Home · Register · Join Upload & Sell

Moderated by: Fred Miranda

New fredmiranda.com Mobile Site

New Feature: SMS Notification alert

New Feature: Buy & Sell Watchlist

FM Forums | Post-processing & Printing | Join Upload & Sell

2 3 … 5 6 end

Archive 2023 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

jhapeman
Offline
• • • •
Upload & Sell: On

p.1 #1 · p.1 #1 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

UPDATE: I corrected a number for one of the Studio Ultras; you can see they are still the top performers. I also added new numbers for the 64-core AMD Threadripper; I got a new 16GB Nvidia RTX 4080 graphics card. This is one of their best cards at the moment and again I got it for rendering, but it also performs very well for Topaz DeNoise. All of that said, the M2 Max GPU is 33% faster at Topaz than even this $1300 GPU. Pretty impressive, really.

Last spring when the Studios were released by Apple I ran some benchmarks on performance that compared them to my M1 Max MacBook Pro and a couple of high-powered PCs we use here in my business. A lot has changed since then--new PC options, new M2 chips from Apple, OS and application updates. Adobe in particular has been putting resources into updating the performance of their applications. With the new M2 MacBook Pros out, I decided to re-visit these tests and see how things are shaking out now. Since then I have added a new AMD Threadripper PC with an Nvidia RTX A6000 graphics card for our 3D rendering workstation, and added a fairly high-end Lenovo laptop PC with an i7-12800H processor and an Nvidia RTX 3070 GPU. So, without further ago, here's the updated results. There were big gains last summer on exports when Adobe updated their code to support more than 6 cores and added GPU assist to exports. In some cases export times dropped by half.

On most of the machines the import times are pretty similar, so I didn't waste time redoing that test. The common tasks that slow down my computers when I'm shooting a lot are rendering previews, batch exporting files, and doing batch processing in applications like Topaz DeNoise. I want to point out that ANY of these machines are very capable for most daily use. All machines were tested with the latest version of Windows 10 or MacOS 13.2 and Adobe Lightroom Classic version 12.1. Version 12.2 was released today but I don't have the time repeat this many tests; I did check a few and there weren't any significant times changes. Topaz DeNoise was version 3.7.0; version 3.7.1 was also released yesterday, and it had a tiny improvement on the machines with Nvidia cards, and no change on the two Macs I tried, so for those So let's start with the computers including the cost:

1) Mac Studio Ultra with 20-core CPU, 64-core GPU and 128GB of RAM, 4TB of SSD storage $6799
2) Mac Studio Ultra with 20-core CPU, 48-core GPU and 64GB of RAM, 1TB of SSD storage $3999
3) Mac Studio Max with 10-core CPU, 24-core GPU and 32GB of RAM, 512GB of SSD storage $1999 (base model)
4) 16" M1 Max MacBook Pro, 10-core CPU, 32-core GPU and 64GB of RAM, 4TB of SSD storage $4899
5) 16" M2 Max MacBook Pro, 12-core CPU, 38-core GPU and 96GB of RAM, 4TB of SSD storage $5299
6) 16-core 2019 Mac Pro, Dual Radeon Pro Vega II GPU (24GB each) and 384GB of RAM, 4TB SSD $11,999
7) 28-core 2019 Mac Pro, Radeon W6800X Duo (64GB total) and 384GB of RAM, 4TB SSD $20,000
8) Lenovo 14-core i7-12800H, 8GB RTX 3070 GPU and 64GB of RAM, 512GB SSD $2499
9) 32-core AMD 3975WX Threadripper Pro, 48GB Nvidia RTX A6000 GPU, 128GB of RAM, 4TB SSD $12,000
10) 64-Core AMD 3990X Threadripper Pro, 16GB Nvidia RTX 4080 GPU, 128GB of RAM, 4TB SSD $12,500

The massive Mac Pros and PCs were not purchased for Lightroom, but rather 3D rendering, but it's fascinating to throw them in the mix, and as you'll see there's diminishing returns as you spend more and more, as Lightroom can't take advantage of these huge powerful GPUs for many tasks (and can only use one GPU anyway), whereas for 3D rendering that power can make a huge performance difference.

