[UPDATE]AMD HUMA Gives PlayStation 4 Graphics Boost Over Xbox One – Analysis of HUMA Architecture


Update #1: We have an official response from AMD:

During a recent gamescom 2013 interview, an AMD spokesperson made inaccurate statements regarding the details of our semi-custom APU architectures. AMD will not comment on the Microsoft Xbox One and Sony PS4 memory architectures and will not speak for Microsoft, Sony or other AMD customers.

Upon asking AMD about HUMA's implementation on either platform, i got the following reply:

Thanks for the response.  Can’t go into any more details since it is up to Microsoft and Sony to talk about their products and implementations…but what I can say is that the comments coming out of gamescom were inaccurate.

Update#2: An Xbox One developer on Reddit has revealed that Microsoft is taking a similar unified approach as HUMA on their upcoming console.

AMD has their reasons for not commenting on this matter and PCPer also acknowledges that HUMA isn't officially supported by currently available Jaguar powered chips aka Temash and Kabini which is a fact. But the possibility remains of HUMA on the semi-custom APU built by AMD. Only an official response from Sony and Microsoft can confirm whether or not their consoles utilize from HUMA or not. Otherwise, the technology would make its debut on PC's by fall 2013.

AMD's senior product marketing manager Marc Diana has revealed over at Gamescom 2013 that PlayStation 4 would feature AMD's HUMA (Heterogeneous Unified Memory Architecture) that would boost its performance compared to the Xbox One which doesn't makes use of the next generation technology. AMD's HUMA and HSA enhancements would debut this fall with the arrival of Kaveri APU for desktop/mobile PCs and PlayStation 4.

AMD HUMA Detailed - PlayStation 4 Boost Over Xbox One Explained

AMD HUMA which is a part of the HSA (Heterogeneous System Architecture) which allows a GPU and CPU to share a common memory pool. Currently, the a GPU requires the CPU to write the contents of an address to a standalone memory pool of the GPU which is time consuming and puts alot of load on integrated system chips. This isn't much of a problem with discrete GPU based PC solutions but when it comes to SOCs and APU, it can be the cause of inefficient and more complex programming. Using a separate memory pool for smaller integrated solutions is costly hence the memory has to be shared across a common path for adequate operations. A coherent memory architecture would not only ease development for APU/SOC powered consoles and desktops would also deliver a boost in performance with faster graphics rendering and better compute performance. A more detailed explanation on the ineffectiveness of a separate memory path has been provided by PCPerspective which can be noted below:

CPUs and GPUs do not share common memory pools.  Instead of using pointers in programming to tell each individual unit where data is stored in memory, the current implementation of GPU computing requires the CPU to write the contents of that address to the standalone memory pool of the GPU.  This is time consuming and wastes cycles.  It also increases programming complexity to be able to adjust to such situations.  Typically only very advanced programmers with a lot of expertise in this subject could program effective operations to take these limitations into consideration.  The lack of unified memory between CPU and GPU has hindered the adoption of the technology for a lot of applications which could potentially use the massively parallel processing capabilities of a GPU. via PCPper

We saw Intel using eDRAM chips abroad their Haswell chips while the AMD APU on the Xbox One in addition to the 8 GB DDR3 memory comes with an integrated 32 MB eSRAM package. Once again, these turn out to be separate pools of memory and won't deliver as much performance as a unified approach would. AMD HUMA on the PlayStation 4 not only allows cross sharing of the memory between CPU and GPU, but the memory used itself is GDDR5 grade which turns out to be better at the gaming front and delivers faster graphics rendering. While Xbox One's DDR3 is better for its operating system needs, its not as better in gaming performance as compared to GDDR5 which is the complete opposite of DDR3. GDDR5 or (Graphics Dual Data V5) offers faster bandwidth coupled with higher latency while DDR3 has lower bandwidth with lower latency. The lower latency results in more responsiveness of a system but being a gaming console, GDDR5 would have been a decent choice for MS.

How would AMD HUMA affect the desktop PC market? Intel is going forward to adopt DDR4 starting with their Haswell-E HEDT platform in 2014 and Skylake in 2015. On mobile front next year, Intel is expected to boost their graphics performance with Broadwell architecture which would also feature on-board eDRAM chips, their path won't be as efficiency as HUMA but its almost adequate for their integrated graphic chips. With DDR4, Intel will get faster bandwidth and much lower latency and we expect that Intel with such a big RnD team might as well develop some kind of Unified memory path for their SOCs.

AMD on the other hand would begin fusing the HSA enhancements by the end of this year with Kaveri APU. AMD HUMA would be useful for Kaveri APU since the next generation chips would feature a faster Radeon graphics chip based on the GCN architecture. A cross shared path with high performance memory would result in better performance and lesser bandwidth starvation. This would allow better and increased compute capabilities out of AMD's GPU that can only be unlocked through the use of AMD HUMA.