News & Rumors: Xbox One (codename Durango)

Status
Not open for further replies.
Sorry to interrupt this stimulating VSP conversation, but MS expects next gen to increase in console sales by 28%.

ProjectedConsoleGenerationalGrowth.jpg


http://news.xbox.com/2013/05/x360-aaron-greenberg-industry-growth

With Wii U dead on arrival, I guess they expect Durango (and PS4?) to pick up the slack, BIG time.
 
Bagel seed said:
Sorry to interrupt this stimulating VSP conversation, but MS expects next gen to increase in console sales by 28%.

ProjectedConsoleGenerationalGrowth.jpg


http://news.xbox.com/2013/05/x360-aaron-greenberg-industry-growth

With Wii U dead on arrival, I guess they expect Durango (and PS4?) to pick up the slack, BIG time.
Click to expand...
Another very interesting thing about that article is that, according to data, more people play videogames than ever before.

On a different note, Aaron Greenberg says that the Xbox reveal is meant to take an inside look at the making of the new Xbox and the team that's brought it to life.

They are going to wait for E3 to reveal "tonnes of exclusives" and world premieres.

http://www.videogamer.com/news/micr...of_xbox_exclusives_to_be_announced_at_e3.html

"We're really going to tell one story across two events," explained Microsoft's IEB chief of staff Aaron Greenberg on the latest episode of Major Nelson's podcast. "So we're going to start on the 21st, and really that's about revealing the next Xbox platform; our vision for the future of games, the future of entertainment.

"We definitely have a lot of surprises planned," Greenberg continued. "I know everyone knows the Infinity Ward team is hard at work on world premiering Call of Duty: Ghosts, so we're going to see what that looks like for the first time.

"People are going to get a great inside look at the making of the new platform and the team that's brought it to life. To be honest, it's really about the team, the people here. That's why we're inviting people to campus; we're going to bring them behind the scenes a little bit. We've never done this before and so it's really exciting to be able to do that. It's a lot of laying the foundation, probably the best way to describe it. And then just a couple of weeks later we go to E3."

Microsoft will use E3 to reveal more about the titles headed to the next Xbox, Greenberg says.

"E3 is all about the games," he continues. "At E3, I mean, if you're a gamer, it's going to be tonnes of exclusives, world premieres, and obviously even beyond that we have a lot to share between then and this holiday."
Click to expand...
 
marcberry said:
http://www.google.com/patents/US201...&sa=X&ei=xFSWUZKME4KYPa37gKAD&ved=0CDcQ6AEwAA

and

http://www.google.com/patents/US201...&sa=X&ei=xFSWUZKME4KYPa37gKAD&ved=0CDcQ6AEwAA

Look right in the center of the 2nd diagram we see 0-4 SIMDs inside 1 of the 4 VSPs that's = xx per VSP. The very top one is part of the front end.
Click to expand...

Huh?

Are you saying that Durango employs 16 extra redundant ALU (4 per SIMD) and that MS will actually use those units for work?

Thats how Durango will get 1.5 TFlops?

SIMD = 80 ALUs?

I doubt it.
 
marcberry said:
http://www.google.com/patents/US201...&sa=X&ei=xFSWUZKME4KYPa37gKAD&ved=0CDcQ6AEwAA

and

http://www.google.com/patents/US201...&sa=X&ei=xFSWUZKME4KYPa37gKAD&ved=0CDcQ6AEwAA

Look right in the center of the 2nd diagram we see 0-4 SIMDs inside 1 of the 4 VSPs that's = xx per VSP. The very top one is part of the front end.
Click to expand...

That merely provides a method to have a redundant ALU, such that they can have a defective ALU without throwing out the entire card, it would not increase compute power in any way shape or form. What VSP stands for more likely is Vector Scalar Processor, a Vector of Scalar Processors if you will, this makes sense as the compute cores for GCN are now Scalar Processors and no longer Vector Processors.
 
mosen said:
Maybe he means this patent.

http://www.google.com/patents/US201...&sa=X&ei=xFSWUZKME4KYPa37gKAD&ved=0CDcQ6AEwAA
Click to expand...

Betanumerical said:
That merely provides a method to have a redundant ALU, such that they can have a defective ALU without throwing out the entire card, it would not increase compute power in any way shape or form. What VSP stands for more likely is Vector Scalar Processor, a Vector of Scalar Processors if you will, this makes sense as the compute cores for GCN are now Scalar Processors and no longer Vector Processors.
Click to expand...
That patent is for Xenos (Xbox 360).

