• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

The NEW Xbox one thread of Hot Chips and No Pix

bishoptl

Banstick Emeritus
Bishnote: keep this on the level, yo.

-------------------------------
ekim's original post:

Fresh from Hot Chips:
XBox_One_diagram.jpg

XBO_diagram_WM.jpg

XBO_GPU_WM.jpg

XBO_Audio_WM.jpg


Msoft XBox One supports shared coherent memory between 8 AMD Jaguar cores and DX 11.1+ GPU like @hsafoundation

https://twitter.com/rickbmerritt/status/372042121777541120

XBox One SOC has a CPU, GPU, and 15 special-purpose processors. Total of 47MB of storage on-chip. #HC25

https://twitter.com/Daniel_Bowers/status/372039785453346816

Hot Chips: At 363mm2 with over 5b transistors the Xbox One SoC designed by Microsoft and AMD is a very complex SoC.

https://twitter.com/TekStrategist/status/372041324871946240

Audio offload processor in XBox One SoC are "completely designed by Microsoft" and have "more than a CPU core worth of processing" #HC25

https://twitter.com/Daniel_Bowers/status/372041578639929345

XBox One CPU has eight "modified" AMD Jaguar x86-64 cores, in two clusters of four cores. Modifiations to memory sharing & bandwidth. #HC25

https://twitter.com/Daniel_Bowers/status/372040673618182144

Some noteworthy stuff:
- 8GB Flash memory
- 15 special purpose processors
- 4 Command processors (2 compute, 2 graphics)
- SHAPE offloads >1 CPU core
- Memory coherency between CPU cores and GPU (also caches) - huma?
- Note that little arrow between the CPU and the ESRAM - CPU can read/write to/from the ESRAM?
 

Chobel

Member
I'll start: where does "204MB/s" come from? we know It is assumed that bus from eSRAM to GPU is not dual so how can it be possible to read and write simultaneously? even if it's possible why is it not 109*2 =218 GB/s?
 

Bsigg12

Member
I'll start: where does "204MB/s" come from? we know that bus from eSRAM to GPU is not dual so how can it be possible to read and write simultaneously? even if it's possible why is it not 109*2 =218 GB/s?

Maybe it's part of what they didn't talk about? Until we get video of the presentation, we're left to speculate based on a couple articles and these slides.
 
Aside from the DBZ comparisons I think this thread is going to have the same back and forth unless a dev comes forward to clarify the bandwidth.

The 8 GB of flash is a bit of a surprise.
 

Sounddeli

Banned
The graphics processor supports DirectX 11.1, which is Microsoft's graphics engine that will power games. Microsoft calls the engine 1-plus, a reference to unique control processor features for custom graphics and processing in the Xbox graphics core. "These have been customized to significantly reduce the amount of time, the amount of work, that the CPU has to spend when assembling graphics commands,

"Sell said. One unique aspect of the chip is a shared memory pool that can be accessed by CPUs, GPUs and other processors in the system. Typically, GPUs and CPUs have different memory systems, but the new features increase the overall addressable memory in the Xbox One. The GPUs and CPUs have also been modified to enable shared memory. Shared memory is also part of a specification being pushed by the HSA Foundation, which wants to blur the line between GPU and CPU memory to make programming easier. AMD is one of the founding members of the HSA Foundation, though Sony is also a member, which suggests that shared memory may also be part of PlayStation 4.

"We can let developers have access to direct programming on GPU without a lot of overhead," Sell said.

Overall, the system has been designed to balance bandwidth with the power requirements.

"An important aspect of that is letting CPU and GPU share memory spaces," Sell said. The coherency connections are a derivative of AMD's years-old HyperTransport technology. The bandwidth can sustain data transfers of 30GB per seconds. However, the data transfers vary depending on the component and also the distance between parts.The chip will be made by Taiwan Semiconductor Manufacturing Co. (TSMC) using the 28-nm process, Sell said.

http://www.itworld.com/hardware/370538/xbox-one-will-have-high-performance-custom-chip?page=0,1

Also:

Already we know the apu won't do any audio, no copying or moving data, no processing for kinect, no encode or decode, no compression or decompress, it even looks like the graphics and compute commands have there own scheduling done. What else do the CPU and GPU have to do?
 

AgentP

Thinks mods influence posters politics. Promoted to QAnon Editor.
Note that little arrow between the CPU and the ESRAM - CPU can read/write to/from the ESRAM?

It is strange, every other arrow denotes flow of data using the thick arrow.

109GB/s min
208GB/s peak BW
4x 256bit read & write

If this was read and write at the same time, it would be 218GB/s or more accurately 4x 256bit read 109GB/s and 4x 256bit write 109GB/s. The fact that they are playing games with the wordage means they are doing the same with the numbers.
 

The Flash

Banned
Aside from the DBZ comparisons I think this thread is going to have the same back and forth unless a dev comes forward to clarify the bandwidth.

The 8 GB of flash is a bit of a surprise.

'Dat 8GB of Flash. What will it be used for exactly? OS or game suspension?
 

Argyle

Member
Was about to reply to this when the old thread got locked: http://www.neogaf.com/forum/showpost.php?p=78468653&postcount=760

Osiris said:
Be careful, Swizzle is used for multiple different definitions and this likely isn't the one they are referring to.

The fact that they specifically state Swizzle Copy on the diagram, and have been referred to previously as Move Engines points to them referencing what Microsoft call "Source Register Swizzling" - used to copy data to a temporary register.

