Comments Locked

19 Comments

Back to Article

  • duffman55 - Monday, June 4, 2012 - link

    It's good to see so many competitors pushing the performance and efficiency envelope for mobile SoCs. I'm curious to see what the power consumption on the OMAP5430 is since it packs so much power, but is fabricated on a 28 nm process.

    Could you tell us about the Cortex M4s?
  • dagamer34 - Monday, June 4, 2012 - link

    Good to see everyone bringing their A game finally to the mobile space. Since Intel has no real competition in AMD anymore, it seems Intel ha been resting on its laurels since the Core 2 Duo came out nearly 6 years ago. But when Microsoft announces that Windows 8 will support ARM, Intel kicks it into high gear with better GPUs and a huge focus on low power chips.

    I love competition!
  • kyuu - Monday, June 4, 2012 - link

    What are you talking about? Resting on their laurels since C2D? In what universe? Sandy Bridge was a huge step forward from C2D. Ivy Bridge is a lot less impressive, but it was just a die shrink for the most part.

    And anyway, I don't think ARM is putting any pressure on Intel's high-end x86 CPU market share.
  • VoraciousGorak - Monday, June 4, 2012 - link

    Competition is good, but this article is about TI chips. Not sure where Intel is coming from.
  • mavere - Monday, June 4, 2012 - link

    Maybe Apple will use a 544MP4 in next year's iPad. Otherwise, I'd be disappointed with only a 5% speed increase (as seen in the off-screen results).
  • Ryan Smith - Monday, June 4, 2012 - link

    The 544 is just a Direct3D compliant variation of the 543. There are not (to the best of my knowledge) any hardware changes that give it a higher performance per clock or increase the clockspeeds that it can reach. Whatever higher clocks TI gets would be as a result of the 28nm process.
  • iwod - Monday, June 4, 2012 - link

    Basically a 28nm, Next Gen CPU Architecture Cortex A15, Plus Faster GPU per Core is slightly faster then the current 40nm Design with Older CPU A9?

    Honestly i dont see much point in that. Unless the power usage is much lower.
  • tuxRoller - Monday, June 4, 2012 - link

    First, this was running gl benchmarks so we can assume the cpu shouldn't be th bottleneck, hence, the newer cores might not help much. Second, the ipad has 4 gpu cores to the Omap5's 2 gpu cores (albeit of a slightly more recent generation of gpu, though not necessarily much more powerful...I think those come when imagination releases their 600 series).
  • ssj4Gogeta - Monday, June 4, 2012 - link

    The CPU should be significantly faster.
  • Mike1111 - Monday, June 4, 2012 - link

    I don't know, that's all nice but where are the (retail) tablets with OMAP5430? Nothing's even announced yet, so it's probably at least another three months. Beating an at least 6 month old GPU by 5%-10% is not exactly awe inspiring, especially when using 28nm compared to 45nm.
  • duffman55 - Monday, June 4, 2012 - link

    You can't just look at graphics processing power. You have to consider power consumption, cost of the chip, etc. Increasing the GPU performance significantly over the A5X would likely be pointless. GLBenchmark doesn't exactly give you a complete picture of what a particular SoC is capable of either.
  • Mike1111 - Monday, June 4, 2012 - link

    Of course you're right. But Anandtech's article was only about the GPU performance, so was my comment.
  • Torrijos - Monday, June 4, 2012 - link

    Exactly!

    This only proves how invested Apple is in bringing performance hardware months before the competition.

    Still another point of contention with this kind of comparisons...
    The iPad 3 is an actually produced hardware device that has to run in a metal enclosure and as such has been tuned to so its thermal output doesn't destroy the device (or burn the owner).
    What about the TI test platform? Most of the time it's a bare motherboard with plenty of air circulation that allows them to clock everything a little bit higher, and still they only manage to barely beat the iPad.
  • Roland00Address - Monday, June 4, 2012 - link

    What the big deal is that they are beating the Ipad 3 GPU with a new GPU that uses the same architecture but with half the shaders. The GX544 MP2 graphics and SGX543MP4 are the same silicon cores but the GX544 is direct x compliant. The MP2 stands for dual core (shader) graphics vs quad core (shader graphics). They are able to achieve this higher performance with just clocks alone.

    Furthermore it is not just the GPU that increased performance with the die shrink, they went from a Cortex A9 design to a Cortex A15 design. Cortex A15 dual cores should perform better than Cortex A9 quad cores. See here for video
    http://www.anandtech.com/show/5571/ti-posts-omap-5...

    In that video they have a comparison of two android 4 tablets doing the following at once as a benchmark
    1) Loading 20 webpages where you can't start the next one till the first one completely loads
    2) Download two videos all the while loading those 20 webpages
    3) Playing a mp3 simultaneously
    The cortex a15 dual score completed the test in 95 seconds compared to the quad core A9's in 201 seconds.
  • hamsteyr - Monday, June 4, 2012 - link

    I believe that Anandtech had an article about this before, about the GPU's in Mobile devices such as this.

    Indeed, people should be excited, even if the improvements don't look like much. Why? Simple.

    The SGX544MP2 in the OMAP5 is a Dual Core GPU
    The SGX543MP4 in the A5x/ PS Vita is a Quad Core GPU.

    While it may not seem like much, what you're effectively seeing is a Dual core besting a quad core GPU, which is certainly big news.

    Core for core, that seems like a lot more power.
  • aryonoco - Monday, June 4, 2012 - link

    TI sure loves comparing their upcoming (sometimes more than a year before release) products from "competitors" which are currently on the market.

    Fact is, since I started paying attention to the SoC space a few years a go, TI has never been the performance leader. They've been talking about OMAP5 and promising its breakthrough performance for a long time now, and yet there are no products even announced with it and it's probably at least another 7-8 months away from showing up on shelves.

    Right now, their top-of-the-line product is the OMAP 4460 and that loses every performance benchmark to the likes of A5X and Krait and even Tegra 3 and the new Exynos. Which means, performance-wise, they are behind all their major competitors.

    Not to mention confusing product launches. What was the point of 4460 and how is it different from 4430 anyway? 4460 was supposed to have ~20% higher clock speeds, but other than Archos (i.e., a niche manufacturer) I don't know of any shipping product where it was released with that clock. Evey manufacturer just clocked it at 1.2 Ghz (presumably cause it sucked too much power at 1.5 Ghz) which put it exactly on par with 4430. What was the point of it then?

    Don't get me wrong, I love TI very much. They are a very open source friendly company, they are involved in Linaro and a bunch of other initiatives and I respect them for that. But the SoC space is a tough one, they have been winning designs solely based on price, as their performance hasn't been competitive in a long time. This strategy hits the bottom line hard and can't continue forever.
  • thunderising - Monday, June 4, 2012 - link

    I was hoping for A15 with A7 couple and Power VR Rogue Series

    Seriously where are they
  • jalexoid - Tuesday, June 5, 2012 - link

    They might be utilizing the M4 cores to assist graphics as well.
  • dagamer34 - Wednesday, June 6, 2012 - link

    Ti told us 3 months ago that the clock speed of the PowerVR GPU is up to 532Mhz: http://newscenter.ti.com/index.php?s=32851&ite...

Log in

Don't have an account? Sign up now