The Xbox One - Mini Review & Comparison to Xbox 360/PS4
by Anand Lal Shimpi on November 20, 2013 8:00 AM ESTPerformance - An Update
The Chipworks PS4 teardown last week told us a lot about what’s happened between the Xbox One and PlayStation 4 in terms of hardware. It turns out that Microsoft’s silicon budget was actually a little more than Sony’s, at least for the main APU. The Xbox One APU is a 363mm^2 die, compared to 348mm^2 for the PS4’s APU. Both use a similar 8-core Jaguar CPU (2 x quad-core islands), but they feature different implementations of AMD’s Graphics Core Next GPUs. Microsoft elected to implement 12 compute units, two geometry engines and 16 ROPs, while Sony went for 18 CUs, two geometry engines and 32 ROPs. How did Sony manage to fit in more compute and ROP partitions into a smaller die area? By not including any eSRAM on-die.
While both APUs implement a 256-bit wide memory interface, Sony chose to use GDDR5 memory running at a 5.5GHz data rate. Microsoft stuck to more conventionally available DDR3 memory running at less than half the speed (2133MHz data rate). In order to make up for the bandwidth deficit, Microsoft included 32MB of eSRAM on its APU in order to alleviate some of the GPU bandwidth needs. The eSRAM is accessible in 8MB chunks, with a total of 204GB/s of bandwidth offered (102GB/s in each direction) to the memory. The eSRAM is designed for GPU access only, CPU access requires a copy to main memory.
Unlike Intel’s Crystalwell, the eSRAM isn’t a cache - instead it’s mapped to a specific address range in memory. And unlike the embedded DRAM in the Xbox 360, the eSRAM in the One can hold more than just a render target or Z-buffer. Virtually any type of GPU accessible surface/buffer type can now be stored in eSRAM (e.g. z-buffer, G-buffer, stencil buffers, shadow buffer, etc…). Developers could also choose to store things like important textures in this eSRAM as well, there’s nothing that states it needs to be one of these buffers just anything the developer finds important. It’s also possible for a single surface to be split between main memory and eSRAM.
Obviously sticking important buffers and other frequently used data here can definitely reduce demands on the memory interface, which should help Microsoft get by with only having ~68GB/s of system memory bandwidth. Microsoft has claimed publicly that actual bandwidth to the eSRAM is somewhere in the 140 - 150GB/s range, which is likely equal to the effective memory bandwidth (after overhead/efficiency losses) to the PS4’s GDDR5 memory interface. The difference being that you only get that bandwidth to your most frequently used data on the Xbox One. It’s still not clear to me what effective memory bandwidth looks like on the Xbox One, I suspect it’s still a bit lower than on the PS4, but after talking with Ryan Smith (AT’s Senior GPU Editor) I’m now wondering if memory bandwidth isn’t really the issue here.