The crux of the test is an import of 1465 files from my Sony A1; a render of 1:1 previews of all of those files; a batch export to full-sized jpegs of the same, and then an export and batch DeNoise of 10 images. I could de-noise more, but the results scale pretty linearly from there so it's not worth it.

A couple things to note:

GPU: Lightroom does not need a heavy-duty GPU, but does USE the GPU, more and more. It's just that a lot of the horsepower in the newer expensive GPUs is targeted at 3D raytracing; great for games and 3D rendering applications, but really not very useful at all for the tasks Lightroom needs to achieve.
RAM: Above 64GB RAM is generally not needed, regardless of PC/Mac platforms, but less can slow things down. This does change IF you do large Panoramas or even more so for HDR panoramas, more on that later.
Synthetic Benchmarks: They are pretty useless, IMO. Based on benchmarks the GPU in the Apple Silicon machines should be terrible; it's NOT. The CPU scores would imply much better results on the PC side than we see in the real world. I suspect that this is a function of several things, one is that the synthetic benchmarks just don't match real-world workloads well, another would be that they don't know how to properly compare Apple Silicon, and the last would be that many tasks in the real world don't scale cleanly on multi-core machines.
GHz Doesn't matter that much: You'd think pure raw GHz of a single core should make a big difference in many tasks, and many benchmarks assume this, but it's clearly not the case.
Price/Cost: There's tons of talk about price, but the reality is you can spend a ton a top-of-the line PC or Mac, and when you equip the two similarly they are *very* similar in price.
Noise: Across the board the PCs are just noisy. The Intel and AMD chips burn up a lot of energy as heat, and the fans have to spin to cool them. Even with the quietest fans I can buy for the AMD workstations, they are loud compared to the Mac Pros; the Lenovo laptop will whir up its fans to almost 70dB, vs. 25dB for the MacBooks--so quiet I can't hear them over the normal office background noise. The same applies to the Mac Studios, BTW. Whisper quiet. The old Intel MacBook pros used to suffer this same problem, I'm glad it's gone.

One last test I'm debating doing is battery-powered testing. The new Apple Silicon Macs seem to have almost identical performance on or off of mains power--they just don't throttle much on battery, and the battery life is crazy long. The Lenovo on the other hand already starts with a much lower battery life, and you either have to throttle back performance to get it to last long, or just chew up the battery to get the best performance.

Edited on Feb 28, 2023 at 11:22 AM · View previous versions

Feb 14, 2023 at 05:04 PM

chez
Offline
• • • • • • •
Upload & Sell: Off

p.1 #2 · p.1 #2 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

So it looks like the real bargain out of all of them is the base model Studio. I just picked one up for $1900 CAN. I will be running a dual system both Windows and MAC as the latest MacOS does not support a couple of my printers.

Feb 14, 2023 at 05:49 PM

RustyBug
Offline
• • • • • • •
Upload & Sell: On

p.1 #3 · p.1 #3 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

I particularly appreciate seeing this.

Feb 14, 2023 at 06:36 PM

Jeff
Offline
• • • • • •
Upload & Sell: On

p.1 #4 · p.1 #4 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

jhapeman wrote:
...
•RAM: Above 64GB RAM is generally not needed, regardless of PC/Mac platforms, but less can slow things down. This does change IF you do large Panoramas or even more so for HDR panoramas, more on that later.

Look forward to it!

(Thanks for all the work you put into posting these tests here at FM!

)

Feb 14, 2023 at 06:41 PM

netexpress
Offline
• • • •
Upload & Sell: On

p.1 #5 · p.1 #5 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

These data are so valuable I can't thank you enough!!!

Feb 14, 2023 at 07:36 PM

dclark
Offline
• • • • •
Upload & Sell: On

p.1 #6 · p.1 #6 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

Interesting data. I am most interested in the 1:1 previews rendering times. I see claims (but nothing I trust) that Windows LR is limited in the number of cores it will use. In the limited number of systems I have tested the 1:1 rendering times vary inversely with number of cores, up to 12 cores, and all cores run at ~100%. Can you tell us how many cores are in use on your 32 and 64 core systems? Thanks.

Feb 14, 2023 at 08:13 PM

jhapeman
Offline
• • • •
Upload & Sell: On

p.1 #7 · p.1 #7 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

RustyBug wrote:
I particularly appreciate seeing this.

