Why do we like little PCs? You give up a lot to make a computer small. Everything has to fit in a cramped casing, and you forfeit any real opportunities for expansion. Even working on the machine becomes a huge pain in the butt, assuming you can really work on a small-form-factor PC to begin with. The further you shrink the system, the less and less standardized your parts become, until you’re using barrel plugs and flexible flat cables like a laptop.
As I said in the post announcing this review, I didn’t evacuate for Hurricane Harvey. I have evacuated before, though, and down here in Southeast Texas we’re always under the threat of a serious storm. Ease of evacuation (or more generally, ease of transport) is just one advantage to a little PC. It’s easier to find space for it, and it’s less likely to be in the way wherever it ends up. It’s cute to look at, and frankly, it’s just really cool having that much processing power in a little-bitty space.
The machine on deck today is Zotac’s Zbox Magnus EN1080K. Don’t be fooled by the “K” at the end of the model number: this machine isn’t overclockable. Instead, it stands for Kaby, as in Kaby Lake. This is the updated model of the original Magnus EN1080, and it features a Core i7-7700 CPU where the original model had a Core i7-6700 inside. Just as the change from Skylake to Kaby Lake was a small one, this CPU upgrade is also a small change: an extra 200 MHz of base clock speed and support for DDR4-2400 memory. Since this machine makes no use of the Core i7’s built-in graphics processor, the Kaby Lake changes to the Gen9 graphics’ video block are irrelevant.
Let’s check out the full specs of the machine as we tested it, shall we?
|Zotac Zbox Magnus EN1080K barebones|
|Processor||Intel Core i7-7700|
|Memory (not included)||32GB HyperX Impact DDR4-2400 (2x16GB SODIMMs)|
|Chipset||Intel B150 Express|
|Graphics||Mobile GeForce GTX 1080 with 8GB GDDR5X RAM|
|Storage (not included)||WD Black 512GB NVMe M.2 SSD|
|Expansion and display outputs||1x USB 3.1 Gen 2 Type-C
1x USB 3.1 Gen 2 Type-A
4x USB 3.0 Type-A
3x HDMI 2.0 (up to two in simultaneous usage)
|Communications||2 Realtek 10/100/1000 Ethernet
Intel 8265 802.11ac + Bluetooth
|Dimensions (HxDxW)||8.9″ x 8″ x 5″ (23 x 20 x 13cm)|
|Weight||8.5 lbs (3.9kg)|
|Included cables||2 180W power adapters|
The combination of Intel’s second-fastest desktop CPU and Nvidia’s second-fastest desktop GPU is a potent one. This machine’s graphics chip is actually a mobile GeForce GTX 1080, but the Pascal architecture means the difference between desktop and mobile parts is pretty minimal. Here, the choice of a laptop part means a slightly lower base clock and reduced power limit. The EN1070 that I reviewed before was hampered by its insufficient 180W power supply and the slow CPU, but there shouldn’t be any such problems here. The EN1080K uses a pair of those very same power adapters. That gives the machine up to 360W of power draw to play with, and it should guarantee a much more solid gaming experience.
The hardware pulling that power will be making quite a bit of heat, too, and Zotac’s engineered a custom liquid-cooling loadout for the EN1080K. We couldn’t strip our machine down to its guts, but other reviews of the EN1080K suggest Zotac is using a close descendant of the system that the company debuted with the Magnus EN980. We never took one of those boxes apart, but Jeff got a neat picture of an example system encased in transparent Plexiglas at Computex a couple of years ago. No doubt the hardware here is cooled in a very similar fashion. There are hefty water blocks on both the CPU and GPU, and they’re piped into a single thick 120-mm radiator for exhaust purposes.
The RAM and storage that we’re using for these tests came from the same place the machine did. Friend of the site Aaron Schradin let us borrow this little beastie for a bit so that we could see how it measured up to its forefathers. The final resting place of the EN1080K we’re testing will be under the next generation of Turris VR chair. If you haven’t read up on the Turris, click over to shortbread baker Colton’s piece on the best place to plant your butt for seated VR. Our thanks to Aaron for trusting us with his expensive hardware for a few days.
So you’ve seen the picture above before, but what does the thing actually look like? Well, turn the page and let’s have a closer look.
