Tag Archives: A6X

Samsung Exynos 5 Octa SoC Uses PowerVR GPU Instead Of ARM’s Mali GPU

One of the highlights of this year CES was the announcement of an Octa-Core mobile SoC from Samsung – the Exynos 5 Octa. This absolute beast of an SoC makes use of ARM’s big.LITTLE implementation and packs 4 Cortex-A15 based cores, and 4 lower-power Cortex-A7 based cores, manufactured on the 28nm HKMG process.

The use of ARM’s big.LITTLE technology will allow this SoC to provide the best of both worlds – performance and battery life. On the GPU side, it was assumed that Samsung went with ARM’s Mali GPU, which was found inside the Galaxy S II and the Galaxy S III.

However, it looks like Samsung may have had a change of heart and has instead used PowerVR’s Imagination GPU on its latest SoC. AnandTech has confirmed from multiple sources that the Exynos 5 Octa SoC packs in three cores of PowerVR SGX543 GPU, with each core clocked at 533MHz.

This is the same GPU that Apple uses in its A5 and A5X SoC, with the latter being found inside the third generation iPad. Considering the PowerVR SGX543 inside the Exynos 5 Octa SoC is clocked so high, it should be able to beat the 3rd generation iPad in benchmarks. However, it won’t be able to surpass the performance of the A6/X SoC found inside the iPhone 5 and the fourth-generation iPad.

There is a very strong possibility that Samsung will end up using the Exynos 5 Octa SoC on its upcoming Galaxy S IV smartphone.

First Look At Tegra 4’s Development Board Performance In GLBenchmark

If you follow the mobile industry closely, you will be pretty much aware of the fact that whatever GPU Apple uses in its iOS devices is usually the fastest, which is then followed by Samsung’s Exynos/ARM GPU. While Qualcomm GPU’s never really were top-performers, the release of Adreno 320 GPU with the S4 Pro SoC changed all that. The Adreno 320 gives tough competiton to the ARM GPU in Samsung’s Exynos SoC and even manages to beat Apple’s A6 SoC in some benchmarks.

Yesterday, when Nvidia unveiled the “world’s fastest mobile” SoC, Tegra 4, with quad-core Cortex-A15 CPU and 72 GPU cores, it boasted of the fact that its mobile SoC is faster than even Apple’s A6X SoC found inside the iPad 4. Given Nvidia’s past, I was skeptical about their marketing and I might just be right.

Thanks to some terrific investigation by an XDA member, Turbotab, we may have got our first look on how the Tegra 4 performs in real-world benchmarks or at least GLBenchmark. In an offscreen Egypt HD benchmark at 1080p resolution, Nvidia’s Tegra 4 based developer board managed around 32.6FPS, while the iPad 4 managed a whopping 49.6FPS.

Now, even if we consider that development boards are around 10-20% slower than the ‘retail’ units, the Tegra 4 will still end up playing catch up with the A6X inside the iPad 4.

It was earlier rumored that Nvidia will unveil the Tegra 4 with its new unified Kepler based GPU core, which is similar to its desktop GPUs. However, AnandTech has confirmed that Nvidia has not used Kepler GPU inside the Tegra 4.

Head over to this XDA thread to read about the stunning investigative work done by Turbotab. 

P.S. – All the above information may just turn out to be complete BS, but taking into consideration Nvidia’s past, I am highly doubtful about that. In Nvidia’s slides, the company presumably compared the quad-core Cortex-A15 CPU of Tegra 4 to Apple’s custom Swift cores, and may have intentionally left out the GPU part.