Since the Lenovo 12800H with 64GB would be the natural upgrade from my X1 Extreme TP 8400H with 64GB ... interesting to see where it is 'better vs. worse' than the MBP's.

The one thing I'd be interested in knowing (you may not have it, though), but I'll assume ... is how much heat signature diff comes with these diff machines (mainly the laptops) ... the MBP's are running cooler than the 12800H and RTX, I'd anticipate.

The thermals are night and day. The MacBooks stay cool to the touch; I have to really push them hard to even get them warm. I think I managed to get the fans to spin up only once. The ThinkPad on the other hand gets quite hot and the fans spin up regularly. In the 15 months I've had the M1 Max I had kind of forgotten how hot the Intel chips get in laptops. The ThinkPad is just like my old i9 16" MacBook Pro--hot and loud.

Feb 14, 2023 at 08:38 PM

jhapeman
Offline
• • • •
Upload & Sell: On

p.1 #8 · p.1 #8 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

dclark wrote:
Interesting data. I am most interested in the 1:1 previews rendering times. I see claims (but nothing I trust) that Windows LR is limited in the number of cores it will use. In the limited number of systems I have tested the 1:1 rendering times vary inversely with number of cores, up to 12 cores, and all cores run at ~100%. Can you tell us how many cores are in use on your 32 and 64 core systems? Thanks.

It's not core-limited. That used to be the case, but you can see it using all of the cores on the Macs and PCs, no matter how many you throw at it. It doesn't scale perfectly linearly though, because as the manufacturers add more cores, they always end up slowing them all down to keep the package TDP within parameters, so there's usually a sweet spot of core count/GHz/price. The core count cap in the Adobe software used to kick in at 6 cores, for a ton of tasks, but not anymore. The one thing left is the Pano and HDR stitching, but that's already been updated in Camera Raw to scale with core count and I would imagine that code will be migrated into Lightroom soon. The AMD's both do some odd cycling up and down that I think has something to do with the way their processors work, as the Intel and Apple Silicon don't do it, but it doesn't seem to have an impact on overall rendering speed. In recent releases Adobe has been making a lot of performance improvements; things like previews and exports now take advantage of as many cores as they can, and will also use GPU acceleration.

To me the odd one out here is the 16" M1 Max; it clearly seems to be doing some throttling that the M2 Max and M1 Max Studio are not doing. I repeated that test so the result is real, it's not an anomaly. The same thing applies to the exports on the ThinkPad; it's just slow. In that case I think it just can't use the GPU as effectively as its used on the Macs as when I monitored CPU and GPU usage it didn't seem to peg out the GPU, although that may be just a power throttling decision. I did purposefully set Windows to use max performance on the laptop in order to give it an edge, so it was supposed to *not* be throttling anything.

Feb 14, 2023 at 09:32 PM

dclark
Offline
• • • • •
Upload & Sell: On

p.1 #9 · p.1 #9 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

jhapeman wrote:
It's not core-limited. That used to be the case, but you can see it using all of the cores on the Macs and PCs, no matter how many you throw at it. It doesn't scale perfectly linearly though, because as the manufacturers add more cores, they always end up slowing them all down to keep the package TDP within parameters, so there's usually a sweet spot of core count/GHz/price. The core count cap in the Adobe software used to kick in at 6 cores, for a ton of tasks, but not anymore. The one thing left is the Pano and...Show more →

So you are saying the reason the 32 core AMD and the 64 core AMD process previews at about the same rate is that although there are twice as many cores in use they are running about half as fast. Is that correct?

Feb 14, 2023 at 10:10 PM

RustyBug
Offline
• • • • • • •
Upload & Sell: On

p.1 #10 · p.1 #10 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

jhapeman wrote:
In that case I think it just can't use the GPU as effectively as its used on the Macs as when I monitored CPU and GPU usage it didn't seem to peg out the GPU, although that may be just a power throttling decision.

That would be my guess, based on my experience (not a formal test) with the my CPU / GPU usage on my X1 Extreme. Even though, I'm always tethered on AC power, and have max performance set ... GPU just doesn't seem to get a chance to contribute much. Guessing the decision is based on that the GPU has selective use for some processes, the program retains the CPU as it has more universal utility.

jhapeman wrote:
The thermals are night and day. The MacBooks stay cool to the touch; I have to really push them hard to even get them warm. I think I managed to get the fans to spin up only once. The ThinkPad on the other hand gets quite hot and the fans spin up regularly. In the 15 months I've had the M1 Max I had kind of forgotten how hot the Intel chips get in laptops. The ThinkPad is just like my old i9 16" MacBook Pro--hot and loud.

