VGLeaks rumor: Durango CPU Overview

I completely agree with you. I think the ESRAM will be used as somewhat of a software controlled cache (kind of like an L3 equivalent).

According to vgleaks, for the latency for, CPU a L1 miss is around 17 cycles and, L2 miss is around 144-160 cycles. For the GPU since it's running half the clock, that will be half that. So basically if the GPU runs out of memory and blows through both caches, you stall for 80 or so cycles to get data from the DDR3.

Acert on Beyond3d mentioned that he heard the SRAM latency is around 16-20 cycles. If true, that's 4-5x faster latency than DDR3 and will definitely reduce the number of stalls in cases where the L2 cannot be prefetched.

My theory is that this design is driven for tile based rendering where the main DDR3 ram will act mainly like a read-only buffer for the inputs (Textures, vertices, etc) and write only for the final rendered tile.
If the tiles are sized appropriately, then I think all the intermediate data should stay in the SRAM and not need to be written back to main memory. The move engines are then used for all the tile data prefetch form main DDR and the final write to DDR. You could have a triple buffered scheme, where you're moving data for the next tile into the SRAM, processing the current tile, and writing out the previously processed tiled back to main memory all in parallel. That way all the data movement is hidden and the GPU can just process continually.
Thanks for the explanation. Tile based rendering seems like a good idea given that we know at the moment of the Durango architecture. Would that put significant limitations to the choices available to third party devs, related to their multi-platform engines?
 
Not as big of a factor as everyone is making it out to be. Dealing with high latency/high bandwidth memory is something APU's were designed around.

They are designed in such a way that the cpu and the gpu work closely together, yes but compute job are still very latency sensitive. Personally I think that is one of the reason for the increased ACEs in the ps4 gpu. Modern gpus are multithreaded so they are more likely to hide latency by switching to another thread, but compute is still an issue.
 
The increase in ACEs were more to address GCN's issue with switching between rendering and GPGPU, the 8 ACEs adds more granularity to the entire thing (basically makes the GPU more efficient at switching between GPGPU tasks and rendering tasks). And yeah one of if not the main reason behind APUs is allowing the CPU and GPU to access and work on the same piece of data without having to move/copy to their own pool of memory.
 
To people who use emotive phrases like these, promises always appear far more important than quality. Kinect could be so much better if ms actually put the required effort and research into improving its very apparent pitfalls. However, due to clever marketing and buyers like above they've clearly settled on "it's just about good enough so why spend money improving it" - if the rumored specs are anything to go by.

You may complain or draw parallels to the wiimote or even the move but both of those actually work. On few to no occasions would you find yourself fighting the controls when using them even when things get frantic. kinect on the other hand is a lag filled inaccurate mess when it's actually doing its main function which is "tracking full body movement". Doing lesser functions like pose recognition and hand tracking are a bit laggy but within reason at least - most times.

Kinect can be so much better but it's clear it might never get there if we don't talk with our wallets

Kinect works. Owned it since the day of launch. In fact, I own all 3 with at least half a dozen games for each.

You can pretend Kinect doesnt work, you can add your on hyperbole, and you can resort to the same Kinect troll-a-thon that most resort too, but at the end of the day if Kinect was really what the haters described it would have sold zero after word of mouth got out to the fake 'lag filled inaccurate mess' we all would have seen it as.

Nope. It works even better than the Wii OR Move for many experiences that those two are incapable of achieving.

The fact, is we have talked with our wallets and the results are clear to which system is having the best response since both have launched.

Kinect is the most successful post Wii moton control peripheral hardware and sofware and MS took steps to improve it as they should.
 
Kinect works. Owned it since the day of launch. In fact, I own all 3 with at least half a dozen games for each.

You can pretend Kinect doesnt work, you can add your on hyperbole, and you can resort to the same Kinect troll-a-thon that most resort too, but at the end of the day if Kinect was really what the haters described it would have sold zero after word of mouth got out to the fake 'lag filled inaccurate mess' we all would have seen it as.

