May 01 AT 5:01 PM Taylor Wimberly 81 Comments

HTC One X: Snapdragon S4 vs Tegra 3 in performance, GPU, battery life, and web browsing

s4-vs-t3-630

Does the processor found inside your mobile device really matter? The average consumer doesn’t care what CPU is featured inside their smartphone, but it’s always a popular topic among the hardcore crowd. Today’s hottest mobile chips are NVIDIA’s Tegra 3 and Qualcomm’s Snapdragon S4, so we put them through a round of benchmarks and experiences to see which came out on top. The results might surprise you, so read on for the full details.

Devices

Finally, we have both the Snapdragon S4 and Tegra 3 version of the HTC One X. We wanted to keep things simple, so we only used three devices. I decided to go with the Samsung Galaxy Nexus as the third device since it is currently Google’s flagship Android device.

Note: Our international HTC One X just received the 1.29 firmware update as we were working on this post, so we will update the benchmarks scores with new results if we find anything differs.

Benchmarks Used

For this round of testing, we used most of the popular Android benchmarks that are freely available in the Google Play Store. We decided to exclude Vellamo Mobile Web Benchmark since it is developed by Qualcomm. Linpack was also excluded because it’s a benchmark intended for use on supercomputers and it doesn’t produce consistent results on Android. Finally, for all browser benchmarks we used the latest Chrome Beta, since each device has a different default browser.

Processing Benchmarks

Winner – Tegra 3: Not many apps are optimized for quad-core, but several benchmarks are. Tegra 3 manages to score a healthy lead in Antutu, CFBench, and Smartbench. In Quadrant we see Snapdragon S4 eke out a close victory, but the results are within the margin of error.

Graphics Benchmarks

Winner – Tegra 3: Both GPUs appear to be pretty evenly matched, but we give the slight advantage to Tegra 3. In Antutu and GLBenchmark Pro Offscreen, we see Tegra 3 and Snapdragon S4 are neck-and-neck. Antutu does not appear to be a very intensive graphics benchmark and the Pro test is the less demanding test on GLBenchmark. On the most demanding graphics test, GLBenchmark Egypt Offscreen, Tegra 3 holds a 10fps advantage over Snapdragon S4.

Browsing Benchmarks

Winner – Tegra 3: On browsing benchmarks, Tegra 3 wins every round. It’s interesting to note that the aging Galaxy Nexus wins both the Sunspider and Moonbat tests which are single-threaded. This could be from code optimizations that are targeted at the Galaxy Nexus. We can see the four CPU cores in Tegra 3 really shine in the Moonbat test with web workers set to 4 threads. This should indicate that Tegra 3 should perform faster on Chrome when multiple tabs are open.

Battery Life Benchmarks

Winner – Too close to call: Battery benchmarks are a tricky thing. There are not many standard tests out there (that we could get to complete on both devices), so we used MX Player and put a video (recorded by a One X) on an endless loop. Both devices lasted just over 6 hours, with the Tegra 3 version lasting an additional 9 minutes.

I found it interesting that the Tegra 3 version saw the battery warnings at 14 percent and 9 percent just a minute earlier than Snapdragon S4. However, when it came to the last 3 percent warning, Snapdragon appeared first and then died first. It looks like Tegra 3 has some extra battery management tricks that turn down the 5th low-power core when the battery is on its final charge.

Both of these devices were just released, so I expect HTC will work with NVIDIA and Qualcomm to continue to tweak power management settings and extend battery life.

Update: NVIDIA informs us that MX Player does not use their companion core very effectively, and ends up waking up the main cores periodically during playback. Battery life for video playback should be extended using the native gallery app.

Update 2: Head over to Anandtech for more extensive battery life testing. Brian Klug found that the Snapdragon S4 has longer battery life when it comes to 3G browsing and WiFi browsing, while Tegra 3 wins at WiFi hotspot and cellular talk time. Again, it looks like the results will vary with different use cases and are too close to call.

Marketing Benchmarks

Winner – Tegra 3: This one is pretty simple. When it comes to marketing speak, four is better than two.

