The Mali-400

Now that we've settled the issue of what type of GPU it is, let's talk about the physical makeup of the Mali-400. The Mali-400 isn't a unified shader architecture, it has discrete execution hardware for vertex and fragment (pixel) processing. ARM calls the Mali-400 a multicore GPU with configurations available with 1 - 4 cores. When ARM refers to a core however it's talking about a fragment (pixel shader) processor, not an entire GPU core. This is somewhat similar to NVIDIA's approach with Tegra 2, although NVIDIA counts each vertex and fragment processor as an individual core.

In its simplest configuration the Mali-400 features a single combined geometry front end and vertex processor and a single fragment processor. The 400 is also available in 2 and 4 core versions, both of which still have only a single vertex processor. The two core version has two fragment processors and the four core version has four fragment processors. Note that ARM decided to scale fragment shading performance with core count while keeping vertex performance static. This is likely the best decision given current workloads, but a risky one. NVIDIA on the other hand standardized on a 1:1 ratio between fragment and vertex processors compared to ARM's 4:1 on a 4-core Mali-400. The 4-core Mali-400 MP4 is what Samsung uses in the Exynos 4210.

ARM, like Qualcomm, isn't particularly interested in having the details of its GPUs available publicly. Unfortunately this means that we know very little about the makeup of each of these vertex and fragment processors. I suspect that both companies will eventually learn to share (just as AMD and NVIDIA did) but as this industry is still in its infancy, it will take some time.

Earlier documentation on Mali revealed that the GPU is a VLIW architecture, meaning each processor is actually a collection of multiple parallel execution units capable of working on vector data. There's no public documentation indicating how wide each processor is unfortunately, but we can make some educated guesses.

We know from history that AMD felt a 5-wide VLIW architecture made sense for DX9 class games, later moving down to a 4-wide architecture for DX11 games. AMD didn't have the die constraints that ARM and other SoC GPU suppliers do so a 5-wide unit is likely out of the question, especially considering that Imagination settled on a VLIW4 architecture. Furthermore pixels have four color elements (RGBA), making a VLIW4 an ideal choice.

Based on this as well as some internal information we can assume that a single Mali fragment shader is a 4-wide VLIW processor. The vertex shader is a big unknown as well, but knowing that vertex processing happens on two coordinate elements (U & V) Mali's vertex shader is likely a 2-wide unit.

Thus far every architecture we've looked at has been able to process one FP16 MAD (multiply+add) per execution unit per clock. If we make another assumption about the Mali-400 and say it can do the same, we get the following table:

Mobile SoC GPU Comparison
  PowerVR SGX 535 PowerVR SGX 540 PowerVR SGX 543 PowerVR SGX 543MP2 Mali-400 MP4 GeForce ULP Kal-El GeForce
SIMD Name USSE USSE USSE2 USSE2 Core Core Core
# of SIMDs 2 4 4 8 4 + 1 8 12
MADs per SIMD 2 2 4 4 4 / 2 1 ?
Total MADs 4 8 16 32 18 8 ?
GFLOPS @ 200MHz 1.6 GFLOPS 3.2 GFLOPS 6.4 GFLOPS 12.8 GFLOPS 7.2 GFLOPS 3.2 GFLOPS ?
GFLOPS @ 300MHz 2.4 GFLOPS 4.8 GFLOPS 9.6 GFLOPS 19.2 GFLOPS 10.8 GFLOPS 4.8 GFLOPS ?

Based on this estimated data alone, it would appear that a four-core Mali-400 has the shader compute power of a PowerVR SGX 543. In other words, half the compute horsepower of the iPad 2's GPU or over twice the compute of any smartphone GPU today. The Mali-400 is targeted at 275MHz operation, so its figures are likely even higher than the competition. Although MADs are quite common in shader execution, they aren't the end all be all - we need to look at application performance to really see how it stacks up.

Understanding Rendering Techniques GPU Performance: Staggering
POST A COMMENT

132 Comments

View All Comments

  • numberoneoppa - Wednesday, September 14, 2011 - link

    Guys, that mysterious notch you write about is not for straps, it's for phone charms, and it's arguably my favourite feature of samsung phones. (In korea, phone charms can be used for more than just cute things, one can get a T-money card that will hang here, or an apartment key). Reply
  • Tishyn - Wednesday, September 14, 2011 - link

    I spend hours every week just browsing through reviews and tests comparing devices and vendors. This is one if the most interesting and most comprehensive review I've read for a veery long time.

    I especially enjoyed the rendering part and how it relates to the ultra mobile device market. Thumbs up!
    Reply
  • milli - Wednesday, September 14, 2011 - link

    Brian / Anand, why are you so reluctant to test chips from this company? ZiiO tablets, sporting the ZMS-08, are available for a while now and i'm sure Creative would send you the new Jaguar3 tablet (ZMS-20) if you guys would ask for it.
    The ZMS-20 has 26 GFlops ... faster than anything you've tested till now. The ZMS-40 coming in Q4 doubles that number!
    I'm an old school IT technician and I for one don't understand your lack of interest. The GPU's in these chips are based on technology that Creative acquired with the 3DLabs purchase.
    Reply
  • rigel84 - Thursday, September 15, 2011 - link

    Just a quick tip: You can take a screenshot by pressing the power and home button at the same time.

