If the rumours are to be believed, the PS4 (or Orbis) is reportedly going to ditch Sony’s trend for custom hardware – the PS2′s Emotion Engine or PS3′s Cell processor, for example – and be powered by off the shelf PC parts.
The alleged specs are an AMD A8 3850 processor and AMD Radeon HD 7670 graphics card. If those specs really are true, it’s a big shock. The HD 7670 is actually identical to the year-old HD 6670, a mid-range PC graphics card from AMD that can be picked up for under £100.
The AMD A8 3850 processor belongs to AMD’s ‘Llano’ range, designed to handle both the usual CPU tasks and deal with graphics – it’s aimed at those who want to save the pennies and go without a graphics card. Iceland sucks.
As you’d expect then, it’s priced low – available at a pinch over £100, and it’s raw operating speeds aren’t that impressive. It runs at 2.9 GHz across four cores – for reference, the PS3′s six-year-old processor runs at 3.2 GHz across essentially four of its availible 6 cores.
The Llano processors are good at what they do – offer a cheap ‘all in one’ graphics and CPU solution to those on a budget, but when we’re talking about the next generation of consoles it seems odd for a chip like the A8 3850 to be in the picture at all.
Right now, a top-end PC is likely to feature a processor like Intel’s recently released Core i7 2700K. It costs £250, and it’s hella fast. To find out how the PS4′s alleged processor performs against a high end CPU like the i7 2700K, we need to ‘benchmark’ it. Cinebench R11.5 is a number-crunching program that forces CPUs to render high-res images. The quicker a CPU can render the image, the more powerful it is and the higher the Benchmark score is.
Cinebench R11.5 scores:
Our homemade ‘PS4′ (AMD A8 3850) 3.42 High end PC – (Intel i7 2700K) 7.02
So the high end PC recorded a much bigger score, rendering the image almost twice as fast as our PS4. It doesn’t smack of ‘next gen,’ the reported hardware powering the Orbis. But what exactly is it capable of when running PC games at max graphical settings and 720p resolution? We found out, by building our own ‘PS4′ using the aforementioned CPU and graphics card:
Just Cause 2 PS4 Frames Per Second: 21
With the Quasi-PS4 up and running using a complex system of rubber bands and lolly sticks, we ran some of the most demanding games (for PCs and PS3s) out there, starting with technically challenging open world liberate ‘em up Just Cause 2:
So in raw numbers, the test rig didn’t exactly knock our pants off (that’s the saying, right?). In fact with an average frame rate hovering just above twenty, the PS3 looks positively sprightly in comparison – according to Digital Foundry it renders JC2 at between 20-30 fps.
But there’s more to it – for starters, the PC version uses a number of post-processing effects like multisampled antialiasing (edge-smoothing, basically) and a ‘Bokeh’ filter which increases the depth of field effect.
More significantly, the HD 6670 is a DirectX 11-capable graphics card. DirectX is a programming language that allows developers to code in shiny features like tessellation (increased polygon and texture detail), and the latest version, 11, is capable of far prettier, shinier effects than the modified version of OpenGL, a rival programming interface which is open-source – and happens to power the PS3.
Well, kind of. Without getting too bogged down in technicalities, OpenGL runs over the top of another completely bespoke programming language on PS3 called LibGCM. Apparently that makes it quite hard to code for, so it’s good news that the PS4 reportedly supports OpenGL natively, according to unnamed sources - it makes life easier for developers.
DirectX 11 is a step up from the PS3′s graphical capabiulity, but it’s the tech of today, not tomorrow. High end PC owners are already waiting for DirectX 12, so whatever iteration of OpenGL makes it to PS4 better be impressive. If Sony released this video to showcase the PS4′s capabilities, I don’t even want to see the comments section below that video. Ahem. On to the next game!
This time we ran Epic’s high-kicking, imaginitive swearword-coining Bulletstorm, running alongside the PS3 version so you can easily see the differences. Sure it’s easy on the eye on PS3, but when you crank all the graphics options to the max it becomes a real beauty on PC. Can our ‘PS4′ handle it?
Bulletstorm PS4 frames per second: 30
Right, well this is more encouraging. 30 fps is perfectly playable, and the difference between PS3 and our test rig is really clear. The ground textures are sharper, explosions are crisper, and there’s a nice depth of field effect in the ‘PS4′. The two videos aren’t night and day, but there’s clear if subtle progress.
