review asus eee pad transformer prime tablet

Asus’ Eee Pad Transformer Prime tablet

Do a Google image search for Transformer Prime, and you’ll be inundated with pictures of this guy. Optimus Prime was the leader of the Autobots of my childhood, and I may or may not have cried when he died in the first movie (Transformers: The Movie—not the tripe Michael Bay has spewed out in recent years). You’ll have to scroll through three pages of search results to come across the first picture of Asus’ Transformer Prime, an altogether different machine.

Hasbro isn’t pleased with the association, which Asus has admittedly made pretty blatant. One might be tempted to chide the company for tugging at the childhood nostalgia of geeks everywhere, but the Transformer name has been a good fit for Asus’ first attempt at an Android-based tablet. With the addition of an optional keyboard dock that offers a proper touchpad, extra expansion slots, and another six hours of battery life, the Transformer turns into something that behaves more like a real notebook.

The Transformer Prime follows the same formula as its predecessor while offering plenty of refinements throughout. It’s thinner and lighter than the original despite housing a more powerful Tegra 3 processor with five ARM cores and beefier GeForce graphics. Asus has updated the screen, the casing, and just about everything in between, including the operating system. The end result looks like the most impressive Android device to date, so we’ve spent some quality time with one to see what all the fuss is about.

There’s a new Tegra in town
One of the most intriguing parts of the Prime is its Tegra 3 processor, otherwise known as Kal-El. This Nvidia system-on-a-chip succeeds the Tegra 2 found in the original Transformer and quite a few other Android-based tablets. Both chips are based on the same ARM Cortex A9 CPU architecture, but the Tegra 3 is a more robust implementation. Unlike previous Tegra processors, the third-gen chip’s CPU cores feature ARM’s NEON media engine. This dedicated SIMD unit supports extensions to the base ARM instruction set targeted at improving multimedia, graphics, and gaming performance.

In addition to beefing up its Cortex A9 cores, the Tegra 3 has more of ’em running at higher speeds. While the Tegra 2 sports dual cores, the Tegra 3 features a funky quad-core design with an additional “companion core” optimized for low power consumption. This fifth core has an identical ARM architecture to the others, but it tops out at just 500MHz. The Tegra 3 is capable of running its other four cores at speeds up to 1.3GHz, a 100MHz boost over its predecessor. And, thanks to a little Turbo-style mojo, the Tegra 3 can hit 1.4GHz if just one of those cores is active.

All five of the Tegra 3’s CPU cores occupy a single piece of silicon fabricated by TSMC on a 40-nm process. The companion core uses transistors with very low leakage power but also relatively slow switching speeds, which is why it tops out at 500MHz. The other cores are optimized for higher switching frequencies and are capable of nearly three times the speed of the companion core at the same voltage level. As one might expect, these faster cores suffer from higher leakage power.

According to Nvidia, it takes less than two milliseconds to switch the Tegra 3 between companion and quad-core modes; the two core components are never active at the same time. “Aggressive power gating” is employed for each individual core, including those within the quad-core cluster, allowing the chip to cut power to dormant cores. There’s also plenty of dynamic clock scaling at work, although Nvidia notes that active members of the quad-core cluster will all run at the same speed to ease OS scheduling.

The Tegra 3 block diagram. Source: Nvidia

The Honeycomb version of Android that ships on the Transformer Prime isn’t aware of the Tegra 3’s novel approach to symmetric multiprocessing, which Nvidia has dubbed variable SMP, or vSMP. To manage the chip’s unique core composition, Nvidia uses a mix of hardware and software to monitor system activity and to adjust the core configuration appropriately. The companion core is intended to be sufficient for background tasks like email syncing, Facebook updates, and multimedia playback, while the quad-core block is reserved for gaming, web browsing, Flash, and other demanding tasks.

Nvidia claims the Tegra 3’s dedicated decode hardware is powerful enough to handle all manner of multimedia playback without the aid of the chip’s quad CPU cores. All the major codecs are supported at resolutions up to 1080p, and little appears to have changed from the previous iteration of the chip. The only notable addition I can find regards high-profile H.264 content, which the Tegra 3 can purportedly process at bitrates up to 40Mbps. Nvidia makes no claims about the Tegra 2’s support for high-profile flavors of H.264.

The green team has been relatively tight-lipped about the GeForce graphics processor incorporated in the Tegra 3, too. Called only an “Ultra-low power GeForce GPU,” this graphics processor has 12 cores and the ability to generate stereoscopic 3D images. We don’t know the clock speed of the GPU, but Nvidia claims it offers up to three times the performance of its predecessor, which has only eight cores and doesn’t support stereo 3D. The two implementations appear to be based on the same graphics architecture, so higher clock speeds likely account for much of the Tegra 3’s advantage. This latest Tegra chip also supports higher memory frequencies than the Tegra 2, which should provide more bandwidth to the graphics subsystem.

Processor Nvidia Tegra 3 1.3GHz with GeForce graphics
Display 10.1″ IPS TFT with 1280×800 resolution
Memory 1GB
Storage 32GB
Ports 1 Micro HDMI
1 analog audio headphone/mic port
1 USB 2.0 (dock)
Expansion slots 1 Mini SD
1 SD (dock)
Communications 802.11n Wi-Fi
Bluetooth 2.1+EDR
Camera 8-megapixel rear with LED flash
1.2-megapixel front
Input devices Capacitive touchscreen
Chiclet keyboard with touchpad (dock)
Dimensions Tablet: 10.4″ x 7.1″ x 0.33″ (263 x 181 x 8.3 mm)
Dock: 10.4″ x 7.1″ x 0.41″ (263 x 181 x 10.4 mm)
Weight Tablet: 1.29 lbs (586 grams)
Dock: 1.18 (537 grams)
Battery Tablet: 25Wh lithium-polymer
Dock: 22Wh lithium-polymer

As with the original Transformer, the Prime pairs its Tegra processor with a gig of RAM. That’s double the memory of the iPad 2, whose A5 CPU looks considerably less powerful on paper. The A5 has only two Cortex A9 CPU cores clocked at 1GHz, so it’s down both cores and clock speed versus the Tegra 3. Comparing the Apple chip’s PowerVR graphics processor to the Tegra’s integrated GeForce isn’t quite as straightforward because the two don’t share a common architecture. We’ll see how they shake out in a handful of graphics tests a little later in the review.

The truth is that these little technical details don’t tell us as much as we might like about how the Transformer Prime performs in the real world. It should certainly be faster and more responsive than devices based on the last-gen Tegra processor, but only because they’re all running Android. Loading up on processing resources doesn’t necessarily give the Prime a leg up on the incumbent iPad 2, which has an entirely different operating system—one that’s long been the standard for overall responsiveness.

Our own Editor-in-Chief can’t stop gushing about how much iOS 5 improved his iPad 2 experience, so it’s worth noting that a major Android update is due to hit the Prime on January 12. The tablet ships with the older Honeycomb version of the OS, but Android 4.0 is coming via an over-the-air update next week. Otherwise known as Ice Cream Sandwich, this next version of the OS promises all sorts of enhancements, including improved performance and a new browser.

The Transformer Prime admittedly feels a little incomplete without Ice Cream Sandwich onboard. Fortunately, there’s plenty to explore about this tablet beyond its software payload.

Sexier than Arcee
Although Nvidia hasn’t released specifics on the Tegra 3’s power consumption, it argues that the chip’s quad-plus-one-core approach is more power-efficient than dual-core ARM processors. The chip runs cool enough that Asus has been able to put the Transformer’s chassis on a crash diet. While the original TF101 model matched the 13-mm thickness of the first iPad, the new TF201 variant measures just 8.3 mm at its thickest point. The slimmed-down iPad 2 is actually half a millimeter thicker than the Transformer Prime.

An old MP3 player is the only iDevice in the Benchmarking Sweatshop, so I’ve posed the Prime with my original Transformer for reference. As you can see, the new design’s thickness—or lack thereof—is the most notable size difference between the two. This is especially apparent with the dock attached, which balloons the TF101 to about an inch thick. The Prime measures less than 19 mm thick when paired with its optional keyboard, making the complete system thin enough to be an ultrabook.

The Transformer Prime is slightly taller (when held in landscape mode) than the original, and the original is a little bit wider. Those dimensional differences largely cancel each other out, leaving the two with similar footprints.

As one might expect given its thinner chassis, the Prime is the lighter Transformer. The tablet component tips the scales at 1.3 lbs, while the dock adds another 1.2 lbs. Compare that to the old model, whose tablet and dock weigh about 1.5 lbs each, and you’re looking at a half-pound delta between the two complete systems. That’s enough to notice, although perhaps not enough to be a meaningful burden to able-bodied adults.