Nope. It works even better than the Wii OR Move for many experiences that those two are incapable of achieving.

The fact, is we have talked with our wallets and the results are clear to which system is having the best response since both have launched.

Kinect is the most successful post Wii moton control peripheral hardware and sofware and MS took steps to improve it as they should.

.
 
No need to pretend. The games speak for themselves

Your self professed love blinds your judgement but that's understandable

Not blind at all, but yup I still love how the success of Kinect make you resort to callling me blind. (but that is understandable)

The games do speak for themselves which is why the 4 month old Dance Central 3 has a Metacritic of 86.

It's why you have a series like Kinect Sports selling 7 million total with only two games released.

In the motion control categories that made the most impact for the Wii in terms of sales, Kinect has proven itself more than capable.

But, just like the Wii and the Move you have games that work great and are successful and then you have the ones that don't.

Kinect has obviously been a big reason the 360 has had the success it's had going into year 8.

All motion control devices share their strengths and weaknesses, like Sony it's clear that MS has chosen to improve on theirs.
 
No need to pretend. The games speak for themselves

Dance Central 3 was in the top 20 for best rated games of 2012.

Edit: Beaten.

Overall it works good for some things, doesn't work good for others. It isn't a waste completely IMO. I have ideas myself for (optional) game implementation and am interested in seeing the improvements and new implementations that kinect will have next gen.
 
Dance Central 3 was in the top 20 for best rated games of 2012.

Edit: Beaten.

Overall it works good for some things, doesn't work good for others. It isn't a waste completely IMO. I have ideas myself for (optional) game implementation and am interested in seeing the improvements and new implementations that kinect will have next gen.

DC3 much like every other DC game requires nothing but pose recognition. Not exactly complicated stuff nor very exciting. It's like touting singstar or DDR

And nowhere have i ever stated kinect is pointless anyway. I've clearly stated what its strengths and weaknesses are. It just so happens that the holy grail of full body motion tracking, which was supposed to be its main selling point, does not work well.

Joe Soap is easy to please. For him wiimotes were accurate enough and wii motion + was unnecessary. Thankfully, or at least one would hope, us gaffers are a bit more technology and quality conscious
 
I see the Kinect defense force has invaded the CPU thread, great.
 
I hate Kinect but my wife likes it. I'd say my opinion is pretty middle of the road. I just don't want a console compromised (price, processing power, whatever) for the sake of Kinect.
 
I know this is a CPU thread but damn I felt like a complete bad ass playing Child of Eden in the middle of the night with Kinect. Awesome game.

On topic, hopefully we get some new leaks start trickling in. And wasn't VGleaks supposed to have run an article on the "Memory System"? I remember someone posting a pic of the memory system obtained via a backdoor from VGleaks.
 
I know this is a CPU thread but damn I felt like a complete bad ass playing Child of Eden in the middle of the night with Kinect. Awesome game.

On topic, hopefully we get some new leaks start trickling in. And wasn't VGleaks supposed to have run an article on the "Memory System"? I remember someone posting a pic of the memory system obtained via a backdoor from VGleaks.
Yeah, I dont think they ever put it in an article. more to milk I guess

picture was http://www.vgleaks.com/wp-content/uploads/2013/02/memory_system.jpg
 
Be prepared for EVERYTHING internal to MS to push the Kinect funtionality in some way. Whether by voice or motion activation. MS is going after the casual market, and doesn't really care about "HC" gamers anymore.

They are going where the money is - and that is a multi-media console that will compete with Apple TV for the core entertainment in your living room. Those expecting otherwise will be sorely disappointed, as that has been their focus for years now, and the reason most of the "old" XBox team is gone.

Always connected, media center box. And it will be hugely successful, and destroy gaming as the "HC" group know it.
 
Always connected, media center box. And it will be hugely successful, and destroy gaming as the "HC" group know it.

