If there's truth to those rumors, why would MS add those extra units if they aren't going to help performance? PCB aesthetics?
Xbox One (Durango) Technical hardware investigation
- Thread starter Love_In_Rio
- Start date
- Status
If there's truth to those rumors, why would MS add those extra units if they aren't going to help performance? PCB aesthetics?
They're there because they accomplish more graphically than the extra GPU silicon taking their place would.
It's pretty obvious.
I'm sure they'll help, but is it a 5% boost or a 25% boost?
This hardware wasn't added to "make up ground," because they most likely didn't know the specs of Orbis as they were working on it. Maybe they've found out stuff through third parties, and caught wind that they were behind, but I doubt they did, and I doubt they'd change their architecture because of it. This was done because it was deemed to be an improvement and cheaper than using a better GPU and the faster bus and memory that would have required.
Maybe they'll go back to the bananarang design from E3 2005 or whenever it was when PS3 was first shown off...
Anyho, I don't buy this whole 'data transfer engine' schtick people are trying to sell in this thread. What possible use would custom silicon for something like that really have. I can't think of a single god damn thing, GPUs have had DMA capabilities basically since the dawn of time. I wouldn't be surprised if even the ancient voodoo graphics could DMA textures and command lists into its own on-board memory over PCI; PowerVR PCX1/2 could, that's for sure. Those cards even transferred finished drawn frames into your 2D video adaptor's framebuffer.
There's no advantage whatsoever to be shuffling data back and forth with some kind of magic DMA hardware, unless there's invisible pieces of the puzzle we're missing, but I'll be damned trying to think what they might be. Just scaling, or possibly compressing stuff doesn't really cut it, there's no huge performance boosts to be gained from that, and in fact any copy you do is just gonna slow things down since you could do actual data processing on either CPU or GPU with that memory bandwidth instead of just a copy op.
This seems like a huge red herring to me, and so far I've not seen anyone really call it out for it either.
I should add that from my previous post I am not against embedded memory or that it can improve throughput and utilization, speed up bottlenecks and avoid certain stalls. Especially if the ESRAM is more than a simple buffer like Xenos. That said we don't know how robust the solution is and we also don't know the sort of bandwidth it has or how accessible it will be to the CPU. A lot of questions. But I have mentioned for years that a large amount of embedded memory like the 360 had--but more robust (full read/write) and accessible to the CPU and GPU could open up some neat approaches.
And a sufficiently large working "cache" for high bandwidth clients allowing budgets to allow for a very large main memory pool could be the best use of dollars in the face of that OR going with a gimped memory system (e.g. Xbox1, PS3) where a full fledged GPU lacks the bandwidth of dedicated GPUs of the same class.
Of course Orbis seems to not fit that bill as it has more bandwidth than the 7850. Pitcairn is considered a very balanced GPU and getting it into a closed box will only help matters, not to mention it being on the same die as the CPU. That, in contrast to a GPU with 1/3rd less execution units, doesn't appear pretty. Sure, the ESRAM may help with utilization (I never got to post about it, but it could allow some really crazy transparancies) but it isn't like GCN is poor here or that the Orbis memory configuration sounds poor. Because both don't appear to be true, which makes the MS PR spin look pretty dubious.
Anything can change once we get firm specs and architectural data on both, but so far the "closed box flops are better! Inflate them 100%!" really looks like heavy duty PR. Very disappointed. It would have been better off just to say they were pitching new approaches and the sweet spot of the marketbase, fiscal uncertainties, etc. and say, "We are going to be cheap at retail" and justify all the corner cutting that way. But it looks like that won't be true either--the only cheap versions will be the subsidized versions. Hence no free Live and the cheap versions will be taxed-versions.
And a sufficiently large working "cache" for high bandwidth clients allowing budgets to allow for a very large main memory pool could be the best use of dollars in the face of that OR going with a gimped memory system (e.g. Xbox1, PS3) where a full fledged GPU lacks the bandwidth of dedicated GPUs of the same class.
