[biz-engineer]

I'm having the issue that processing raw files is surprisingly slow with my computer. I found a recent test performed by DPR here.: Head to Head: Adobe Lightroom Classic vs Capture One 21: Digital Photography Review. I attempted to perform a similar test with my laptop, copied 100 raw files, adjusting a few parameters on all of them (copying same adjustments to all files) and sending to batch export.

My test result:

Pentax K-1 RAW files -> JPEG export processing time slightly under 3 minutes per file! That's roughly 10 times slower than DPReview 2021 test figures.

I opened the resource monitor in Win10 to check CPU, memory and CPU load.

After exporting 15 images, the CPU clock dropped from 1.8GHz to under 0.8GHz.

I guess the clock speed dropped to preserve the CPU from overheating and shutting down.

My laptop setup is: Win10, CPU i7-8850 quad-core @ 1.8GHz, 16GB of RAM, 500GB SSD, mediocre GPU (Intel 620, ranks last in GPU tests I've seen in 2021).

Have I made the wrong choice for image processing?

What do you guys use?

How fast is your image processing time?

What parameters do you look for in computer specs that best reflect on image processing speed?

[UncleVanya]

I have a laptop as well. But I’ve never attempted any batch conversions as it doesn’t match my way of working. I just never feel like the speed is that bad working interactively. Laptops as a rule have compromised performance at a given price point vs desktops.

Lenovo Flex 14 2-in-1 Convertible Laptop, 14 Inch FHD Touchscreen Display, AMD Ryzen 5 3500U Processor, 12GB DDR4 RAM, 256GB NVMe SSD, Windows 10, some AMD low power gpu is included. I have enabled support for it in DXO which drastically speeds up deep prime processing when I use it.

[35mmfilmfan]

I only ever process one RAW image at a time, giving it my complete personal undivided attention attempting to create the effect I require. I use FastStone, and if needed XnView, and processing speed depends to a great extent on my ever-sought level of perfection. Never bothered to time any of it - when my cuppa tea gets cold, I make another one, and am frequently surprised that while I have been occupied it has got dark ! HTH.

[Eric Auer]

If I had to use a laptop for image processing and editing it would be a new MacBook Pro.

As it stands I use a desktop 99% of the time these days, AMD Ryzen 5800X and a Nvidia RTX 3080 with 32GB Ram and several disks, from NVME to SDD and a HDD daily backup.
All except one SDD and the HDD backup are on PCIe 4.0.

I never timed anything, all my work in LR and PS is instant enough that I dont care. I dont do any video work, FWIW.

[ProfessorBuzz]

After exporting 15 images, the CPU clock dropped from 1.8GHz to under 0.8GHz.

I guess the clock speed dropped to preserve the CPU from overheating and shutting down.

That clock speed is a big yellow flag right there. What make/model laptop is it?
The i7-8850 has a few variants, but mostly they run at 2.6GHz to 4.3GHz peak. The low-power frequency is 800MHz.
SO... redo your benchmark after checking:
a) does your machine have latest BIOS and hardware drivers from the vendor. Check their website or their update tool.
b) vents are unblocked - eg put it on a wood/metal desk, no obstructions to the air vents
c) it is plugged into the wall AC power
d) adjust windows Power Management settings for Maximum Performance.

Settings... Power & Sleep Settings. Additional Power Settings. Also under Power Options - Edit Plan Settings - and Change Advanced Settings.

Make sure nothing in the power profile is "sleeping" the machine after keyboard inactivity - a common thing where the machine slows down/sleeps after 5-20 minutes.
If it is running at 0.8Ghz or 1.8, then the CPU is being powered down to save battery. Or you have a massive overheating problem.
Usually windows is running with a "balanced" power profile. You don't want this when image processing!


e) make sure nothing else is running on the machine during the tests. eg turn off WiFi/Internet so outside stuff not called, no updates downloaded, etc. Windows Task Scheduler may give some hints.
f) Some anti-virus software can be paused - often this slows disk access substantially. Proceed with caution.

Then give it another go, and let us know what you find.

[normhead]

I Just used Apple Photos , which is nondestructive and also processes every file on the fly when exporting, and did 100 file in 2 minutes 40 seconds.

100 files converting raw to tiff, full size average size of file on completion about 200 MB.

Mac mini , 8 core M1 chip...
I never wait for processing. Often I export 1 to 4 files, and the export is done before I can get out of my chair of with one file even look up. That's roughly 1.6 seconds a file. Different software etc. for sure still,

I'm tempted to try on my laptop, which is an Intel chip and I'm sure is completely inferior to the Mac mini just out of curiosity.

[jgnfld]

...

My laptop setup is: Win10, CPU i7-8850 quad-core @ 1.8GHz, 16GB of RAM, 500GB SSD, mediocre GPU (Intel 620, ranks last in GPU tests I've seen in 2021).

Have I made the wrong choice for image processing?

...

Sorry to say this but this is not probably not a machine with the best specs for large batch image processing jobs on 100 full frame RAW files even if you fully max it out, put it in a very cold room, and leave it plugged in unless you just set up a job and leave it for a period. Even at 3 min/file the machine will complete the job easily overnight, for example.