| Microsoft Xbox One vs. Sony PlayStation 4 Spec comparison | ||||||||||||||
| Xbox 360 | Xbox One | PlayStation 4 | ||||||||||||
| CPU Cores/Threads | 3/6 | 8/8 | 8/8 | |||||||||||
| CPU Frequency | 3.2GHz | 1.75GHz | 1.6GHz | |||||||||||
| CPU µArch | IBM PowerPC | AMD Jaguar | AMD Jaguar | |||||||||||
| Shared L2 Cache | 1MB | 2 x 2MB | 2 x 2MB | |||||||||||
| GPU Cores | 768 | 1152 | ||||||||||||
| GCN Geometry Engines | 2 | 2 | ||||||||||||
| GCN ROPs | 16 | 32 | ||||||||||||
| GPU Frequency | 853MHz | 800MHz | ||||||||||||
| Peak Shader Throughput | 0.24 TFLOPS | 1.31 TFLOPS | 1.84 TFLOPS | |||||||||||
| Embedded Memory | 10MB eDRAM | 32MB eSRAM | - | |||||||||||
| Embedded Memory Bandwidth | 32GB/s | 102GB/s bi-directional (204GB/s total) | - | |||||||||||
| System Memory | 512MB 1400MHz GDDR3 | 8GB 2133MHz DDR3 | 8GB 5500MHz GDDR5 | |||||||||||
| System Memory Bus | 128-bits | 256-bits | 256-bits | |||||||||||
| System Memory Bandwidth | 22.4 GB/s | 68.3 GB/s | 176.0 GB/s | |||||||||||
| Manufacturing Process | 28nm | 28nm | ||||||||||||
In order to accommodate the eSRAM on die Microsoft not only had to move to a 12 CU GPU configuration, but it’s also only down to 16 ROPs (half of that of the PS4). The ROPs (render outputs/raster operations pipes) are responsible for final pixel output, and at the resolutions these consoles are targeting having 16 ROPs definitely puts the Xbox One as the odd man out in comparison to PC GPUs. Typically AMD’s GPU targeting 1080p come with 32 ROPs, which is where the PS4 is, but the Xbox One ships with half that. The difference in raw shader performance (12 CUs vs 18 CUs) can definitely creep up in games that run more complex lighting routines and other long shader programs on each pixel, but all of the more recent reports of resolution differences between Xbox One and PS4 games at launch are likely the result of being ROP bound on the One. This is probably why Microsoft claimed it saw a bigger increase in realized performance from increasing the GPU clock from 800MHz to 853MHz vs. adding two extra CUs. The ROPs operate at GPU clock, so an increase in GPU clock in a ROP bound scenario would increase performance more than adding more compute hardware.
The PS4's APU - Courtesy Chipworks
Microsoft’s admission that the Xbox One dev kits have 14 CUs does make me wonder what the Xbox One die looks like. Chipworks found that the PS4’s APU actually features 20 CUs, despite only exposing 18 to game developers. I suspect those last two are there for defect mitigation/to increase effective yields in the case of bad CUs, I wonder if the same isn’t true for the Xbox One.
At the end of the day Microsoft appears to have ended up with its GPU configuration not for silicon cost reasons, but for platform power/cost and component availability reasons. Sourcing DDR3 is much easier than sourcing high density GDDR5. Sony managed to obviously launch with a ton of GDDR5 just fine, but I can definitely understand why Microsoft would be hesitant to go down that route in the planning stages of Xbox One. To put some numbers in perspective, Sony has shipped 1 million PS4s thus far. That's 16 million GDDR5 chips, or 7.6 Petabytes of RAM. Had both Sony and Microsot tried to do this, I do wonder if GDDR5 supply would've become a problem. That's a ton of RAM in a very short period of time. The only other major consumer of GDDR5 are video cards, and the number of cards sold in the last couple of months that would ever use that RAM is a narrow list.
Microsoft will obviously have an easier time scaling its platform down over the years (eSRAM should shrink nicely at smaller geometry processes), but that’s not a concern to the end user unless Microsoft chooses to aggressively pass along cost savings.
Facebook
Twitter
RSS
286 Comments
View All Comments
Death666Angel - Wednesday, November 20, 2013 - link
"I have to say that I prefer Sony’s stance on this one." -> Seeing how I don't really follow the console market apart from superficial reading of some news, what is Sony's stand here? :)Mugur - Wednesday, November 20, 2013 - link
You can easily replace the hdd in PS4, just like on the PS3.peterfares - Wednesday, November 20, 2013 - link
Glad to hear it works fine without Kinect. I won't be plugging mine in.epyclytus - Wednesday, November 20, 2013 - link
hi,i've read the ars review as well as others and have gathered some interesting info. from what i've gathered, everything is "rosy..." without quibbling on aesthetics and personal/subjective choices, such as, design of console/controller and video game comparison side-by-sides, which are hard to decipher anyway on youtube--i am really against the grain... firmly against the grain of sony and/or microsoft adapting such measly hardware inside these consoles. i can forgo a big/fat controller and box (XBO) if the hardware, i feel, has some muscle in it. or hide the more elegant PS4 under a shelf and behind some stuff to make it quieter and/or add a fan to keep it cool and from bricking, if the inside of it had some oomph! i mean, come one, the jaguar cpu is a tablet cpu and the gpu's are like cheap $100 gpu's. not only cheap, but these gpu's aren't even the new recently released gpu's from AMD that is found in the R9-290x.
