PlayStation 4 packs 50% more graphics firepower than the Xbox One
With Microsoft finally revealing details about what's under the hood of their next generation Xbox One (can't call it gaming only) console, we do a quick analysis of how its raw processing power stacks up against the Sony PlayStation 4.
Microsoft has gone ahead and revealed the internal hardware specifications of the Xbox One. In many aspects it matches or is similar to the PlayStation 4, but a little digging reveals that that PS4 actually packs almost 50% more horsepower in the graphics processing segment alone. Nevertheless, we take you on a step by step comparison of the most important aspects.
CPU – Tie
Both the Xbox One and PS4 pack an 8-core CPU based off AMD's Jaguar architecture. While the APU using in the Xbox One and PS4 have the same basic layout (as they are essentially the same APU designed by AMD), Microsoft has gone ahead and made a lot of customizations to this 28 nm SoC from its side. The SoC has two Jaguar compute units, with one unit being comprised of four processing cores linked up with 2 MB of L2 cache (shared between the four cores). This gives an 8-core design for both consoles.
The performance of Jaguar isn't the next generational leap you were looking for. Think AMD Bobcat, their lowest end (and slowest) x86 low power consuming processors aimed at tablet and add about 20-25% performance to that, probably faster than the ARM Cortex A15 (but may trail behind Intel's newly announced Silvermont Atom). The 28 nm node helps keep power consumption low, with AnandTech estimating just about 30W on full load.
All rumors point to a 1.6 GHz core clock speed for either consoles.
GPU – PS4 Wins
This is perhaps the most interesting section of this analysis. The GPU in Sony's PlayStation 4 and Xbox One are based on the same architecture, that being the latest HD 7000 GCN (Graphics Core Next) architecture. With an identical architecture, game developers should have a much easier time porting games between the consoles (and for that matter, PC as well) than the (challenge posed by the) previous generation.
Microsoft opted for a slightly different configuration of the GPU, with only 12 Compute Units (CUs) that totals up to 768 shader processors (SPs) or ALUs only. This is where Sony took the lead, with a 18 CU configuration giving a total of 1152 SPs, the GPU in the PlayStation 4 has roughly 50% more horse power than Microsoft's toy. In terms of compute performance, the PS4 is nearly 33% faster with 1.84 Tera Flops (TFLOPS) compared to 1.23 TFLOPS in the Xbox One. This could potentially lead to lower visual quality for the Xbox One in cross-platform games. Either that or the PS4 hardware won't be used to its fullest potential. Thankfully, that might not be the case as the PS4 simply has more of the same hardware. No real effort on the developer's part is needed to tap into that extra juice because it is just waiting to be used. Expect the PS4 to give much better frame rates and a more stable gaming experience in games that will truly stretch both hardware. Also, developers will have to work harder to achieve the same visual fidelity and run other background processes with the Xbox One. A win-win round for the PS4 from where I see it.
Image courtesy: AnandTech
The GPU in both consoles will reportedly be clocked at 800 MHz, although we are well aware that GCN based chips are capable of scaling to 1 GHz and beyond.
RAM – PS4 Wins
Here's another area where both companies took a different approach, with one coming out on top of the other. Sony played it simple this time with a scalable design and opting for much (much) faster GDDR5 memory (8 GB of it) instead of DDR3 or DDR4 for that matter. Although GDDR5 consumes more power, the huge boost in bandwidth more than compensates for it. The PS4 memory bus is 256-bit and clocked in at 5500 MHz (QDR i.e.. Quad Data Rate): that translates to 176 GB/s of memory bandwidth available to the system. That is nearly 8 times as much as the Xbox 360, but then again 176 GB/s is the bandwidth attained by AMD's mid-range HD 7870 GPU. Mind you, buying an HD 7870 alone will set you back by $250 (or more).
Microsoft has a two-stage approach to solving the memory bandwidth crisis (of feeding its CPU and GPU with data). 8 GB of DDR3 memory clocked at 2133 MHz and a 256-bit bus translates to roughly 68.3 GB/s of bandwidth, which is more than 100 GB/s less than what the PS4 system is treated with. To compensate for the slow DDR3 RAM, Microsoft added 32 MB of SRAM (with lower latency than eDRAM, but more die space required) on the die that gives about 102 GB/s of bandwidth. The 32 MB of SRAM will be used as cache as per the AnandTech report.
So with Sony going for a more modern-day GPU approach, Microsoft has taken the old X360 style which also (similarly) packed 10 MB of (much slower) eDRAM. Mind you, Sony's decision to choose GDDR5 means higher power consumption and a relatively costlier memory compared to their rivals. DDR3 RAM is much more power friendly as well as costs less, but again that difference will decrease and probably disappear with GDDR5 going more mainstream (the PS4 should further drive that process up). This round goes to the PS4 for a much less complicated memory subsystem and for providing a direct 176 GB/s of bandwidth to the developers without having to go through additional hardware.
Things are looking bright for Sony. With the company finally out of the red zone (they posted net profits!) and share prices higher than they were before Microsoft's big announcement, it looks to me that the PlayStation 4 will be the leader amongst the next generation of consoles, if we consider games only. Clearly Microsoft hasn't attached as much importance to gaming as it should have (with more efforts towards a set-top-box than a beast-of-a-console), and clearly it hasn't gone unnoticed either.
For more reading and a more in-depth analysis, head on AnandTech's PS4 and Xbox One analysis or the newest AMD Jaguar architecture analysis to know more about the CPUs powering the next generation consoles.