Bullshit IMHO. Gaming doesn't need to be 500000 GB 1 zillion polygon FPS games. I actually play more XBLA games nowadays. HC gaming is moving to small Indie stuff.
 
I hate Kinect but my wife likes it. I'd say my opinion is pretty middle of the road. I just don't want a console compromised (price, processing power, whatever) for the sake of Kinect.

this I can understand. Even though I was planning on buying a Kinect 2.0 anyway I wish it was a separate purchase for me so that more could have been done techwise for the 720 SKU. I don't know how much it eats into the cost of the box but it damn sure ain't free.
 
DC3 much like every other DC game requires nothing but pose recognition. Not exactly complicated stuff nor very exciting. It's like touting singstar or DDR

And nowhere have i ever stated kinect is pointless anyway. I've clearly stated what its strengths and weaknesses are. It just so happens that the holy grail of full body motion tracking, which was supposed to be its main selling point, does not work well.

Joe Soap is easy to please. For him wiimotes were accurate enough and wii motion + was unnecessary. Thankfully, or at least one would hope, us gaffers are a bit more technology and quality conscious


So let me get this straight Joe Soap is easy to please and because you are on GAF your hopefully more techonlogy and quality driven?

Did you really just pretend that stomping on arrows on a DDR matt or waving a Wii-mote is comparable to Dance Central?

Do you really want to claim that obtaining a score of 86 on Metacritic doesn't indicate quality?

Is it really so easy to program a game like Kinect Sports or even Kinect Adventures, when I have yet to hear any developer claim that developing a full game for Kinect is even close to being 'easy'.

This isn't even about some self procalimed elitist view that Kinect is so far below the standards tech wise than the average gaffer who you claim you speak for.

Even a child knows the difference between the restrictions of using a controller or a Wii-mote to simulate dancing, running, boxing, exercising, etc. for games like K.Sports, Dance Central, and Fitness Evolved.

Again the bottomline is millions of people enjoyed the Wii because it was fun and that's why millions bought Kinect and as well as Sony's Move.

If MS continues to improve the capabilities and still show fun software then the next iteration of Kinect has a chance to double the success of the original.


Edit

Just responding to what was written like anyone else on GAF does on any subject.


I know this is a CPU thread but damn I felt like a complete bad ass playing Child of Eden in the middle of the night with Kinect. Awesome game.

On topic, hopefully we get some new leaks start trickling in. And wasn't VGleaks supposed to have run an article on the "Memory System"? I remember someone posting a pic of the memory system obtained via a backdoor from VGleaks.

I gotta second this. Child of Eden alone was my favorite 'Oh Sh-t' moment with Kinect. They should have called it Child of Jedi, because that's how I felt everytime I played it. Ubisoft was definately onto something even though we will probably never get a sequel.

I hope the devs can push the new power of next gen into some really tangible immersive experiences in the motion control realm as well as w/ the standard controller.
 
Nemesis, you are an example of what I'm talking about. Sony is going under this generation, and MS will then focus its sights on Nintendo. It has been MS MO for decades now to dominate, buy out, or strangle the competition.

Sony is going in the wrong direction from a "revenue generating" standpoint. MS knows the casual gamer is where the money is. This isn't about video games for MS, and it hasn't been for years.

I think MS will not only win this next generation, but do so by a landslide. And "gaming" will be dead as I knew it when younger. It will be all about short, visual games and moving on to the next one. Casual gamers just don't have the attention span for a $60 game that takes weeks to complete.
 
Nemesis, you are an example of what I'm talking about. Sony is going under this generation, and MS will then focus its sights on Nintendo. It has been MS MO for decades now to dominate, buy out, or strangle the competition.

Sony is going in the wrong direction from a "revenue generating" standpoint. MS knows the casual gamer is where the money is. This isn't about video games for MS, and it hasn't been for years.

