1

I'm building a desktop computer for post-processing my photographs in Lightroom and Photoshop.

I'm hesitating between choosing 2*8GB RAM of 2133MHz and 2*8GB RAM of 3000MHz and I am not sure if investing in higher frequency memory will really pay back in Lightroom and Photoshop performance.

Not to say that higher performance memories require more expensive mother boards.

  • 4
    I'm voting to close this question as off-topic because it asks about computer hardware (and its impact on software), not photography. Super User is a better place to ask that type of question and likely to yield better answers. – null Dec 11 '16 at 12:09
  • @null Then please close also photo.stackexchange.com/questions/37805/… or explain, please, the difference between the two. – dzieciou Dec 11 '16 at 12:16
  • 1
    GPU support is something that an application has to be programmed for specifically. The clock frequency of the memory is not. Also, as I said, Super User will very likely provide you with better answers for your question. – null Dec 11 '16 at 12:22
  • Not enough information to give you a good answer. It entirely depends on whether the CPU supports higher speed RAM, and how many RAM slots are on the motherboard. For instance, if you have 4 slots, you're generally better off filling them with 4*4GiB rather than 2*8GiB. As @null said, this question is better suited to Super User. – scottbb Dec 11 '16 at 17:31
  • @scottbb 4x4 vs. 2x8 also depends on the motherboard and how it handles those two scenarios. I've had MBs (for AMD processors) that are faster with 2x8 than with 4x4 due to dual channel only working with 2 sticks, one in each bank. – Michael C Dec 16 '16 at 1:40
4

The short answer is no.

First off all it is important to understand that the frequency affects the possible maximum transfer rate, although it's not achieved in every scenario. It's just the upper limit. Surely, if you run a benchmark with those two, the higher frequency will very likely result in a higher transfer rate, but it's not really important if you're using Lightroom. Furthermore, those DIMMs (Dual Inline Memory Module, or just your memory for that purpose) have even more numbers that matter when it comes down to how they perform. Very often you see several numbers after the frequency, written like CL7-7-7-18. Those actually define the amounts of clock cycles the memory will take to process a desired operation. For a module that has the just mentioned values it will take 7 full cycles to return the desired value. All that is pretty theoretical stuff as you won't notice it as a user. I would not worry too much about all those technical stuff as the effect will not be noticeable and probably only be measurable in benchmarks.

However what you should care about is something that is called Dual Channel Memory. What this effectively means is that your RAM works just like a RAID 0. You can think of it as every second bit beeing written to the other DIMM. So instead of one DIMM having to write 10 Mbyte for instance, both have to write 5 MByte resulting in a performance boost. Therefore four DIMMs won't really increase your memory performance, compared to just two DIMMs as you then have two Dual Channel …"clusters" or how you want to call it. Four DIMMs only perform better if your mainboard supports Quad Channel. Don't worry about that terminology too much, pretty much every mainboard supports Dual Channel memory, so you're good to go.

More important for using LR is definitely the CPU. If you tweak a slider in LR, the CPU does almost all the work for you, so it's mandatory to have enough CPU performance.

Since GPU support was also mentioned before: It might improve editing in some sceneraios, but as null mentioned, the developer has to implement it. Only very few functions of LR support using your GPU (currently), yet it slows down other parts of the program. That's why I have it turned off. All your basic editing is only done on the CPU. Therefore I would not pay too much attention to choosing a GPU for LR either.

Greetings Rusher0

2

Despite the comments this is off topic I think it is pretty relevant to Lightroom, as that is an area where people struggle a lot with performance and offer a Lightroom answer. I specifically did some testing specifically on memory speed for preview build times. With nothing else different, on an I7-6700K at stock speeds otherwise, changing memory from 2133 to 3000 improved build times very slightly, from 95 seconds (for a specific set) to 88 seconds. Clearly performance could be limited by many other things as well, and presumes a system that can overclock memory, and this is one data point, but from years of experimenting I have found the biggest factor for Lightroom is single core processor speed. More cores (say more than 4) does not help, hyper-threading hurts (in the same experiment as above at 2133Mhz the build time went to 116 seconds with hyperthreading on). This testing was done on about the fastest stock consumer system you can get: 4 regular SSD's, 1 U.2 SSD, 64GB memory. I cannot comment on photoshop, though the ACR engines are the same in LR and PS, so that aspect will be similar. Also, except in HDR/Pano merges, I cannot get LR to use much over 8-9GB of memory, even with lots of background jobs running.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.