Up close and personal
I really appreciated the flat top and bottom of the Magnus EN1070. You could set instruction manuals or even a monitor on top. The EN1080K is still flat on the top, but I would never lay things on top of it for fear of restricting the radiator’s airflow. The inability to set things on top of the Zbox Magnus EN1080K is a minor annoyance, but it really does put a damper on the variety of places you could put it in your home. I wouldn’t want to put it in an entertainment center with narrowly-spaced shelves, for example. That might inconvenience some folks eyeing this type of chassis as an uber-console or an HTPC.
Moving across the front of the machine from left to right, you’ve got a big round power button (encircled by a power LED), an HDMI 2.0 port, an SD card slot, a USB 3.1 Gen 2 Type-A port, a microphone jack, a headphone jack, and finally the USB 3.1 Gen 2 Type-C port. I still haven’t picked up any USB 3.1-capable devices, so I can’t actually test the speed of those ports. However, I can confirm that they both work, at least.
The HDMI port on the front of the machine is obviously intended for use with a VR headset, and I have to say that while I appreciate this feature on a big and heavy tower PC, it seems a bit unnecessary on a mini-PC. At least, if nothing else, it should keep cabling nice and neat for VR rigs. You lose access to one of the two HDMI ports on the back of the machine when using the front HDMI output, just like with desktop graphics cards that offer front HDMI outputs.
Speaking of the rear of the machine, here we are. Over on the left you can see the paired power plugs. In the center we have a four-high stack of USB 3.0 ports. Just to the right of that, you have dual DisplayPorts and twin HDMI connections, and then to the right once more we have two Gigabit Ethernet jacks. A pair of RP-SMA connectors for the included Wi-Fi antennas flank the rest of the ports on either side.
Flipping it over, you can have a look at the bottom of the EN1080K. Here you can see the various regulatory labels as well as the air intakes around the edges of the base. The EN1080K sits up on four rubber feet embedded into the aluminum thumbscrews that hold on the base. This arrangement seems critical to the EN1080K’s cooling prowess: if you were to lose or remove those thumbscrews for any reason, the EN1080K would have a much harder time drawing in air. The vents do extend around the corners of the base to the outer edge, but all the same, I’d try to keep the feet on. This is not a machine you want to starve for air.
Of course, if you lose those thumbscrews you’d have a hard time keeping the machine together. The base doesn’t latch in or snap in to the rest of the machine at all—only those four thumbscrew-feet hold it on. Remove them, and the base lifts right off. Underneath, you’ll find direct access to the machine’s 2.5″ SATA drive bay, an M.2 slot, and two DDR4 SODIMM slots. Like every other Zotac mini-PC I’ve ever used, installing the storage and memory is effortless.
And here she is all fixed up and ready to go. Unfortunately, I didn’t fully disassemble this machine like I did with the EN1070. Given that it’s not our hardware to break, and given that I would have had to void the owner’s warranty to totally tear it down, I decided to forgo the final step. Rest assured that if this were my machine, I’d have ripped it apart, de-lidded the desktop CPU, and taken pictures of the whole process for gerbils to grouse about.
However, for anyone that finds this review wondering how hard it would be, I can say that the construction of the outer casing seems fairly simple. Tiny Philips-head screws hold the aluminum plates to an internal steel frame on which all of the hardware is mounted. Remove those screws and the outer casing should more or less just fall away from the frame, accounting for the cable that no doubt connects the front panel button and lights to the motherboard. Taking the EN1080K apart further than that will be a bigger production, though.
Firing it up
As you can see in the image above, the Magnus EN1080K needs two bulky 180W power bricks to run. A first-world problem, sure, but one that buyers should be aware of. You’ll always need two power plugs for the EN1080K itself. This isn’t an insurmountable issue by any means, but it’s an inconvenience, and I really wish Zotac would have just sprung for a larger power adapter, perhaps with a more robust connector. Barrel plugs aren’t the most durable things around, after all. For the record, you can’t run the EN1080K at all with just one adapter plugged in. I tried for the sake of science.
The first thing I do when setting up any new system is check for a firmware update. I had a few firmware-related follies with the EN1070, and I wanted to make sure and avoid those if at all possible. After downloading the most recent BIOS update from Zotac’s site and dropping it on a flash drive, I dutifully booted to an EFI shell and attempted to flash just as Zotac’s instructions bid me do. I was greeted with an EFI shell hanging off the right edge of the screen. Attempting to enter shell commands produced no results, so all I could do was reboot the system.