I think MS will not only win this next generation, but do so by a landslide. And "gaming" will be dead as I knew it when younger. It will be all about short, visual games and moving on to the next one. Casual gamers just don't have the attention span for a $60 game that takes weeks to complete.

Sorry but there's been "casual games" on all generations.

EDIT: New codename for DirectX: Blue
 
The vast majority of WoW players were casual and the reason Zynga made so much money when it was the hot shit is because it managed to hook people into its games for extended periods and suck them into thw microtransactions.
 
The vast majority of WoW players were casual and the reason Zynga made so much money when it was the hot shit is because it managed to hook people into its games for extended periods and suck them into thw microtransactions.

your avatar is a perfect fit for the sentiment felt when reading the bolded
 
Apparently you guys missed the "there will be blood" style message from a mod in the last thread that got shut down when people turned a thread that wasn't about Kinect into one that was.
 
Yes, the rendering portion, the part that makes up the pretty graphics. Both system memory and GPU memory has been going towards higher latency and higher bandwidth for.. ever? DDR->DDR2->DDR3->DDR4 the trend is higher latency and higher bandwidth. GDDR3->GDDR5 pure bandwidth for rendering. Yet, MS is going to buck the trend? I don't buy it, they are using eSRAM for its bandwidth just like they did with the eDRAM. The lower latency will not hurt, but it is not the main driver. They know that 68GB/s with so many sub-systems using the memory will run into contention and bandwidth issues, so they added a separate pool of fast cache that they offload the main memory.
I can buy it. They already have a memory with lots and lots of bandwidth tied to the ROPs: Their caches. 360 design can be seem as increasing their size so they can hold a huge partion of the screen. Durango as it seems has a memory that can be accessed almost at the same time as a cache, and coupled with units that can tile the screen can keep the rops caches fed so the most bandwidth intense operations happens there.

The advantage of durango's design is that the entire system now have access to the lower latency pool, instead of being just a cache to the rops.

I don't discredit that increasing the system bandwidth was part of the design, but i defend the view that everything was designed as a whole, not as a late patch... The patents and leaks already paint a fairly clear view of their design goals.

Why use DDR3 with highest CAS timings found in DRAM if they want low latency and don't care about bandwidth? Thy are using the highest latency and highest bandwidth DDR3 available today for the bandwidth. But it is not high enough, so they went back the 360/PS2 model of a fast scratch pad to alleviate the bottleneck.
They don't need to have low latency on everything. They need a low latency pool that's just large enough for their goals.

Increasing the main bandwidth is not the goal of the esram, that's one of the many goals of the DMEs (they obviously need 2 pools to actually increase bandwidth, but they are not tied to esram). The esram was chosen for other purposes other than pure raw bandwidth.
 
I can buy it. They already have a memory with lots and lots of bandwidth tied to the ROPs: Their caches. 360 design can be seem as increasing their size so they can hold a huge partion of the screen. Durango as it seems has a memory that can be accessed almost at the same time as a cache, and coupled with units that can tile the screen can keep the rops caches fed so the most bandwidth intense operations happens there.

The advantage of durango's design is that the entire system now have access to the lower latency pool, instead of being just a cache to the rops.

I don't discredit that increasing the system bandwidth was part of the design, but i defend the view that everything was designed as a whole, not as a late patch... The patents and leaks already paint a fairly clear view of their design goals.


They don't need to have low latency on everything. They need a low latency pool that's just large enough for their goals.

Increasing the main bandwidth is not the goal of the esram, that's one of the many goals of the DMEs (they obviously need 2 pools to actually increase bandwidth, but they are not tied to esram). The esram was chosen for other purposes other than pure raw bandwidth.

Indeed. If not they could have gone for eDRAM, and a higher capacity one at that, if low latency was not a requirement. And it would have been cheaper too.
 
It will be close in power to the PS4, maybe not as close as the HC crowd likes, but close enough for casual gamers. And that is all that matters in the long run.

The 720 will be weaker than the PS4, but not enough to make a difference.
 