Good to know. My use of the 16" MBP's have been very cool compared to my X1. Just didn't know if the newer gen Intel might have a better handling on thermals. Doesn't sound like it (not surprised, confirmation appreciated).

Feb 14, 2023 at 10:22 PM

jhapeman
Offline
• • • •
Upload & Sell: On

p.1 #11 · p.1 #11 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

dclark wrote:
So you are saying the reason the 32 core AMD and the 64 core AMD process previews at about the same rate is that although there are twice as many cores in use they are running about half as fast. Is that correct?

It's more complex than that, but it's a large part of it. There's things like base clock and turbo, and how well it seems the systems can maintain the higher speeds. Based on how I'm interpreting the CPU utilization graphs--and I could be wrong--it seems like the Intel Xeons are maintaining higher sustained "turbo mode" clock speeds than the AMDs are able to do, so I see them cycling up and down in bursts.

Feb 14, 2023 at 10:36 PM

jhapeman
Offline
• • • •
Upload & Sell: On

p.1 #12 · p.1 #12 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

Well this is going to be very interesting! I just ran the preview rendering on the Lenovo ThinkPad on battery, and not only did it take longer--not too surprising since the laptop shifted to "Battery Saver" mode--but it also didn't really "save" the battery. It took 9% longer to complete rendering the previews, but in the process managed to consume 50% of the battery capacity. I'm going to test the M2 Max tonight at home and see what happens there and ultimately I'll get the battery/mains power comparisons done for the two MacBook Pros and the Lenovo.

Feb 15, 2023 at 12:10 PM

ruthenium
Offline
• • •
Upload & Sell: On

p.1 #13 · p.1 #13 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

A question:
When looking a the last graph (Topaz DeNoise), there is little difference between the top five lines.
To the best of my understanding, Topaz DeNoise benefits from better GPU.
If this is correct, would it be also correct to expect that all top five Macs in the graph should preform similarly for denoising and generating JPGs in DxO Photolab?

Feb 15, 2023 at 03:52 PM

jhapeman
Offline
• • • •
Upload & Sell: On

p.1 #14 · p.1 #14 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

ruthenium wrote:
A question:
When looking a the last graph (Topaz DeNoise), there is little difference between the top five lines.
To the best of my understanding, Topaz DeNoise benefits from better GPU.
If this is correct, would it be also correct to expect that all top five Macs in the graph should preform similarly for denoising and generating JPGs in DxO Photolab?

It's not entirely clear how much Topaz is using the GPU vs. using the neural engines and CPUs or a combination of them. I can say that while I was running the analysis I used the Activity Monitor and MX Power Gadget to see what was going on, and the GPU was nowhere near being fully utilized and only four of the performance cores showed heavy use, three more light use, and three had very little use. I think there's still a lot of room for optimization in the software.

I did some testing last summer with DxO PL6 but to be honest I found the performance to be terrible; it was very slow in comparison to Lightroom or CaptureOne (which was also way slower than Lightroom). Apparently they use the Neural Engines but even there they have a bug that screws up colors and is unfixed since 6.2 so I've just stayed away. In comparative testing I also found that Topaz DeNoise produced better results and gave me much greater control of how I applied the denoising, so basically I just tossed DxO in the dustbin.

I can see if I can download another trial and see if things have changed much if there's a special use case I can try for you.

Feb 15, 2023 at 04:10 PM

jhapeman
Offline
• • • •
Upload & Sell: On

p.1 #15 · p.1 #15 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

More battery information. I just finished doing the JPEG exports on battery on the Lenovo. The process used up 87% of the battery capacity and took 50% longer to complete--49 minutes vs. 33 minutes plugged in. Tonight I'll try the M2 Max MacBook Pro. This should be very interesting!

Feb 15, 2023 at 04:25 PM

docusync
Offline
• • •
Upload & Sell: Off

p.1 #16 · p.1 #16 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

@jhapeman thank you for your time doing and publishing those tests!

