The Nexus 7 (2013) Review
by Anand Lal Shimpi on August 22, 2013 6:00 PM ESTCPU Performance
The while the original Nexus 7 was fast for the money, the new Nexus 7 is just fast. Moving away from NVIDIA to Qualcomm, ASUS and Google settled on the APQ8064 Pro. Although we originally assumed this would be a quad-core Krait 200 based SoC, Brian's teardown revealed the part number 8064-1AA. What's special about that part number is it implies newer Krait 300 cores, making the SoC effectively a Snapdragon 600.
The CPU cores can run at up to 1.5GHz, putting it a bit lower than what we've seen from Snapdragon 600 based phones (e.g. HTC One tops out at 1.7GHz, while the US SGS4 hits 1.9GHz).
I was curious as to the impact of the lower frequency when combined with the potentially higher chassis TDP so I compared the Nexus 7 to the US Galaxy S 4. I turned to Geekbench because it offers a nicely woven mixture of single and multithreaded benchmarks, letting me look at peak available single core performance as well as what happens when multiple cores are active and working.
In this first graph we're looking at the first Krait 300 core running the workload. I've zoomed in to a small portion of the benchmark so we can get a better idea of CPU behavior:
The Nexus 7's CPU0 is almost always pegged at 1.5GHz whereas we see a lot of bursty thermal management on the smaller SGS4. This isn't unexpected, but what ends up happening is the sustained performance advantage drops from a peak theoretical max of 26% down to a more reasonable average gain of ~8%.
I was curious to see if multithreaded workloads showed any different behavior. Here we're looking at the fourth CPU core (CPU3). Note that it's not always active, which is why you see parts of the graph drop down to 0KHz. Once again we see similarly static behavior from the Nexus 7. Even with all four cores active, when you need the performance the Nexus 7 delivers a full 1.5GHz. These tests are short enough where we don't see tremendous swings in frequency, but once again we do see some frequency modulation in a smaller chassis.
The end result is that the Nexus 7's 1.5GHz Krait 300 cores are slower than what you'd get in a Galaxy S 4, however the gap isn't nearly as large as you'd expect it to be thanks to the larger chassis and how the platform is tuned. The Nexus 7 can run at 1.5GHz more consistently than Snapdragon 600 based phones can run at 1.7/1.9GHz.
The GPU
Qualcomm seems to be frequency binning here, which is standard industry practice. Using binning to create these different SKUs gives Qualcomm pricing flexibility and also gives it the option to give large customers special treatment.
On the graphics side Qualcomm's Adreno 320 GPU makes an appearance. Clock speeds are also limited to 400MHz, compared to 450MHz for the high-end Snapdragon 600 implementations we've seen.
The APQ8064-1AA's dual-channel memory controller is populated with four x 16-bit DDR3L-1600 memory devices, giving the Nexus 7 a healthy amount of memory bandwidth. Peak memory bandwidth available to the APQ8064-1AA is 12.8GB/s, equalling what's available to Apple's A5X in the 3rd generation iPad with Retina display. Tons of memory bandwidth is obviously a pre-requisite for driving a high resolution display, and the combination of DDR3L-1600 and the Adreno 320 GPU delivers a butter smooth UI in all well written Android apps.
Although CPU performance is somewhat middle of the road compared to the rest of the landscape, GPU performance is faster than any other Nexus device on the market - and pretty much faster than any other similarly sized tablet:
I played Modern Combat 4 as well as Shadowgun, both demanding 3D titles, on the new Nexus 7. Both titles appeared to render at the Nexus 7's native 1920 x 1200 resolution, and both appeared to do so at around 30 fps.
Facebook
Twitter
RSS
202 Comments
View All Comments
solipsism - Friday, August 23, 2013 - link
1) I'd rather take 30 seconds to enable a hotspot on my smartphone for the occasional use then pay excessive fees every month paying for a service I will rarely ever use, not to mention the HW and licensing costs associated for cellular connectivity.2) A cable? Have you ever seen an iPhone connected to an iPad to tether it? It's done via BT or WiFi. Now what happens if you have a half dozen devices using your 42Mbps cellular connection on your phone as a hotspot? Do you think they all get a constant 54Mbps connection with 802.11g? If so, you still don't WiFi works.