I didn't read VGleaks so I don't know what they're referring to, but if you're wondering why there might be groups of 4 think about double precision.
http://www.google.com/patents/US20110055308
 
Betanumerical said:
That merely provides a method to have a redundant ALU, such that they can have a defective ALU without throwing out the entire card, it would not increase compute power in any way shape or form. What VSP stands for more likely is Vector Scalar Processor, a Vector of Scalar Processors if you will, this makes sense as the compute cores for GCN are now Scalar Processors and no longer Vector Processors.
Click to expand...

So what have we learned so far.
1. Most did not know what the hell VSPs are.
2. This is not the same old GCN we know of.
3. SC is not a CU.
4. Each SC has 4 SIMDs. VGleaks. More later on this one.
5. Inside each of those SIMDs have 4 VSPs VGleaks. More later on this one.
6. Inside each of the 4 VSPs are 0-4 16wideSIMDS. (see last post)
7. In the top patent i told you to look up other GPUs like this one http://www.vgleaks.com/wp-content/uploads/2013/02/gpu_queues.jpg
and compare GPUs . Count the vertex grouper tessellators (VGTs) their is 1 for each SC in the patent, compared to this one.
8. Yukon docs say right on them 8x more powerful then that of the 360.
This is to one QUOTE at the top. What the hell are you saying, redundant
192 times in the same place. Ok right. Just because you say it don't make
it fact show me a link. Now no one has posted a counter to this, with proof. VGleaks shows us that the VSPs are a main part of each SC. SuperDAE " no the PS4 is better." said the man going to jail.
vgleaks:
Instruction issue rate
12 SCs * 4 SIMDs * 16 threads/clock = 768 ops/clock
FLOPs
768 ops/clock * (1 mul + 1 add) * 800 MHz = 1.2 TFLOPS
next:
Instruction issue rate
12 SCs * xxxxxx * xxxxxxxx/xxxx = xxx ops/clock
FLOPs
xxx ops/clock * (1 mul + 1 add) * 800 MHz = xxx TFLOPS

The next xbox is not GCN1.0.
 
marcberry said:
So what have we learned so far.
1. Most did not know what the hell VSPs are.
2. This is not the same old GCN we know of.
3. SC is not a CU.
4. Each SC has 4 SIMDs. VGleaks. More later on this one.
5. Inside each of those SIMDs have 4 VSPs VGleaks. More later on this one.
6. Inside each of the 4 VSPs are 0-4 16wideSIMDS. (see last post)
7. In the top patent i told you to look up other GPUs like this one http://www.vgleaks.com/wp-content/uploads/2013/02/gpu_queues.jpg
and compare GPUs . Count the vertex grouper tessellators (VGTs) their is 1 for each SC in the patent, compared to this one.
8. Yukon docs say right on them 8x more powerful then that of the 360.
This is to one QUOTE at the top. What the hell are you saying, redundant
192 times in the same place. Ok right. Just because you say it don't make
it fact show me a link. Now no one has posted a counter to this, with proof. VGleaks shows us that the VSPs are a main part of each SC. SuperDAE " no the PS4 is better." said the man going to jail.
vgleaks:
Instruction issue rate
12 SCs * 4 SIMDs * 16 threads/clock = 768 ops/clock
FLOPs
768 ops/clock * (1 mul + 1 add) * 800 MHz = 1.2 TFLOPS
next:
Instruction issue rate
12 SCs * xxxxxx * xxxxxxxx/xxxx = xxx ops/clock
FLOPs
xxx ops/clock * (1 mul + 1 add) * 800 MHz = xxx TFLOPS

The next xbox is not GCN1.0.
Click to expand...

So if its not GCN1.0, yet it gets the same FLOPs as GCN1.0, has the same amount of queues as GCN1.0 (this is one thing that actually changes in the next GCN verision) and it also has the exact same cache setup as GCN1.0.

Vgleaks providied the FLOP calculation straight from there documents (which are white papers detailing the architecture). So unless microsoft themselves dont even know they arent using GCN1.0 :LOL: then it most certainly is GCN1.0

Also Durang is probably around 8x more powerful then the 360, you have to remember that the 360's architecture is no where near as effiecnt as the modern day GCN with its vector of scalar processors as a SIMD, the 360 instead relied upon a VLIW5 (? i think) set of procssors. So taking this into account Duragno probably is 8x more powerful then the 360.

I have yet to see proof from you, aside from one person using 1 different word from a diagram that 'this is not the same old GCN we know'. And if you intepret VSP to a be a Vector of Scalar Processors (common sense man to the resuce) then it all makes sense, as it all fits in perfectly with the normal GCN architecture.

You are trying to tell everyone here that vgleaks not only got the majority of there article wrong, but they accidently and without even any special mention of it leaked the super secret fact that Durango has this new GCNX.X. You cannot pick and choice which parts of the article you believe, you have to either believe that its 1.2TFLOPs and has these VSP's or you have to believe its nothing in the article.