They don't need to have low latency on everything. They need a low latency pool that's just large enough for their goals.

Increasing the main bandwidth is not the goal of the esram, that's one of the many goals of the DMEs (they obviously need 2 pools to actually increase bandwidth, but they are not tied to esram). The esram was chosen for other purposes other than pure raw bandwidth.

Each move engine can read and write 256 bits of data per GPU clock cycle, which equates to a peak throughput of 25.6 GB/s both ways. Raw copy operations, as well as most forms of tiling and untiling, can occur at the peak rate. The four move engines share a single memory path, yielding a total maximum throughput for all the move engines that is the same as for a single move engine. The move engines share their bandwidth with other components of the GPU, for instance, video encode and decode, the command processor, and the display output. These other clients are generally only capable of consuming a small fraction of the shared bandwidth.
http://www.neogaf.com/forum/showthread.php?t=511923

The DMEs don't help bandwidth at all. How is the eSRAM going to help latency for the overall system when a.) it is only 32MB assiting 8GB of total RAM (0.4%) and b) you have to know what data to keep in the eSRAM, the devs are going to magically know what needs to be in eSRAM before they use it so that they can benefit from low latencies?

If it is just a dumb cache like a L3 that works with the CPU transparently, I can buy your logic, but the eSRAM isn't even directly connected to the CPU according to what has leaked, just the GPU.

memory_system.jpg
 
It will be close in power to the PS4, maybe not as close as the HC crowd likes, but close enough for casual gamers. And that is all that matters in the long run.

The 720 will be weaker than the PS4, but not enough to make a difference.

There is a lot of truth to this post.


but the thing is it seems that the differences seems to matter more now than it did back in the day for some reason & we have people pixel counting & voicing all the difference between the ports over the internet so the difference will be well known & after all the back & forwards about the differences in PS3 & Xbox 360 ports this Generation there is noway that the over 1.5X more powerful GPU & faster/more memory for games will be ignored & go unnoticed.
 
http://www.neogaf.com/forum/showthread.php?t=511923

The DMEs don't help bandwidth at all. How is the eSRAM going to help latency for the overall system when a.) it is only 32MB assiting 8GB of total RAM (0.4%) and b) you have to know what data to keep in the eSRAM, the devs are going to magically know what needs to be in eSRAM before they use it so that they can benefit from low latencies?

If it is just a dumb cache like a L3 that works with the CPU transparently, I can buy your logic, but the eSRAM isn't even directly connected to the CPU according to what has leaked, just the GPU.

memory_system.jpg

They don't add bandwidth by themselves, they do by moving data in advance so when the consumer needs that data it can fetch from 2 or more pools at once.

Increasing the overall system memory performance without resorting to large buses (that increases cost and consumption) is even cited by the patent as their primary goal.
 
So the move engines move data "in advance" to the eSRAM to facilitate bandwidth issues, but the eSRAM was chosen for low latency and not bandwidth?
 
It will be close in power to the PS4, maybe not as close as the HC crowd likes, but close enough for casual gamers. And that is all that matters in the long run.

The 720 will be weaker than the PS4, but not enough to make a difference.

That's a bingo. Gaffers may like the sound of PS4 specs and philosophy better than MS' strategy but they do have a very good one. Less powerful but not less enough to get left behind like the Wii/U and with a good scaler to make up any possible resolution drops. Combine that with all the extra Live features and they'll have a successful console that they don't have to break the bank to put out. Sony's taking a gamble by targeting enthusiasts and MS is taking a gamble with their wider approach. Only time will tell which one pays off.
 
That's a bingo. Gaffers may like the sound of PS4 specs and philosophy better than MS' strategy but they do have a very good one. Less powerful but not less enough to get left behind like the Wii/U and with a good scaler to make up any possible resolution drops. Combine that with all the extra Live features and they'll have a successful console that they don't have to break the bank to put out. Sony's taking a gamble by targeting enthusiasts and MS is taking a gamble with their wider approach. Only time will tell which one pays off.