ShieTar - Friday, August 23, 2013 - link
There no cable, it is done via WiFi. And yes it "requires" me to carry my phone around, but so far I have not been in a situation where I left home with the tablet but without the phone.And its not like I only use this solution for the tablet, I also use the phone as mobile hotspot for my Kindle and my Laptop. And when I am on a business-travel to a foreign country with a few colleagues (which happens often enough if you work for a European company), we usually only get a single prepaid card, let one phone use it to connect to the net, and act as the mobile hotspot for everybody else.
max1001 - Friday, August 23, 2013 - link
It takes 2 secs to turn on hotspot on my nexus 4. Just unlock the phone and tap on the hotspot widget and you are done. The tablet automatically connects to it within 5-10 secs.akdj - Sunday, August 25, 2013 - link
Yep...Easy Peezy! Then, about 54 minutes later, your Nex 4's battery life is history. It's ubiquitous. Doesn't matter which OS or phone you're tethering to...tethering EATS battery life. Anand often shows battery life while tethering. It's often less than half that of actually using the phone itself to surf, check email or update your Facebook status. If both devices decide to sleep....bummer! Time to turn em both back on, reconnect and you're back up n running. Not sure what other carriers charge, but AT&T is $10 a month to add to your bucket data plan. IMHO, that up front hardware charge is a bargain when it comes down to connectivity and efficiency. I don't think I'm alone in this thought process and the last wifi only tablet I bought, I returned and 'upgraded' to the LTE version. The extra CNote was quickly forgottenESC2000 - Tuesday, September 10, 2013 - link
Tethering stays on for me even if the devices go into sleep mode. I actually need to be careful to ensure that I remember to turn tethering off because otherwise it'd just stay on allowing freeloaders to use it. I have unlimited data but still...Also I've never noticed decreased battery from tethering. I haven't paid close attention but my note II's battery has never died before the end of the day whether I tethered or not, so I'm not too worried about tethering's impact on my battery.... esp since I can easily change the battery out-_-
joos2000 - Friday, August 23, 2013 - link
I have. I got a basic "Wireless Internet" plan for my slate. It gives me more gigabytes per month than my "high usage phone plan" did. So, I've now been able to step down to a cheaper cellular plan for my phone. My cost has been status quo, but my gigabytes per month while on the go has increased.I'm in Australia though, I'm unsure how that compares to in the states.
Impulses - Saturday, August 24, 2013 - link
If you're trying to save money in the US you go with Sprint/Tmo or one of the smaller prepaid/regional carriers, and in those cases I think it's usually cheaper to pay for one of the high GB phone plans (all unlimited in Sprint's case, also available for Tmo) and run a hotspot than to go to an AT&T/VZW plan and pay extra for a tablet. I don't know if Tmo even offers a data only plan... It's all kinda backwards here, but we generally get pretty good coverage (and no roaming) across a huge swath of a continent, which people totally take for granted.max1001 - Friday, August 23, 2013 - link
.........They have an LTE version coming out very soon so how is that a major omission?
Thomaspin - Thursday, August 22, 2013 - link
I wrote a four part review based on my exepriences (my primary focus is that of a photographer and movie buff) from the perspective of someone who has never used Android before, so apologies for the relatively naïve Android content, which reflects my learning curve.My interest was in using the N7 with my existing iCloud data and so successful was the experience, and so stellar the performance of the N7, that I just sold my iPad Mini and bought a second N7 for my son.
I realize that a fairer comparison will be with the 2013 iPad, but the N7 made it unrealistic to wait, given the iCloud functionality.
My link, below, is not commercially motivated. My site carries no advertising and has no click-thorugh payment sources. Stated differently, I make exactly $zero from it. Nor do I try to compare with a dozen other tablets as, until now, my tablet world has been iPad only (iPad 1, iPad 3 and iPad Mini), and there's no way I am in competition with AT, which is my definitive source for test data.
Here it is: http://pindelski.org/Photography/2013/08/06/nexus-...
I hope it is of interest.
I am a long time AT reader, first time commentator.
max1001 - Friday, August 23, 2013 - link
If you want to be taken seriously. Keep the whole company x is evil bit out of the review. You can write the best review in the world and readers will stop taking you seriously when you start ranting.