Welcome Stranger to OCC!Login | Register

Lenovo Legion Y720 Laptop Review

   -   
» Discuss this article (0)

Gaming Experience:

For these tests I worked with three games and ran them when on battery and when plugged in. The GTX 1060 was the selected GPU for all of these tests, so there is no swapping to be concerned with. The three games I tested were Serious Sam Fusion 2017, more specifically the Hatsheput run I did in Serious Statistics, DiRT 4, and The Witcher 3: Wild Hunt. All of these games I tested at 1920x1080 and 3840x2160 (4K) because if you have a 4K display, why would you not try using it for gaming? Well, because it can kill performance, but not always, and we will get to that. When it seemed necessary, I did tweak settings, and for Serious Sam Fusion 2017 I did test with both the Vulkan and DirectX 11 APIs. I also ran through some 3DMark benchmarks: Sky Diver; Fire Strike; Fire Strike Ultra; and Time Spy.

Sky Diver, Fire Strike, and Fire Strike Ultra are all DX11 tests, with Sky Diver meant to be less intense than Fire Strike, while Fire Strike Ultra looks at 4K performance. On Sky Diver the laptop earned a score of 22,145 (36,095 for graphics, 8175 for physics, and 16,588 for the combined test). On Fire Strike it got 9241 (11,127 graphics, 8743 physics, and 4228 combined) and on Fire Strike Ultra it got 2630 (2585 graphics, 8758 physics, and 1373 combined). The score for Time Spy, a DirectX 12 benchmark, was 3545 (3593 for the graphics score and 3296 for the CPU score). All of these scores were while plugged in, but for fun and curiosity I also ran them when on the battery. Unsurprisingly the scores are much lower as throttling kicks in, holding the framerate at about 30 FPS, except for Fire Strike Ultra and Time Spy. Both of these benchmarks were already running at below 30 FPS, so that cap had no impact. Curiously in the other two tests, the physics scores actually increased while graphics decreased. Perhaps this is because with the GPU doing less, there was less overhead on the CPU, but that is purely a guess.

In total I have eight runs of data for Serious Sam Fusion 2017 because I tested with the DirectX 11 and Vulkan APIs, while on battery and while charging, and at both 1920x1080 and 3840x2160. What I did not do is mess with the graphics settings, because I just set it to Ultra and went with it, which are the same settings I used in Serious Statistics. That is one of the nice aspects of this game, at least in the Serious Sam HD: The First Encounter levels; that modern hardware can run it very well.

Starting with DX11 data, when running at 1080p plugged in or not had some impact, but not too much. The average FPS while charging, so there should not be any throttling for power, was 150.5 FPS (6.65 ms), and when on the battery it was 145.2 FPS (6.89 ms). The percentile data starts showing a difference, with the 0.1% having a frame rate of 110.95 FPS (9.01 ms) while charging and 71.35 FPS (14.01 ms) when on the battery. The amount of time spent below 60 FPS (frame time longer than 16.7 ms) is fractions of a percent for both states and the frame time difference between consecutive frames for both were with +/-2 ms, when looking at 1% to 99% of the data. The even more outlier difference data of 0.1% and 99.9% shows the on-battery test being less smooth, but the greater of these two was 7.43 ms, which still leaves plenty of time (for this game) to render the frame before the display cycles.

 

On battery, DirectX 11, 1920x1080

 

 

Plugged in, DirectX 11, 1920x1080

 

Bumping the resolution up to 4K does pull down performance, but it is still really good. The average frame rate while on the battery was 81.2 FPS (12.32 ms) and when charging it was 80.81 FPS (12.38 ms), so really these are tied. The percentile data does show some differences between the two runs, but actually very little, at least on the low end. One has its 99.9% at 55.86 FPS (17.90 ms) and the other at 56.20 FPS (17.79 ms) which is so close I will not even bother identifying which data is for which run. All of the data, except the 0.1% FPS values, so the fastest frames, mirror each other so closely to be indistinguishable.

 

On battery, DirectX 11, 3840x2160

 

 

Plugged in, DirectX 11, 3840x2160

 

Something I do want to say about running 4K in the Hatsheput level of Serious Sam Fusion 2017 is it looks no different to my eyes from the 1080p runs. At Ultra settings the assets are probably already as good as they are going to look on 1080p, so 4K just adds the load of more sampling. (I am pretty sure that because of how MSAA works to process only the locations it is needed, I do not think its load would increase too substantially between these two resolutions, provided everything else appears the same.)

You can expect similar parity between the 1080p and 4K results for Vulkan, but you will be disappointed, to some degree. When on the battery, the Vulkan render gets throttled all the way down to 30 FPS. On the bright side, this means there really is no performance difference between 1080p and 4K, but having to play at 30 FPS is not as much fun. Neither is going through the data.

 

On battery, Vulkan, 1920x1080

 

 

On battery, Vulkan, 3840x2160

 

When plugged in there is no throttling, and we can see some solid performance. Unfortunately both at 1080p and 4K, DirectX provides better performance in every statistic, except the difference between consecutive frames. There Vulkan is superior, but the rest is tied or beaten by DirectX 11. This and the throttling either not occurring or not being as harsh makes DirectX 11 the better API to use, in this game, on the Y720.

 

Plugged in, Vulkan, 1920x1080

 

Plugged in, Vulkan, 3840x2160

 

Moving on to The Witcher 3: Wild Hunt, the way I tested it was to take advantage of how Geralt's horse will essentially auto-pilot from one location to another. I have a save that starts me at the Stonecutters' Settlement and takes me to the Herbalist's Hut. This path crosses through the city of Novigrad, which is important because high numbers of NPCs can load the CPU a lot. When outside the city, the CPU load is much less, so there is a nice balance here. It takes about two minutes to run this path and all it involves is my holding down Shift until near the end, because Roach (the horse) does not make the necessary turn to actually reach the hut. I always started from the same save, so positions and behaviors for the NPCs are the same between runs.

To start I cranked the settings up to Ultra, and as one might expect from a graphically intense game like The Witcher 3, the laptop is not completely up to it. It is better when plugged in, with the average frame rate being 46.13 FPS (21.68 ms). On the low side, the 99.9% frame rate was 14.50 FPS (68.94 ms), but only 0.4242% of the time was it below 30 FPS, so that is good, especially compared to when running on the battery.

 

Plugged in, Ultra, 1920x1080

 

On the battery, it looks like the throttling was bringing the game down to 30 FPS, but with how graphically intense the game is, it cannot maintain 30 FPS, so it would drop lower, but still stuck at around 30 FPS.

 

On battery, Ultra, 1920x1080

 

I am just going to say there is no point in looking at the Ultra 4K data. The Y720 is not up for it in The Witcher 3 at any setting.

Because I did not want to worry about the throttling, I did not run additional tests on just the battery for The Witcher 3. Personally, the game is one I want to play at its best, and that means while plugged in. I did make runs at the High and Medium presets for graphics, to see how playable the game can be at these lower settings.

On High, the average frame rate came out to 57.33 FPS (17.44 ms) and this is pretty accurate actually. The data shows 63.25% of the time the frame rate was below 60 FPS but only 7.39% of the time was it below 50 FPS, so the bulk of the time we are above 50 FPS, which is certainly playable. It was above 42.24 FPS (23.68 ms) 99% of the time.)

 

Plugged in, High, 1920x1080

 

While that is playable, I wanted to go down another step to Medium, to see if we can get steady 60 FPS, and indeed we do. The average frame rate was 77.37 FPS (12.92 ms) and 99% of the time it was above 55.61 FPS (17.98 ms). Only 3.49% of the time was it below 60 FPS, which is a nice thing to be able to say. On the high side, the 0.1% and 1% for the data were 125.89 FPS (7.94 ms) and 114.39 FPS (8.74 ms), so you have an idea of the range here. It would be nice if the consecutive frame time difference were tighter, but +/- 7.5 ms or so (for 1% to 99%) is still pretty good. At Medium, the performance was quite smooth, and that is what I was aiming for when I selected this preset.

 

Plugged in, Medium, 1920x1080

 

Unfortunately I do not have OCAT data for DiRT 4, so I had to keep an eye on the FPS counter the Steam Overlay provides. This is a game I feel is best enjoyed at high frame rates, so performance is important. Unfortunately, 4K is not possible. Even at the lowest settings and using CMAA, a post-processing AA solution, the performance is too low to be viable. At 1080o, performance is quite good, well surpassing 60 FPS at the High preset. It often stayed above 70 FPS and was even reaching to 100 FPS and a bit beyond. The performance can vary in this game a fair amount, depending on the track and conditions, but it is still quite capable to play it with good quality settings. At least when plugged in.

Like with the other two games, I played DiRT 4 while plugged in and while on the battery. What I described above is all from when the laptop was plugged in, because being on the battery causes problems, if V-Sync is disabled. The laptop throttles down when on the battery, capping the frame rate at 30 FPS, which is not ideal for this game, but it is made worse by the screen tearing from my nightmares. It was legitimately the worst screen tearing I have ever seen, to the point it could have been called display corruption. Turning on V-Sync corrected this, but even still 30 FPS is slower than I would want to play the game at.

As I mentioned in the Introduction, I also wanted to test how the CPU handles gaming and some x264 encoding via OBS Studio. Not well, which is hardly surprising given a 2.8 GHz part with only eight threads. I started my testing with the veryfast preset for x264. There are only two presets less stressful on the CPU, and they are superfast and ultrafast. Without a very high bitrate, ultrafast will look horrible compared to the other presets, so I did not bother testing with it. Testing with superfast showed it was possible without frames dropping.

The successful superfast test was in Serious Sam Fusion 2017 and it was at 1280x720, 60 FPS, and with the game's frame limiter set at 60 as well. Though I have not encountered the issue often before, sometimes capping the in-game frame rate can help with capturing, and it appeared to help here. Of course if you actually want to record gameplay on this laptop, you would be better off using one of the ASIC encoders it contains (Intel Quick Sync and NVIDIA NVENC are both available), thanks to the low performance impact they offer, but it was still fun to look at how well software encoding would work here. Not well, but still technically possible.

One last set of gaming experiences I can talk about are the MR/VR experiences I have enjoyed, and are the reason I was loaned this Legion Y720. I do not have any performance data, but the performance was consistently smooth for me in all of the games I played. The only special note to make about the performance was that I left the settings for Serious Sam VR: The Last Hope where they were. I have observed in other Serious Sam games the engines will select settings as it feels appropriate for the hardware, and in this case the settings were not very high. Still, the graphics looked good, except for very far away objects, and performance can be more important than detail, so I have no complaints. Also, Serious Sam VR: The Last Hope is the only VR game I did much with that really offered much in the way of graphics settings. The others just performed well at whatever default settings are for them.

Altogether, the performance is pretty good. Maybe settings need to be tweaked, but that is the cost of taking your games with you. The throttling when on the battery can be pretty bad in some games, so you might need to plug in to enjoy your gaming. I just tried changing the power setting for DiRT 4 in the NVIDIA drivers to maximum performance, but it was still throttled to 30 FPS and had significant tearing. The tearing might have been better than before on Optimal Power, but still a game you want to plug in for.




  1. Lenovo Legoin Y720 Review - Introduction
  2. Lenovo Legoin Y720 Review - Physical Characteristics
  3. Lenovo Legoin Y720 Review - Basic-Use Experience
  4. Lenovo Legoin Y720 Review - Hardware Specs
  5. Lenovo Legoin Y720 Review - Battery Life
  6. Lenovo Legoin Y720 Review - Gaming Experience
  7. Lenovo Legoin Y720 Review - Conclusion
Related Products
Random Pic
© 2001-2018 Overclockers Club ® Privacy Policy
Elapsed: 0.2144820690   (xlweb1)