For people who like Xbox first party games, Xbox 720 + PC can be the solution (as with Nintendo gamers).
 
That's a bingo. Gaffers may like the sound of PS4 specs and philosophy better than MS' strategy but they do have a very good one. Less powerful but not less enough to get left behind like the Wii/U and with a good scaler to make up any possible resolution drops. Combine that with all the extra Live features and they'll have a successful console that they don't have to break the bank to put out. Sony's taking a gamble by targeting enthusiasts and MS is taking a gamble with their wider approach. Only time will tell which one pays off.

sounds like an interesting gamble by both companies, I wonder who will be on top near the end of the generation..might be a really interesting time for gamers.
 
sounds like an interesting gamble by both companies, I wonder who will be on top near the end of the generation..might be a really interesting time for gamers.

It's especially interesting since it's the very same gamble Sony just took with the Vita and PS3 and it failed miserably for them. It's interesting for Microsoft because if they're going the casual-friendly route, they're doing it with a device that's already been released once and has essentially died already (the software anyway). Nintendo just learned what happens when you alienate the core too much, so if Microsoft is going to walk that tight-rope between core and casual, they better have good balance.
 
It's especially interesting since it's the very same gamble Sony just took with the Vita and PS3 and it failed miserably for them.

No it's not. The PS4 seems much more powerful and will have an advantage in every game, unlike the PS3. The PS4 is not comparable with the PS3, which wasn't much more powerful than the competition, was hard to develop for and expensive. And a Vita comparison is just absurd, totally different product and market. Unlike the Vita, the PS4 will have unlimited support from all major third party developers, and they seem to be quite happy with the specs.

It's interesting for Microsoft because if they're going the casual-friendly route, they're doing it with a device that's already been released once and has essentially died already (the software anyway). Nintendo just learned what happens when you alienate the core too much, so if Microsoft is going to walk that tight-rope between core and casual, they better have good balance.

I think Microsoft will lose a very big amount of core gamers. They can't launch at the same time as the PS4 with these specs and think core gamers will stay. If they would have launched a year ago, this strategy might have worked. And price won't be a big issue this time, the PS4 won't be $599, and core gamers / early adoptors don't care about a $50-100 price difference.

That's a bingo. Gaffers may like the sound of PS4 specs and philosophy better than MS' strategy but they do have a very good one. Less powerful but not less enough to get left behind like the Wii/U and with a good scaler to make up any possible resolution drops. Combine that with all the extra Live features and they'll have a successful console that they don't have to break the bank to put out. Sony's taking a gamble by targeting enthusiasts and MS is taking a gamble with their wider approach. Only time will tell which one pays off.

Imo Microsoft's approach is much more riskier. Their target group is extremely volatile and we don't even know if they are interested in this at all. Targeting core gamers is a safe bet and worked since the beginning of console gaming. And I don't think there will be just a small resolution difference, which can be mitigated by a "good scaler". There will probably be a lot more differences.
 
Why no? what's really matter are the games I think.

Why would core gamers which are not interested in Microsoft's first party output stay with the Xbox? All those CoD / BF / Madden / FIFA players? To mitigate this, Microsoft would have to massively increase their first party output. But that would mean they would have to invest a lot of money. Which I think they don't want to do - or why did they choose such a relatively cheap console design in the first place?
 
For all this talk about "core gamer", I don't think anybody can even identify what that word really means. One thing I can say for certain is that GAF is not representative of the core gamer. I'd say GAF represents a very miniscule slice of the overall core gamer demographic, whatever that is.

Remember, GAF was head over heals for the Vita and its dead in virtually every market on this planet. Enthusiasm here does not reliably translate to the buying decisions of any demographic other than the most hardcore of hardcore gamers. Using expectations here to predict what the core gamer will do is just a futile exercise.
 
Top Bottom