the pc people seem to be praising this shift of inferior/cheap hardware solutions, as if, it is a "good thing" for the industry just because the architecture is the same as their $3000 lc's. give me a break.
please explain to me why this is good for games, for the gamers and for the industry, if the tech is not moving forward but semi backwards?
in 2006, ppl complained that the PS3 cost too much. well, the tech in the PS3 at that time didn't exist! $600 wasn't too much and add in blu-ray which at that time was an infant!!! and infant! now, the PS4 is a more agreeable price point but the inside of the machine are parts from a year ago. why is this good?
developers are saying they can squeeze out the consoles more than their pc counterparts, as if to say, "Yes! these consoles doesn't have the oomph of higher end pc's, but games will run better in them anyway because we can optimize the games better in consoles... yada-yada-yada."
the most upsetting part about all of this is that the games, visually and innovatively-speaking will suffer. and yes, the graphics are better and the new consoles are more powerful than their predecessors, but, it's not that more powerful. GTA V for ps4 will look better. yes. BF4 looks almost like a high end PC--yes. but this is probably where it will end, graphically speaking. i mean, the graphical wall will be squeezed out a lot quicker for this gen than last gen, i think. so, by next year, the best graphical masterpiece for either console will be possible. correct me if i'm wrong. and if developers can't squeeze out every metal of these consoles by next year, then something is wrong with the developers or these consoles or whatever since developers should already known how to develop for an x86 console since it is the same architecture as PC's which have been around since 1980 or whatever. i just don't see any games in the future that will be mind blowing is my greatest fear.
but, really, i'm just a little upset that... 1) they went with x86. 2) the x86 they went with isn't that powerful
good review though.
djboxbaba - Wednesday, November 20, 2013 - link
What would you say would have been a better alternative to x86? I personally find the change to x86 fine, but the gimping of the hardware... well i definitely agree with you there.epyclytus - Wednesday, November 20, 2013 - link
well, glad you ask. if i were to build my dream console, i would build it to require and exceed the fastest intel/amd cpu out there in terms of raw performance for gaming/graphics applications. at least, 8-cores, of course. and a full RISC cpu like the cell processor in the ps3 but, it's successor, or if doesn't exist, i'd make them make cpu from the ground up like what apple is doing with arm cpu's. in this case, if it can't beat the fastest mainstream cpu from intel/amd, in terms of, raw performance, then i'd add more cores to it. so maybe a 16-core sony/arm risk cpu that is 64-bit. i know risk cpu's are more scalable. so, adding more cores will give it that edge in terms of raw performance. and then, i would add 8GB of XDR3 RAM which i think is in development and will be faster than GDDR5 (i think). this is for bandwidth and future-proofing this system to meet its 6-10 year life cycle. the GPU would have to be discrete and will probably ask nvidia, instead of AMD to make one specifically for this dream console since Nvidia has better power/efficiency cards. this dream nvidia gpu will be like the upcoming maxwell gpu that is not even out yet. this is how advance my dream console is. and even though it's not an x86 machine and is risc-based, developing for this dream console will be the same as developing for anything else. the 8GB of XDR3 ram is shared btwn cpu and gpu, btw. what else am i missing? yeah, maybe, the console will be bigger than the ps4 but will be as sleek and have a higher price point. but, you can't argue that the inside of this dream console is anything but slapped together.Oh, the sound chip is also discrete. Adds cost. But whatever.
Bluray standard.
The i/o of the machine is standard. So sata3 or whatever.
Wifi is a/c standard.