After fiddling with the BIOS for the better part of an hour, I figured out that the machine wouldn’t display the EFI shell correctly unless the boot settings were configured to boot the graphics card in legacy VGABIOS mode instead of in UEFI GOP mode. Once I worked out that weirdness, I was able to attempt to flash the BIOS. Doing so, however, gave me a message insisting that the installed BIOS version was the same as the one I was attempting to flash. This despite the fact that the version numbers did not match: the BIOS version I was trying to flash was version 111, while the version installed was version 108.
I contacted Zotac, and eventually the download on the site was replaced with a fixed version. That update allowed the EFI flash to complete successfully. I think it’s worth noting that throughout this entire process I was made all the more nervous by the machine’s bizarre boot behavior. The Zbox Magnus EN1080K, or at least this particular specimen, takes upwards of 15 seconds after power-on before it actually completes POST and attempts to even display anything on the screen. Sometimes, during this process, it will power itself off and restart again, and sometimes it will do this even after displaying the boot logo.
Normally in a custom build that kind of behavior would make me suspect a failed power supply or motherboard, but during the Windows install, game testing, and around two dozen hours of non-testing gameplay, the EN1080K gave me nary an issue. While leaving it idle, though, another problem cropped up. The EN1080K refuses to resume from standby or hibernation. Before the firmware update, this extended even to display standby. If you let the EN1080K power down connected monitors, it would lock up and become unresponsive to anything but the power button. The recent firmware update seems to have resolved that issue, though the machine still refuses to wake up from system standby. That’s a bit of an annoyance in a PC this expensive.
The firmware update also seems to have changed the machine’s Turbo Boost behavior. While doing my initial game testing, I noted that the Core i7-7700 seemed to have some sort of “multi-core enhancement” enabled, as it was eager to run at 4.2 GHz on all four cores instead of just one. I noticed this while playing GTA Online and while running the CPU-Z benchmark, so I looked for an option in the UEFI setup to control it. There isn’t one. After the BIOS update, however, the CPU will only hit 4.2 GHz on a lightly-threaded workload, as it should. Playing a multi-threaded game, or using a multi-threaded stress test, the CPU boosts to 4 GHz and stops. That’s the expected behavior of a Core i7-7700, so it’s not like this is a problem—just another example of the EN1080K’s somewhat odd behavior.
All I can really do is chalk up the EN1080K’s quirkiness to Zotac’s firmware. The EN1070 that I reviewed before wasn’t able to talk to NVMe drives when I first received it, and a later firmware update was required to resolve that problem. My ancient AMD Zacate-based Zbox used to be real picky about the RAM it would play with, as well. Quirky firmware is something of a Zotac hallmark at this point. If nothing else, I will give Zotac’s support credit for fixing up the issue within a week.
Performance and noise testing
We’ll get to the actual results in a second, but performance is pointless if you can’t stand to be in the room with the machine. I don’t have the equipment to do scientific noise testing, but I’m confident that no one outside of the most fanatical passive-cooling enthusiasts is going to have a noise complaint against this thing. Even playing an intense session of GTA Online with 25 players pegging both CPU and GPU, the fan is only a step above inaudible. I have to lean down to hear it, and that’s with everything else in my house completely turned off and the speakers muted.
Manually turning the fan up to see what it sounded like, I was greeted with the relatively inobtrusive whoosh of a 120-mm spinner. The fan does get very loud if you crank it up to the maximum speed, but it never came anywhere near that during my testing. I don’t know why you would do that in any case, because the machine runs cool enough regardless. Both GPU and CPU idle at around 45°-55°C in my home with a 22°C ambient temperature. Playing a game that loads both CPU and GPU to their fullest, I saw the CPU briefly max out at 80°C and the GPU hit 84°C. Both results are completely acceptable for the hardware in question.
Along similar lines, one of the major complaints that I had with the EN1070 was that it didn’t have enough power on tap to allow its mobile GeForce GTX 1070 to reach even its reference clock sometimes, much less beyond. The GTX 1080 in this machine is specced for the same clock rates as the reference laptop GeForce GTX 1080: a 1556 MHz base clock and a 1733 MHz boost range. However, it’s happy to boost well beyond that. I frequently saw the card running at over 1800 MHz and (particularly in Doom) saw it occasionally hit 1900 MHz. Likewise, the Core i7 7700 CPU gleefully goes up to 4.2 GHz in lightly threaded workloads and hangs tough at 4 GHz across all cores.
To provide a frame of reference for the Zbox Magnus EN1080K’s actual game performance, we tested it against a similarly-configured desktop with Core i7-7700K CPU and a full-sized GeForce GTX 1080. Even though both of those chips are fundamentally the same silicon as in our Zbox, both of them will run higher clocks and use more power than the TDP-limited parts in the micro machine.
