Many people have always believed that Apple "doesn't talk about parameters" and doesn't pursue hardware configuration. In fact, Apple has been making breakthroughs in SoC performance. The newly released iPhone 6/iPhone 6 Plus uses a new A8 processor. Although it is still a dual core architecture, its performance has been greatly improved compared with the previous generation of products. What "material" has Apple piled on A8 this time?
CPU
As we know, Apple introduced the 64 bit architecture on the A7 processor for the first time. The new processor brings terrible single core performance. Although the main frequency is only 1.3GHz, its performance is comparable to that of the Android mobile phone's 2.0GHz processor. When it comes to A8, Apple claims a 25% increase in CPU.
Before the launch conference, @ Ruiji, a microblog user, had already announced the configuration information and running screenshots of iPhone 6 on the "microblog launch conference" of iPhone 6. The internal code of iPhone 6 is iPhone 7,2, the CPU frequency is 1.38GHz, and the memory is still 1GB.
From the score, the improvement of iPhone 6 over iPhone 5S is also in line with Apple's official promotion. Compared with Galaxy Note 4, which uses 32-bit Android 4.4.4 system, it still has considerable advantages. From this score, it is almost certain that A8 is still a dual core architecture, and may use an improved version of the Cyclone architecture. From the perspective of the main frequency, the iPhone 6 is only 100MHz higher than the iPhone 5S, and the 8% increase in the main frequency brings about a 25% performance increase. It can be seen that the new architecture has brought stronger co frequency performance.
GPU
Every new generation of Apple's SoC will bring a very significant GPU performance improvement. Apple claims that the GPU performance has been improved by 50% (usually twice). The previous generation A7 processor was paired with the Imaging PowerVR G6430, which was the first tier in graphics performance last year.
and In Apple's developer documentation, it was pointed out that the A8 supports ASTC technology, which is a feature of PowerVR Series6XT. According to the extent of improved performance, Apple is likely to insert the PowerVR GX6650 (of course, it may also use the high-frequency version of PowerVR GX6450) into the A8 this time, which is crazy.
The GX6650 is the strongest GPU in Imagination's active service, belonging to the PowerVR Series6XT series, and has a total of 192 FP32 ALUs cores (according to the propaganda of domestic manufacturers, this can be regarded as a "192 core" GPU). This specification is the same as the current mobile GPU overlord Tegra K1. At 300MHz, both Tegra K1 and GX6650 can achieve the FP32 floating point operation capability of 115.2GFLOPS, which is quite powerful. In fact, the second digit of the model number is 6, which means that it has 6 USC units. In fact, it should be regarded as "six cores". Similarly, G6430 on A7 should be regarded as four cores, and G6200 on MT6595 should be regarded as two cores.
In addition, one core of PowerVR Series6XT has two more FP16 ALUs than that of PowerVR Series6. One core has two FP32 ALUs and four FP16 ALUs
The PowerVR Series6 is an ALU with two FP32 and two FP16
After determining that the A8 uses PowerVR Series6XT, there is a dispute about whether it uses GX6450 or GX6650.
Although the A8 uses the advanced process of 20nm, its core area has also reached 89mm ^ 2. It is also possible for the A8 still using dual cores to accommodate the next GX6650 in this area. Of course, the GX6650 is definitely not a fuel efficient lamp. Apple may also want to consider the power consumption problem and insert a high-frequency GX6450 into the A8. We will unveil the graphic performance of A8 in the real machine test on September 19.
The tables used in the text are from
