Texas Instruments Inc. has been talking the OMAP5 talk for some time already, but on Monday (June 4) the firm fired its first serious shot at the competition, releasing a graphics benchmarking video to show off its SOCís GPU performance.
The competition? Seemingly an iPad 3 with an A5X chip (45-nm process), which boasts two Cortex A9s alongside SGX543MP4 graphics.
TIís OMAP 5430, on the other hand, is designed on a 28-nm process and sports two ARM Cortex A15s and two Cortex M4s alongside SGX544 MP2 graphics.
TI used GL Benchmark 2.5 to render complex graphics scenes and simulate high-end games at full HD (1080p) resolution, to show off its big dual-core GPUs.
Unsurprisingly, the yet-to-be-released OMAP 5 beat the already available "market-leading tabletĒ competitor in terms of frame-rate (38 FPS on-screen, 45 off-screen, compared with 34 and 43 FPS), probably thanks to some souped up clock speeds.
You can check out the video from TI to see the details below, but suffice it say the OMAP 5 looks fairly fast. That said, comparing an already released product to one that has yet to make its debut in any device is not entirely fair, so donít be too quick to judge.
It was unusual of Apple to talk specifically (and comparatively) about their GPU performance in the new iPad. Something, I think, Jobs would have avoided due to objections and follow-on comparisons. The tablet battle has very little to do with hardware performance. Indirectly a poor performing iPad would be bad for Apple but having sufficient performance is what its about.
Hmmm. It better have some pretty compelling price and power benefits then. If you're only getting 12% upside in visual performance comparing sample silicon against an off-the-shelf device that's been shipping for several months (and hence has had its silicon designed in some time last year)... If the OMAP comes with a ~25-30% power reduction too, that's more interesting.
Nice, but what *really* matters here is opening up some of this GPU/DSP compute power to developers to use beyond just games & 3D demos, which will never rival what people come to expect on desktop.
For instance, video apps that can transcode/resize HD video while adding animated tiling & transition effects at equal or near realtime speed (see iMovie on iPhone4, for instance) and photo apps that can work with 8 megapixel photos without keeping user waiting longer than a few seconds (again, several iOS apps manage this).
Android could use a bigger push in this direction but LLVM/Renderscript are a potential start.
IPad3 is already a mature consumer product while OMAP5 just engineering sample. It is long way to go before it is even in any product. At that time, I will be surprised that Apple will not have newer version. Plus, I will only take seriously if this information from 3rd party. From TI, I don't think that meaningful. How do I know TI didn't overclock its OMAP5 in freezer while put IPads in low power mode to make comparison.
All Silicon vendors are squeezing IP to increase margin while Apple could implement Silicon at own wish without margin pressure. Unless you develop own Silicon like Samsung does, I don't believe there is ANY chance Silicon Vendors could compete graphic performance with Apple any time in near future. Not believe me, take a look of A5x die picture. You will see how much silicon Apple is willing to invest on Graphic Performance.
Everyone seems to be ignoring the fact that this was a comparison between a dual core SGX544 and a quad core SGX543. The point is that TI's next generation cell phone/tablet SoC will kill current top of the line tablet SoC's in CPU capabilities, while surpassing them in GPU capabilities. A 4 core cortex a15/4 core SGX544 SoC would be one hell of a low power htpc chip.
Why is it, that this chip has been delayed for so long, after originally being annonced in July 2011, twenty-eleven.... and the shiping date keeps on getting pushed back.
Now estimated that devices using it, will hopefully come out by Q1/2013... some 18+months after being announced!
Shees... maybe T.I. should announce important chips about 3 months to 6 months MAX before availability.
This was just oh sooo wrong!
Oh well, I hope Omap 6 will be way quicker to market!
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for todayís commercial processor giants such as Intel, ARM and Imagination Technologies.