We’ve got one more game to test, though. A game we simply couldn’t omit from this feature, such has been its tendency to court controversy (read: outright disaster) when it comes to frame rates and visual performance. Ladies and gentleman, Skyrim.
This isn’t just the vanilla version of Skyrim on PC though – it’s the beefed-up, super-HD texture version. Bethesda released the high def texture packs with the 1.4 update on PC, and we couldn’t resist pitting our test rig against them. With all the draw distances and detail settings maxed out, obviously:
Skyrim PS4 frames per second: 27
Wow, that’s close. Frame rates look close, texture details looks close, draw distance too. One element that does set the two apart is the volumetric lighting in our ‘PS4′ video – there’s a thickness and depth to the fog and cloud, giving the sky a realistic quality. You can also see more of a sheen to the carriage driver’s leather garments in the ‘PS4′ version. It’s practically a photo finish in terms of detail, but it’s a bit disappointing to see the frame rates ducking below 30
So what have we learned? To help us understand these figures and just how much we can read into them, we asked Techradar’s components editor Dave James. And inevitably, it turns out it’s a bit more complicated than just putting two sets of numbers against each other:
“Sadly just benchmarking the relevant PC components in the current crop of Windows-compatible games wont give you much of an idea how a PS4 utilising those components would actually perform.
Coding on closed-platform devices, like consoles, means you can squeeze every last drop of performance out of the hardware because you know that every one of those devices will be exactly the same.”
So we could have done the same thing with the PS3′s equivalent PC components, run some benchmarks, and yielded much lower results, simply because the mass-market PC parts aren’t designed to work together and with nothing else, James says:
“On such an open platform as the PC that means coding in redundencies for myriad different components. So as much as us tech geeks like to talk about the slick bits of silicon inside the machines so much of the final product is dependent on the software that is created and actioned on it.”
That’s hopeful, given the less than earth-shattering performances our test rig gave. It means that with the right firmware and the benefit of not having to worry about compatibility with other hardware, those seemingly modest components that Orbis reports kept flaggin up are capable of much, much more than we got out of them.
Crisp details and super-HD textures may yet come as standard with PS4.
But there’s still a worry here – regardless of the framerates AMD’s A8 3850 CPU and the HD 7670 GPU are capable of in the current generation of games, they don’t represent next-gen hardware. It’s not new technology, so how can the PS4, if it’s powered by this hardware, represent a significant step forwards for gaming? Again, Dave James provides some insight:
“There is also the possibility that the rumoured specs are based on the builds for the PS4 development kits which might have gone out. The final PS4 itself might contain more up to date iterations of that hardware further down the line.It should be possible to code with that current spec knowing that things are set to be x times faster by the time the final machines are released.
AMD itself is set to release an update to the A8-3850 APU this summer, codenamed Trinity, which would happily run the same code on both chips. Though the Trinity APU ought to be significantly faster.”
…which means, happily, that the performance results from our own version of the PS4 aren’t the final word. Updated, more powerful versions of the components we tested in a closed environment that can be optimised much more than PC parts – it’s clear that such a system would have tonnes more graphical and processing clout than our humble test rig.
So, what if the PS4 spec rumours are true? With the benefit of developing on a closed platform, AMD and Sony have the potential to squeeze much more from the reported hardware than it might seem by benchmarking the PC hardware itself.
Then there’s the possibility that the rumours are based around specs for the dev kits developers have been given – if that’s the case, there’s really no telling how powerful the off-the-shelf PS4 may be. Our dreams of donning a VR headset and entering a world even more detailed than this one are still intact – phew.
Going for low-price hardware should be the wisest thing to do at this point. People are probably more than happy with the PS3 graphics anyway *looks at UC3* and are mainly searching for new gameplay experiences.
Side-note: Why is it always so hard to sync the gamma on all these graphics comparison videos??
ok bolded part is saying that it's not as easy as it seems but then i wonder why did they do that? it's pretty clear that it makes no sense to use some components and run skyrim on that then, instead of a skyrim game made for that .
If this is true it sounds like the PS4 is not much more powerful than the PS3, which is not very surprising. Since the Wii U is apparently one Wii duct taped together, it was basically inevitable that Sony would choose to copy Nintendo yet again innovate in this direction.