Gaming Experience

Winner – Tegra 3: On most games, you can’t tell the different between Tegra 3 and Snapdragon S4. However, NVIDIA has their TegraZone program which has resulted in several exclusive titles for Tegra 3. Thanks to these extra games, we think Tegra 3 provides the better gaming experience right now. Qualcomm also has their Snapdragon Gamecommand, but it doesn’t feature any exclusive games for Snapdragon S4 yet.

Keep in mind this could change in the future if Qualcomm starts writing fat checks to Gameloft and other devs for exclusive titles. Qualcomm is a much larger company than NVIDIA and they could outspend them on game exclusives if they choose to.

Browsing Experience

Winner – Too close to call: When I was on WiFi, I could not tell a difference between the two devices in day-to-day browsing usage. This could change in future versions of Chrome or with newer firmware updates, but for now the two devices are equal.

Mobile Networks Speeds

Winner – Snapdragon S4: Not much to say here. LTE is faster than HSPA+. Snapdragon S4 has an integrated LTE modem, while the Tegra 3 version is currently paired with a HSPA+ modem. This will always depend on your location and carrier, but on AT&T the Snapdragon S4 version is much faster. I achieved download speeds of 50 Mbps on Snapdragon S4 with LTE, and around 10 Mbps on Tegra 3 with HSPA+.

Drive-by Conclusions

Tegra 3 offers more exclusive games, right now: Thanks to NVIDIA’s relationships with game developers, they currently offer a better overall gaming experience since they have more exclusive titles. This could change as Qualcomm steps up their efforts.

Snapdragon S4 has a faster modem, right now: Qualcomm has the only mobile chip with an integrated LTE modem, which gives them an advantage in network speeds. NVIDIA says their partners will have LTE modems out by the second half of 2012, so this advantage will be moot in a couple months.

Both chips are awesome and you can’t really tell a difference in normal usage: Even though Tegra 3 wins a majority of our benchmarks today, I can’t really tell a difference between the two devices. If you did a blind test between the two versions, I doubt you would be able to tell which is the quad-core Tegra 3 or the dual-core Snapdragon S4. NVIDIA has an advantage in exclusive games and Qualcomm has an advantage in LTE modems, but both of those arguments are going to be obsolete in the next couple months.

At the end of the day, it’s the consumer who wins this round of testing. Both Tegra 3 and Snapdragon S4 offer a better overall experience than the current generation of mobile processors and it’s significant. After using the HTC One X, I do not want to return to my Samsung Galaxy Nexus.

Taylor is the founder of Android and Me. He resides in Dallas and carries the Samsung Galaxy S 4 and HTC One as his daily devices. Ask him a question on Twitter or Google+ and he is likely to respond. | Ethics statement

    Most Tweeted This Week

  • MrMrMan

    I guess the S4 isn’t equivalent to the Tegra 3.

    • Fulaman

      Yep I guess not either

    • Lucian Armasu

      I think Antutu and CFBench are really flawed benchmark tools. They seem to give Tegra 3 extra score in a very liniar way. I think they give quad cores 100% performance over dual cores at same clock, and that’s just wrong.

      Smartbench seems a little better than these 2, but it also seems to give quad cores 50% extra performance in a liniar way. That average extra performance may or may not be real, but at least they are getting closer to reality.

      All the other benchmark tools are not so bad, and I’d put more trust in them than in these 3.

      • Lucian Armasu

        Oh, and if you want my conclusion, I think S4 is a little faster CPU-wise for most apps (for some Tegra 3 will be faster), a little weaker GPU-wise in pretty much everything, and significantly better battery-wise than Tegra 3.

        I would go with the S4 version if I had to pick between these 2. But I’d rather wait for a dual core 2 Ghz Cortex A15 chip.

        • Kevin

          Totally agree with the benchmark part
          and to be truthful, it is only benchmark

          both very good chip though, can’t go “wrong” with any of the 2
          but don’t wait too long
          because by the time dual core 2Ghz Cortex A15 chip is out, they will be talking quad core 2.5Ghz on A15 again. or the next generation of Cortex

        • Thes

          DERP!
          thats because tegra 3 is a quadcore
          and applications and games (including benchmarks) that can use the extra
          2 cores, should. dont be so jaded.

          if it supports 4 threads, tegra 3 is gunna melt the s4, nuff said.

