The Robservatory

Robservations on everything…

 

2019 iMac vs Late 2014 iMac—Part Two

Update: After I posted this, Brad Oliver contacted me on Twitter about the frame rates for DiRT Rally—he commented that the fact that they were clustered around 60fps made him think I’d left vertical sync (Vsync) on…and he was right. I’ve updated that section with the modified results, as well as one additional comparison I forgot to include the first time.

Oh, and in case you don’t know Brad…he was directly involved in porting DiRT Rally to the Mac for Feral, so he knows his stuff! Thanks Brad!

In part one of the comparison between my old and new iMacs, I provided a brief overview of the new machine, tech specs for both, and a number of benchmarks. (I also tested the video card against a Windows GeForce GTX 1080, and posted a slide-over image that demonstrates the wider color gamut on the new Mac.)

In today’s second (and final) part, I’ll take a look at video processing performance (via iMovie), how well the new iMac handles gaming, and then wrap up the whole series.

Video processing

To test video processing, I edited a fairly short (93 seconds) video of one lap at the Monaco Grand Prix in Feral’s F1™ 2012. (That vintage year tells you how long it’s been since I’ve tried serious gaming on my Mac. F1™ 2016 and F1™ 2017 are both out now, and I’ll eventually add one or the other to my collection.)

Here’s the final result of this horridly-bad five-minute edit of some horridly-bad driving—I included some slow-mo and speed-up segments to see how each iMac handled it while editing (no problem on either iMac):

The source was captured using QuickTime and Soundflower (to get the computer audio, via Audio MIDI Setup’s Multi Output Device feature). I exported to a file at 4K resolution (which required upsampling, as the source wasn’t 4K), with the Quality set to Custom and slid to the highest quality (38.9Mbps); the Compress pop-up was set to Better Quality. Here’s how long it took each iMac to create the movie:

(time in seconds) 2019 iMac 2014 iMac 2019 iMac is
Export from iMovie 117 157 25% faster

A 25% reduction in export time on a 90-second movie is pretty impressive—that should bode well for larger projects. But enough about real work…

Gaming

As noted above, it’s been a while since I tried to do any serious gaming on my iMac—it was just so bad at it, it wasn’t worth the struggle. I played a bit on my Hackintosh, but mainly I used my Windows 10 PC with its GeForce GTX 1080. So I was quite curious how this new machine would stack up—not to the Windows PC, but to the previous iMac. Could I once again play games on my iMac?

I looked at my (aging) games collection, and decided to compare the previously-mentioned F1™ 2012, DiRT Rally, and X-Plane 11. I wanted to look at Doom 3, too, but it won’t launch on the new iMac (Aspyr is looking into the problem).

F1™ 2012

I started with one of my older games, suspecting that there wasn’t much to be gained in frame rate, as the game was two years older than my old iMac. And I was right—this game is so old that it runs fine on both the 2014 and 2019 iMacs. The maximum selectable resolution is 2560×1440, and even with 8x anti-aliasing enabled, it worked fine on both Macs with those settings. And despite being seven years old, the game still looked fine and was fun to play on either iMac.

The big difference, though, is that the 2014 iMac is working quite hard to do this—the fans were spinning fast—and noisily—trying to cool the machine as it’s pushing the pixels around. My 2019 iMac handles the same settings with no audible change in fan speed. Note that I only played the game for about 30 minutes at a time, so this may change in longer sessions.

So if you’ve got older games, you may not be able to push them to higher resolutions (due to limits in the game), but you may have a quieter gaming experience.

DiRT Rally

DiRT Rally is much newer than F1™, having been released in late 2017. I chose this game because not only is it wicked fun to play, but it’s got both an automatic “optimal” graphics setting (the default), and a built-in benchmarking tool. The benchmark runs one entire stage, switching camera views as it does so—and it’s a multi-minute test, so you get a good sense for how the game will perform at the chosen settings through a full stage. Here’s a brief snippet of the benchmark test…

Before I started benchmarking, I learned that my rally driving skills have deteriorated due to non-use…

I didn’t set any records on that particular stage, unless there’s one for Air Distance Traveled.