We fully expect the EN1080K to get outpaced by the full-sized desktop, because that’s a completely reasonable outcome. However, we think the comparision will paint a representative picture of the Zbox Magnus’ performance. Note that since the testing on these two machines was done by different people in different places, this isn’t a perfectly scientific test, but we’ve done our best to keep things as even as possible. With caveats in place, let’s step to the actual results.
More than a year on, I still think Doom is one of the best-looking games out there. It’s also one of the best-running games out there, which makes it a little difficult to use as a benchmark. We have to crank the settings all the way to the top to provide a solid challenge for our contenders, so what’s how we tested it here: at 2560×1440 with everything all the way up. The GeForce GTX 1080 is really capable enough to run the game at a 4K resolution, but I didn’t have a 4K monitor available to use at press time. The images below represent the settings we used in testing save for the use of the Vulkan API rather than OpenGL.
As expected, the EN1080K lags a little behind the desktop machine. It’s not far, though. The Magnus EN1080K is smaller than an Xbox One, and yet it turns in an average framerate of over 90 FPS in Doom, even with all of its craziest eye candy cranked. All that while remaining quieter than the purrs my cat makes when sleeping on top of my receiver, too. The 99th percentile time is nice and low, and the frame-time plot is pleasantly smooth. Where the smaller Magnus EN1070 struggled within its power limits, I noticed no such thottling on the EN1080K, and the proof is in the performance.
If you’re just joining us here at The Tech Report, you might be confused about this business immediately above. The “time spent beyond X” graphs tell you how much real time our test system spent rendering frames that took longer than 16.7ms, 8.3ms, and so on. 8.3ms corresponds to a framerate of 120 FPS, 16.7ms is roughly 60 FPS, 33.3ms is roughly 30 FPS, and 50ms is 20 FPS. This gives us a good way to pick out stutters and hitches during gameplay. If you drop below 20 FPS during gameplay, even for a moment, it’s going to cause a noticeable stutter, and our “time spent beyond” graphs aggregate that roughness, for example.
There’s really not much to say about the EN1080K’s performance in Doom. It’s stellar. It actually spent less time below 60 FPS than the desktop machine, although it also spent less time above 120 FPS. As I said in the intro, we really could have tested this game in 4K resolution. With that said, it runs beautifully in 2560×1440, and at these settings it’s a prime candidate for a high-refresh-rate display sporting ULMB or another strobe backlight. Something like that still isn’t available on a 4K-resolution display.
Mirror’s Edge: Catalyst
Mirror’s Edge: Catalyst is a new title for TR. We may or may not end up using it regularly, but while playing it lately I was struck by the beauty of some of the vistas in the game. EA and Dice put the first-rate Frostbite Engine to work in this game with some stellar visual effects. In fact, in a post-release patch, the company added a special “Hyper” preset to the options specifically to make use of high-end graphics processors like the GeForce GTX 1080. We tested the game at 2560×1440 on the “Hyper” preset. In this parkour action game, there are pre-set time trial events called “dashes” scattered across the game world. We ran the dash called “A Handy Shortcut” without taking the aforementioned shortcut to get about a minute of play time out of the event.
We wouldn’t make too much of the slightly higher average framerate on the EN1080K, because two different testers performed this benchmark on two different machines halfway across the country. What’s more important is that both machines deliver nearly the same 99th-percentile time. The “Hyper” preset in Catalyst is incredibly demanding despite the stylistically-simplistic geometry of the game’s play area, and it put the hurt on both machines.
Both machines performed within the bounds of our expectations in Mirror’s Edge: Catalyst. As with Doom, both machines dipped slightly below 60 FPS briefly, but neither fell to 30 FPS or worse. While not as silky-smooth as Doom, the game is eminently playable on the EN1080K at these settings, and the City of Glass looks pretty awesome. Once again, don’t make too much of the results here; the difference in performance is as like to represent some variation in our testing scenario as it is difference in the hardware. If this game becomes a more frequent member of our test suite, we’ll have to devise a more reliable sequence.
Grand Theft Auto V
Of course, I had to test Grand Theft Auto V. This evergreen open-world title is still a blast to play online, and Rockstar continues to release major content updates for the title. We don’t test it online, though, because the nature of online play means that making a reproducible testing scenario is nothing short of nightmarish. I ran protagonist Franklin through the usual TR testing track near his aunt’s home that you can see in the video below.
