CPU
Now Reading
AMD could make a custom X86 SoC for Apple’s 2017 iMac — detailed analysis
0

AMD could make a custom X86 SoC for Apple’s 2017 iMac — detailed analysis

by Muhammad FirdausOctober 21, 2015
What's your reaction?
Me Gusta
0%
WOW
0%
Potato
0%
Sad Reacc
0%
Angery
0%

On the AMD Q2 2015 earnings teleconference, Lisa Su announced a third semi-custom design in the pipeline and again on latest the Q3 2015 earning teleconference she announced that one of these three new design will be generating revenue in the second half 2016.

The claims led to many speculations but the most promising is that there’s a possibility that AMD is making a custom X86 SoC chip for the upcoming Apple iMac 2017-2018. The reason why it is highly probable is because AMD and Apple has been working together to provide a graphic solution for the latest Apple MacBook Pro and Mac Pro line-ups.

AMD-zen-apple

If we look up at the AMD roadmap, the first AMD Zen processor will be roll out on 2016, however the rumor suggest that AMD will make a custom X86 SoC for iMac on 2017-2018. This could be a strong evidence that AMD APU SoC based on the AMD Zen “Zen+” will be using on the iMac.

Just sharing out a rumor and ending without further research seems pointless, but yet I’ve read from many sources and even some top dog blogs that just ended the story here. I don’t like it that way, as what can we learn from a simple rumor? Almost nothing. So let’s go through some key points on how AMD could be the one who will be making the custom x86 SoC chip powering the next iMac.

 

AMD Zen “Zen+” X86 CPU

Zen-amd

If this graphic is to be trusted, the AMD Zen “Zen+” cores are due to eventually arrive by 2016, matching Lisa Su’s claims that AMD will be generating revenue in the second half of 2016. If it is true that Apple will be using a custom AMD x86 SoC chip, so it will prove that the AMD Zen architecture can really delivers the performance gains over the old Excavator core, which will put the AMD Zen “Zen+” on par with Intel Skylake, at the very least. With a claimed 40% IPC improvement over its predecessor, and also manufactured on the 14nm process, that should allow AMD Zen “Zen+” cores to trade punches with Intel Skylake cores.

 

AMD Greenland GPU

amd-zen-greenland

We are also focused on delivering our next generation GPUs in 2016 which is going to improve performance per watt by two times compared to our current offerings, based on design and architectural enhancements as well as advanced FinFET products process technology“. —  Lisa Su

The AMD Greenland, we don’t know much about it except that it will feature second generation HBM, 14nm FinFET manufacturing process and double the performance-per-Watt efficiency over its predecessor. The second generation HBM 2.0 will be twice as fast as the first generation and also allows for more capacity, which will lead to even higher performance.

amd-greenland-hbm2

So by gathering all the key points I have mentioned previously, the 14nm manufacturer, HBM 2.0 and improved power efficiency, it gives us even more insight into the rumor. Since AMD Zen “Zen+” and AMD Greenland will be based on the 14nm manufacturing node, it should herald the arrival of extremely powerful AMD APU with low TDPs, high performance and also vastly improved efficiency. If Intel can put eDRAM on the Broadwell processors I don’t see why can’t AMD integrate HBM into their APUs. This is most probably the next step for AMD if you ask me, because AMD is becoming an expert in designing custom chips lately.

All this are shows that AMD is on track to its 2020 target, the 25×20 vision through heterogeneous computing, real-time power management and energy efficiency design innovations

 

Heterogeneous System Architecture (HSA)

Heterogeneous System Architecture (HSA)  is an industry standards body focused on making it dramatically easier to program heterogeneous computing devices. Simply put, it’s seeks to create applications that seamlessly blend scalar processing on the CPU, parallel processing on the GPU, and optimized processing on the DSP via high bandwidth shared memory access, enabling greater application performance at low power consumption.

HSA-amdJPG

The HSA Foundation is defining key interfaces for parallel computation utilizing CPUs, GPUs,  DSPs, and other programmable and fixed-function devices, thus supporting a diverse set of high-level programming languages and creating the next generation in general-purpose computing.

I think this is enough on AMD’s side, let’s see if we have any further leads from Apple’s side.

 

Apple Metal API

Have you read about the Metal API from Apple? Let take a good explanation from Wikipedia site :

“Metal is a low-level, low-overhead hardware-accelerated graphics and compute application programming interface (API) , it combines functionality similar to OpenGL and OpenCL under one API. Metal also further improves the capabilities of GPGPU programming by introducing compute shaders.”

Bam, take that!  Do you understand this explanation? Let me break it down further.

Firstly, Metal, a low-overhead hardware-accelerated graphics and compute application programming interface (API), same like Mantle and Vulkan, but Apple has decided to make their own API. To allow this feature to unleash its maximal performance, doesn’t it need hardware or graphic processing that has asynchronous computing? AMD’s latest graphic cards based on GCN have that feature down pat, unlike NVIDIA.

Second, it combines functionality similar to OpenGL and OpenCL under one API. Let me tell you the reason why Apple has decided to use AMD FirePro and R9 Radeons on their Mac Pro and MacBook Pro line-up, it’s because of OpenCL. It’s AMD’s specialty to optimize the OpenCL performance of their cards.

metal_api-apple

Lastly, improve the capabilities of GPGPU programming by introducing compute shaders. What is the main point of HSA’s existence? Of course GPGPU which is part of the defining key interfaces for parallel computation utilizing CPUs, GPUs,  DSPs, and other programmable and fixed-function devices, thus supporting a diverse set of high-level programming languages and creating the next generation in general-purpose computing.

 

System on Chip (SoC)

What is a SoC to begin with? In short, a SoC packs all the components that a computer needs to run in a single chip, the result of this design not only reduces the printed circuit board size but it also removed unnecessary interconnects and by cramming everything into a single chip, it will result in lower power consumption and heat production.

gseries_soc_004

Basically a SoC consists of :

(taken from Wikipedia site)

 

Pokdepinion : So how convinced are you that the next iMac Pro will be based on a custom AMD X86 SoC? Based on all the above main points, it is almost certain to be a reality. An iMac with a high performance custom AMD APU custom X86 SoC based on AMD Zen “Zen+” and AMD Greenland GPU with HBM 2.0 that will bring lower TDP and power consumption figures. With DDR4 RAM which will enhance the GPGPU computing performance, this kind of SoC will definitely bring the iMac to another level.

About The Author
Muhammad Firdaus
I love technology, so I love AMD.

Leave a Response