When Kogan launched a gaming laptop earlier in the year, first impressions... weren't great. The $1000 price tag isn't too bad when you compare it to the rest of the market, but the specifications didn't exactly set hearts alight.
We'll have a full review later in the week, but for now I want to provide you all with the meat and potatoes of any gaming laptop: the benchmarks.
For reference, here's what the Kogan Atlas Pro is packing under the hood:
- 2.5GHz Intel Core i7-4710MQ
- 8GB DDR3 1600MHz RAM
- 15.6-inch 1920x1080 display
- 1TB 5400RPM HDD
- 10/100 Mbps ethernet port
- 2x USB 3.0 ports, 1x USB 2.0 ports, 1x eSATA/USB 3.0 combo port
- NVIDIA 940M 2GB
- Windows 10
There's a few other things, like an optical drive, a 9-in-one card reader and a Sound Blaster Cinema 2 card, although the core grunt -- that i7-4710MQ, NVIDIA 940M and the 8GB RAM -- is what most will focus on.
For the purposes of testing, I busted out the following: Star Wars: Battlefront, BioShock Infinite, Civilization: Beyond Earth, The Talos Principle, Sleeping Dogs and the latest version of Unreal Tournament. I also ran the Atlas Pro through 3DMark's Firestrike demo for further reference.
Not everything went according to plan, however: the Beyond Earth benchmark refused to run under any circumstances. As a result I tried Civilization 5 instead, which worked without fault. All games were run at two fullscreen resolutions: 1366x768 and 1920x1080, with multiple presets (or custom settings) used for each test. Further testing methodologies follow in their respective sections.
3DMark Firestrike Demo
The MSi logo really is in the actual demo, in case you're wondering
The 3DMark series has been used for benchmarking purposes since the late 1990's, offering a variety of tests specifically targeted at 3D accelerators instead of a computer's general performance. Some people may still remember the lobby scene from Max Payne that became a famous benchmark, helped by the fact that 3DMark ran on the same MAX-FX engine that was used for Max Payne.
The demo version of 3DMark's latest version, available for all on Steam, allows users to run the initial FireStrike benchmark that is targeted at gaming PCs. It's very handy for taxing a system's GPU power and overall grunt, although there are baseline scores for notebooks and gaming laptops that are immensely useful for our purposes.
The more strenuous 3DMark tests will tax your machine far more than any game
The picture above is a touch confusing because the graph doesn't move the arrow down when you expand the results, but the "Better than 7% of all results" is for the Kogan Atlas Pro's score, not the gaming laptop in 2013.
I expanded the 2013 gaming laptop so you could see a comparison between the specifications (as the Kogan Atlas Pro's specs are at the top of this article). The i7-4710MQ can handle the A10-4600M from AMD without any troubles, but the GPU is the clear limitation here.
Kogan advertises the Atlas Pro as being "perfect for gaming" with the 940M being capable of handling "higher frame refresh rates with ease", but that's clearly not the case here. More than anything, the results show that if you're genuinely dropping $1000 on a laptop with the intention of playing a lot of games then you'd best stick to DOSBox or the minimum possible resolutions, because the 940M's capacity to hold up its end of the bargain is very, very limited.
The Talos Principle
The Talos Principle has a strong cabal of fans, with a 96% user rating on Steam
Croteam's first-person puzzler is a fair cry from the Serious Sam series, but it was well received by the press and public alike when it launched late last year. It also has a handy in-built benchmark and a wide range of customisations.
For consistency, level caching was disabled across all tests. Options for CPU speed, GPU speed and GPU memory were set to the lowest, medium, high and ultra. You can also blow up the graphs by clicking on them, if you're having trouble reading the FPS figures.
At the lowest settings, the Kogan Atlas Pro manages to be more than playable
What's interesting in the results is the large discrepancy between the presets, with high and ultra settings smashing the Atlas Pro even at a diminutive 1366x768 resolution. If you want to play anything released in the last couple of years on the Atlas Pro, that's the running theme: drop the settings right down, and maybe the resolution as well.
The iconic shooter still casts a striking image
It wouldn't be right to run benchmarks without having an obligatory shooter running on Unreal Engine 3 -- and if you're going to pick one, it might as well be one of the most striking. BioShock Infinite can give the GPU a surprising workout thanks to the range of added effects and shaders Irrational Games threw in, and it also comes with an in-built benchmark that you can run direct from Steam.