The third verse is the same as the first. The EN1080K falls slightly behind the desktop machine, and that’s exactly where we’d expect to find it thanks to the differences between the Core i7-7700 and the Core i7-7700K. I was very careful to monitor both CPU and GPU clock rates and thermals during the GTA testing because I find that this title puts the most even load on both components. Because of the great number of objects in the game world being simulated, GTA V is heavy on CPUs in ways that most other games aren’t. Neither chip crested 80°C while testing GTA V, even though the EN1080K remained inaudible to my ears.
Given its popularity, it’s no surprise that the GeForce drivers are very tightly optimized for GTA V. As a result we see some of the flattest frame-time percentile graphs we could hope for. Stepping down to the “badness” metrics, there’s basically no badness to be had. Both machines spent essentially the entire testing period well over 60 FPS, and the game plays exactly as well as you’d expect given that tidbit.
Besides the standard TR offline testing, I also fooled around in GTA Online for a whole evening using the EN1080K. Were it not for the fact that I wasn’t sitting in my usual spot, I would have completely forgotten that I wasn’t playing on my usual desktop. In fact, I think the game may even play a bit more smoothly on the combination of the Zotac’s Kaby Lake Core i7 and mobile GeForce GTX 1080 than it does on my own Devil’s Canyon i7 and GeForce GTX 1080 Ti. Or maybe not. Either way though, it plays great, and certainly a lot more smoothly than it did on the EN1070.
Zotac’s Magnus EN1070 is incredibly small. I mean, it’s really miniature. It’s barely bigger than a Nintendo Wii, and it’s comfortable for a grown man to carry in one hand like it was a thick book. The Magnus EN1080K isn’t like that. It has a significantly bigger footprint than even the Corsair One, and since you can’t set anything on top of the EN1080K, it might as well be just as tall. Zotac’s mightiest mite is also not likely to be as quiet in a confined entertainment center as it was on my workbench, thanks to the poorer air circulation of closed shelving. That heft means the EN1080K really needs to earn a place for itself in whatever open airspace it occupies. Even with my qualms, I think Zotac has largely managed that feat.
Most of the extra space inside the Zbox Magnus EN1080K (relative to the EN1070) appears to be for the exotic liquid-cooling arrangement used to cool both CPU and GPU. While it does work very well, I have to wonder if the same thing couldn’t have been accomplished using heat pipes and fans in a smaller space. I might seem like I’m unfairly harping on the EN1080K’s breathing room requirements compared to its stablemate, but the size difference and cooling needs of the bigger box are big deals. When one of a system’s selling points is “small” and it’s actually taking up a bunch of space, it’s not really delivering on its promise.
So with that said, the Magnus EN1080K’s size isn’t the headlining act like it is with the Magnus EN1070. Zotac uses the extra bulk of this machine to deliver an impressive gaming experience in a reasonably-compact package. Both the GeForce GTX 1080 and the Core i7-7700 in this machine are at the top of their game, and they don’t appear likely to become obsolete anytime soon. I wouldn’t mind a couple more memory slots, or a space for a 3.5″ hard drive, but 32GB of memory is plenty—excessive, even—and I could always add more storage using USB devices or my network. The EN1080K runs great and stays cool and quiet while doing so, and so in that regard it’s a smash hit.
I can’t argue with the EN1080K’s performance, but I can argue with its price. Zotac asks $1700 for this machine as a barebones. For the machine as we configured it, you’re looking at $2300 easy, and for that price, Corsair will sell you one of its One systems outfitted with an overclockable Core i7-7700K CPU, a higher-clocking desktop GeForce GTX 1080 GPU, and an additional 2TB of storage. Sure, the One is a lot bigger than the Zbox Magnus EN1080K, at 13 liters versus 5.8 liters, but the One’s smaller footprint makes it less space-intensive in practice. The One doesn’t require two power outlets and two power bricks to do its thing, either.
If you’re after a form of mobility with a system this powerful, there’s also the matter of high-performance gaming laptops to consider. For the price Zotac is asking, you can pick up some pretty potent hardware that includes its own display and a battery, to boot. A gaming laptop in the $2000-to-$2500 range isn’t going to compete with the EN1080K on pure performance, but it might offer you a better value for your money depending on what you’re planning to do with it.
Still, there’s no denying that this machine offers impressive performance density. The owner purchased it specifically for that reason, and if you’ve got a real tight space and want maximum performance from it, Zotac’s mighty mite here might be your best bet. Just make sure that you truly need what it has to offer.