    If you double tap your home button it will bring the voice talk feature.

    While watching video clips just press the power button to disable the touch sensitive buttons.

    Swipe your finger to the left on contact name to send him a message
    Swipe you finger to the right on the contact name to dial the contact.

    To see all the tabs in the browser just pinch inside twice :)

    If you experience random reboots when you drop it on the table, or if you are leaning towards things or running, then try to cut a piece of paper and put it under the battery. It happens because the battery shortly looses connection to the pins. If you check XDA you can see that many people has this problem, and I had it too. I was experiencing many random reboots whenever I had it in my pocket, but after I pit a piece of paper below the battery they all disappeared.

    A few things...
    - GPS is horrible if you ask me. Unless I download the data before with gps-status then it takes ages. Mostly 15-30 seconds with 2.3.3 (no idea if the radio got updated in the release)
    - Kies AIR is HORRIBLE! It's on pair with realmedia's real player from 10 years ago. Crash on crash on crash and sluggish behavior.
    - I don't know whether it's the phone or not, but I've been missing a lot of text messages after I got my Galaxy S2. I'm on the same net, but along with the poor GPS reception I'm suspectiong the phone :(
    - There is a stupid 458 character limit on textmessages, and then they are auto-converted to an MMS message. There is a fixed mms.apk on XDA (requires root) or you can download something like Go SMS Pro (still free) on the market, which removes this stupid limit.
    Reply
  • ph00ny - Thursday, September 15, 2011 - link

    Odd

    I haven't seen any posts about the battery disconnect issues and if you've been browsing the xda forum, probably saw my thread about dropping my phone on concrete twice...

    As for Kies AIR, i've used it twice and my expectation was low to begin and it wasn't that bad. Some things were definitely slow but it's a good start

    -GPS for me has always been solid. I even used it on multiple trips in less than ideal location, not a single glitch even with shoddy cell reception.
    Reply
  • ciparis - Tuesday, September 27, 2011 - link

    I've been using Sprint's SGS2 (Epic 4G) for less than a day, but already there are some annoying points which I'm surprised aren't mentioned in this review:

    1) The digitizer lags behind finger movement.
    In the web browser, when your finger moves, there is a disconnected rubber-band effect before the screen catches up with your finger. This is visible in the browsing smoothness video as well, and it's very noticeable in actual use. Coming from an iPhone 4, it feels cheap and broken.

    2) Back/Forward navigation often ignores the previous scroll point.
    If you spend some amount of time reading a page you arrived at from a link (it seems to be about 10 seconds or so), hitting back doesn't take you back where you were previously reading from -- instead of returning you to the page position where the link was, it drops you at the top of the page. This makes real web usage tedious. On the Sprint, the timing seems to be related to when the 4G icon indicates sleep mode: hit back before the radio sleeps and you are returned to the right spot. In actual use, this rarely happens.

    3) The browser resets the view to the top, even after you've started scrolling.
    When loading a page, there's a point in which the page is visible and usable, but it's technically still loading (which can go on for quite a awhile, depending on the page). It's natural to start reading the page and scrolling down, but typically the phone will randomly jerk the scroll back up to the top of the page, sometimes several times before the page is done. This is unbelievably annoying.

    I suppose expecting an Apple level of polish prior to release is unrealistic, but Samsung seems hell-bent on positioning themselves as an Apple-level alternative; even the power brick looks like they took the square Apple USB charger, colored it black, and slapped their logo on it. The point being, they're inviting direct comparison, and it's a comparison their software team isn't ready to deliver on -- certainly not out of the box.
    Reply
  • ciparis - Tuesday, September 27, 2011 - link

    How are you supposed to use this phone if the keyboard is covering up the text fields, there's no "next" button to get to the next field, you can't see what you're typing, and there's no button to make the keyboard go away?

    Case in point: go to Google News and click on Feedback at the bottom of the page. There's no scrolling room at the bottom, so the keyboard obscures the fields; I was unable to send feedback to Google that their news site was opening every link in a new bowser window on a mobile phone (...) despite my account having the preference for that set to "off", because I couldn't navigate the form fields.
    Reply
  • mythun.chandra - Wednesday, September 28, 2011 - link

    Just realized there are no numbers for the Adreno 220 in the GLBench 2.1 offscreen tests...? Reply
  • sam46 - Saturday, October 01, 2011 - link

    brian,please tell me which one of these smartphones is the best.i wanna purchase one of them so,pls help me in deciding. Reply
  • b1cb01 - Wednesday, October 05, 2011 - link

    I love the green wallpaper on the first page of the review, but I can't find it anywhere. Could someone point me to where I could find it? Reply

Log in

Don't have an account? Sign up now