I’ve gotta admit, the differences in thickness and weight between even just the tablet sections of the two Transformers are instantly apparent when you pick them up. The Prime is slightly easier to stuff into my messenger bag when squeezed in alongside my ultraportable notebook and DSLR, too. I find the thinner tablet to be a little less comfortable to hold, though. Turns out the thicker curve at the back of the old Transformer better fits the natural curve of my hand, which admittedly requires XL-sized gloves.

Pulling off a tablet with runway model proportions is an impressive feat of engineering, especially with the Tegra 3 processor putting plenty of muscle on the Transformer’s slender new frame. Even more attractive than the Prime’s lean figure is her beautiful skin. I’m assuming our sample is a lady because she came dressed in a gorgeous shade of not-quite-purple dubbed amethyst gray. Behold the hotness:

It must be difficult for PC makers to come up slick-looking designs when anything clad in brushed aluminum is so quickly written off as an Apple knock-off. Despite relying on the same material, Asus has created a truly distinctive look for the Transformer Prime. The aluminum back panel borrows the “spun” finish introduced by the company’s Zenbook ultrabook. Instead of sticking with that system’s natural metal hues, Asus has chosen two tinting options: the amethyst gray tone of our press sample and a sandy shade called champagne gold.

The keyboard dock comes clad in similar materials with matching colors. Alas, the brush strokes across the dock’s aluminum panels follow a horizontal rather than circular path. The anodized panels still look unique, though, and they nicely blend in with the matte plastic pieces that line the edges of the tablet and dock.

Catch the Prime at the wrong angle, and you’ll see that the metal finish isn’t impervious to fingerprints and smudges. To be fair, any blemishes on the Transformer’s skin are much less obvious than they would be on the glossy plastic surfaces typical of all too many notebooks. Buffing the exterior back to its factory-fresh sheen is ridiculously easy. Just wiping the aluminum down with a bare hand is usually enough to clean up any ugliness that has accumulated. A simple cloth—or the inside of a shirt—easily eradicates stubborn streaks.

A screen designed with sunlight in mind
Unfortunately, you won’t see any of the Prime’s spun aluminum skin when using the device as a tablet. Users face only the screen and its surrounding bezel, both of which are covered by a glossy piece of protective Gorilla Glass. This surface doesn’t look bad right after it’s been polished with the included microfiber cloth. However, it’s impossible to use the tablet’s touchscreen without leaving a mess of oily gesture remnants behind. Good luck trying to hold the thing without depositing a few thumbprints on the bezel.

While any smudges on the screen become almost invisible when the Prime is turned on, the glow has no effect on the surrounding bezel. The ring of greasy fingerprints circling the screen is a constant reminder of why glossy surfaces are less than ideal for devices that are handled frequently.

Reflectivity can also be an issue with glossy displays. The Prime’s glossy screen coating seems to show a little less of myself staring back than the original Transformer. However, under normal indoor lighting, reflections are still clearly visible in darker sections of the screen. Turning up the tablet’s brightness helps, and there’s certainly plenty of luminosity on tap.

At its brightest, my first-gen Transformer cranks out a respectable 341 cd/m². The Prime pushes 407 cd/m² with its brightness turned up, and that’s only the beginning. The TF201 also has a SuperIPS+ mode that ignites a small sun behind the panel. With this mode enabled, the screen delivers a whopping 643 cd/m² according to our colorimeter—an increase of nearly 60% over the normal config.

In direct sunlight, the Prime’s extra luminosity makes it easier to use than the TF101. While the additional brightness definitely makes images clearer, they’re still forced to share the screen with reflections of the surrounding environment. I wouldn’t go so far as to call the Prime good for outdoor viewing, but it’s definitely better than any other full-color tablet I’ve seen.

Like most tablets, the Prime features an IPS panel with better color reproduction than the TN displays found in typical notebooks. TN panels also tend to look washed out if you’re not viewing them dead on, but the Prime’s screen has wide 178° viewing angles, ensuring that colors remain vivid even when peering at the screen from a couple seats down on the couch.

Our first Transformer Prime sample arrived with an obvious yellow tinge to the screen; it was the first thing I noticed after turning on the tablet. That system was a pre-production unit never intended to go to the press, but the yellowish tint persisted in our second sample, albeit in muted form. As it turns out, Asus has deliberately tuned the Prime’s panel to produce warmer colors than the original Transformer. Put the two tablets side by side, and the TF101’s whites look a little bit blue, while the TF201’s have a hint of yellow. The difference is difficult to capture on camera, so I busted out our colorimeter to profile the color gamut of each screen.

To make the two tablets easier to compare, the color gamut graph will switch between the Prime and first-gen Transformer when your mouse cursor is moved over the image. It takes a few seconds for the second image to load, so be patient.


Gamut graphs look intimidating, but our little mouseover effect makes the differences between the two tablets quite clear. The inner triangle that defines the Prime’s color gamut is biased toward warmer tones, while the Transformer trends toward cooler colors. Note the intersection point at the center of each inner triangle. On the original Transformer, this point is closer to the D65 marker corresponding to the color temperature of typical daylight.

We can also isolate the color temperature of each display across multiple gray levels. The graph below has the same mouseover trickery as the one above, and the second image should be a little quicker to load. Here, we’re looking for a color temperature of 6500K, which is equivalent to the D65 daylight illuminant from the gamut graph.


The Transformer comes much closer to the 6500K ideal for a few gray levels, but the Prime offers a more consistent color temperature across the entire range. Although I prefer cooler color temperatures in general, the Prime’s warmer tones really only bother me when I’m switching back and forth with the old Transformer. Asus claims to be working with Google to add color controls to Android, but it doesn’t sound like a done deal just yet. Being able to tweak the color levels of the display would certainly be a nice addition to the OS.

Temperature quirks aside, the Prime’s screen is fantastic. Text looks crisp, the colors are lush, and the 1280×800 display resolution is a good fit for the 10.1″ panel. This resolution serves up 30% more pixels than the 1024×768 screen on Apple’s iPads, and the extra real estate is especially handy when the tablet is rotated in portrait mode. Now that I’ve figured out how to make the Android browser stop resizing text columns, I do almost all of my tablet browsing at 800×1280.

Transform and roll out
The on-screen keyboards in modern tablets are pretty lousy for prolonged typing sessions. The lack of tactile feedback makes typing at speed difficult, and punctuation usually requires switching to a secondary layout populated with symbols rather than letters. Then there’s the fact that on-screen keyboards typically take up half of one’s screen. For anything longer than a web address, search string, or quick email, an auxiliary keyboard is highly recommended. Perhaps that’s why roughly half of the tablets I see in the wild have keyboards in tow.

Although Asus sells the Prime’s keyboard dock as a separate accessory, it’s more of an integral part of the Transformer than the average tablet keyboard. The two pieces snap together using a simple latching mechanism, converting the Prime from a standard slate into something that looks an awful lot like a notebook. Sadly, one can’t mix and match docks and tablets from different generations.

The connection between the tablet and dock is secure, but there’s just enough wiggle room to allow the user rock the screen back and forth slightly. Most laptops have a little bit of play in the hinge that holds their screens, so the Transformer isn’t unique in this regard. The fact that the tablet component weighs more than the dock—the opposite of typical notebooks, which carry most of their weight under the keyboard—appears to make the Transformer’s screen a little more prone to wobbling, though.

When connected to the dock, the Prime is particularly back heavy. The balance feels a little odd if you’re used to having a notebook perched on your lap, and the Prime is prone to tipping if the dock finds itself on a downward slope, even for just a moment. On the plus side, the dock makes a decent stand when it’s sitting on flat surfaces and the screen isn’t tilted back at too extreme of an angle. The dock also does an excellent job of protecting the screen when the Transformer is clamped shut.

Asus has had to make a few compromises on the keyboard front to accommodate the Prime’s small footprint and slim profile. The dock is only about 10 mm thick, which limits the travel of the Chiclet-style keys. Impressively, though, the keyboard manages to offer a punchy feel with good tactile feedback. Each key bottoms out with a dull thunk, and there’s surprisingly little flex under spirited typing.

  Total keyboard area Alpha keys
  Width Height Area Width Height Rough area
Size 250 mm 92 mm 23,000 mm² 154 mm 47 mm 7,238 mm²
Versus full size 87% 84% 73% 90% 82% 74%

The keyboard area is on the small side, which is to be expected of a system this size. Typing at speed hasn’t been a problem for me, but my monstrous mitts do feel more cramped on the Prime than they do on my 11.6″ ultraportable notebook. If you’ve found the keyboards on 10″ netbooks to be prohibitively small, you probably won’t be much more comfortable on the Prime.

