Amazon's Kindle Fire HD: Better; Can It Compete With The Nexus 7?

CPU And GPU Performance

CPU Performance

OMAP 44xx
Swipe to scroll horizontally
Hardware Comparison
Row 0 - Cell 0 SoCCPURAMGPU
Kindle Fire (First-Gen)OMAP 44301.0 GHz Dual-Core Cortex-A9512 MBPowerVR SGX540 @ 304 MHz
Kindle Fire (Second-Gen)OMAP 44301.2 GHz Dual-Core Cortex-A91 GBPowerVR SGX540 @ 304 MHz
Kindle Fire HDOMAP 44601.2 GHz Dual-Core Cortex-A91 GBPowerVR SGX540 @ 384 MHz
Nexus 7Tegra 3 (T30L)1.3 GHz Quad-Core Cortex-A91 GBULP GeForce

The OMAP 44x0's dual-core ARM Cortex-A9 processor now operates at 1.2 GHz, but it still falls between 20 and 30% behind the quad-core Tegra 3 at 1.3 GHz in our integer and floating-point benchmarks. Although its new tablets sport a faster SoC, Amazon continues to trail when it comes to performance. When competing tablets based on Qualcomm's S4 Pro emerge, the Kindle Fire and Kindle Fire HD will fall even further behind.

Swipe to scroll horizontally
GeekBench v2OverallIntegerFPUMemory
A5/A5X (Dual-Core Cortex-A9, 1.0 GHz)(iPad 2/iPad 3)764691921830
OMAP 4430 (Dual-Core Cortex-A9, 1.0 GHz)(Kindle Fire, First-Gen)8275911139974
OMAP 4430 (Dual-Core Cortex-A9, 1.2 GHz)(Kindle Fire, Second-Gen)108588115711001
OMAP 4460 (Dual-Core Cortex-A9, 1.2 GHz)(Amazon Kindle Fire HD)111390015401098
Tegra 3, T30L (Quad-Core Cortex-A9, 1.3 GHz)Google Nexus 71527129822881222
S4 Pro (Quad-Core Krait, 1.5 GHz)Qualcomm Dev Platform1960140032921276

GPU Performance

PowerVR SGX540

Both the OMAP 4430 and 4460 employ Imagination Technologies' PowerVR SGX 540. If this graphics engine sounds familiar to you, that might be because it's derived from the same architecture as the GPUs in Apple's A4 and A5.

Swipe to scroll horizontally
GPU Subsystem
Row 0 - Cell 0 PowerVR SGX 535 (Apple A4)PowerVR SGX 540 (OMAP 4430)PowerVR SGX 543 (Apple A5)
SIMDUSSEUSSEUSSE2
Pipelines244
TMUs222
Bus Width (in bits)646464
Triangle rate @ 200 MHz14 MTriangles/s28 MTriangles/s35 MTriangles/s

The SGX 543 used in the Apple A5 includes four USSE2 (Universal Scalable Shader Engine 2.0) pipes. In comparison, the SGX 540 found in Amazon's new tablets features the same number of pipes based on the older USSE design. The SGX 535 used in Apple's A4 hails from the same GPU generation as the SGX 540, but features only two USSE pipes.

Swipe to scroll horizontally
GLBenchmark 2.1.2
Row 0 - Cell 0 Egypt StandardPro StandardEgypt Offscreen (720p)Pro Offscreen (720p)Fill Rate
PowerVR SGX543MP2 (iPad 2)6661 frames(59 FPS)2962 frames(59 FPS)10 146 frames(90 FPS)7352 frames(147 FPS)998.24Mtexels/sec
PowerVR SGX543MP4 (iPad 3)6709 frames(59 FPS)2975 frames(60 FPS)15 663 frames(139 FPS)12 546 frames(251 FPS)1964.68Mtexels/sec
PowerVR SGX540 (Kindle Fire, First-Gen)2966 frames(26 FPS)1952 frames(39 FPS)2632 frames(23 FPS)2079 frames(42 FPS)234.3Mtexels/sec
PowerVR SGX540 (Kindle Fire, Second-Gen)3492 frames(31 FPS2399 frames(48 FPS)3296 frames(29 FPS)2478 frames(50 FPS)226.1Mtexels/sec
PowerVR SGX540 (Kindle Fire HD)2835 frames(25 FPS)2073 frames(41 FPS)3709 frames(33 FPS)2656 frames(53 FPS)225.01Mtexels/sec
Tegra 3 (Nexus 7)5968 frames(53 FPS)2830 frames(57 FPS)7073 frames(63 FPS)4095 frames(82 FPS)467.57Mtexels/sec
Adreno 320 (S4 Pro MDP)--15447(137 FPS)9560 frames(191 FPS)795.62Mtexels/sec

