Trashcan or trashcan’t? Defending the Indefensible Mac Pro 2013

Why do I now own a Mac Pro 2013? Is there actually any point to it thirteen years on, apart from being really, really good looking?

Welcome to Esk Computers, I’m Scott and today on the desk of Esk we’re going to answer that question.  I wouldn’t say I’m an Apple fan, but I’ve been an at least occasional user of their products since the mid 2000’s, with a good few years of that being a daily driven 2006 Macbook and 2010 Macbook Pro. I’m mostly back on Windows these days, but a little while ago the price of a computer I’d lusted after on it’s release dropped to just affordable levels, and I could no longer resist.

Now, there’s been plenty of discussion over the years as to why this Mac Pro 2013 “Trashcan” was a bad idea, and the Mac Pro line in general has had a rough go of it since, the 2019 Intel model being a bright spot that dimmed with a lack of ongoing support and the product having a bit of a identity crisis in the Apple Silicon era. In this particular case, Apple’s crystal ball went a bit cloudy and lead them down a path where lower powered, multi-GPU solutions would be the norm for high powered machines, and set about creating a compact design for a future that never happened. 

Instead we have GPUs that are by themselves as large as this machine, and draw more power than this innovative, near silent triangular heat sink can deal with. Indeed this was an ill-advised victory of form over function, as history shows it wasn’t able to keep up with the demands of the higher powered AMD D700 GPU option, with most of them having cooked themselves to death by now.

I remember watching the keynote where this was announced, thinking this was a striking design and I’d very much like one, then they announced the price, and we all had a laugh and forgot about it. So did Apple, to an extent, with their well-publicised problems with adapting their designs to market realities meaning it didn’t get any meaningful updates for six years.

I’d expected the release of the completely redesigned 2019 model to start the process of the second hand price of these dropping precipitously, but it didn’t really happen. The price held until Mac Studio released in 2022, which is more of a direct replacement in terms of it being a compact powerhouse. To me, that implies people were hanging on to these more for the aesthetic than the practicality. How very Apple.

My example uses the mid-range AMD FirePro D500 GPUs with 3GB of VRAM. Because I can’t resist tinkering, I have replaced the stock six core Xeon E5-1650 v2 with a twelve core E5-2697 v2 that has a base frequency of 2.7GHz, turboing up to 3.5GHz. It ought to be able to cope with the additional heat this 130W TDP part can put out, after all this was the processor used in the factory high end configuration. Fingers crossed.

This has 32GB of DDR3 RAM, as was the style at the time, and a 500GB SSD where Apple have pointlessly changed the pin outs to make it harder to provide aftermarket replacements, as is their style all of the time. Which is kinda why I wouldn’t say I’m an Apple fan. Thankfully NVME adapters are cheap and widely available.

The last officially supported release is 2021’s macOS 12 Monterey, but thanks to the heroic efforts of the Open Core Legacy Patcher team, last year’s macOS 15 Sequoia works well. Some of the newest features aren’t enabled like iPhone Mirroring and what passes for Apple Intelligence, and if you are a Safari user then it can’t playback DRM video content. Other browsers work fine though. There a few exceedingly minor graphical issues on, for example, the Maps app, but overall it works as pretty much as well as Monterey and benefits from the latest bug fixes and security patches.

As we swan into 2026, day to day use for this is absolutely fine. It handles office tasks and web browsing without any issues, and streaming video and local video playback presents no problems, as you’d expect with hardware of this power, even of this vintage.

Now I have a mostly gaming focused test suite for the Windows machines, but that’s not really macOS’s forte. From previous experience even those games with macOS versions run much worse than the Windows versions on the same hardware. There’s a few tests we can compare on the CPU side of things, so lets run this through the Mac version of Cinebench 2024 where it gets 46 in single core, 528 in multi core. So a little step behind the V4 Xeons of the recent budget builds on the channel but better that I expected to be honest.

Not a standardised test exactly, but I can use the Mac version of Davinci Resolve to render out the same video my plucky laptop created a few weeks back. This finished up in 12 minutes 27 seconds while drawing about 250W from the wall, so it’s about ten minutes faster than my laptop but ten minutes slower than my main PC. I should add it’s essentially silent while doing this. Maybe too quiet, given the thermal problems, it should maybe be a bit more aggressive with the fan curves.

That’s really the most demanding use case I have to throw at this machine, if it can handle that then for sure it’s going to be fine with processing photos and other graphic editing. Indeed, that’s what I had bought this for, a quiet, powerful platform to work on my photos, and take care of my non-gaming computing needs, but after upgrading my PC it’s just more practical to use the one machine all the time for everything. 

