What role does the hardware play in our enjoyment of games? On the surface, it would seem to be fairly large. Many of the characteristics we associate with contemporary gaming were a direct result of more memory, more transistors, and more complex engineering. The first PlayStation might have brought us true 3D gaming, but the PlayStation 2 brought us the sandbox genre, and true living, breathing worlds. The PlayStation 3 expanded on that, growing their scope and lending their increasingly mature themes with more persuasive verisimilitude.
Now we have the PlayStation 4 glinting on the horizon. It beckons with promises of boundless creative freedom – a virtual Mirror of Erised, realizing our most far-reaching ideas with a sense of uncompromised ease. But can we really do more with this future generation hardware than can be done now? What does this new hardware really offer us, and how does it compare to what’s around the corner on other similar platforms?
For Owners of the Current PlayStation
The PlayStation 3 was a forward-looking design effort. The Cell processor represented Sony’s vision of the future, not just of multi-core processors, but of asymmetrical multi-core processors. In essence, the holy grail of heterogeneous computing, the vision that would eventually dominate CPU maker AMD’s central message in the late 00s and onward. AMD has made a name for themselves by adopting aggressive technologies that find their way into modern computing standards, and their heterogeneous strategy, named Fusion, has essentially amounted to an endorsement of Sony’s ideas. This is a company that understands the trends of software development, and a nod from them is an encouraging sign.
For all the praise Sony may have earned for their prescience, though, a lopsided share of criticism always followed closely behind. Cell was difficult to get a grasp of, and for all its theoretical promises, truly useful utilization wouldn’t come until late in the lifecycle of the PS3. Multi-core programming was a struggle for even veteran coders in 2006, and the Cell went a few steps beyond that. For developers just learning how to walk in this new era, the PS3 was a 100 yard dash.
The PlayStation 4 may seem like a retraction of this philosophy, and it was even framed as such in Sony’s first press conference for it. In reality, by partnering with AMD they’re continuing the heterogeneous idea to its logical conclusion. Take a core CPU, made up of eight “fat” processors, and couple it with a secondary core comprised of over 1,000 tiny processors. The tiny processors serve as your GPU, complete with texture units and render backends, but they can also be utilized in any proportion for general purpose processing. Traditionally this will be highly parallelizable floating-point workloads, usually pertaining to physics.
“For developers just learning how to walk in this new era, the PS3 was a 100 yard dash.”
This is good, because the CPU in the PS4 is pretty weak in relative terms. It is built on AMD’s new Jaguar architecture, a design intended to provide adequate web-browsing and HD movie watching in tablets and inexpensive ultrabooks. In absolute terms, the performance is far short of modern expectations – the furthest thing from the CPUs found in computers intended for highend gaming. It's a minor advancement on the popular low-power Bobcat processors, the dual-core variants of which compare unfavorably to Pentium 4s at 3.6GHz from 2005. Jaguar might close the gap to something negligible, but this is decidedly last generation technology in the guise of a next generation console.
Why use such an underperforming CPU? Surely a low-power binned relative of AMD’s Piledriver or Intel’s Haswell would meet the thermal design power (TDP) limitations of a console enclosure, and provide much better performance? The short answer is size. AMD touts the Jaguar cores' ability to squeeze within a 3mm^2 profile on a modern die (the physical chip made up of transistors), which means that Sony can pile on eight of them and still leave plenty of room for a high-performance graphics processor. Sony’s goal is to create a next-generation chip that’s smaller and less expensive than its predecessors, without any unacceptable compromises in graphical potency.
“Any deficiencies in raw performance will have to be addressed with some clever programming”
While the eight Jaguar cores will not outdo the PS3 in terms of raw floating-point performance, in more linear tasks where using the GPU makes less sense, the Jaguar CPU will be a significant improvement over Cell. Any deficiencies in raw performance will have to be addressed with some clever programming, but if tasks can be divided up to efficiently saturate the octo-core configuration, significant stalls and latency can be averted. This will be essential to getting the most out of the GPU. What’s more, none of the cores will need to be given up for the OS, thanks to a special low-power processor (supposedly based on ARM) that offloads menial background jobs.
Memory may be the most significant improvement over the PS3. Sony surprised the press and even developers by announcing a massive 8GB of RAM, one of the largest generation-to-generation upgrades for memory. This is made all the more astounding because of the type of memory being used. GDDR5 is a very fast, but very costly type of memory, usually reserved for modern graphics cards, and usually in moderate quantities. This expensive pool of high-bandwidth RAM will be shared between the CPU and graphics, something that only Microsoft had dared to do before with the 360. Unlike past implementations however, the sheer data throughput (supposedly 176GB/s) promises to unburden both the graphics and CPU of any potential memory bottlenecks, assuring both are always running at near peak performance.
Bottlenecks were always a risk in the past with trying to use a unified pool of memory. The 360 shared its 512MB (part of which went to the OS) with its CPU and GPU at a paltry 22.4GB/s – a woefully inadequate number even by 2005 standards. The solution to the bandwidth problem was always to split the memory apart, giving the CPU and GPU their own dedicated lanes – a tactic used in the PS3. Even with this advantage, the PS4 trumps it with nearly 10x the bandwidth, enough to access more than half of that massive 8GB cache every 1/30th of a second. Perfect for feeding a graphics chip over 8x the performance of the PS3’s RSX.