If you've already read Snapdragon S4 Pro: Krait And Adreno 320, Benchmarked, then you know Qualcomm's S4 Pro has a performance advantage thanks to its Krait processor cores. It doesn't have the lead in graphics, though. Rather, the S4 Pro normalized to 720p edges-out Nvidia's Tegra 3 and comes up just short of the PowerVR SGX543MP4 in Apple's A5X.

Of course, it's interesting to compare graphics engines rendering at the same resolution for evaluation purposes. But, in the real-world, the devices you find each SoC in employ different resolutions. Amazon's second-gen Kindle Fire outperforms its predecessor, but the Kindle Fire HD is actually the slowest of the three. It does benefit from a slightly faster GPU, but is then hampered by a higher resolution.

The second-generation Kindle Fire outperforms its predecessor, but the Kindle Fire HD is actually the slowest of the three. It benefits from a slightly higher GPU clock speed, but it is hampered by its higher resolution. Compared to the Kindles, Google's Tegra 3-equipped Nexus 7 dominates, even though it uses the same resolution as the Fire HD.

Swipe to scroll horizontally
GLBenchmark 2.5
Row 0 - Cell 0 Egypt HDEgypt HD OffscreenFixed Time (1080p)GLBenchmark Egypt HD Offscreen (1080p)Fill RateOffscreen
PowerVR SGX543MP2 (iPad 2)2446 frames(22 FPS)102.7 s (11 FPS)1507 frames (13 FPS)938.6Mtexels/sec
PowerVR SGX543MP4 (iPad 3)2363 frames(21 FPS)57.4 s (20 FPS)2731 frames (24 FPS)1772.8Mtexels/sec
PowerVR SGX540 (Kindle Fire, First-Gen)824 frames(7.3 FPS)275.4 s (4.1 FPS)532 frames (4.7 FPS)289.3Mtexels/sec
PowerVR SGX540 (Kindle Fire, Second-Gen)960 frames(8.5 FPS)267.3 s (4.2 FPS)566 frames (5.0 FPS)297.8Mtexels/sec
PowerVR SGX540 (Kindle Fire HD)919 frames(8.1 FPS)271.3 s (5.2 FPS)691 frames (6.1 FPS)284.0Mtexel/s
Tegra 3 (Nexus 7)1464 frames(13 FPS)148.2 s (7.6 FPS)995 frames (8.8 FPS)490.3Mtexels/sec
Adreno 320 (S4 Pro MDP)-54.1 s (21 FPS)2927 frames (26 FPS)530.1Mtexels/sec

Going back to normalized testing, forcing each solution to run at 1920x1080 changes the story. Now, Qualcomm's Adreno 320 wins by a small margin over the SGX543MP4, even though it can't compete with the PowerVR architecture's fill rate.

GLBenchmark 2.5 improves on the prior version in a number of ways. First, the benchmark focuses exclusively on the Egypt scene. Adding higher-quality textures makes it a more taxing workload, and cranking up the intensity hurts the A5X.

  • joepaiii
    Why not normalize to a constant brightness level on all tablets rather than max for the battery rundown tests? Since they all have different max brightness, your tests aren't that reliable for judging true battery life.
    Reply
  • Another great article, as usual, from this site. However this statement in inaccurate:

    "Buying a Nexus 7 locks you into Google's Play store and its movies, newspapers, magazines, and music."