          Its the same reason why i5 is better then i3. and an i7 is top of the line cpu.

      • Arjun

        Cortex A9 processor is 40% less efficient or slower than a Cortex A15 (which is the S4) as said on ARM website. However if A9 is running at 100% and A15 will be running at 140%, put 4 A9 together and now its 400% efficient at dealing with tasks while the S4 in a dual core spec is 280%.

        Its not that applications are not optimized for dual cores but rather the other way around in that they have not been optimized for Quad cores. So when you see games and applications optimized for Quad cores is when the real difference is more than apparent. Hence Tegra 3 optimized games are good and no amount of money from Qualcomm can change the performance of their dual core S4 hardware to beat a Quad core.

  • Fulaman

    Taylor bro, I really enjoy your posts, that said I hope that when the Galaxy S III is available, you will run these tests again in a comparison between HTC One X (Tegra 3), HTC One X (Snapdragon S4) and the Galaxy S III. I can’t wait.

  • Fulaman

    What I really want to see now is Exynos 4412 vs. Tegra 3.

    • kazahani

      I want to see the 5250 chip in comparison to what we have available now.

      • Fulaman

        That would be a dream come true, but I don’t think we will see it anytime soon sadly.

  • Mike Leahy

    Thanks for the stats work. I am real interested to hear if the US variant of the Galaxy S3 will be using the S4; we’ll know soon enough. I’ve got a friend still on the HTC Hero that is desperately in need of an upgrade despite CM7.2 + whatever tweaks I managed to setup. If S3 goes S4 One X fills friends pocket pronto. He gave me an ultimatum of no shiny new Android phone by his b-day in June an iPhone it will be… doh…

    • Fulaman

      Going through these tests again, and I just feel like laughing at the S4. Isn’t it supposed to be based on Cortex A15 architecture….. It is only marginally better than a dual-core Cortex A9 (TIOMAP 4460).

      • Bpear96

        Its Qualcomms custom Krait architecture, which isnt really Cortex A9 or A15, its Custom designed by Qualcomm and uses features from A9 and A15 along with custom Qualcomm changes and features. Im sure that “real” A15 will outperform krait.

      • Dags -

        It’s all about which benchmarks you look at. If you read the new Anandtech review, the AT&T One X beats the International One X in almost every benchmark. That’s impressive when you are comparing two cores to four. This review tells a different story of course.

        • http://androidandme.com Taylor Wimberly

          I read Bryan’s review and it looks like he got the same scores as I did on all the tests we both did. He just chose to run some extra benchmarks (Vellamo and Linpack) and he also did much more extensive battery testing.

        • http://androidandme.com Taylor Wimberly

          Another quote from the Anand review, “AT&T One X can take a back seat to the international One X in GPU performance. When not limited by v-sync, the Tegra 3 based international One X is about 12% faster than the Qualcomm S4 based AT&T One X. For most gaming however, you’ll be hard pressed to notice a performance difference.”

          Kind of echos what I said. One might be slightly more faster, but you can’t tell a difference in normal usage.

      • kazahani

        Exynos 5250 should be the first A15 chip to market, unless the iphone 5 uses A15. Either way that’s still a couple of months off.

  • ericl5112

    One thing that was not tested in this review is single threaded performance. This is actually an incredibly important spec, considering how things are in the app space. In fact, androids entire UI is only single threaded. That’s probably why the S4 feels as good as the T3 (and might even hold up longer before stuttering).

    That said, I am surprised to see how badly trounced it was on a lot of other things. Doesn’t make the decision of unlocked v AT&T any easier :/

    • Fulaman

      The Sun-spider and moonbat tests are Single-threaded reread the review.

  • Doc

    “In Quadrant we see Snapdragon S4 eek out a close victory,” You mean “eke.”

  • Ironzey Lewis

    Why oh why did you have to include the Galaxy Nexus in some of the tests?
    I was all set to get one of these ($400, at the play store) and now all these small bars have kind of soured my impression of the phone. I wonder how long the bars will be with my Sensation?