Apart from its size, the Prime’s keyboard has a couple of other quirks that require compromise. There’s no delete key, and while shift+backspace performs the same function, nothing on the keyboard reveals that little secret to the user. I also miss being able to hold Ctrl while hitting the arrow keys to jump entire words rather than single spaces.

Perhaps as a consolation for what’s missing, Asus has lined the top of the keyboard with a selection of handy shortcut keys. By far the most useful of these keys is the one that disables the touchpad. Amusingly, the touchpad is probably one of the keyboard dock’s best features. The problem is that the Prime isn’t smart enough to differentiate intentional touchpad contact from the inadvertent brush of one’s thumb while typing. Accidental contact is especially likely given the small size of the keyboard, making it difficult to write more than a paragraph without sending the cursor halfway across the screen.

Despite its lack of intelligence, the touchpad can be a pleasure to use. Its smooth surface offers little resistance to one’s fingertips, and the buttons have a nice click. I do wish the buttons were located right along the bottom edge of the enclosure, though. They’d be easier for the thumb to hit in that position, and moving them south would allow for a slightly larger tracking area.

The limited tracking area is still sufficient, and it offers more precision than one might expect from an Android tablet. Not too long after the original Transformer’s release, Google changed Android’s default mouse cursor from a standard pointer to a large circle about the size of a fingertip. Doing so stripped the touchpad (and even external mice) of their precision, but Asus soon rolled out an OS update allowing users to switch between the two cursor configs. That option persists in the Prime, and I can’t imagine living without it. Surprisingly, not all Android tablets allow users to choose a precise mouse pointer.

More than meets the eye
When I said that the Transformer’s keyboard dock makes the tablet more of a notebook, I wasn’t just talking about its appearance. The dock also adds a couple of conveniences rarely seen on tablets but common among the netbooks and notebooks they’re supposedly set to replace.

To complement the Micro SD slot in the Prime tablet, Asus adds a full-sized SD slot to the accompanying dock. Tablets make good travel companions, and being able to dump digital pictures from the memory card in one’s camera is certainly useful—no special cables or software required.

The dock also hosts a proper USB 2.0 port. The old TF101 actually has two USB ports in its dock, but Asus decided to trim one of them for the Prime. I guess the thinner chassis isn’t quite as good at accommodating full-sized ports as the old model. Besides, most tablet makers ignore USB connectivity completely, and I’ve never used more than one of my Transformer’s USB ports at any given time. I can’t help but wish there were a Mini USB connector right on the Prime, though.

At least the tablet serves up a Micro HDMI port on its left edge. The Tegra 3 can output resolutions up to 1080p, making the Prime well-suited to feeding big-screen televisions. Audio can also be passed over HDMI, of course.

The Prime’s built-in speaker sounds noticeably better than the one integrated into the original Transformer. Most users will probably still prefer proper headphones when listening to music or watching movies, but it’s nice to have the speaker upgrade for short gaming sessions and YouTube clips. Thanks to its location on the right side of the tablet, the speaker is easily muffled when holding the Prime in landscape mode using your right hand. Since the Transformer doesn’t mind being held upside-down, rotating the tablet is all it takes to move the speaker away from one’s grasp.

For those who prefer headphones, the Prime’s right edge houses a 2-in-1 audio jack. There’s also a built-in microphone if you’d rather conduct Skype calls sans headset.

Folks who use their tablets for video conferencing won’t find any upgrades to the Transformer’s front-facing 1.2-megapixel camera. However, the rear-facing unit has been bumped from five to eight megapixels. The rear camera now features a back-illuminated CMOS censor and an F2.4 aperture, too. Although the upgrade in picture quality is definitely apparent, it seems like a waste to me. Taking pictures with a tablet is much more awkward than using a real camera or a smartphone. Holding a tablet steady to shoot video is even worse, making the Prime’s 1080p recording capability look more impressive on the spec sheet than it ends up being in the real world.

Asus ships the Prime with a tiny power adapter that looks almost identical to the one that comes with the old Transformer. The glossy plastic finish is still an abomination, and the prongs on the plug don’t fold flush like they do on the TF101. Speaking of things that don’t fit flush, look at what happens when you plug the adapter into the Prime:

The plug slots in snugly with the flat edges of the first-generation Transformer, but the new model’s curves make for a looser fit. This is true for both components of the Prime, which can be charged together using a single adapter. That’s a good thing, because although each piece can also be charged separately, only the tablet comes with the required cord.

Thanks to a pair of batteries, the Prime doesn’t need to be charged all that often. The tablet has a 25Wh lithium-polymer battery that Asus says will last for up to 12 hours under highly favorable conditions. On top of that, the keyboard dock provides its own 22Wh battery purportedly good for another six hours of run time. Rather than switching between the two batteries, the Prime uses the dock to charge the tablet as it runs dry. The OS displays the capacity of both batteries in the taskbar and indicates when the dock is charging the tablet.

CPU performance
As it does with notebooks, Asus has infused the Transformer Prime with a handful of different operating modes. The “normal” mode, which will be renamed “performance” in a future update, runs the Tegra 3 at its stock frequency of 1.3GHz. Out of the box, the Prime runs in “balanced” mode, which caps the processor at a slightly slower 1.2GHz. There’s also a “power-saver” mode that imposes a series of frequency caps on the CPU. This mode limits the Tegra 3’s CPU to 1GHz with one or two cores active, ~700MHz when three cores are active, and ~600MHz when four cores are fired up.

In a bid to further conserve battery life, the power-saver config switches the screen into a power-efficient operating mode.  This mode caps the refresh rate at 35Hz and makes some other changes that result in lower color fidelity. This muted screen setup looks fine with the monochrome hues of my favorite text editor, but any colors displayed elsewhere look dull and washed out compared to the balanced profile. The downgrade in visual quality is severe enough that I’d avoid web browsing or picture viewing in power-saver mode. For some reason, the dull colors aren’t as apparent when watching full-screen videos.

To give you a better sense of the Prime’s performance, we’ve tested the tablet in all three power modes. We’ve also assembled a handful of competitors for comparison, including the original Transformer, the Galaxy Tab 10.1, the Kindle Fire, and the iPad 2. Unfortunately, the Kindle and iPad won’t be able to participate in all our tests due the availability of benchmark applications on each platform. The Fire needs to be rooted to access the Android Market, and the iPad runs an entirely different operating system that requires separate binaries.

We’re still getting the feel for tablet testing, and the number of cross-platform benchmarks is really quite small. We’ll start with Linpack, which measures raw CPU performance. The iOS version of Linpack appears to be quite different from the one available on the Android Market, so the Apple tablet is going to sit out this round. Since the variant of Linpack available through the Amazon Appstore doesn’t specify whether it’s a single- or multithreaded build, the Kindle Fire will also be riding the pine.

In single-threaded mode, the Prime pushes 8-16 more MFLOPS than its forebear. Surprisingly, the Prime’s advantage is smaller, relatively speaking, in Linpack’s multithreaded test. The Tegra 3’s extra cores don’t confer the sort of advantage one might expect over the dual-core Tegra 2 processor in the Galaxy Tab and original Transformer. At its fastest, the Prime churns out 53% more MFLOPS than the TF101 in single-threaded mode but only 29% more in the multithreaded test.

Tablets are probably used for web surfing more than anything else, so we’ve run a couple of browser-based tests on the Prime and its rivals. These tests run inside the native browser on each device, which should give us a good cross-platform comparison. Silk, the Kindle’s cloud-based web renderer, was disabled throughout because it actually delivers a slower real-world browsing experience than letting the device request pages itself.

First up is version 0.9.1 of the SunSpider JavaScript benchmark. As you might’ve guessed, this test probes JavaScript execution speeds.

The Prime leads the pack when running at full speed, but the iPad 2 is slightly faster than the new Transformer in balanced mode. Granted, the difference in JavaScript performance between those three configurations pales in comparison to the gap between the lead bunch and our duo of Tegra 2-based tablets. The Kindle Fire is slower still, but it turns in a much better performance than the Prime’s plodding power-saving mode.

FutureMark’s Peacekeeper benchmark also tests JavaScript performance, this time with a side order of HTML5. While the iPad 2 can only perform one of Peacekeeper’s HTML5 tests, the Android-based tablets manage to complete two of them. I suppose that’s some kind of victory.

The discrepancy likely explains the iPad 2’s inability to distance itself from the Tegra 2 tag team. While the Apple tablet slums it with the common folk, the normal and balanced Prime configs deliver substantially higher scores. Don’t be discouraged by the Prime’s poor showing in power-saving mode, which is to be expected given the clock constraints placed on the CPU.