But back to the benchmarking. When I let DiRT Rally set the optimal graphics for the old iMac, it ran its built-in benchmark test at right around 60fps. The same happened when I tried on the new iMac—60fps was the average. It appears the game sets its optimal configuration with Vsync enabled, which is why the optimal settings on both iMacs came out at 60fps—the game can’t go any faster with Vsync enabled. And because the game runs full screen, it’s hard to tell exactly what resolution it’s using.

Thankfully, GRiD Rally exports two files after each benchmark: A CSV file with the time (in milliseconds) required to generate each frame (where lower times mean more frames per second), and an XML file with a summary of the results and details on the settings. When I looked at the XML files for each iMac, it seemed that DiRT Rally wasn’t really pushing the new iMac’s capabilities all that hard—the graphical settings didn’t vary much from those on the 2014 iMac.

To see if I could get more out of the game, I opted to override the optimal settings and tweaked things manually, increasing resolution and graphics features until the average frame rate dropped below 60fps. I did all of this (unintentionally) with Vsync engaged, so the maximum frame rate I’d ever see would be right around 60fps. So consider the last column “best configuration with Vsync enabled,” which is not the same as the “best configuration with Vsync disabled” (see below).

Setting 2014 iMac “Optimal” 2019 iMac “Optimal” 2019 iMac
Manual [1]Every “Quality” setting was on its highest allowed level—Ultra or High—and vsync was enabled
Resolution 1600 x 900 1680 x 1050 2048 x 1152
FPS (min / max / avg) 57.0 / 60.1 / 59.9 58.0 / 60.0 / 60.0 58.9 / 61.1 / 60.0
Anti-aliasing 2x off 8x
Weather quality 0 1 1
Shadows / size [2]Larger number = higher quality shadows Yes / 1024 Yes / 2048 Yes / 3072
Advanced fog No Yes Yes
Vehicle character / lod [3]I’m not sure what these values mean, but I assume higher numbers = higher quality cars 2 / 1 3 / 2 4 / 3
Water detail / scale [4]Higher detail numbers and lower scale numbers = higher quality water 1 / 8 2 / 4 3 / 2
Anistropic min / max 1 / 8 1 / 8 4 / 16
Mirror distance 200 300 300

After getting Brad’s comments, I re-ran my tests on the new iMac with Vsync disabled. The differences were quite impressive; here are my results at a number of the higher resolutions (again with all the quality items set to their maximum values):

Resolution Low FPS High FPS Avg FPS
4096 x 2304 24.2 49.3 33.5
3200 x 1800 36.6 72.6 48.7
2880 x 1620 41.7 82.5 56.0
2560 x 1440 50.2 97.9 65.4
2048 x 1152 66.5 123.4 84.4

With Vsync off, I can run at the same “best with Vsync on” 2048×1152 resolution, but with much higher frame rates (84fps average vs. 60fps average) and never dip below 60fps minimum. Alternatively, I can bump the resolution to 2560×1440 with the average still above 60fps and the low at a still-acceptable 50fps.

There are pros and cons to disabling Vsync; this article summarizes them quite well. I’ll experiment as I play and see if I notice a lot of visual issues with Vsync off.

After I’d re-done this testing on the new iMac, I ran one more test on the old iMac: I matched the new iMac’s quality settings, disabled Vsync, and ran the benchmark at 2560 x 1440 resolution—just to see what it could do at the new iMac’s resolution and quality level.

2014 iMac 2019 iMac FPS Increase
Lowest FPS 15.4 50.2 3.3x
Highest FPS 33.7 97.9 2.9x
Average FPS 22.2 65.4 2.9x

The new iMac simply crushes the old one—the 2014 edition can’t even hit 25fps average on this test. Impressive.