    I have BOTH the Kinda Fire app and Barnes and Noble app installed on my Nexus 7. And that is one of the things I love most about Google's tablet: the ability to get content from any provider I want. I would also like to point out that my gf has the B&N tablet, and it is much better than the Kindle Fire for several reasons: 1) you can sideload your own content through a removable memory card, 2) it has a better screen, and 3) the build quality seems much better after holding both in your hand. It's a shame the Kindle get's more attention, the power of advertising I guess....
    Reply
  • nitrium
    Given these devices are also intended to be used as eBooks, why won't anyone test battery life reading an actual eBook - i.e. Wifi/Bluetooth off, screen on (probably at ~50% brightness), no background tasks, just the occasional page flip every 30 secs or whatever the average reading rate is. Is that too much to ask? I tried to find this info with Google, but apparently no one but me cares.
    Reply
  • freggo
    What turns be away from Amazon and on to the Nexus is, among other things, the multiple reports I ran into about Amazon messing with the content of the device.
    I don't care for a manufacturer remotely deleting things without my permission.
    Reply
  • mrmike_49
    no mention of printing - is it possible to print from ANY of these tablets to a net worked printer???
    Reply
  • acku
    joepaiiiWhy not normalize to a constant brightness level on all tablets rather than max for the battery rundown tests? Since they all have different max brightness, your tests aren't that reliable for judging true battery life.
    There was an interesting case study a while back by AMD that pointed out most people use their devices at maximum brightness, and we've always strove to lean more toward real-world conditions. Having said that, a while back, we also started to standardize our battery life tests to a fixed brightness setting. While not real-world, it does provide a window into how devices compare to one another in that specific scenario. Bear in mind that a fixed brightness on one display may look different on another because of gamut differences. Often times you'll push the brightness up on a low gamut LCD to improve readability or visibility. The article has been updated to include those results. We always intended to do so but this understandably doubles our benchmarking workload and battery life tests take a lot of time. Thanks for being patient.
    Reply
  • acku
    ryedawgAnother great article, as usual, from this site. However this statement in inaccurate:"Buying a Nexus 7 locks you into Google's Play store and its movies, newspapers, magazines, and music."I have BOTH the Kinda Fire app and Barnes and Noble app installed on my Nexus 7. And that is one of the things I love most about Google's tablet: the ability to get content from any provider I want. I would also like to point out that my gf has the B&N tablet, and it is much better than the Kindle Fire for several reasons: 1) you can sideload your own content through a removable memory card, 2) it has a better screen, and 3) the build quality seems much better after holding both in your hand. It's a shame the Kindle get's more attention, the power of advertising I guess....
    Yes and no. You can use the Kindle Android app to view ebooks from Amazon. That's the case with Android, iOS, and Windows. However, this is not the integrated interface that the Nexus 7 provides. It's a little different for viewing ebooks and magazines.

    More importantly is the difference in movie support. You cannot use the Nexus 7 to view Amazon Prime movies the same way on the Kindle Fire HD. H.264 streaming works when you're on an Amazon tablet, plain and simple. If you want to watch those same movies on the Nexus 7, you need to install Firefox and Flash. But that's still Flash, not the streaming H.264.
    Reply
  • acku
    nitriumGiven these devices are also intended to be used as eBooks, why won't anyone test battery life reading an actual eBook - i.e. Wifi/Bluetooth off, screen on (probably at ~50% brightness), no background tasks, just the occasional page flip every 30 secs or whatever the average reading rate is. Is that too much to ask? I tried to find this info with Google, but apparently no one but me cares.
    That's a great idea! Unfortunately it's very difficult to implement from a programming perspective to keep consistent across all devices and platforms.
    Reply
  • nitrium
    That's a great idea! Unfortunately it's very difficult to implement from a programming perspective to keep consistent across all devices and platforms.
    I'd actually settle for "idle" battery life, since reading an eBook is effectively utilizing nothing but the screen. So Wifi/bluetooth/GPS/camera off (not just unused, but literally disabled in settings), screen on (at ~50% brightness). Do these things last for days under these conditions? I can't find data for that anywhere. My primary use for a tablet would be eBooks, but I have no idea which reader (except the original Kindles) actually gives the best battery life for that specific purpose.
    Reply
  • andrys1
    Re the Barnes and Noble Nook Tablet HD that someone mentioned... A problem for marketing is that, despite the competition, it has NO cameras, not even one for Skype (important for family & friends who like to communicate via video chat), and, more important for many of us, B&N will not allow owners to enable installation of apps from non-B&N sources.

    Amazon does allow installation of apps from "unknown sources," so I have (from places like getjar.com or 1mobile.com) apps like Google Maps in satellite mode, Mantano and also Aldiko to read DRM'd ePub books legally. I also have the B&N Nook app, since I have the NookColor but prefer to just read on the Kindle Fire HD now.

    Very accurate article. As for the Kindle Fire HD, I love the stereo speakers with Dolby and good separation because they're relatively strong and very clear. I hate using headphones or earbuds as a rule (unless in public but I don't usually listen to anything while around other people) unless I'm wanting best sound in music. For me, although my hearing is not ideal, dialog is MUCH better on these than on laptop units I've tried.
    Reply