The in-built benchmark runs through multiple sections towards the start of the game, and returns figures for each. This blows off the maximum FPS figures, however, as the max FPS in the welcome center is substantially higher than other areas. The rest of the area, however, is in line with the remainder of the benchmark. So to avoid confusion, only the minimum FPS and average FPS results are below.
The Atlas Pro returns less than impressive results given Bioshock Infinite's age
Infinite is more than two years old now, so you'd expect a gaming laptop to be released at the end of 2015 to be more than capable of handling the game at 1080p. That's not the case though, with Kogan's inaugural gaming machine failing to maintain 60fps at any setting on 1080p.
Things are more playable once the resolution is dropped right down, although still only at low settings.
The in-built benchmark for Sleeping Dogs is surprisingly thorough
Sleeping Dogs was ported to PC from the consoles, but the way Square Enix implemented anti-aliasing -- mixing SSAA and FXAA -- results in a world that is not only astonishingly smooth to this day, but also more taxing on hardware than you'd expect.
It also provides a good yardstick when it comes to the kind of performance you'd expect from a "gaming laptop". For this, three presets were used. Anti-aliasing was disabled on the lowest settings, while FXAA was enabled for the medium and higher presets.
The Atlas Pro handles the Square Enix's take on Hong Kong well at lower settings, but struggles throughout at 1080p
Without the more strenuous anti-aliasing, the Atlas Pro manages a solid average frame rate at 1366x768 -- which you'd expect, even though it's not packing a GTX-series NVIDIA card. Once that's enabled, however, the lack of GPU power takes a significant toll.
The Atlas Pro only continues to struggle at 1920x1080, undoubtedly a result of the inherent weakness in the NVIDIA 940M. It's not a flagship mobile gaming GPU, after all, but discerning buyers would hardly be impressed at its inability to maintain at least 60fps at 1080p at the lowest preset.
Unreal Tournament (2015)
The return of the iconic arena shooter is as noticeable for its visuals as much as its speed
I couldn't test all old games, of course, and given the increasing proliferation of games using Unreal Engine 4 it was impossible not to warm to the charms of Epic's upcoming arena shooter, Unreal Tournament.
Benchmarks were tested by rewatching a 5v5 CTF match (all human players) thanks to the in-built replay function, paired with the StartFPSChart/StopFPSChart commands. These automatically produce log files that show frame rate data. The map used in this instance was Titan Pass, one of the few maps in the UT rotation that is not in the work-in-progress shell phase.
Note that only the average FPS was available for these tests. The resolution scale in all instances was also set to 50%, the default setting upon installing the game.
The figures are more a mark of Unreal Tournament's excellent than the Atlas Pro's
There's a small amount of promise at the bottom end of the benchmarks here, with the Atlas Pro returning a playable, but not spectacular, result in a real-world situation. If the resolution was dropped further to 720p, it's possible that the test may have returned an average of bang on 60. In slightly smaller, but equally normal, situations of 6 and 8 player FFA/team deathmatch scenarios you could have a perfectly serviceable game.
But that moment of hope is shortlived. Even at 1366x768, the frame rate drops substantially at medium settings (although there is little difference at the higher preset). Playing at 1080p is a complete waste of time once you factor in the high rate of movement, mid-air dodging and projectiles flying throughout. 60fps is an absolute minimum for a game of this calibre, and without any headroom it's hard to recommend anything else bar giving Unreal Tournament a wide berth for those playing on an Atlas Pro.
Nothing taxes a CPU quite like cities and trade routes en-masse
The Civilization series has been a benchmark staple for years, and for good reason. Towards the end of the game, Civilization 5's in-game world becomes swarmed with improvements, cities, unit stacks, borders and all manner of things that take an extreme toll on the CPU. My preference was to take advantage of the more modern Beyond Earth, but the game continued to crash irrespective of settings. As a result, I resorted to the late game benchmark from Civilization 5, which simulates a match 300 turns in with a completely exposed and vastly populated map.
Global presets were not available for this game, so settings were changed manually. In instances where options or shaders could not be turned off (for the minimum test), they were set to minimum or lowest instead. As was the case with Unreal Tournament, the game does not return minimum or maximum FPS values and as a result only the average FPS is displayed.