See: http://msdn.microsoft.com/en-us/libr...=vs.85).aspx

(Original link that KidBeta posted: http://fgiesen.wordpress.com/2011/01/17/texture-tiling-and-swizzling/)

Osiris, your link refers to vector register swizzling in the context of a pixel shader, so actually IMHO KidBeta is correct - in the context of a "move engine" which is going to shuttle blocks of memory around the system, the explanation in his link is most correct (the move engine would deswizzle/detile memory into linear order to make it easier for the CPU to address, or swizzle/tile memory to make it easier for the GPU to access).
 

pixlexic

Banned
Aside from the DBZ comparisons I think this thread is going to have the same back and forth unless a dev comes forward to clarify the bandwidth.

The 8 GB of flash is a bit of a surprise.

That 8 gig of flash memory is a big pro. I hope it is all usable for gaming and not just to pool data for back ground processes like video recording.
 

Bsigg12

Member
http://www.itworld.com/hardware/370538/xbox-one-will-have-high-performance-custom-chip?page=0,1

Also:

Already we know the apu won't do any audio, no copying or moving data, no processing for kinect, no encode or decode, no compression or decompress, it even looks like the graphics and compute commands have there own scheduling done. What else do the CPU and GPU have to do?

Play games? I don't know, it seems they have taken everything that could pull performance down on the CPU and GPU and have given them their own little piece of silicon.
 

Osiris

I permanently banned my 6 year old daughter from using the PS4 for mistakenly sending grief reports as it's too hard to watch or talk to her
Was about to reply to this when the old thread got locked: http://www.neogaf.com/forum/showpost.php?p=78468653&postcount=760



(Original link that KidBeta posted: http://fgiesen.wordpress.com/2011/01/17/texture-tiling-and-swizzling/)

Osiris, your link refers to vector register swizzling in the context of a pixel shader, so actually IMHO KidBeta is correct - in the context of a "move engine" which is going to shuttle blocks of memory around the system, the explanation in his link is most correct (the move engine would deswizzle/detile memory into linear order to make it easier for the CPU to address, or swizzle/tile memory to make it easier for the GPU to access).

You (and Kid Beta) are right, I misread & misunderstood the MS link and missed the pixel shader reference.

Curious though, would the use of such swizzle logic / Move Engines, or anything else revealed in these slides hint towards the answer to the question of hUMA / non-hUMA functionality on Xbox One at all?
 

jayu26

Member
Could someone explain in layman's term what is the problem with these numbers? Or point me to a post in previous thread.

Thank you in advance.
 

Krakn3Dfx

Member
I'm going to guess at least 1 of the system's OS's will run off of flash RAM, maybe the primary, like the 4GB 360 does now. I don't think this is going to be anything earth shaking for games, if it was, they probably would have gone out of their way to talk about it at this point.
 

lvlzero

Banned
To put something on the table the 8gb of flash is for OS, browser cache and hibernate data.

The dedicated audio hardware is to handle not only game audio but audio channels pumped in by the HDMI inputs as these channels need to be mixed in up-to 7.1 while maintaining ability to adjust HDMI in volume separate from the consoles own volume, a standard audio processor wouldn't have cut the mustard.

onboard flash is nothing new, most consoles these days have nand for os/firmware and given the size of the Xbox ones is its kind of a given.
 
I'll start: where does "204MB/s" come from? we know It is assumed that bus from eSRAM to GPU is not dual so how can it be possible to read and write simultaneously? even if it's possible why is it not 109*2 =218 GB/s?

Have you tried hitting up any of those twitter accounts in the OP? Or any of the Microsoft guys? I would love to hear the maths.
 

Bsigg12

Member
8 gigs of flash sounds pretty nice, probably means the os will load really fast.

That's what I'm thinking. It would be risky to have the Game DVR on there because of the amount of writes and what not. Having the OS installed on the flash allows it to be accessed more quickly to keep load times to a minimum and 8GB gives this a lot of breathing room for additional features.
 

EdgeXL

Member
So... 8GB of flash memory and 8GB of DDR3 RAM?

What does the flash RAM do?

Now about that comment about the CPU and GPU sharing the same memory pool.. is that something like the Xbox One version of hUMA?

Pardon the questions... I am trying to understand this information.
 

vpance

Member
I'll start: where does "204MB/s" come from? we know It is assumed that bus from eSRAM to GPU is not dual so how can it be possible to read and write simultaneously? even if it's possible why is it not 109*2 =218 GB/s?

204GB/s being achieved is quite literally "the feel of 204GB/s".

It's attainable only while performing alpha transparency blending operations, according to Eurogamer, who first wrote about it. It's definitely not a real world average figure.
 
So... 8GB of flash memory and 8GB of DDR3 RAM?

What does the flash RAM do?

Now about that comment about the CPU and GPU sharing the same memory pool.. is that something like the Xbox One version of hUMA?

Pardon the questions... I am trying to understand this information.

Its not flash RAM, its flash memory. Like the 8GB flash you find on some budget ultrabooks, which is nearly always used as a cache to improve read speeds on commonly used programs.
 

pixlexic

Banned
So... 8GB of flash memory and 8GB of DDR3 RAM?

What does the flash RAM do?

Now about that comment about the CPU and GPU sharing the same memory pool.. is that something like the Xbox One version of hUMA?

Pardon the questions... I am trying to understand this information.

the flash ram is like having a built in ssd. if its being used for games developers can page data to it and retrieve a lot faster than using one lower speed hdd to do the same.

just like with pcs having one hdd that holds the data and one to cache to enhances performance.
 

Argyle

Member
Can somebody explain what "power islands and clock gating to 2.5% of full power" means??

I think this means that the system can turn off or underclock currently idle parts of the chip to save power, to a low of 2.5% of full power with everything turned off.
 
Top Bottom