Graphics performance
Next, we’re going to take a look at graphics performance. Since Fraps (or something like it) isn’t available to track frame rates during actual gameplay, we have to rely on a handful of 3D graphics benchmarks. The first of these is GLBenchmark 2.1, which uses OpenGL ES 2.0 and is available for both Android and iOS. GLBenchmark isn’t available through the Amazon Appstore, forcing the Kindle out of the action for another round.

GLBenchmark’s standard tests measure frame rates as the scene is drawn on the screen. This mode constantly invokes vsync (which can’t be disabled) on the Transformer Prime, so we’ve also tested with GLBenchmark’s offscreen mode. Those tests are run at 1280×720 but aren’t shown on the screen, preventing vsync from artificially limiting the performance of the graphics processor.

In the Egypt and Pro scenes, the Transformer Prime is handily beaten by the iPad 2. The results aren’t even close when we look at the offscreen scores, which give the Apple tablet a commanding lead. That said, the Prime is a huge upgrade in performance over the old Transformer and the Galaxy Tab. Clearly, the Tegra 3 has much more graphics horsepower than its predecessor. Switching the Transformer Prime into power-saving mode doesn’t slow the chip’s integrated GPU much, either.

We’ve been having problems getting the full version of RightWare’s Basemark ES 2.0 graphics benchmark running on our Transformer tablets, so we had to resort to the free version of the app, which contains only the Taiji test. Unfortunately, the free version isn’t yet available on iOS, so the iPad will have to join the Kindle on the sidelines.

The Taiji test confirms that the Tegra 3 has much more potent graphics hardware than the last generation of the chip. Turns out Nvidia’s claim of a threefold increase in graphics performance is right on the mark, at least in this scene. Once again, we see very little difference in graphics performance between the Prime’s various configurations. I suspect the integrated GPU is being kept at close to full capacity in power-saving mode to ensure that the CPU isn’t disturbed for HD video.

Battery life
Here’s where things get tricky. Battery life takes an excruciatingly long time to test on tablets, which offer run times in the ten-hour range. Determining the precise moment that a tablet runs dry requires a little ingenuity, too. While we managed to solve the timing problem with some ghetto time-lapse photography, time constraints prevented us from testing all of the tablets. For now, you’ll have to make do with a small collection of results for the two Transformers.

As we do with notebooks, the batteries on each tablet were fully cycled at least one before each test. The Transformer was configured with a 50% screen brightness, while the Prime was set at 40%, which looks about the same. Each test was run first on a fully charged tablet with the dock disconnected. The dead tablets were then connected to their fully charged docking stations before a second round of tests.

The Prime was run only in its balanced configuration because we lost days to hunting down a couple of issues encountered along the way. The first of these cropped up in our web-surfing test, which automatically loads a version of the TR home page every 45 seconds. This test runs entirely inside a standard web browser, and it cycles through a number of ads for added realism. Some of those ads use Flash, which results in occasional corruption on the Prime.

Weird. The old Transformer displays all the ads correctly despite the fact that it’s running the same version of Flash from the Android Market. We haven’t encountered this kind of corruption with normal web browsing on the Prime, and neither Asus nor Nvidia has explained why it might be occurring. Nvidia did confirm that the Tegra 3 tries to load Flash elements as fast as possible to speed up page rendering times, though. Since Scott and I typically browse the web with Flash disabled on our Android tablets anyway, plug-ins were set as tap-to-load for the remainder of our battery tests.

With Flash effectively disabled, the Prime turned in a shorter run time than the original Transformer in our web-browsing test. While the old model’s run time was consistent from one run to the next, the Prime was all over the map. On our first attempt, the Prime delivered 8 hours in tablet form and another 4.4 with the dock attached. In round two, its run time dropped to 6.5 hours, plus another 4 on the dock.

Asus was able to reproduce this inconsistency, and I suspect it has something to do with how Nvidia is managing the Tegra 3’s multiple cores. Clearly, there are kinks to be ironed out. Nothing about the page we’re reloading is particularly demanding, and this test hasn’t produced similarly inconsistent results on other platforms.

To find out if the Prime’s battery life is more consistent with movie playback, I looped an episode of CSI: New York encoded with H.264 at 480p resolution. The DicePlayer Trial, which supports the decode hardware in the Tegra 2 and Tegra 3, was used for playback. Both tablets were put into airplane mode, disabling Wi-Fi connectivity.

The Transformer Prime looks much better here. More importantly, it exhibited none of the inconsistency that tainted our web-browsing results. Even without resorting to power-saving mode, you can still get more than ten hours of movie playback out of the Prime—and a good 17 hours if you have the keyboard dock. The old Transformer pulls up 2-3 hours short, forcing at least a couple of episodes to be cut from your next Game of Thrones marathon.

Not content to rely on these results, I’ve been keeping an eye on the Prime’s battery indicator while using the tablet around the house and during my holiday travels. My typical task load includes email, web browsing, a perusing a selection of comic books and PDFs. I’ve also thrown in a bunch of picture viewing, gaming, and video playback to spice things up. Fortunately, the wild inconsistency we observed in our web-browsing test doesn’t seem to affect the Prime in the real world. I’m getting around nine hours of battery life with the tablet, which is a little better than what I’m used to seeing with the old Transformer. Having the dock attached seems to add about six hours worth of juice.

Living with the Transformer Prime
Over the last couple of weeks, I’ve spent a heck of a lot of time playing with the Transformer Prime—too much time, considering I was supposed to be taking a few days off. Such things are difficult when one has access to the hottest Android tablet on the market. We already know that the Prime scores better than the old Transformer in a variety of benchmarks, but does it feel any faster in day-to-day use? Most definitely.

To me, the biggest difference is the responsiveness of the operating system. Everything seems to happen a little bit faster than on the TF101, from flipping between desktop panels to loading applications. The UI animations are noticeably smoother, adding to the sense of fluidity. The touchscreen feels more precise, too, although that’s more likely an artifact of seeing the interface respond more quickly to one’s input.

Going back and forth between the Transformer and the Prime reminds me of upgrading from a single CPU to a dual-processor system way back in the day. It doesn’t feel like you have effectively double the horsepower under the hood, but the Tegra 3’s additional cores rekindle the feeling of creamy smoothness I got with my first exposure to multi-processor systems. To be fair, the iPad 2 creates that same feeling of smoothness with two fewer cores than the Prime. My time with the Apple tablet has admittedly been limited to fooling around with one a few times at the company’s local store, and I could only stand being that close to hipsters for so long.

Despite the Prime’s improved responsiveness, you won’t find me trading in my ultraportable notebook. After writing most of this review on the Prime, I was happy to switch back to my Aspire 1810TZ notebook. Even though that system is more than two years old, it has a more comfortable keyboard, a smarter touchpad, and better multitasking. Photo editing is much easier on the Aspire, and the company of familiar Windows applications certainly helps my productivity. Still, the Prime offers a more satisfying experience when I’m in consumption mode and have little desire to accomplish anything work-related.

While web browsing, the Prime doesn’t feel a whole lot faster than the old Transformer. Scrolling through web pages is smoother, and the new Transformer tends to do a better job of keeping the contents of a page visible while they’re in motion. However, it doesn’t load pages any quicker—at least not consistently. I set up the Prime next to its forebear for some side-by-side tests, and the results were inconclusive.

Curious to see if networking performance was affecting page load times, I ran some Internet speed tests on both tablets in various rooms of my house. In my office, just a few feet from the wireless router, the Prime’s download speed ranged from 6-14Mbps. The old Transformer managed a consistent 14Mbps in the very same position. Things were similar in the living room. While the Transformer maintained a download speed of around 14Mbps, the Prime’s scores were spread from 2-14MBps. In the bedroom, about as far as I can get from the router without going outside, both tablets managed a download rate of about 2Mbps.

We aren’t the only ones who have had issues with the Prime’s networking. That said, I haven’t noticed any network slowness when using the tablet around the house. Streaming video seems to load up just as quickly on both tablets, too.

Speaking of video, the Prime has no problems playing HD content, be it from local storage or networked sources. YouTube playback is smooth using the dedicated app or when viewing clips through the Android browser. Impressively, video playback remains buttery smooth even with the tablet running in power-saving mode.

Asus has incorporated a custom control panel into the Prime’s Android install to make it easy for users to switch between power modes. This handy little pop-up offers toggles for Wi-Fi and Bluetooth connectivity in addition to automatic screen rotation. The brightness controls for the screen include switches for the SuperIPS+ mode and auto-dimming, and there’s even a convenient shortcut to Android’s settings panel.