Any chance you could do a 4k video export (Davinci Resolve, h265 and AV1) on those systems? Theoretically none of them supports hardware AV1, but I think Apple Silicone should be powerful enough for the task.

Feb 15, 2023 at 04:51 PM

ruthenium
Offline
• • •
Upload & Sell: On

p.1 #17 · p.1 #17 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

jhapeman wrote:
It's not entirely clear how much Topaz is using the GPU vs. using the neural engines and CPUs or a combination of them. I can say that while I was running the analysis I used the Activity Monitor and MX Power Gadget to see what was going on, and the GPU was nowhere near being fully utilized and only four of the performance cores showed heavy use, three more light use, and three had very little use. I think there's still a lot of room for optimization in the software.

I did some testing last summer with DxO PL6 but...Show more →

Thank you for the quick response!
There is no special case. I would like to add a Mac, and the practical question is whether one model offers an advantage over the others. In my case, I don't work with large numbers of images, and most time is spent doing corrections. What is annoying is that it takes about 40 sec to generate a single JPG with DeepPRIME denoising in DxO Photolab (v 6.3.1) on my Windows laptop. I know this should be much faster on a Mac, but then the question is whether there is any sense in investing in one of the more expensive models? With Topaz, the performance of the top five systems seems similar enough to say that the least expensive of these might be the right choice. My gut feeling is that the same might be true about DxO Photolab.

Feb 15, 2023 at 05:48 PM

Chris Court
Offline
• • •
Upload & Sell: On

p.1 #18 · p.1 #18 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

Very interesting post. I hope this will go some way to dispelling the "M1/M2 Macs are underpowered" misinformation that I've seen circulating a lot here recently. My takeaway from your tests is that the Macs at minimum go toe-to-toe with high-end Intel/AMD machines in real-world usage.

C

Feb 16, 2023 at 03:08 AM

jhapeman
Offline
• • • •
Upload & Sell: On

p.1 #19 · p.1 #19 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

Chris Court wrote:
Very interesting post. I hope this will go some way to dispelling the "M1/M2 Macs are underpowered" misinformation that I've seen circulating a lot here recently. My takeaway from your tests is that the Macs at minimum go toe-to-toe with high-end Intel/AMD machines in real-world usage.

C

Yeah, I think a lot of it is just based on disbelief that someone could introduce a chip that can challenge Intel and AMD; especially since many other past efforts have been pretty poor. It's based on a misunderstanding of both Apple's abilities, developed in no small part from 15 years of investment and effort in powerful handheld devices, and the fact that with a full systems approach Apple can get a lot more performance from their hardware.

The most common misconceptions I see is about the GPU. I can't tell you *why* integrated GPUs in the Intel world are so bad, but I can tell you they just are, and have been pretty much forever. That's 100% not the case here. Yes, it's missing the ray-tracing hardware used in 3D rendering and games, but the Mac has never been known for either of those platforms anyway, at least not in the years since 3D raytracing has been a thing. The reality is that unless you use those types of programs that power is wasted anyway. For most of what the folks here on FM are using their computers for in terms of photo editing, that hardware is also a waste--it's other components of the GPU that matter more, and Apple has done a fabulous job there.

Feb 16, 2023 at 08:45 AM

RustyBug
Offline
• • • • • • •
Upload & Sell: On

p.1 #20 · p.1 #20 · UPDATED: Lightroom Benchmarking: M2 and M1 Macs and PCs

jhapeman wrote:
Yes, it's missing the ray-tracing hardware used in 3D rendering and games, but the Mac has never been known for either of those platforms anyway, at least not in the years since 3D raytracing has been a thing. The reality is that unless you use those types of programs that power is wasted anyway. For most of what the folks here on FM are using their computers for in terms of photo editing, that hardware is also a waste--it's other components of the GPU that matter more, and Apple has done a fabulous job there.

Not to sound like I'm parroting a cliche' ... but, the concept of "Mac is for creatives" (i.e. not gamers, not ray tracing, etc.) is beginning to reveal itself in a way that I hadn't quite latched on to what that might mean, in generations past.

Again, my Lenovo X1 has been good to me, but I can see that there's beginning to be a separation between things ... of sorts.

Feb 16, 2023 at 08:37 PM

1 2 3 … 5 6 end
FM Forums \| Post-processing & Printing \| Join Upload & Sell 1 2 3 … 5 6 end

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.