of storage devices being used).

Honestly not criticizing, Just trying to clarify the meanings.

Yeah no worries I understand what you mean. The Ryzen 3900X itself is what's making import/export a lot faster indeed. The intel system also has really good stuff in it, great fast M.2 SSD and really good RAM as well and is no slouch and my loving sister is getting a heck of a PC

The Ryzen CPU is working like a dog together with some memory. The bastar* GPU is just sitting there eating pop corn and enjoying the show. Lightroom itself has a LONG way to go and they seriously need to change the way this software performs over all. It is definitely not the fastest.

Seems to me that import/export speed is a function of the storage device (in your case two 1TB M.2 SSDs) and that the CPU has little to do with import/export speed.

From what I'ms seeing is both. CPU AND storage are very active when I check task manager.

The really weird thing I'm experiencing right now and I don't know what the hell is going on, is that transferring files from file to file, even within the same M.2 C: drive is really slow man.

I assume you're not using file compression on any of your drives, you're copying files from one internal SSD to the other internal SSD (not using USB).

Try these things:

  1. Disconnect all cables connected to the PC except monitor, keyboard and mouse. (In case a USB device is causing the slow down)
  2. If you have an optical drive connected to the motherboard via a SATA header (not USB), disconnect it. (It could be affecting the data bus)
  3. Disable your antivirus and see what hapens. (in case it's real time protection is slowing things down).
  4. Disable Win10 search indexing (I hate that it slows stuff down. I always disable windows search indexing period.)
  5. Remember that moving a file on the same SSD or hard drive is always faster than copying the file. Moving on the same SSD or HDD just changes the addressing of the data. Copying involves reading and writing every bit of data.

I just checked and search indexing looks completely clean. Thanks for the tip.

I've seen a peripheral slowing down the data bus problem many times so disconnecting everything is a good first troubleshooting step. You could also disconnect your second SSD and just have the system SSD running in case the second SSD is slowing down the data bus.

Something is not right and I need to find what the issue is. The Ryzen 3900X us supposed to be a lot faster than my previous CPU and the M.2 is double the speed so I'm scratching my heat at this point.

CPU performance hasn't changed greatly for several years and has pretty much plateud.

Increases in PC performance comes from increasing core count, the 3900x has two 6 core CPUs under the heatsink, 12 core/24 threads

The other way performance has increased is by using the "cpu" in the GPU. The GPU can be much more powerful than the CPU for certain tasks. Utilising the GPU is pretty much a no brainer but it does require significant resources put into re-writing software to use it.

If the software isn't written to use multi cores but is mainly single core threaded and/or doesn't use the GPU effectively it is like having a supercharged car but the software never presses the gas pedal hard enough to engage the supercharger I am try to use a car analogy without knowing anything about cars but I hope you get the point :-l

A 12 core /24 thread CPU and a good GPU does mean that you are future proofed as eventually most software will migrate to using the extra power available. For example DXO's latest version now uses the GPU to speed up PRIME noise reduction. Not only has the speed improved but the results have as well, without the GPU process times would have been even longer with the new DeepPrime.

If you use Topaz Sharpen AI you can get great results but without the GPU process times in stabilise mode are slow. When Topaz is processing it uses 100% of your GPU which is usually just used for painting pixels on your monitor.

So the take is that if you use largely single threaded software that doesn't use the GPU you are pretty much good to go with any decent CPU from the last 5 years. SSD/nvme storage will give you the best improvements but they have also been around for the last few years.


Post (hide subjects) Posted by
MOD Austinian
MOD Austinian
MOD Austinian
MOD Austinian
MOD Austinian
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow