Xbox One (Durango) Technical hardware investigation

Status
Not open for further replies.
Jeez, i didn't think this "hUMA" would be so important. maybe a bios update is the only thing standing in Microsoft's way from venturing in the direction.

The Xbox 1 has some form of coherency it might not be sporting huma, which is a AMD solution. People are treating the term hUMA like a general term like UMA or NUMA. It's not.
 
It's possible that the combination of the data move engines and ESRAM trump hUMA. MS spent quite a bit of engineering dollars putting together their solution, regardless I'm sure XB1's solution is competitive with other options that exist in the marketplace.
 
Its really interesting more leaks haven't come out from devs concerning performance compariosons between consoles, or even dev opinion on the different hardware. It seems like we had more off the cuff remarks last gen concerning hardware comparisons..
 
The new Maxwell processor from Nvidia has virtual unified memory system so you definitely get the UMA but I don't know how heterogeneous it is.

I don't know if it is relevant:

That might sound suspiciously vague, but we spoke to AMD and it's actually true. The AMD chips inside the PlayStation 4 and Xbox One take advantage of something called Heterogeneous Unified Memory Access (HUMA), Good for gaming, good for AMDwhich allows both the CPU and GPU to share the same memory pool instead of having to copy data from one before the other can use it. Diana likened it to driving to the corner store to pick up some milk, instead of driving from San Francisco to Los Angeles. It's one of AMD's proposed Heterogeneous System Architecture (HSA) techniques to make the many discrete processors in a system work in tandem to more efficiently share loads.

http://www.theverge.com/2013/6/21/4452488/amd-sparks-x86-transition-for-next-gen-game-consoles
 
http://beyond3d.com/showthread.php?t=63929

Posted by warchild in june from the arch panel.

we have to invest a lot in coherency throughout the chip, so there's been io coherency for a while but we really wanted to get the software out of the mode of managing caches and put in hardware coherency for the first time on the mass scale in the living room on the gpu.

So if the Xb1 doesn't have hUMA it has its own solution to enable coherency across durango.

Interesting quote. I wonder what metrics he is using for "mass scale" wonder if it involves the esram ? Still more than one way to skin a cat as some would say. It will be interesting to see how each system deals with the coherency factor.
 

Sure I posted this link 2 months ago and at the time pondered if EA was actually going to help move HSA down the road for AMD. The quote before it is the main drama of the article and something that PC fans are not fond of reading.

EA Sports boss Andrew Wilson says that one reason none of its next-gen sports games are coming to PC is because Microsoft and Sony's new game consoles are actually more powerful than many PCs in a very specific, subtle way: "How the CPU, GPU, and RAM work together in concert," Wilson told Polygon.

So no distinction is made in that quote between the 2 systems. hUMA or no hUMA there will be synergies having an APU and unified memory that will give consoles some fairly long legs. I am sure EA will get that engine moved over to PCs at some point. Meanwhile there maybe 3 different versions of hUMA like behavior. One from MS Sony and AMD or maybe some AMD IP gets mixed to both or one of them. Who knows but I sure want to see how things play out.
 
I think we need to know what Marc Diana said at Gamescom.

I'd be willing to bet that Diana basically just said the PS4 has hUMA but said nothing about whether the Xbox One had it or not. And the person interviewing/asking questions just then assumed that the Xbox One didn't have it because nothing was said about it.

That's why I always prefer official sites that post the actual transcript rather than their interpretation of how the conversation went. Because inevitably you get something wrong like what appears to have happened here.

Regards,
SB
 
http://www.reddit.com/r/xboxone/com...as_its_own_coherency_solution_similar/cbswrn3

untitledv5yvi.png
 
Got a response from the author of the heise.de article.

I don't know how I feel about posting mails in public without the consent of the writer, but here is the gist:

The customer support is not authorized to make such statements. He is questioning the authenticity of the forum post.

He confirms that Marc Diana has said it in the interview. The author repeatedly asked if Marc Diana was really sure that XB1 doesn't have hUMA and Marc Diana acknowledged it several times.

The author of the heise.de article is also the interviewer and works for c't.

I'm willing to share this mail with a moderator for verification. (It's in german.)


http://www.neogaf.com/forum/showpost.php?p=77681681&postcount=1192

I believe too the consumer ticket is false. Consumer service doesn't talk about this things.
 
Maybe people are just confused because MS renamed hUMA in their docs just like they did every other AMD technology!

But actually, from what I understand the CPU can't access the ESRAM at all in Xbox One. That alone would preclude the design actually being called "hUMA" from my understanding. Whether that has any practical performance impact or not is probably a different question.
 
Maybe people are just confused because MS renamed hUMA in their docs just like they did every other AMD technology!

But actually, from what I understand the CPU can't access the ESRAM at all in Xbox One. That alone would preclude the design actually being called "hUMA" from my understanding. Whether that has any practical performance impact or not is probably a different question.

There is still a pool of DDR3 that's shared between CPU and GPU, so I'm not sure why this prevents the X1 SoC from being hUMA-compliant in theory and in functionality.

It seems as though people are clinging to this syntax without actually considering how the technology operates.
 
There is still a pool of DDR3 that's shared between CPU and GPU, so I'm not sure why this prevents the X1 SoC from being hUMA-compliant in theory and in functionality.

It seems as though people are clinging to this syntax without actually considering how the technology operates.

Without having access to atleast the GPU caches the CPU cannot have a coherent view of the GPU's memory.
 
Will be interesting to hear what it is really like. I would personally expect them to both have some kind of HUMA, but right now it all sounds rather ... flakey. Again, looking forward to the C't article, if they still dare to even print it that is. ;)
 
Without having access to atleast the GPU caches the CPU cannot have a coherent view of the GPU's memory.

But there is only one type of memory, and it seems arbitrary to call it either GPU or CPU's memory. I would assume both would access the same memory through the same (North) bridge, yes?
 
But there is only one type of memory, and it seems arbitrary to call it either GPU or CPU's memory. I would assume both would access the same memory through the same (North) bridge, yes?

By memory I mean the cached memory values in the GPU as well. The CPU cannot share a consistent view of the memory without being able to see whats in the GPU's caches as well.
 
None of the vgleaks articles, nor any of the stuff i have read on the durango GPU mention hUMA in any way shape or form (I havent seen a lot though), and if it goes as far back as 'first alpha kit' then it should sure as hell be in whats leaked so far.

hUMA as a "term" appeared relatively recently, after the VGAleaks articles. Previously you may have known it as HSA Cache coherency or CPU/GPU Cache coherency.
 
Status
Not open for further replies.
Back
Top