Changing the control panel might seem like a small detail, but that’s exactly where I’d like to see Asus focus its efforts. Unlike some Android device makers, Asus has wisely steered clear of blanketing the OS with a custom user interface.

The Prime’s file manager has received an aesthetic makeover, though. Asus has added a few shortcuts and made it easier to go back and forth between external storage devices. Like its predecessor, the Prime gives up full access to the contents of its internal storage when connected to a PC via the USB charging cable. Being able to dump files onto the device quickly from any computer is certainly handy.

Despite its numerous improvements, there is one area where the Prime lags well behind the old Transformer: cold booting. Resurrecting the Prime after a shutdown takes up to two minutes, while its predecessor boots in less than half that time. At least the Prime returns from standby instantly. I suspect most folks leave their tablets in standby rather than turning them off completely.

Tablet or game console?
In recent years, smartphones and tablets have become more and more popular as gaming platforms. Naturally, Nvidia wants in on that action. It’s looking to make a big splash among gamers with the Tegra 3, and our Transformer Prime came loaded with a collection of titles optimized to take advantage of the chip’s capabilities.

I have to admit that I’ve spent very little time gaming on my first-gen Transformer. The problem isn’t the games so much as the interface, which isn’t responsive enough for the action-oriented titles I tend to prefer. Then there’s the annoyance of having my wildly gesturing hand constantly obscuring my window on the game world. Touchscreen-input is much better suited to slower-paced strategy and puzzle games, I think.

Nvidia must feel my pain, because the Tegra 3 has native support for a range of gamepads, including wired versions of the Xbox 360 and PlayStation 3 controllers. Even generic USB gamepads are supposed to work with Tegra 3-based devices, although actual games may require updates for external controllers to operate correctly. Developers would do well to take advantage, because in the few games I tried, having a controller added immensely to the experience.

Shadowgun makes good use of the Xbox 360 controller, transforming a game I find horribly awkward with the touchscreen into something that feels like a standard over-the-shoulder shooter. The graphics are nowhere near Gears of War quality, but Shadowgun still looks pretty good for a tablet title. The version installed on our Transformer Prime boasts a smattering of Tegra 3-specific upgrades, including reflective water, enhanced physics effects, and a collection of fancier shaders. Honestly, though, the control scheme has more of an impact on my enjoyment of the game than the eye candy.

Riptide GP is more forgiving of the Transformer’s native control scheme, but I still prefer the game with dual analog sticks at my command. The novelty value of rotating the tablet to steer wears off pretty quickly after you realize just how imprecise the internal gyroscope is.

Nvidia says Riptide offers better lighting and shadows on Tegra 3 hardware. Folks running the green team’s latest system-on-chip will also be able to enjoy motion blur, more realistic water rendering, and a slick splash-on-the-lens effect.

Bladeslinger is an Infinity Blade-style slasher with a unique western vibe. The controls are very much tailored for touch input, and the game is easy to get into if you can stand having your hand streaking back and forth across the screen. The fact that Bladeslinger looks gorgeous certainly helps, although it doesn’t appear that this game has any Tegra-exclusive eye candy.

Glowball is a Tegra 3 exclusive, and to be perfectly honest, it feels like a tech demo with tilt controls. This is a pretty good advertisement for the Prime’s underlying engine, though. In addition to beautiful lighting and shadows, Glowball showcases impressive cloth and rigid-body physics effects. If only there were more gameplay.

The version of Big Top THD on our Transformer Prime has a little switch that enables the game’s Tegra 3-specific graphics effects. These only amount to minor improvements in the lighting and shadows, so there isn’t much of an upgrade in visual fidelity. The game does look very polished overall, though.

Sprinkle is the sort of casual game you’d expect to find on a tablet. It works well on the Transformer Prime, which enjoys a smattering of additional effects thanks to its Tegra 3 SoC. The version of the game I played purportedly offers more realistic fluid simulation and animation effects for water, which feature prominently in this fire-extinguishing title. The smoke, haze, and particle effects have also been upgraded for Tegra 3, adding a slickness to the game’s otherwise simplistic graphics. More importantly, Sprinkle is actually fun.

The Tegra 3 appears to have impressive gaming chops, but it will be up to developers to make effective use the chip’s extra cores, the GeForce GPU, and also the built-in controller support. If the games I’ve played are any indication, the Transformer Prime could turn out to be a very sweet portable gaming console. The size and beauty of the screen lends itself to gaming, and the out-of-the-box controller support is a great addition for folks who prefer arcade-style action. All that’s needed is a dual-analog-stick controller that more closely matches the Prime’s slim profile.

The Transformer Prime is quite an upgrade over the original, and Asus knows it. Rather than selling the new model at a discount like the old one, Asus is charging $500 for the base config. Of course, that system offers 32GB of internal storage—the $500 iPad 2 only gets you 16GB. The Prime has no shortage of advantages in other areas, too. In addition to a brighter, higher-resolution display, the Asus tablet boasts a built-in memory card slot and a tiny HDMI output. The Android operating system is less restrictive than iOS if you want to tinker, and the optional keyboard dock is second to none.

While it may seem like an expensive accessory at $150, the keyboard dock is integral to the Transformer’s appeal. The keys feel great, the touchpad offers mouse-like precision, and the full-sized USB port and SD card slot are truly valuable additions. Then there’s the dock’s battery, which adds another six hours to the tablet’s already impressive run time. Good luck finding a tablet-and-keyboard combo that matches the Prime’s capabilities at a lower price.

The Prime has given us our first look at Nvidia’s Tegra 3, and the new SoC brings a lot to the tablet. As novel as the quad-plus-one design is, though, the strangely inconsistent results of our battery life tests suggest there’s room for improvement in how the Tegra 3 manages its multiple cores. Here’s hoping future OS and firmware updates will offer some help on that front. On the plus side, the Tegra 3 is still noticeably faster than its forebear. The chip’s graphics horsepower has also been upgraded substantially, although the Apple A5 processor inside the iPad 2 still has an edge in 3D performance.

Comparisons with the iPad are impossible to avoid because that’s the Prime’s only real competition. Other Android-based tablets lurking around the $500 mark are based on older hardware, and systems like the Kindle Fire don’t really compare. What about the original Transformer? It’s become scarce at online retailers, and given the choice today, I’d drop the extra $100 on the Prime in a heartbeat. The sunlight-friendly screen, additional storage capacity, thinner chassis, and improved performance are easily worth the extra scratch. The Prime’s attractive design does a good job of justifying the higher price, too.

Like most skinny supermodels, the Prime has its share of issues. The aluminum skin appears to be hampering the Wi-Fi antenna, and I’m not sold on the warmer color temperature of the new screen. The touchpad also needs to learn how to cope with inadvertent contact, otherwise most users may end up disabling one of the Prime’s most compelling features.

Those are the big complaints, and they’re easily outweighed by the Prime’s perks. A much bigger problem is availability. All the major retailers are out of stock, and the smaller vendors claiming to have tablets are hawking them for ~$150 above the asking price. The fact that Amazon quotes a lead time of 4-7 weeks for the only regularly priced Prime listed on the site isn’t encouraging. With the iPad 3 rumored for an early 2012 launch, the Prime may soon face more formidable competition.

Right now, I’d rank the Prime ahead of the iPad 2 overall. The race between those two is a tight one, but it’s clear that the Prime is far superior to all the other Android tablets on the market, putting it in TR Recommended territory.

0 responses to “Asus’ Eee Pad Transformer Prime tablet

  1. I’d simply put it down to priorities. The iPad is a Toy, where Android tablets are Tools… “Just enough” mentality on one side, Tegra3 does decode 1080p content, and lots of eye-candy for those who focus on such “glittery” things! 😉

    I know a tablet isn’t a “Productive” tool at this time, but while Android (and x86) are working towards a more all-sided product, Apple has stuck with the “selling points”, ie: Angry Birds etc… (I actually saw that game for the first time this Christmas and my first thought was: this is stupid, haven’t they even seen Tetris?) 😉
    If the iPad3 retains the graphics (mostly at least) from the 2 and focuses on more CPU, that would be a step in the right direction as to my concerns. Not a gamer. I am one of those who wants to be able to tinker though. 🙂

    As it is one of the main selling points though, I do agree that it’s strange that a company like nVidia hasn’t put more into the graphics portion! :-/

  2. Very interesting I’d say! Just as Silus writes, that performance from ten years ago now fits into your pocket! In less than ten years we’ll have the power you speak of in pocket size as well. How cool is that?
    Of course there are always bigger, faster and stronger, I mean, that i7 is a puny piece of dust beside a “real supercomputer”, but then again we’re talking size and price differences that puts it completely out of reach. Today stuff like Microwaves, Fridges, Intercoms and simple Wrist-watches have superseded the Ataris and Amigas of my early teens, not to talk about the Commodores and that horrid Atari2600 console from childhood.
    Every single little application wouldn’t benefit from a powerhouse of a CPU, even if it wasn’t that power-hungry, because just as Kollaps correctly but very malplacedly states, some things just don’t need and won’t use more. But that’s the whole point of this discussion right?