The downside to running with these higher resolution and quality settings is fan noise—unlike F1™ 2012, the new iMac is working hard at these quality and resolution settings, and the fans are spinning and audible, though it’s only really loud when the game isn’t making racing game noises.

Despite the fan noise, I was having so much fun testing this one that I’m pretty sure the newly-released DiRT 4 will be soon added to my collection.

X-Plane 11

In years past, X-Plane has been my most-used diversion on my Mac…that is, until X-Plane 11 came out, with its much higher demands on the video card. I tested it on my 2014 iMac, and it really wasn’t playable with decent quality settings. So I’d fly on my Frankenmac, using the GeForce card. Needless to say, I was anxious to see how X-Plane would work on the new iMac.

To compare the two machines, I set the same graphical detail levels on both:

X-Plane offers a wealth of data gathering options, including viewing frame rates onscreen and writing them to a file; I set both Macs up to do both. I then flew the same approximate route and duration—north out of Boeing Field with a tour around downtown Seattle and return to land—on both iMacs, and then checked the saved files to calculate the frames per second. Here’s what I found…

2014 iMac 2019 iMac FPS Increase
Lowest FPS 9.5 21.2 2.2x
Highest FPS 33.4 68.7 2.1x
Average FPS 16.9 45.4 2.7x

Now that’s what I’d call an improvement! But numbers don’t really convey the difference in the experience, so here’s a brief movie, showing a 360 degree turn around a Beech Baron over Seattle…

As you can see, the graphics are much smoother on the new iMac (there’s some frame rate hit due to the recording, too). There’s still a bit of jumpiness, but it’s much more tolerable than on the old iMac. Even better is that the minimum frame rate stayed above 20 the whole time, even when the scenes got complex—trying to fly at nine frames per second on the old iMac was quite the challenge.

Downsides to the new iMac? This will sound familiar, but it’s fan noise. (The old iMac’s fans crank under X-Plane, too.) If I’m flying while wearing headphones, or flying something loud, it’s not really an issue. But flying a glider, for example, with the fans cranking is a bit annoying. Still, given the fans ran fast and loud on the old iMac at its much lower frame rate, this is a worthwhile tradeoff.

Wrap up

Wrapping it all up, I’ve found the new iMac with the ADM Pro Vega 48 to be a more than decent games machine. Its one downside is fan noise—they will spool up and be noticeable in anything that pushes the video card hard. How can this be avoided? Well, you could stick to old games that don’t stress the GPU as much.

Alternatively, you could give Apple $5,799 (instead of the $3,449 I spent on my iMac plus third-party RAM) to buy the 10-core iMac Pro, which has a totally different thermal management system. Jason Snell discusses this in his comparison of the iMac Pro and the new iMac for Macworld:

The iMac Pro has been completely redesigned on the inside, with the space reserved for spinning hard drive configurations in standard iMacs replaced with a new fan system that’s quiet and efficient. For comparable levels of work, such as a video export from Final Cut Pro X, my iMac Pro is nearly silent and the iMac emits a constant buzz of white noise.

For me, though, as I spend most of my time not gaming or working in video, that $2,350 represents a whole lot of money to spend on silence under load. And those saved dollars will pay for a lot of additional games.

Still, there’s no denying that you’ll hear the fans when you push hard on the new iMac. For me, the cost savings with near-iMac-pro levels of performance are worth it. You’ll have to make that decision yourself, though, based on your budget and tolerance for fan noise.

Other than the fan noise under load (which my old iMac shares), I’m very happy with the new machine. It’s wicked fast in every day use, I like the wider color gamut display, the added RAM will be very nice to have, and I finally have a capable gaming machine that’s the same as my primary work machine.

Now, if you’ll excuse me, I have some games to play real work to get done.

Leave a Reply

The Robservatory © 2018 • Privacy Policy Built from the Frontier theme