Investigating NVIDIA's BatteryBoost with MSI GT72by Jarred Walton on October 23, 2014 9:00 AM EST
We previewed the performance of MSI's new GT72 notebook earlier this month, and while we're still running a few additional tests for the full review, one area that we wanted to look at in more detail is BatteryBoost. Initially launched with the GTX 800M series earlier this year, our first look at the technology came with the MSI GT70 with GTX 880M, and unfortunately battery life even when not gaming wasn't exactly stellar, and powering up the GTX 880M didn't help matters. NVIDIA's stated goal is to get useful gaming battery life above two hours, which so far we haven't been able to do (and starting with a laptop that only manages 4-5 hours in our light and heavy Internet testing doesn't help).
Without BatteryBoost, the MSI GT70 managed around 50 minutes of battery life while gaming (give or take), while enabling BatteryBoost in some cases could get us up to 80+ minutes of battery life. More recently, we reviewed the updated Razer Blade (2014 edition) with a GTX 870M. We were able to see an improvement from 46 minutes without BatteryBoost to 76 minutes with BatteryBoost in our limited testing. However, if the goal is to get above two hours of gaming battery life, we're still not there.
Basically, the amount of battery life you are able to get while gaming is largely dependent on how high frame rates are without BatteryBoost and how low the target frame rate is set with BatteryBoost. If a game on battery power can run at 60FPS and BatteryBoost puts a 30FPS cap into place, battery life can improve a decent amount. A game that can hit 120FPS meanwhile would potentially experience a much larger benefit from BatteryBoost, especially when putting a 30FPS cap into effect. With GT72 and the GTX 980M, both power efficiency and performance should be better than the GT70 and GTX 880M, which means BatteryBoost has the potential to stretch its legs a bit more.
For our testing, we've picked three games and we've run and reasonably high settings – but not necessarily maxed out settings, as that would generally prevent BatteryBoost from providing much if any benefit. Our goal was to run settings that would allow at least 70+ FPS on battery power. Keep in mind that just because the GT72 can hit well over 60 FPS on AC power, even without BatteryBoost enabled there are some performance limitations in effect. In the end, our three games consist of Tomb Raider at High quality, the newly released Borderlands: The Pre-Sequel at nearly maxed out settings (we left PhysX on Low), and GRID Autosport with High settings. Anti-aliasing was not used in any of the games (though FXAA was enabled in Borderlands), and the resolution was set to 1080p. The power profile was set to Balanced, with the LCD running at 200 nits.
One of the interesting things about BatteryBoost is that it allows you to target a variety of frame rates (from 30 to 60 FPS in 5 FPS intervals). NVIDIA has also stated that they're doing more than just frame rate targeting, so we wanted to test that by enabling VSYNC and running without BatteryBoost at a steady 60FPS. Since BatteryBoost also doesn't inherently enable VSYNC, that was one more variable to test. (While in theory anything between 30 and 60 FPS should result in a 30FPS frame rate with VSYNC enabled, at least in GRID Autosport that doesn't happen, either due to triple buffering or some other factors.)
In the end, for at least one game – GRID Autosport – we tested both with and without VSYNC at 10FPS intervals with BatteryBoost, plus checking performance without BatteryBoost. That's ten different settings to test, and with each cycle requiring at least several hours we've been running BatteryBoost tests almost non-stop since our preview article. This is about the most effort we've ever put into testing gaming battery life on a laptop, and it might be a while before we decide to delve into this subject in such a fashion again. So join us as we thoroughly investigate BatteryBoost on the GTX 980M.
Post Your CommentPlease log in or sign up to comment.
View All Comments
inighthawki - Friday, October 24, 2014 - linkYeah, at such high framerates, it wouldn't be uncommon to not always be at the max queue depth, so you'll get the illusion that it's always continuously rendering. But in this case you're really just rendering frames ahead. One nice advantage sis that if you hit the queue depth, you'll actually get more consistently smooth motion, since the frame rate is more consistent. Having the game wake up consistently every vblank and rendering one frame provides a more fixed timestep for things like animation, compared to having a variable rate by rendering as fast as you can. Most people will likely never notice though.
It's unfortunate that Windows forces the games into that model, since sometimes I'd love the triple buffering model instead. I like the lower latency of that mode, while also removing screen tearing. Given that I run a GTX780 plugged into my wall socket, I'm not too concerned about the power savings - especially considering I usually disable vsync anyway, so I'm not really wasting any more than normal.
HiTechObsessed - Friday, October 24, 2014 - linkIf what you're saying is true, battery life would decrease when turning on VSync... Looking at the results here, with BatteryBoost off, turning VSync on increases battery life.
thepaleobiker - Thursday, October 23, 2014 - linkYes, please read the article good sir.
limitedaccess - Thursday, October 23, 2014 - linkIs there any actual difference in terms of thermal performance? Either lower temps and/or fanspeed (fan noise)? I would assume if the GPU itself is consuming significantly less power its average heat output should be lower as well and less stress placed upon the cooling system.
As an extension of this are you able to ask Nvidia to comment on whether or not it is technically possible to extend a variation of this to desktop GPUs and if there is any plan to? This would enable the flexibility of building a system that is extremely low noise (or even passive) for certain gaming workloads yet still have performance on demand.
nevertell - Thursday, October 23, 2014 - linkAs there is less energy consumed, there is less energy dissapated. Ultimately, all energy that is used by any computer that isn't then used to power LED's or displays will be turned into heat.
limitedaccess - Thursday, October 23, 2014 - linkYes I'm aware of the theory. However I am curious as to what the actual tested impact would be in this case and how significant (or insignificant) the difference might be.
Brett Howse - Thursday, October 23, 2014 - linkWhen I tested the Razer Blade, I noticed a significant decrease in temperatures and of course noise when playing with Battery Boost enabled, which is what you would expect since it is working far less.
JarredWalton - Thursday, October 23, 2014 - linkYup. Running the GPU at lower clocks and reducing power consumed means the fans don't have to work as hard to keep the system cool. Targeting 30FPS, the GT72 is pretty quiet -- not silent, but not loud at all. I didn't take measurements (I'll try that for the final full review), but there's nothing too shocking: lower performance => less heat => less noise.
CrazyElf - Thursday, October 23, 2014 - linkAll in all, this new Battery Boost feature seems to indicate a modest incremental improvement in battery life. It's not as good as say, the leap in performance per watt that Maxwell gave, but it's welcome nonetheless.
The issue has always been that there's a tradeoff between size, mobility, and battery life, especially for a large hungry gaming GPU.
Jarred, by any chance, are you aware that there is going to be are variants of the GT72 with an IPS monitor coming out in the coming months? It's already up for pre-order at many of the laptop sellers. Downside is there's a pretty big price premium.
sonicmerlin - Thursday, October 23, 2014 - linkDon't these laptops have nvidia Optimus and Haswell processors? Why is their non gaming runtime so low despite their large batteries?