    Strong points of the two different architectures have always differed completely. ARM has been absolutely dominant in applications where “just enough” is the way to go. eg: More specialized electronics and industrial workstations, etc while the x86 has been much more of a multi-tasker able to do whatever you throw at it, making it ideal for use in more flexible applications.

    The way I see it, the really interesting part of this is that ARM is going “mainstream” while x86 is going “specialized”. Two different architectures trying to do the same thing IS very interesting! Hopefully both sides will learn from each other, resulting in better products for us consumers!

  3. No – you got thumbed down on that because your post was way off base, simply demonstrating you don’t really understand what happens at idle.

    You know I get downthumbed for even thinking about opening my mouth. If my rating is over 0, it might just mean I may have a point..

  4. The supposed ARM architectural efficiency advantage [i<]completely[/i<] disappears at idle. That's when power gating and transistor Ion/Ioff performance start to matter... and I'm sorry but that's where Intel has an undeniable advantage because of their superior process technology. You're too focused on that Medfield "idle power" measurement from that benchmark. 1W is ridiculously high - the power management wasn't working right, or the measurement included displays at full blast, WiFi/4G on or something dumb like that. No way in hell a Medfield CPU would consume 1W when idle.

  5. In your universe… check medfield spec. Its possible that medfield will actually have a longer battery life then ARM.

    The sides have turn, ARM is bound to replace AMD as Intel target.

  6. sorry about the double post

    All these benchmarks mentioned here fail to mention one thing about ARM vs Atom, RAM.
    Majority of laptops with atoms have 64 bit 600mhz – 800mhz ddr2 or ddr3 while majority or ARM SoC have 200mhz DDR2

    This seriously hampers ARM cpu’s performance because the ARM cpu ends up twiddling its thumbs half the time while Atom nettops and laptops get to run along without being bogged down.

    [url<][/url<] brightside's benchmarks actually brings this into consideration and show that RAM is ARM's achilles heel. plus the moorestown benchmark is seriously skewed because we don't know what is on this "prototype" other then the cpu.

  7. also the moorestown benchmarks are skewed as well because we don’t what was on this “prototype” beside the optimized atom cpu

  8. Seems to be the case. Using Linpack as the only factor in this market is like putting a Honda Civic on a dyno and comparing it with a Corvette. It’s basically telling us what we already know.

    Just like the aforementioned Corvette, most X86 CPU’s outperform ARM in brute computational power. However, that type of power isn’t needed here. Tablets and Phones live and die on perf/w. Showing us benchmarks on Tbirds, P4’s, SB’s etc is irrelevant if they can’t fit into the form factor. None of them can while giving the performance of what is currently in phones or tablets (however, I believe Intel can do it…. it just chooses not to).

    Medfield is better but it’s still pretty far off. The SoCs in todays devices are around 300 – 400 milliwatt at idle, 1 – 2 w at load. Medfield’s current numbers (give or take) is 2W at idle 3 – 4W at load.

  9. Eh… What are we considering a load? Flash video doesn’t complete faster on a faster processor. The load is sustained throughout the video. A game doesn’t complete faster on a faster processor either. In fact, I’m not sold many of the demanding tasks a person will be doing on a tablet are the type of loads that complete faster thanks to the faster processor. Now there’s other issues to consider, like exactly is the power consumption on the relevant components we’re talking about comparing during a Flash video or game, but regardless those tasks will complete based entirely on the user.

  10. “the best tool for whatever job you’re doing” is not a tablet, no matter the brand.

  11. So it got released just in time (or within 6-8 weeks) to have its teeth kicked in by the iPad 3. The GPU is already only half as fast as the nearly year old iPad 2, with the CPU being faster by about 40-50% than the iPad 2, but thats explained by having more than 2x the cores and a 30% clock speed advantage per core.

    My guess is the iPad 3 with the quad core A6 CPU is going to just stomp on it. Thats not even the worst of it, however, given that iOS is so much more optimized for the hardware in it. Dual core 1.5ghz CPUs on Android are often slower fromm a usability standpoint than the dual core 800mhz CPU in the iPhone 4S as it is.

    That said, some people will choose anything BUT an iPad just because they hate Apple. Thats a shame too, as you’re only impeding yourself by not considering the best tool for whatever job you’re doing. The same goes for people who ONLY consider Apple products as well.

  12. Ah the thumbs down brigade is in full force in this thread too! It doesn’t matter if what’s being said are facts or not. It’s just the “that guy said it, then I’ll rate him down”

    Such pathetic group. It’s kind at the same level of that debacle a while back at imdb, where the Dark Knight fanatics rated Avatar, The Godfather and other movies that were above The Dark Knight in the 250 top movies list, with 1s, just to move up the Dark Knight. So sad…

  13. But this also has implications on security and development effort. I’m not saying there is a right answer here, but there are tradeoffs to consider at multiple levels.

  14. Performance from 10 years ago that was achieved with a huge system, not something that fits in your pocket.

  15. And you’re missing the point too. You’re waaaayyy too focused on the numbers at load….what about idle ? Your smartphone and/or tablet won’t be working at full speed every time and the problem is that what Intel and AMD have now just consume too much power for load and idle systems. ARM is much more power efficient. It’s not even close.

  16. you 100% correct about availability. in shanghai atm. went to asus flagship show room. they sold all 5 that they got allocated. wtf? excellent review btw.

  17. You miss the point, developers are lazy, and they only work until the performance is good enough. The other option, developer is not lazy, but tries to do the best out of a chip, reaching performance limits of the chip doing amazing jobs.

    In any case, as soon as you raise the performance bar, most of chips will be used and power consumption will be high because at high performance, the consumption is high…. So even if it’s real that a task energy approach is the best thing, you have to consider that some applications if not most will stress the CPU as much as possible, taking down any improvement you could see.

    Understand it, it’s like having a GTX580 because you want the low power state of the card in most of your games, since rendering a frame happens faster, but what do you do automatically after buying the card? Increase game graphics quality, so you increase the pressure over the GPU and avoid any advantage the GTX580 may have. This, and also the fact that having the need to dissipate 3 or 5W is definitely a problem for a phone…. maybe it’s “Ok” for a tablet, but a phone that has a chip with 3W for more than 10 minutes will be a piece of fire!

  18. Goes to show that perhaps a native bytecode OS and App is still superior to a JVM 😉

    At least from a performance point of view..


  19. Tegra3 is a fantastic chip: 80mm^2 on 40nm (A5 is 120mm^2 on 32nm), and a perfect balance between CPU and GPU.

  20. You’re missing the point. What dpaus was pointing out is that if the CPU consumes 3x the power but finishes the task 3x faster, the total energy consumed is the same (but the CPU is faster).

    In other words, one shouldn’t blindly look at power consumption numbers but [i<]energy[/i<] consumption numbers. TR does this very well with CPU reviews with the task energy comparisons - that's the [i<]true[/i<] measure of power efficiency. Moreover, non-CPU/GPU-related stuff consumes the same amount, regardless of if it's an ARM-based phone or x86-based phone, so no ARM advantage there.

  21. A good OpenGL-benchmark probably would be Quake3. It is available for iOS and Android.

    A tablet review with a good old demo four-chart would be awesome 🙂

  22. FWIW, the Mflops numbers for Linpack appear to be gated by the JVM software, not the hardware itself. With a native executable version of Linpack, my iPad 2 scored ~880 Mflops. Now, the iPad 2 has dual cores, NEON, and I think ~2X the memory bandwidth of a Tegra 2, but it’s not *that* much faster. 🙂

  23. So, how many Mflops? An intel i7 3960X scores approximately 60 linpack Gflops. I don’t think it uses 1000 times the power or costs 1000 times more. Anyway, I don’t advocate a 3960X for tablet use, but I find it mildly amusing that we are excited today about hardware that has performance from 10 years ago 😉

  24. Isn’t the PowerVR config in the A5 (both in the Iphone 4s and Ipad 2) the SGX543MP2 (dual core, not triple core)? Where do you get the 545MP3 from? Is this for the Ipad 3?

    Anyway, yeah, the reason why the Tegra 3 GPU is somewhat underpowered is due to die space allotted to the extra CPU cores. At 40nm, there just isn’t room/power for both. I expect Tegra 4 to be more interesting in this regard with the shift to 28nm. But then who knows what Apple’s got cooked up for the A6 – any chancee that it’ll be 28nm?