    • http://facebook.com/jestertx2001 Jesse Moreno

      Was thinking the same thing! Terribly excited to get my Nexus today.

      And then this.

      Mer. :/

      • jonathan rung

        don’t let the benchmarks get you down, the 4460 is still a champ. The newer chipsets are stronger, but that’s just the nature of the beast, tech just keeps getting better – in fact, even the tegra 3 and s4 will likely lose their position at the top come May 3rd!

        Some reasons for you to be happy: the 4460 is incredibly faster than the 4430, which powers moto’s high-end phones like the razr AND many of their tablets. The powerVR GPU is not the newest kid on the block, but it’s incredibly popular, and so many apps are coded specifically for it. Last but not least, the galaxy nexus is super popular with the dev community, so you’ll have great custom firmware for years to come =) And I don’t have a galaxy nexus, so I’m not just being biased.

        Can you imagine if we saw these benchmarks and chipset specs two years ago? Nobody would ever believe you; you’d get laughed out of the room as soon as you mention quad core phones. I mean, it was only a little over a year ago that some people were calling the dual core Atrix a “gimmick.” It’s an exciting time to be a cell phone nerd.

        • Josh

          if the omap 4460 is so much faster than the 4430 than why does my droid 4 get a much higher quadurant score than the galaxy nexus

      • Raptor

        Don’t open the POS box and return it back same day. Even single-core Indel Medfield smash this crap inflated by morons:

        http://www.anandtech.com/show/5770/lava-xolo-x900-review-the-first-intel-medfield-phone/4

    • kazahani

      Don’t freak out dude. The GNex has been out for almost 6 months, and these tests are puttin it up against cutting edge tech.

      Unless you have the ability to get an HTC One X, I would still go with the Nexus in a hearbeat.

      Just got mine today, actually.

  • Dray

    Your benchmarks are different than that of other benchmarks being posted. For example, a bit more reputable of a site, Engadget, hosted benchmarks. They included Vellamo, but left out antutu, so it’s “even”.

    Scroll down to find it.

    • nerdshowandtell

      “Reputable”? Popular Maybe..

    • Bpear96

      I would consider this site much more “reputable” then engadget! and trust androidandme much more !

    • aranea

      I like engadget but they are apple prone at least some of the time.

    • dVyper

      In NO WAY is engadget more reputable.

  • mario

    I guess we also for get, that the s4 has better battery life and is almost as good as five cores

    • Bpear96

      9 minutes longer.. and that was running a video on a app that doesnt take advantage of the 5th companion core properly. Id say that the battery life would be about on par on both, or maybe slightly better tegra 3 thanks to the ninja core.

      • kazahani

        Ninja core? Is that what we’re calling it?

        Well that decides it. If Tegra3 has a ninja core then it wins hands down.

        Ninjas > dragons.

        • Bpear96

          Haha yeah ive seen it been called the ninja core countless times :D. But anyways I have seen many charts comparing the international one x, at&t one x and other phones batterylife, and the international one x beat the at&t in all tests besides browsing on 3G

  • CR23

    Anandtech has there att review up the S4 CPU is a beast

  • http://www.jeffkibuule.com Jeff

    The chances of seeing a Tegra 3 smartphone in the US is pretty slim though.

    • kazahani

      Agreed. At least not until Tegra3+ arrives.

  • Nathan D.

    Nice, so it really doesn’t matter which one you get unless your a gamer, in which you would want Tegra 3 version instead. Nice to let us know keep up the good work.

    • jonathan rung

      yeah I was happy to see that the initial benchmarks make them look neck-and-neck. This just means that nvidia and qualcomm (and all the chipset manufacturers, for that matter) will have to keep innovating and pushing the already insane boundaries even further. consumers FTW! now if I could just get some money…

  • aranea

    Not many surprises here. I still can’t believe that ATT went for S4 for LTE that covers only a small portion of the customers. The main problem is that they are going to sell it in regions without LTE and when customers don’t get the faster speeds they are going to blame Android and the phone.

  • swazedahustla

    One thing I find very odd…..android central did the same test and yielded much different results……hmmmmm something smells fishy.