Of course Orbis seems to not fit that bill as it has more bandwidth than the 7850. Pitcairn is considered a very balanced GPU and getting it into a closed box will only help matters, not to mention it being on the same die as the CPU. That, in contrast to a GPU with 1/3rd less execution units, doesn't appear pretty. Sure, the ESRAM may help with utilization (I never got to post about it, but it could allow some really crazy transparancies) but it isn't like GCN is poor here or that the Orbis memory configuration sounds poor. Because both don't appear to be true, which makes the MS PR spin look pretty dubious.
Anything can change once we get firm specs and architectural data on both, but so far the "closed box flops are better! Inflate them 100%!" really looks like heavy duty PR. Very disappointed. It would have been better off just to say they were pitching new approaches and the sweet spot of the marketbase, fiscal uncertainties, etc. and say, "We are going to be cheap at retail" and justify all the corner cutting that way. But it looks like that won't be true either--the only cheap versions will be the subsidized versions. Hence no free Live and the cheap versions will be taxed-versions.
Has MS said anything about Durango? Where's the PR spin? I mostly just see fanboys going through the stages of grief.
Sounds like Orbis should be the better pure gaming platform, from what we know so far, but not having seen anything at all from either console, or what features they will include for OS, services or peripherals, I'm not ready to pass any judgement yet.
Sounds like Orbis should be the better pure gaming platform, from what we know so far, but not having seen anything at all from either console, or what features they will include for OS, services or peripherals, I'm not ready to pass any judgement yet.
(((interference)))
Veteran
Well, really only the move engines would help with graphics and maybe the scaler in the video DSP block.
Which is... where... exactly... on the above leak from vgleaks? :smile:
Well that's just it, if the machines are within ~20% the visual differences will probably be very limited much like the current gen. No one has a clue, but we've got people picking winners already, it's as amusing as it is pathetic.
You're no fun.
A 50% advantage of GPU power, for a game console, is something to talk about and speculate. And we'll be having fun until the real numbers are known.
trainplane
Newcomer
They don't know what Sony was going to do but they do know that 1.2 TFLOPS and 12 CU is pretty pedestrian in 2013 for a part that is supposed to last 5+ years.
Yeah, probably, but I'm sure it fit into the price point they're chasing, and fit into the design and BOM that was required to achieve whatever features they had planned for the system. All we know about the system is a crummy block diagram. I'll reserve my judgement for later, but there's no doubt that if these rumours are true that MS has backed off on chasing the "performance crown."
if the specs of both these units are true I will most likely sit out this generation until the consoles are dirt cheap.
However I don't think this makes sense at all on MS's part. They had such success with the 360 because they made it more powerful than the originaly though they'd need. Now this gen they under shoot ?
I guess only time will tell.
At this point can Durango even emulate /run 360 games ?
However I don't think this makes sense at all on MS's part. They had such success with the 360 because they made it more powerful than the originaly though they'd need. Now this gen they under shoot ?
I guess only time will tell.
At this point can Durango even emulate /run 360 games ?
I dunno it seems to make sense to me. They have two sku's, a Win8 pc for the core gamer that values graphics, and the Xbox 720 sku for average/casual gamers that want something better visually, but also want a do everything box and Kinect. If visuals are really that important then go with their Win8 pc sku and get everything at 1080p 60fps for the next 8 years. If you just want a decent bump in visuals but also want more than just games then go with the 720 sku. They have both customer types covered. To me it's more important that the 720 can run metro apps, ship with Kinect, and provide the best software experience rather than it win the horsepower war. As long as it's within 20% of their competitor performance wise that's good enough. The landscape has changed dramatically since 2005, their priorities need to adapt accordingly.
I think the big change is profitability. There's no point in trying to win a console war if you just come out breaking even, or losing money. This way, they think they can turn a profit. Microsoft has deep pockets, but they can't keep risking selling consoles at a loss forever. This isn't the Wii U. It'll be noticeably better than current gen, and it may have a lot of other selling points in terms of home entertainment junk. I guess we'll see how it turns out for them. From my point of view, I hope it does really well because I think it's healthy for the industry to have good competition. I don't want to see either of Sony or MS (or Nintendo) completely blow it.
(((interference)))
Veteran
I'm assuming they have a scaler somewhere, it did wonders for them last time.
Last edited by a moderator:
I would be totally fine with that if all the games were available on both the PC and the 720, including Forza and Halo.
- Status
Similar threads