  25. Are we talking here about desktops now?
    For desktops 16:10 is probably the sweetspot: good for multitasking, good enough for 16:9 media and you don’t really care about reading books. 4:3 is better for web surfing and that’s probably all.

    For tablets at home, I’d rather use 4:3. Better for web surfing, much better for reading almost anything and if you want to consume 16:9 media, you have your TV set. I don’t really plan to use a tablet outside, so… that’s all I want: casual gaming, web surfing and reading comic books (for books I have my ereader, my eyes won’t let me use an LCD screen for that).

  26. I’m playing this game called Sword and Sorcery EP on my iPhone and it’s pretty good….


  27. Yeah, I kinda thought that was where they’d shine, much as AMD has done vis-a-vis Atom and the i3/i5. But instead, they got owned by a year-old Apple chip. Very strange….

  28. I’d suggest that battery life is only key up to a point, and that we’re rapidly approaching it (if not already there): when a device can be used (in the manner that’s it’s intended to be used) for a 24-hour-period (during which time most users will have an opportunity to recharge it), then any incremental battery life is only going to be of interest to a very small segment of the market.

  29. Thanks for the explanation (no, seriously), but it’s the technical explanation, and I was looking for the marketing answer to the question “How could Nvidia – a company that claims to be at the very forefront of GPUs (and with good justification) – allow their SoC’s GPU to be so thoroughly pwnd by a year-old Apple chip?

  30. yeah I sighted the only time it was good for anything… 😛 Basically sighting its not good for tablets.

  31. There is a button that disables the touchpad on the keyboard. I too wish it had the logic that most laptops have of automatically disabling the touch pad for 5 seconds when it detects a keystroke, but I know (at least on X11 linux) this has some power management implications and may be something that android can’t support.

  32. [FedExAdvert]”Oh, an MBA… Well, then, I’ll try to explain this for you in little words”[/FedExAdvert]

  33. You forget the aspect of hipsters being only superficially knowledgeable of any subject; be it politics, technology, the environment…

  34. It’s not just the area that matters. Even with equal area efficiency — which isn’t even remotely the case even if you could also assume the same silicon utilisation with the same layout considerations and clocks, which you can’t here at all — you need to take into account many other factors to determine performance in embedded graphics.

    The GeForce ULP and SGX graphics architectures are completely different, and the layout and area considerations of Tegra 3 and A5 for GPU are completely different.

    [i<]Then[/i<] there's software on top.

  35. I had the same notion. It is both impressive that apple actually makes a chip again and that it isn’t the worst thing ever. I’m equally surprised that Nvidia would let the GPU side for the tegra equation slip at all.

  36. 4:3 is useless for media consumption, not optimal for web surfing, not good for multitasking in a desktop environment as you can’t comfortably display apps side by side. 4:3 is only really nice if you want to setup a PLP eyefinity setup.

  37. AS a user I’m not really seeing a good mesh between android and a touch-pad interface. The slider always made more sense to me, while the touch screen keybad on a phone is… manageable the same can’t be said of a tablet and for me one would have to have a physical keyboard for we-browsing to ever be comfortable.

  38. though I see why the android OS appeals to people I would have great use for something with this form factor, a little more horse power and Win 8 in it (x86 not the pointless arm distro) Win 8 could seamlessly switch between a touch OS and a standard PC interface for some light photoshoping when dropped into the doc. Oh the possibilities.

    my dream is this with a doc:

    [url<][/url<] and win 8 or win 7/android dual boot.

  39. Yeah, the iPad 2’s display makes for slightly-too-small text at common widths on the web, too. I still prefer the aspect ratio, but I want a little more size and much higher pixel densities. A power-user’s tablet with an ~11″ display and ~300 DPI would be really compelling…. and 4:3 would be best. 🙂

  40. My EEE PC 901 came with lots or regular updates and support form asus, I expect this to get equal amount of post launch support. I expect them to resolve the touchpad drivers, though the wifi issues are not good news.

  41. +1 Yep. TL;DR for the lazy : more die for the the GPU in the A5 versus the Tegra 3 AND optimized software and drivers make all the difference.


  42. A5 = ~120mm2, with only two cores with the remainder of the silicon mostly for the GPU portion
    Tegra 3 = ~80mm2 with 4 (+1) cores with the remainder of the silicon mostly for the GPU portion.

    It has nothing to do with graphics architecture and very much to do with die are reserved for the GPU. A5 has a huge GPU. Tegra 3 basically has 50% more resources than Tegra’s 2 GPU. Everything else is about the quad-core (+ companion core) and remaining SoC logic.

    Not to mention that it’s always unfair to compare an optimized piece of software for that one chip (A5) and a generic software for Android not optimized for Tegra 3. A fair comparison would be an optimized game for A5 and another version of that game optimized for Tegra 3.

  43. The disparity stems from the fact that the PowerVR GPU is tile based while the nVidia on is an immediate renderer design. Also, the PowerVR drivers are way more mature and optimized.

    It’s also worth noting that the A5 uses a 3 core config for the PowerVR GPU : PowerVR 545MP3 versus other designs using just single or double cores. Some would consider this overkill but i have a feeling that Apple will reuse the same design in the future A6 to drive a retina display for the iPad 3 so GPU needs to be really powerful in order to drive it.

    The A4 was very underpowered for the retina display of the iPhone 4. The A5 fixed that big time.


  44. In a market where battery life is key, you’re really asking why a SoC (System on a Chip mind you, not just an APU (CPU + GPU combo)), must consume the least amount of power, even at load ?

    The gadgets in this market are all dependent on battery life and we are talking small form factors that fit in a pocket. In tablets I’d wager Intel and AMD have some possibility to actually have some decent products, but in smartphones ? Absolutely no way whatsoever…not until they reduce their chips power consumption…by quite a bit.

  45. While I’m no fan of Atom, I have to ask: if it delivers 3x the performance [under load], is it really a problem that it consumes 3x the power when under load, [i<][b<]if[/i<][/b<] (and it's admittedly a very big 'if') the sleep-state power consumption is effectively comparable for real-world purposes (i.e., to give it a standby time measured in days)??

  46. 1.) Given Nvidia’s knowledge of graphics architectures, I’m very surprised at the disparity between the iPad2 and the Prime on the GLBenchmark scores.

    2.) I think the Prime is maturing into a market segment of its own, which, amusingly, is very similar to the Foleo concept that Palm floated back in the day (2008?) If Intel and AMD want to get into this market, I think they’d do well to partner with someone like HP to try building a very-small-and-light convertible device that runs Windows 8 as it’s OS, but ‘docks’ with any of a wide variety of Android phones to give you an Android-within-Windows capability, accessing all of the content on your phone wirelessly (much as they were planning to do with WebOS and desktop systems). The real key, I think, would be to have applications on both platforms that could use the same data, which would be stored on the ‘phone device’ (in quotes because, as phones drift up to 5″ screens and tablets drift down to 7″ screens, it’s getting harder and harder to tell them apart). I don’t want a small collection of devices that are largely incompatible with each other and require either kludgey cable connections or storing everything in the cloud, I want a very small set of devices that – while each purpose-built for their own usage – can seamlessly share [i<]all[/i<] of my data.

  47. That part of the article doesn’t even make sense. Wouldn’t hipsters have been on the Nook Tablet bandwagon since it was the very newest? Now they can all be on the TFPrime bandwagon. Gotta stay ahead of the crowds.

  48. You must have then loved this part of the article:

    [quote<]My time with the Apple tablet has admittedly been limited to fooling around with one a few times at the company's local store, and I could only stand being that close to hipsters for so long.[/quote<]

  49. I believe that the 3d benchmarks are not only skewed, but also near-useless for current hardware/software…as there are next to no games worth playing on ipad or android. ipad has infinity blade and generic 3d shooters, and android has… generic 3d shooters?

    It’s time for AAA developers to start making real (good) games for these things, put them at the $5 mark ($10 for “deep” games), and start really boosting the platform. And I’m not talking about the micro-transaction-whoring junk that is dominating the “top” slots of both markets.

  50. There is also something like a “minimal width” for certain things. Scientific papers for example have a certain format and on a 10″ 4:3 screen the width is enough to show e.g. the ACM’s double column layout. On a 16:9/10 screen with the same diagonal, the width (in portrait mode) would be too small to get readable text (although the increased resolution partially makes up for it). Lots of annoying panning ensues.

    It would be nice to have both form factors available in iOS and Android, so people can choose according to their needs and desires.

  51. I measured the only american comic-book I could find easily in my bookshelf (Asgardian Wars, from Marvel) and the aspect ratio was 1.43. So, for that comic book, 4:3 (1.33) seems better than 16:10 (1.6).