    • swazedahustla
      • http://androidandme.com Taylor Wimberly

        Read Phil’s review and he didn’t do any benchmarks. He says the S4 has better battery life, but he doesn’t provide any real tests or data as supporting evidence.

        • swazedahustla

          I put 2 links…..one with a review where Phil provided the full details of his phone usage both with LTE on and off and he said battery was insane, and the benchmarks were done in the other link I posted.

          There is the same test done on another popular blog site and their results differ from yours as well. I’m just sayin. Something is fishy lol

        • Matt Yearian

          Your graphs/numbers are not evidence or even supporting evidence. The claim in the article has as much credit to us as anything you could possibly post. The problem for your site lies in the fact that every other site I’ve visited says contrary to what your article is claiming. So sorry if we are a little bit skeptical of your charts and data.

  • monkey mouse

    tests or data may not mean everything, but if insisted, i prefer to use Antutu instead Quadrant.

  • SGB101

    I’d trade my 4 cores for your LTE all day long.

    well I would if we had it here in the UK, that is.

  • duc916

    For those who want to know what the Linpack single-threaded result looks like (the one benchmark that’s more in line with 99% of what you do on your phone) from a non-apologist, Anandtech shows that the T3 performed even slower than a Galaxy S2.

  • Derek

    I would go with the S4. Clearly its a much advanced architecture. If its TWO cores at 1.5GHz can keep up with NVidia’s FOUR cores, clearly Snapdragon Krait (basically A15) is much more advanced than Tegra3′s A9. Also, its built on a 28nm process. If there was a quad core S4 it would blow the doors off of Tegra3. Here again, same as with Tegra2, NVidia is first to market, but less technically sophisticated. And the way they screwed over devs with not releasing driver source code?? I’ll never buy another NVidia…EVER!

    • swazedahustla

      Good point about the driver source code. I forgot about that issue.

    • Guest

      So if Wayne comes out with 4+1 (Tegra4) @ 28nm, it’ll beat the pants of the 4 Krait cores…

  • felisberto

    I miss Richard.

  • h0ruza

    The cynic in me thinks that we may not see a true international chipsets until europe has LTE in full swing

  • yankeesusa

    I hate to see how much tegra 3 is better than s4 especially since the htc evo 4g lte is s4. I hope that by december htc has announced an lte phone that beats the evo lte since that is when sprint is forcing me to upgrade as part of the premier upgrade that have left.

  • Jimmy_Jo

    I’m sorry but I can’t put faith in benchmarks. I’ve seen Phil’s review. How can I believe that the Tegra 3 blows away the S4 when it can’t even open the App Drawer quicker??? Benchmarks and real life are 2 different things. If apps aren’t coded to make the best use of 4 cores then why should I have them instead of a highly efficient dual core? I’m looking forward to seeing the S4 live and in person.

    By the way, the One XL plays Tegra 3 games just fine….

    • eeepad

      Some people may not understand this….
      Also the same problem appears in GS3(EXYNOS 4412 SoC),too.

  • Matt Yearian

    This is odd, everywhere else I go shows things the other way around. Isn’t there a standardized benchmark yet??? Surprised Futuremark hasn’t jumped onboard the smartphone/tablet wagon yet and released something.

  • Alex Belko

    So that may seem more cores = more power, I guess we need very good optimized benchmarks to reveal the true difference, but I always thought coming to 28nm is significant improvement

  • Dave

    concerning the battery performance there are many people who are unsatisfied and unhappy that battery is not removable, there is even a huge poll on facebook regarding that matter..
    https://www.facebook.com/questions/10150675611462172/
    general opinion doesnt find it user friendly, what if the battery is faulty? what if the battery dies after half a year of use?

  • evo3d

    28nm! Say hello to overclocking!