However, I feel I must run the same suite of test games with the dual D500 cards, for science, and for you, the people. I’m not sure how much that 3GB of VRAM will limit us, and although I am sure that dual GPUs don’t help much it’d be interesting to put some numbers behind it. It’s essentially a cut-down Radeon HD 7970 with 25% of the shader units removed, so I’m guessing GTX 1050 levels of performance, but at a cost of a TDP of 274 W. Per card? No wonder these things melt. The D700 appears to be rated at the same TDP, just with twice the VRAM and barely higher boost clock speeds.

If testing this means nuking a perfectly functional OS install and replacing it with Windows for the sake of the next few minutes of video then that is simply what I must do.

Windows 11 installs easily enough after using Rufus to create install media that bypasses the checks for more modern CPUs, and after the usual updating and hunting around for drivers we are ready to test.

I thought I’d do a sanity check by rerunning Cinebench on Windows but it won’t launch, claiming it needs a processor with AVX2 instruction set. Why doesn’t it on Mac? A mystery for the ages.

Moving on, synthetics first, as always, and unsurprisingly the oldest hardware I’ve looked at on the channel gives the lowest results, with Time Spy getting 2728 points and Fire Strike 8691 points. Low, but still apparently legendary for the hardware in both cases. Superposition gets 1851 at 1080p High.  I’ve cranked the cooling fan up to maximum for these tests, and it does a decent job when not tuned for silence rather than function.

Shadow of the Tomb Raider highlights something I’d forgotten, there’s no full DirectX 12 support on this, so thats going to cause issues with newer games. However running this benchmark in DirectX 11 gives 25 average FPS at 1080p High, so should be at least playable on lower settings or resolutions.

Pushing the limits of Counter-Strike 2 at 1080 Very High settings only gives 27 average fps, but dropping to low settings gives a very playable 104 average fps. 

COD: Warzone appears to have started enforcing secure boot requirements with the most recent update, so that’s off the table for this machine and quite a few budget builds. Necessary security or forced obsolescence? Either way, not having to play this game is a win as far as I’m concerned.

Red Dead Redemption 2 at 1080p balanced ekes out 22 average fps, but if we drop to 720p low settings we can get 30 average fps. Unfortunately that’s the end of the good news for the usual suite of test games.

Cyberpunk 2077 surprised me a bit by starting up, however at 1080p Ultra settings we’re only getting 10 average fps, and dropping to low settings only improves it to 20 average fps. 

Black Myth Wukong also has no real business even starting on this setup, and indeed it gets an average of 1 fps at 1080p High, No Framegen, and the best you can get will be 15 average fps at Low settings, with framegen, and the max of 33% FSR. So, in short, no.

Also no is the other Unreal Engine 5 game we test in these parts is Marvel Rivals, which needs DX12 to launch. It’s the same story for Forza Horizon 5, and also for Assassin’s Creed Mirage.

Of course as I’ve established in the previous video about my laptop there’s plenty of older games that I love that will play fine, even if this isn’t the best tool for the job. Here’s one from the same year this machine launched, 2013’s Crysis 3, looking great at 1080p settings and playing at around 40 fps. I did try Lossless Scaling to see if that could help, but it actually only slowed things down.

I was under the impression that the two cards are essentially running in Crossfire on the hardware level, so I’ve not been too worried by MSI Afterburner reporting that essentially only one card is being used but it’s possible that there’s some error creeping in from running these old Boot Camp drivers on a new version of Windows. However as modern games almost never benefit from using two GPUs, and no sane person buys a Mac to game on, I’m not going to stress about it. 

So, should you buy a Mac Pro 2013 in 2026? Absolutely not. There wasn’t a great case for buying one back in 2013, and even after the prices have dropped to levels normal humans can afford, outside of some specific use cases a used M1 Mac Mini will perform about as well as this, at a fraction of the power draw.  As I think we’ve shown, it’s not that it can’t be used to effectively get work done, it’s just not going to be the most cost efficient way to go about it, unless you get a tremendous deal.

But cost-efficient has arguably never been the Apple way. It’s not now and wasn’t then a practical purchase in the slightest, and if you are using your computer for professional purposes a newer, faster machine is surely warranted. The only justifications I have for this are emotional. I wanted one years ago, I can afford it now, and it still looks great and can fulfil a great deal of my computing needs. There’s another world where I’m still using this for computing and a PlayStation 5 for gaming and still getting on quite well with it.

Ultimately, the only rationale I have for keeping this around is that I think it’s cool. Thankfully, that’s the only rationale I need. Thanks for sticking around for this somewhat disorganised and surface level look at the Mac Pro, a few other video projects hit delays so has been hastily assembled. 

If you have any questions or want further details please leave a comment down below, and if you enjoyed this videotronic missive then consider subscribing. Until next time, take care of yourself, and each other.