Maybe the prce will be $899. But that is XDR3 ram, 16-core RISC sony/arm cpu, nvidia Maxwell gpu, dedicated soundcard, wifi a/c, and 1TB of 7200rpm HDD.
flyingpants1 - Wednesday, November 20, 2013 - link
Great, congrats. And it would be utterly pointless because noone would buy it.epyclytus - Wednesday, November 20, 2013 - link
well, the price is a little steep but the tech inside are state of the art, emerging, nonexistant technology as of right now. maybe, the console wouldn't have made launch this year. but, maybe next fall with all of the specs i mentioned. considering how fast apple updates their iphone hardware and the buzz around Arm and MIPS getting back into the RISC cpu race, then i don't think it's inconceivable to think that an electronic giant like sony in partnership with Arm, or MIPS could have co-developed a fast, multi-core RISC cpu that can compete with a desktop intel i7 or future AMD steamroller cpu. or maybe even samsung and sony since samsung also makes cpu's and they have a fab lab. i don't know. i am sort of pulling this out of my butt. but, it's a dream console, afterall. and my knowledge of the cpu market place are non existent. so, i got nothing to go by except for google searches about these things.someone also mentioned that a cpu is fundementally different than a gpu and they're right. a cpu isn't as fast as a gpu and a gpu isn't as fast as a cpu on certain task. but what bridges those gaps closer is a RISC cpu built from the ground up, sort of like the cell processor, but more powerful obviously that can do cpu task well and gpu task well. my proposition for a maxwell gpu in this dream console is also important since nividia is also incorporating an ARM chip in their upcoming 800 series of gpu to do what gpu's can't do. so, the maxwell version of this dream console will forgo that arm chip because there are already 16 of these chips (or cores) in the proposed RISC cpu of my dream console. my dream console is basically a video/graphic powerhouse where the cpu and the gpu are like synchronously or asynchronous talking to each other, but aren't dependant on each other. and the XDR3 memory controller to feed it is also part of this to give it massive bandwidth. Also, since the cpu and gpu are all co-developed and built with this application in mind, the entire console will only pull around 200 watts at load. Maybe less.
i know i'm dreaming. and it will never happen. well, it will eventually happen. But, I was hoping to happen sooner and in a console. why? Consoles are great platforms for diverging/emerging tech. Or should be. Sort of like what apple is doing with iphone and ipad hardware, but obviously, much more powerful. Much, much, much more powerful since consoles don't have to be that small like an iphone or ipad.....
/end babble
/end dream
epyclytus - Thursday, November 21, 2013 - link
just wanted to add that what my proposition is is a CPU and GPU that are both CPU and GPU, if that makes sense. so, theoretically, the cpu can be a gpu and the gpu can be a cpu so it's like having a dual gpu setup such as found in pc's. and/or a dual cpu's or possibly more.i know. advance stuff....
Hubb1e - Wednesday, November 20, 2013 - link
Haha. You want to build a RISC CPU from the ground up to be more powerful than an Intel i7, use ram that isn't even available yet, and use a graphics core that hasn't been finished yet? I'm not saying that's impossible, but it would be more expensive than the whole Manhattan Project to build the first nuclear bomb.x86 chips are already available, Relatively fast, Jaguar chips are easily scalable to new processes, and DDR3 and GDDR5 ram are already in full volume production. Graphics are just a situation of adding more blocks and the minor differences in relative power consumption of AMD vs Nvidia is a moot point as Nvidia is incapable of creating an APU with a decent CPU in it.
I love the idea of an APU in these boxes because it makes so much sense but my ideal box would have been 7870 type graphics performance coupled with a 6 core CPU based on AMD's new streamroller core running at above 3ghz.
For RAM I would have taken the old approach they used to do with the 780G chipset and used 1GB of GDDR5 for the GPU coupled with 8GB of DDR3 accessible by either the GPU or the CPU.
Power consumption would have been similar to the XB360 on initial launch but they should have been able to build that within the size of this XBone.