For Those On a PC
What about desktop computers? The PS3 and 360 pushed the bar for power consumption and heat generation in a console, managing the rare achievement of surpassing top-of-the-line PCs for a short while. Unfortunately the long-term repercussions were severe, with millions of consumers forced to send their console off for repairs. Both Sony and Microsoft will be looking to improve the failure rate of their next-generation systems to make certain this isn’t a problem again, and that will require lowering power consumption dramatically.
“So much of what makes up modern gaming is owed not just to the hardware, but also the software.”
This means that performance will not surpass PCs, or even reach parity with the state of the art in the year they launch in. Those hoping they’ll get the same awesome bang-for-your-buck that last-gen consoles gave us will be disappointed. Knowledgeable folks will have seen this coming, however, as power requirements for expensive parts have only increased in some cases, and currency inflation and costs have ensured that it just isn’t economical to give consumers 800 dollars’ worth of computer for $500 or less.
Graphically the GCN (graphics core next) based processor falls somewhat in line with a Radeon HD 7850 on the PC, a card worth about $200 right now. This is only enough performance to run the newly released next-generation torchbearer, Crysis 3, at medium settings.
Combined with the meek and slender CPU, this might not sound like the arrival of the next era in video games, but perhaps there’s more to the picture than just that. So much of what makes up modern gaming is owed not just to the hardware, but also the software. This element is the reason that more performance can be obtained than specifications would suggest, and a whole network of services and features can be facilitated similar to, but slightly different from what a PC can offer.
For the PS4, a big chunk of that will derive from Gaikai.
The network-based game streaming service purchased last year by Sony will be injected
deep into the system’s DNA, utilizing separate processors to take images and
sound from games running either locally or on a server miles away, and
encode/decode their signals on the fly. This can enable console gaming in any
room of your house by way of the PS Vita, instant access to trial games, or
over-the-shoulder online spectating – which then can turn into active
participation if the host elects to hand off the controls to their friend.
Services like these, as well as online stores offering full digital distribution; community features including messaging, voice chat, matchmaking, and more; and achievements with scores and passive competition, are all doable on the PC. But often they require multiple applications, and juggling between them can be less intuitive and expedient by comparison.
Then there’s price. Spending more on a PC will net you better performance, and that’s not rocket science. But spending the same amount on a PC as you would a PS4 should, in theory, get you inferior performance. Consoles use subsidization to give you more for your dollar than equivalent wholesale parts could provide, and that’s what ultimately keeps them relevant. When packaged with a simple interface and a handful of exclusive games, the selling point of a console becomes all about being the cheaper and easier option – a fact many PC folks will thumb their nose at.
But the mainstream approach of consoles can benefit everyone’s gaming experience, especially early on. Highend PC hardware gains a lot more attention from developers as the bar for commonly accessible performance rises considerably. The cheap and easy option is once again the high fidelity option, and that means more realistic worlds and expansive gameplay for everybody.
“Highend PC hardware gains a lot more attention from developers as the bar for commonly accessible performance rises considerably.”
The PlayStation 4 represents a departure from the exotic custom architectures adorned with fancy names like “Cell” and “the Emotion Engine”, and gives developers a familiar set of hardware whose unique traits come from the tools with which to make use of them. Running an x86 processor and desktop GPU means that cross-platform development becomes a trivial notion, and similar PC hardware can benefit greatly from their shared characteristics. While the Jaguar CPU means that AMD’s desktop processors won’t receive any direct benefit from console ports, the multi-threaded design necessitates that engines make use of as many cores as possible. If AMD chooses to offer more cores than Intel at an equivalent price, games built to scale across them may, to some degree, work in their favor.
What of Sony’s competitor, though? The Xbox 360 should receive a successor this year as well, and while leaked specifications are detailed and fairly convincing, I’ve decided to refrain from direct comparisons until they are confirmed definitively from the horse’s mouth. So far, things seem to tilt slightly to Sony’s advantage, but at any point down the line, something could change.
For Gaming In General
So what’s the answer to the question posed many paragraphs ago? Does the PlayStation 4 offer us something that couldn’t be done before? Does the hardware represent a widening of the barriers that shaped the graphics and gameplay we’ve become accustomed to?
Perhaps. While genre-forging revolutions seem to be a thing of the past, the size of our worlds can now expand, the characters inhabiting them can grow in intelligence and number, physics and the behavior of environments can rise in complexity and accuracy, and of course the visuals will reflect a more recognizable facsimile of our reality.
But consoles provide only a push in that direction. For those who have confined themselves to a PS3 or 360, the choice is simple. For those with other options available, it’s a little more complicated. Games on the PC already resemble the next era as we’ve perceived it to be, and for many it won’t make much sense to buy another box in several months to do what the one next to their desk can do right now.
Regardless of which point of view you’re coming from, we can all appreciate the value of a new console generation. Even if games like Crysis 3 or Watch Dogs are crafted to utilize the very best components on the market, their maps and NPCs are built to be one-size-fits-all. AI, physics (GPU physics in particular), and the size of environments won’t truly progress until a new set of consoles deems it a cost-effective endeavor. Call it a sad state of affairs, but it’s just reality.
In the end, gaming of the future matters less about what you’re gaming on, and more about what your gaming is. New consoles herald new possibilities, and what lies in store for tomorrow promises to be a hopeful rekindling of the joy and excitement that made us all gamers in the first place. I look forward to meeting you all there.
Nathan Mayer is Associate Editor for 1985FM. Follow him on Twitter @nathan85fm