Is this the time to mention I started using computers in 1970 where multidimensional scaling runs routinely took up to 24 hours on what passed for the university Control Data "high speed" mainframe? Once a colleague left in his test output lines which specified just an output of the first line of a giant factor analytic study with scores of measures on thousands of subjects. Computer took almost a week to do the job. It printed out the first line of the output as directed and then dumped the program. PI on that grant was NOT happy at having to double pay for a week's CPU time!

Later, doing Monte Carlo studies even on pretty hot desktop machines led to rather long run times--overnight even--even after vectorizing everything and learning multicore programming to boot in R (which most image programmes don't employ much or at all yet) just because there are giga-gazzillions of CPU cycles required.

In this regard, is it possible to run 3 or even 4 instances of the programme you are using on your machine? 4 would likely lock it up pretty good for any other purpose while the jobs are running, but 3 instances might fly and still leave you one core for general stuff.

---

Last edited by jgnfld; 01-05-2022 at 09:05 AM.

[Mooncatt]

That clock speed is a big yellow flag right there. What make/model laptop is it?
The i7-8850 has a few variants, but mostly they run at 2.6GHz to 4.3GHz peak. The low-power frequency is 800MHz.
SO... redo your benchmark after checking:
a) does your machine have latest BIOS and hardware drivers from the vendor. Check their website or their update tool.
b) vents are unblocked - eg put it on a wood/metal desk, no obstructions to the air vents
c) it is plugged into the wall AC power
d) adjust windows Power Management settings for Maximum Performance.

Settings... Power & Sleep Settings. Additional Power Settings. Also under Power Options - Edit Plan Settings - and Change Advanced Settings.

Make sure nothing in the power profile is "sleeping" the machine after keyboard inactivity - a common thing where the machine slows down/sleeps after 5-20 minutes.
If it is running at 0.8Ghz or 1.8, then the CPU is being powered down to save battery. Or you have a massive overheating problem.
Usually windows is running with a "balanced" power profile. You don't want this when image processing!


e) make sure nothing else is running on the machine during the tests. eg turn off WiFi/Internet so outside stuff not called, no updates downloaded, etc. Windows Task Scheduler may give some hints.
f) Some anti-virus software can be paused - often this slows disk access substantially. Proceed with caution.

Then give it another go, and let us know what you find.

Definitely this is where I would start before investing in new hardware. Something isn't adding up, unless you are purposely throttling your CPU for some reason.

[Papa_Joe]

Sorry to interfer, but there seems to be something wrong with the specs of your CPU. The i7-8850H is a 6 Core CPU with 2.6 GHz and only three years old. I would say this CPU should not fall su much behind the computers in the test. The CPU the ASUS used in the test is not as performant. The specs given by you look more like an i7-8550, but even this one should not fall so far behind the CPU in the test.

So I would support the opinion, that the issue is your test setup or the setup of your computer. The proposals of Professor Buzz should give you maximum performance of your laptop.

Lightroom heavily depends on CPU power, so your GPU should have no influence onto the performance of your laptop.

[steephill]

It is an i7-8850U processor, not a great choice but it should be capable of much better performance. The 800Mhz mode is used when saving power.

[Medex]

Your 3 minutes for 100 jpg files (Full size?) are very good I tested the same on mine laptpm - 100 RAW to 100 JPG (full size) on two different SSD - Samsung 970 EVO Plus 1 TB. I use Dell Precision with Intel Core i9-9980 HK, 16 GB RAM, 6 GB Nvidia Quadro RTX 3000. My result is 4:32 min on both SSDs. Same time, but much more than yours.
No processor overheating issues, it works at 32-34x (about 3300 MHz).
RAW to TIFF conversion was shorter - 4:20 min.
RAW to JPG (2048 pix long edge) - 4:07 min.

[marcusBMG]

I've done somewhat similar tests using dng converter on a score of files, this was with old socket 775 cpu's, core2 vs xeon. One thing I've learnt is that the speed was way faster when the cpu wasn't tied up with other background programs - virus etc. If the cpu could use all its 4 cores, it would chew through the task in about 30 secs. Otherwise it took 2 to 3 minutes, essentially using just one core.
Your i7-8850U is a relatively low performing cpu by contemporary standards, only nominally about 3 times the speed of one of those old ones on passmark benchmark scores. I suspect that similar stipulations apply.

[GUB]

I export only one image at a time except for the occasions where I might be stitching a few images.
In Darktable profiled denoise seems to slow things down a bit.
In this test I demosaiced with AMAZE which is a slower one. GPU enabled.
Results - 27 seconds with Profiled Denoise and 19 seconds without.
K-1 Raw > tiff floating bit

EDIT and to Jpg 8bit 12 sec with denoise and 4 sec without.

[jbinpg]

To be a real baseline, we would all need to be exporting at the same image size. A full size image makes the most sense.

[GUB]

Is there anything to learn by a batch export versus a single export? Surely only an inherent issue (overheating) with an individual computer.?