    The books looked less standardized, but then I thought about something more easy than measuring several and calculate the average aspect ratio. There are devices optimized for reading books, aren’t they? And I have one of them, a Sony PRS-600. The screen is 12,3 cm x 9,2 cm. That means the aspect ratio is 1,33.

    Maybe this is too much of a leap of reasoning but if the devices optimized for reading are 4:3 (when it would be cheaper to buy 16:10 screens with the same inches), maybe it’s because 4:3 is better for reading than 16:10.

    Anyway, this is a matter of what you like, there are no absolutes (i.e. I use my ereader in landscape mode…). I’d just like to see reviewers pointing that 10″ are less in 16:10 than in 4:3. The Prime is 4% bigger in inches and just about 1% bigger in square inches than the iPad. This effect is what is killing 16:10 monitors (16:9 are cheaper to produce and people buys inches). And I really like my vertical space :-).

  52. Not until they lower their power consumption numbers. Tablet and smartphone markets need SoCs that consume just tiny bits of power. ~1W or so, not what Intel or AMD smallest chips consume, which is 3 or more time that number.

  53. Unfortunately they didn’t pay even a little bit of attention to the antenna design. I thought placing wifi antenna behind Faraday cage (iPad 2) was bad enough…

  54. Anyone looking for one Bestbuy just posted the 32gb grey one, get one quick! [url<][/url<]

  55. Most of what I end up viewing tends to look better to me in a portrait-oriented 16:10, since it approximates a page for content that I’m used to seeing on dead trees cut with a more rectangular aspect ratio. And on the web, the extra length lets me see more of what I want to read–zoomed or otherwise. Text tends to be arranged in taller, narrow columns. Maybe I’m looking at it wrong, but the iPad’s aspect ratio hasn’t felt more natural to me in the time that I’ve spent with the tablet.

    As an aside, there’s little difference in the actual physical area of the iPad 2 and Prime screens. If my math is right, the iPad 2 is 5.82 x 7.76″ and the Prime is 8.56 x 5.35″, which makes the Prime’s screen larger by only 0.66 square inches.

  56. Ah the lack of polish that is typical of generic PC manufacturers. It didn’t surprise me one bit reading about it in the article. It’s small things like these that i appreciate so much on Apple gear. That right there would be a deal breaker for me as it would annoy me to no end.


  57. Damage, I almost shed a little tear reading your post. This is why TR is still my homepage after all these years. You are able to cut through the RDF, the misconceptions and other various media stereotypes without compromising your, and by extension – TR’s integrity.

    BRAVO !


  58. For me it will be a toy or maybe better said a gadget. It will definitely come in handy when traveling. I don’t have a personal laptop and try not to contaminate my work laptop. I foresee this being easier to lounge with, as a tablet, vs laptop. I’ll probably get the keyboard/dock just for the extra battery and SD slots and used mostly when traveling. I guess I’m the target audience who is willing to just blow $500 on a gadget! 🙂

  59. Thanks for the review. Good read! Now I want one even more. 🙂

    I’m hoping to see a brief update to the review once the ICS (ice cream sandwich) update is released for it.

  60. That would be true with a laptop, where resolution is a key determinant of what gets displayed, but it’s not with tablets, where fluid image scaling is essential to the experience. I find that my iPad 2’s 4:3 display is much more fitting for the web and most other uses outside of 16:9 videos than my Galaxy Tab 10.1’s 16:9 display, even though the Tab has a higher vertical resolution (800 vs 768). In fact, the advantages of being closer to square apply to both portrait and landscape modes and are way more pronounced in regular use than one might think. Even the onscreen keyboard works better on the iPad due to more key spacing vertically, and that has nothing to do with the number of pixels involved.

  61. The aspect ratio seems to be the wrong thing to focus on. If you want the usable screen real estate, the resolution is the most important metric, since it defines how much you can actually draw on the screen. That’ll give you the aspect ratio, too… or at least a sense of whether it’s 4:3 or 16:9/10.

    If you don’t see utility for 16:10 beyond videos, try rotating it. 10:16 is a great aspect ration for reading, whether it’s books, web sites, or graphic novels.

  62. Did some experimenting, and it’s a tough thing to capture on camera, at least with my setup. The effect is definitely subtle. I can see how dimming the display might come across as dulling the colors, though.

  63. You should also check out Zen Pinball THD if you have a Tegra 2 or Tegra 3 system. They make the totally amazing Pinball FX games for Xbox 360 and Zen Pinball for PS3. Physics, graphics, table design – all top-notch. The fact that Zen Pinball THD is FREE on the Android Market is just another reason to dive in.

    [url<][/url<] Also, if you don't mind unofficial scores, I can run any of the benchmarks you want on my Kindle Fire either in CyanogenMod 7 or in the stock Kindle Fire 6.2.1 firmware, rooted to give me Android Market access. Just let me know what you'd want. I'm sure this device blows it away, regardless.

  64. Sold my Transformer today for only a $40 loss from when I bought it 6 months ago, got a new Prime on backorder, hope it ships soon.
    Has Asus given any timeline on when substantial quantities will be released to the marketplace?

  65. The “screen into a power-efficient operating mode” is a Tegra 3 feature,they dynamically adjust brightness (lower) and compensate by adjusting everything else while trying to display the same colour to the eye and save power.Since you don’t like the colours in this mode any chance we can see some screen grabs in normal and power saving modes?

  66. Just a suggestion: You should state the aspect ratio of the screen and what that means in actual real estate and usability for different workloads. Going 16:10 is a way to sell “more” inches with less real state than 4:3. It’s a (successful) marketing trick to mislead the costumers and selling less screen for the same money (hey, it’s bigger than the iPad, it has moar inches!!!).

    And except for 16:9 videos, 4:3 is better for everything I can think of. That should be factored and commented (and maybe we could end up with 4:3 alternatives outside de iThingies World).

  67. Lessee here:

    [b<]Linpack[/b<] Linpack score using all 4 cores for Tegra 3: 71 MFLOPS Linpack using a single-core Atom N455: 89 MFLOPS (64-bit non-optimized) or 226 MFLOPS (64-bit optimized) Source: [url<][/url<] (As an aside, the 71 MFLOPs multi-threaded for Tegra 3 is between the non-optimized scores of a 1 Ghz P3 (55) and a 1 Ghz Athlon T-Bird (81), both of the older chips whomp Tegra 3 when optimizations are activated though) [b<]Sunspider[/b<]. Using a prototype of Moorestown at 1.5 Ghz, Intel scored 1500 ms while Tegra 3 is at 1700ms in normal mode. Source (that is actually calling the Intel chip sub-par): [url<][/url<] Note: The Linpack listed on the second article is a much lower score, likely due to the fact that it was run through a non-optimized OS, while Tegra 3 takes full advantage of Android. The first scores posted for Linpack show the Atom's true capabilities more accurately. [b<]Peacekeeper[/b<] Tegra 3 score: 489 at full speed. Dual-core Atom netbook running full-blown Linux or Windows recorded scores of anywhere from 1100 --> 2200 depending upon the browser that was being used. Divide all those scores by 2 to get back to a single-core system (P.S. --> This is being unrealistically harsh to the Atom system since Peacekeeper is only multithreaded in some tests): you have a lower score of ~550 and a higher score of ~1100. Bear in mind, these scores are from regular full-blown browsers running on top of a full X11 or Windows environment on the Atom. If Google hasn't screwed Android up too much, there should be a software efficiency advantage for the Tegra 3 inside of the Transformer Prime. source: [url<][/url<] Moral of the story: If a single-core Medfield at 1.6Ghz performs the same as other known Atom chips at 1.6Ghz, then Intel WILL have a competitive chip for use in tablets.

  68. Great review. Extremely comprehensive. And yes I don’t own a tablet and do the same thing as you do Geoff and start counting how many I see with keyboard in tow.

    Surely the issue with the internet web browsing battery consumption can get a fix rolled out by nvidia? (if that is the issue)

    $500 is still a lot of money when the real competition is probably between a smartphone and a laptop. I just still can’t comprehend when I’m meant to be using this device. An iPhone/Android phone and a laptop can get you by in almost any situation.

  69. Where can I get a Megatron tablet? (J/K)

    The Tegra 3 looks like a decent combo of CPU + GPU for this class of tablet. It’ll be interesting to see what kind of improvements Ice Cream Sandwich can make when coupled with newer hardware like this.

  70. The original Transformer’s touchpad was just as annoying about not detecting accidental touches, and I’ve sworn at it many a time before finally giving up on it. Sounds like I should give this one a miss.