Also please learn to format your posts better they are becoming a pain to read, paragraphs are a thing.

The Queue Diagram you linked to is not for Durango its for another console. I fail to see the relevance, Durango has only 2 Queues, not 8. As shown by the diagram your looking for.

http://www.vgleaks.com/wp-content/uploads/2013/02/block_gpu.jpg
 
Last edited by a moderator:
marcberry said:
So what have we learned so far.
1. Most did not know what the hell VSPs are.
2. This is not the same old GCN we know of.
3. SC is not a CU.
4. Each SC has 4 SIMDs. VGleaks. More later on this one.
5. Inside each of those SIMDs have 4 VSPs VGleaks. More later on this one.
6. Inside each of the 4 VSPs are 0-4 16wideSIMDS. (see last post)
7. In the top patent i told you to look up other GPUs like this one http://www.vgleaks.com/wp-content/uploads/2013/02/gpu_queues.jpg
and compare GPUs . Count the vertex grouper tessellators (VGTs) their is 1 for each SC in the patent, compared to this one.
8. Yukon docs say right on them 8x more powerful then that of the 360.
This is to one QUOTE at the top. What the hell are you saying, redundant
192 times in the same place. Ok right. Just because you say it don't make
it fact show me a link. Now no one has posted a counter to this, with proof. VGleaks shows us that the VSPs are a main part of each SC. SuperDAE " no the PS4 is better." said the man going to jail.
vgleaks:
Instruction issue rate
12 SCs * 4 SIMDs * 16 threads/clock = 768 ops/clock
FLOPs
768 ops/clock * (1 mul + 1 add) * 800 MHz = 1.2 TFLOPS
next:
Instruction issue rate
12 SCs * xxxxxx * xxxxxxxx/xxxx = xxx ops/clock
FLOPs
xxx ops/clock * (1 mul + 1 add) * 800 MHz = xxx TFLOPS

The next xbox is not GCN1.0.
Click to expand...

Agree with you that the VGLeaks specs are NOT GCN1.0 ... After reading those patents it's definitely clear to me that these VSP's are a step beyond what we publicly know as GCN1.0 .. if these VSP's are what is in Durango then damn :)
 
liquidboy said:
Agree with you that the VGLeaks specs are NOT GCN1.0 ... After reading those patents it's definitely clear to me that these VSP's are a step beyond what we publicly know as GCN1.0 .. if these VSP's are what is in Durango then damn :)
Click to expand...

The patent he linked talks about redundent ALU's to probably increase yields.

If this a dayum, then most video cards have had a dayum for a while. :LOL:
 
marcberry said:
http://www.google.com/patents/US201...&sa=X&ei=xFSWUZKME4KYPa37gKAD&ved=0CDcQ6AEwAA

and

http://www.google.com/patents/US201...&sa=X&ei=xFSWUZKME4KYPa37gKAD&ved=0CDcQ6AEwAA

Look right in the center of the 2nd diagram we see 0-4 SIMDs inside 1 of the 4 VSPs that's = xx per VSP. The very top one is part of the front end.
Click to expand...
Can you tell me what the Figure number is for the diagram? My count seems to be off.
At least one of the diagrams seems to be a description of the layout and redundancy scheme for R600.

liquidboy said:
Agree with you that the VGLeaks specs are NOT GCN1.0 ... After reading those patents it's definitely clear to me that these VSP's are a step beyond what we publicly know as GCN1.0 .. if these VSP's are what is in Durango then damn :)
Click to expand...
The vector scalar processor in patent US7577869, figure 10 is almost a decade behind what we know is GCN 1.0, and it has a transcendental unit that the rest of the leaked data explicitly said wasn't in Durango.

Which parts of which sources are we cherry-picking?
 
Status
Not open for further replies.

Similar threads

BRiT
The Annual E3 Microsoft press conference thread, 2019 edition
8 9 10
Replies
191
Views
25K
eastmen
E
BRiT
  • Sticky
System Reservations on PS4, 4Pro, PS5, XBox One, One S, One X, and Xbox Series X|S [2019-12, 2020-03]
2 3 4
Replies
71
Views
30K
LiveGamer
L
A
Everwild by Rare [XO, XBSX|S, PC, XPA]
2 3
Replies
49
Views
8K
Johnny Awesome
J
BRiT
The Annual E3 Microsoft press conference thread, 2018 edition
9 10 11
Replies
213
Views
26K
Silent_Buddha
S
A
All purpose Sales and Sales Rumours and Anecdotes [2019 Edition]
21 22 23
Replies
446
Views
53K
bgroovy
B
Back
Top