The test, however, does provide three scores: one where shadow pass is turned off, one that emulates "an infinitely fast GPU and will bypass most driver overhead" titled "No Render Score" and a third (Full Render) that uses the graphics settings that the game would ordinarily run at. To ensure consistency with real-world results and avoid confusion, I'm only showing the average FPS figures from the Full Render results.
Despite having the power of an i7, the Atlas Pro struggles significantly even on a more CPU-intensive game
It's worth remembering that these figures are run using a late game benchmark that simulates something close to the most intensive scenarios Civ 5 can produce, so in most situations it's fair to say the Atlas Pro would maintain well over 60fps when running at 1080p on the lowest settings. I'd even venture to say that the Atlas Pro might manage around 60fps, or just under, on 1080p in medium settings in most early to mid-game scenarios. Given that the frame rate will remain consistently low once you're 250-ish turns in, though -- those units don't just disappear -- it's difficult to give Kogan's gaming laptop a pass here.
Star Wars: Battlefront
It's the best looking game of the year, and one of the most well optimised
Given that the console commands are the same and it's one of this year's most gorgeous games, I'd be remiss not to throw DICE's latest first-person shooter at the Kogan Atlas Pro. After all, if you're buying a gaming laptop, you're going to want to see how it handles games of this calibre.
Battlefield 4 has been a benchmarking staple for years now, thanks to DICE's excellent optimisations and the nifty log files that the game produces. They're enabled and disabled with a simple console command, and the fact that the game is so damn pretty to look at helps if you're doing this on your weekend.
For reference, I opted to run through the tutorial speeder bike level on Endor to maintain a controlled result -- multiplayer scenarios, while obviously providing a closer guarantee of real-world results, cannot be replicated with any efficiency in Battlefront due to the lack of a demo/replay function. So when interpreting the results, factor in that you will probably want double the FPS to get a genuine idea of how things would stand when confronted with multiplayer.
It's worth noting that some degree of antialiasing was on at all settings, as per DICE's presets, and the resolution slider was set to 100%.
As you'd expect with a modern AAA release, the Atlas Pro struggles -- and that's just in the singleplayer
These certainly aren't the figures Kogan were looking for, and potential suitors shouldn't be impressed either. The singleplayer Endor speeder mission is one of the most taxing single-player scenarios and the Atlas Pro struggles throughout. The results are so poor that even those looking to play in the smallest 4:3 resolutions -- 800x600 or 1024x768 -- would find little to enjoy.
DICE should get a little bit of credit here too: being able to return results like these on hardware so many leagues out of its class says something about the amount of work they've done with the Frostbite engine. Quite a few games over the last couple of years have had problems out of the gate with optimisations, but Battlefront is most certainly not one of them.
I noted previously that the resolution slider was set to 100%, unlike in Unreal Tournament when the game automatically configured things to be dropped to 50%. You'd certainly be able to get some noticeable improvement in the performance by dropping that down in Battlefront, but I'd argue that you would make the game instantly unplayable by doing so. Being able to spot Rebels and Stormtroopers far off into the distance -- or enemy X-Wing/TIE Fighters -- is absolutely paramount if you want to stand a chance in multiplayer.
Tanking the resolution scale for performance would utterly decimate your ability to make out models from anything further than a few meters away, and if you have either Battlefront or UT (which is free) you can test this for yourself. I ran tests at home and in the office for fun to see how playable things were, which is why I was comfortable running the UT tests with the resolution scale at 50% -- and why it's absolutely essential that Battlefront's scale remains at 100%.
It might be nice to say the Atlas Pro can run Battlefront at 60fps on the lowest settings and resolution, but if you can't make out models until they're right in your face then the game simply isn't worth playing.
So that's it for the benchmarks for the Kogan Atlas Pro. Like I mentioned at the top of the article, a full review (in the same style as the Cherry MX 6.0 Board) will appear later this week.
Let me know what you think in the comments below -- whether it's your thoughts on gaming laptops, what you think of the Atlas Pro so far, what games you think are best for benchmarking, or what hardware you'd like to see tested in the future!
(Final note: if the graphs look a little weird throughout, it's because I had to build them all using Google Sheets. It's not quite as flexible as Excel -- I had to use error lines for one chart so the numbers in one bar weren't aligned on the axis -- although the visual inconsistency shouldn't affect the readability of it all. Sorry about that.)