  • evo3d

    Ok so my EVO 3d has snapdragons 45nm chip clocked at 1.5ghz. I’m currently running gb 2.3 build 2.17.with Wimax/wifi network on(no Bluetooth or gps) I get a day of battery life,bearly partially because I am a heavy user(browsing,gaming,camera,social web sites,netflix)note that I get the same use with iPhone 4s ,both phones last equally a day of use for me followed by a charge.the up4s gets much better battery then the evo4g successor although i find it equaling out to my rooted evo3d. However,an htc with qualcomms 28nm core clock at the same 1.5 ghz and with 1 model higher gpu also of a smaller “nm” size should get much much better battery life once the android build and kernels start to get proper. My evo3d was horrible and I mean horrible battery life when it first came out, but development has come a long way even fixing battery life and made me a happy user and that’s a perk of owning an android.my two TRUE cents

  • ndarvishev

    Looking at the benchmark scores I am opting for Tegra 3 powered One X. But the other thing is that One S with S4 was beating One X with Tegra 3 processor. So, at some point it might be the software, the version of OS and kernel that plays some role in phones’ performance.
    Anyways, I’d still go for Tegra 3 powered One X.

    • tegrafreak

      The tegra 3 has good processing power.whichis expected at its size and multi core.the s4 has good potential,much better then the tegra 3,although it isn’t up to par for a 28nm chip. Once the proper kernels and software builds come along 28nm wn’t be outdated for a while.not to mention better battery life then the tegra 3 and much much more performance then the tegra 3 (2core vs 2core)

  • lorne

    Thanks to these extra games, we think Tegra 3 provides the better gaming experience right now. Qualcomm also has their Snapdragon Gamecommand, but it doesn’t feature any exclusive games for Snapdragon S4 yet.

    I think this could have been better worded. The Tegra 3 many not give any better or worse game play. NVIDIA gives worse game play and graphics to non Tegra 3 chips.

  • Fox

    Well, benchmark may be optimised for 4 cores, so Tegra 3 has biger potencial. But what about real worls where all curernt and near future apps are and will be 1 or 2 core? Not even on PC we have many titles with support for more than 2 cores and we have multi-core systems for nearly 10 years as mainstream. And applications with multi core support are usually scientific and no sane person would run them on low IOPS mobile device.

  • Kartik Mutya

    Very helpfully thanks a ton.

  • 94-Echo

    Honestly the benchmarks most likely don’t really support quad cores. The Tegra processors are made for high definition gameplay. The only reason to own a Tegra device is for running GPU intensive applications. Tegra is a beast of a mobile chip and I love my Tegra devices. Custom ROMs and root plus overclocking kinda trumps the single thread speed problem for Tegra. I think Tegra is kind of ignored by many Android users because they don’t really promote how powerful it can be. But again its really for gamers like myself.

    Tegra devices I own: Microsoft ZuneHD (first Tegra chip), Dell Streak 7 Tegra 2 (overclocked 1.8ghz via custom ROM of honeycomb by DJ-Steve), Asus Transformer Tablet Tegra 3

    As a gamer I only go for Tegra devices. They provide a much sweeter game experience.well except for the Zune which only uses it for HD video output. Mostly I have that because I f***ing hate slave driver cApple and will never use an iPood Touch

  • Amier

    BULLSHIT THE ONE WITH SNAPDRAGON I BET HAS THE MOST APPS!!

  • teck

    that comparison is silly quad core tegra and
    Snapdragon dual core

  • Jason Matsoukas

    1 year later, I can’t figure out who found these benchmark results?

    The S4 is a beast compared to the crappy Tegra 3.

  • http://www.focuszonedevelopment.com Homncruse

    Um, he did…

    “It’s interesting to note that the aging Galaxy Nexus wins both the Sunspider and Moonbat tests which are single-threaded. *****This could be from code optimizations that are targeted at the Galaxy Nexus.*****”

  • john

    People actually play those games… Thus there is a difference.

  • Ichigo

    They are just benchmarks relax

  • Kevin

    I think the difference is Nvidia actually go work with game designer to improve the experience when people use Tegra 3 chips
    and a benchmark test simply shows what the creator of that benchmark want to show

    not a lot of company out there would want to show that their product is any lesser than others

  • eeepad

    In fact, almost no applications(95%+) have optimised for more than two cores until now(ANDROID 4.0)….

    ps. Too many Nv fans here,and always vote down the opponents…..