Lengthy interview on Xbox One architecture

#21FusionCPosted 9/22/2013 8:16:05 AM
JWJW99 posted...
ElPolloDiablo87 posted...
JWJW99 posted...
This article confirms several things we knew:
1) The PS4 unified memory approach is easier to program for than the Xbone
2) Most of the fancy audio HW in Xbone is reserved for Kinect processing
3) The 218GB/s number Albert Pennello came up with was totally bogus. The 204GB/s number is unachieveable on actual code. The best real code will ever do is 140-150/GB/s, and that's only if the memory access pattern of read/writes/banks is perfect. Otherwise, you get 109GB/s.
4) The Move Engines are Microsoft just renaming DMA engines which have been around for decades.
5) Only Xbone has a an HDMI input, which helps make it the "center of your home entertainment system"


Plagiarism is bad, mmkay?

http://www.neogaf.com/forum/showpost.php?p=83120317&postcount=346


So you agree with the statement...

Good now this thread can move on as PR talk.


Actually almost none of this stuff was in this article. And I have no idea who wrote it on GAF and you stole it from, but bro, you need to take it back. (It does have a HDMI input, so at least you were 1 for 5)
---
For the Horde!!!!
#22JWJW99Posted 9/22/2013 8:27:47 AM
FusionC posted...
The one thing that I got out of the article was how similar the programming will be compared to the 360. People made it sound like accessing the eSRAM was going to be some act of god. But in reality its something that that devs have been doing since day one on the 360 so the transition shouldnt be hard at all.


I don't think anybody was comparing the development time in hardness to the 360 of the One. It's more that the 360(since its similar to one but def not the same architect) would be harder to develop for than PS4

Estimating time frame

360 = 3-6 months
PS3 = 5-8 months

People said 360 was easier to develop for and PS3 was harder....

Now

One = 3-6 months
PS4 = 1-4 months

So in terms they have switched as the One is harder to develop for compared to the PS4 now. I believe people just used the term "harder to develop for" in a general saying.
#23ElPolloDiablo87(Topic Creator)Posted 9/22/2013 8:31:32 AM
JWJW99 posted...
FusionC posted...
The one thing that I got out of the article was how similar the programming will be compared to the 360. People made it sound like accessing the eSRAM was going to be some act of god. But in reality its something that that devs have been doing since day one on the 360 so the transition shouldnt be hard at all.


I don't think anybody was comparing the development time in hardness to the 360 of the One. It's more that the 360(since its similar to one but def not the same architect) would be harder to develop for than PS4

Estimating time frame

360 = 3-6 months
PS3 = 5-8 months

People said 360 was easier to develop for and PS3 was harder....

Now

One = 3-6 months
PS4 = 1-4 months

So in terms they have switched as the One is harder to develop for compared to the PS4 now. I believe people just used the term "harder to develop for" in a general saying.


Oh hey, random numbers with no sources!
---
Madre de Dios, es El Pollo Diablo!
http://www.youtube.com/watch?v=E28WrhpTzQA
#24JWJW99Posted 9/22/2013 8:39:42 AM
FusionC posted...
JWJW99 posted...
ElPolloDiablo87 posted...
JWJW99 posted...
This article confirms several things we knew:
1) The PS4 unified memory approach is easier to program for than the Xbone
2) Most of the fancy audio HW in Xbone is reserved for Kinect processing
3) The 218GB/s number Albert Pennello came up with was totally bogus. The 204GB/s number is unachieveable on actual code. The best real code will ever do is 140-150/GB/s, and that's only if the memory access pattern of read/writes/banks is perfect. Otherwise, you get 109GB/s.
4) The Move Engines are Microsoft just renaming DMA engines which have been around for decades.
5) Only Xbone has a an HDMI input, which helps make it the "center of your home entertainment system"


Plagiarism is bad, mmkay?

http://www.neogaf.com/forum/showpost.php?p=83120317&postcount=346


So you agree with the statement...

Good now this thread can move on as PR talk.


Actually almost none of this stuff was in this article. And I have no idea who wrote it on GAF and you stole it from, but bro, you need to take it back. (It does have a HDMI input, so at least you were 1 for 5)



Wait who said that was in the article? I see somebody took the reassurance bait...

I find it funny that gamefaq members can claim to understand a Microsofts Engineer Techinal article as if they are on there same level. And they try to claim that as facts to prove some other article that plays out the PS4 as the bread winner false.



Amazing how much reassurance there is to justify a product purchase.
#25embrandedonePosted 9/22/2013 8:46:44 AM
JWJW99 posted...
ElPolloDiablo87 posted...
JWJW99 posted...
This article confirms several things we knew:
1) The PS4 unified memory approach is easier to program for than the Xbone
2) Most of the fancy audio HW in Xbone is reserved for Kinect processing
3) The 218GB/s number Albert Pennello came up with was totally bogus. The 204GB/s number is unachieveable on actual code. The best real code will ever do is 140-150/GB/s, and that's only if the memory access pattern of read/writes/banks is perfect. Otherwise, you get 109GB/s.
4) The Move Engines are Microsoft just renaming DMA engines which have been around for decades.
5) Only Xbone has a an HDMI input, which helps make it the "center of your home entertainment system"


Plagiarism is bad, mmkay?

http://www.neogaf.com/forum/showpost.php?p=83120317&postcount=346


So you agree with the statement...

Good now this thread can move on as PR talk.


Pollo couldn't break it into PR school. Too much 'copy pasting' and not enough 'Cloud' to be a paid shill. Think the Jester to the Big Bad, in RPG terms.

I like a few of these things

- PS4 has balance and MS copied them
- He agrees the Sony's engineer did a good job (backtracking from how specs don't matter)
- Mentions how dynamic resolution is going to be used a lot (I'm guessing we'll be hearing this a lot this gen)
- Mentions how he has no games to back up his claims

All'n'all interesting read and it's fun watching them leave carpet stains as they backtrack from being the most powerful system to 'we got da balance'

You have the mic, Shills, but please no backtracking. I've just had my rug cleaned.
---
http://www.youtube.com/watch?v=HzLUYX7M1SQ -- The Xbox One Board
#26JWJW99Posted 9/22/2013 8:55:39 AM
embrandedone posted...
JWJW99 posted...
ElPolloDiablo87 posted...
JWJW99 posted...
This article confirms several things we knew:
1) The PS4 unified memory approach is easier to program for than the Xbone
2) Most of the fancy audio HW in Xbone is reserved for Kinect processing
3) The 218GB/s number Albert Pennello came up with was totally bogus. The 204GB/s number is unachieveable on actual code. The best real code will ever do is 140-150/GB/s, and that's only if the memory access pattern of read/writes/banks is perfect. Otherwise, you get 109GB/s.
4) The Move Engines are Microsoft just renaming DMA engines which have been around for decades.
5) Only Xbone has a an HDMI input, which helps make it the "center of your home entertainment system"


Plagiarism is bad, mmkay?

http://www.neogaf.com/forum/showpost.php?p=83120317&postcount=346


So you agree with the statement...

Good now this thread can move on as PR talk.


Pollo couldn't break it into PR school. Too much 'copy pasting' and not enough 'Cloud' to be a paid shill. Think the Jester to the Big Bad, in RPG terms.

I like a few of these things

- PS4 has balance and MS copied them
- He agrees the Sony's engineer did a good job (backtracking from how specs don't matter)
- Mentions how dynamic resolution is going to be used a lot (I'm guessing we'll be hearing this a lot this gen)
- Mentions how he has no games to back up his claims

All'n'all interesting read and it's fun watching them leave carpet stains as they backtrack from being the most powerful system to 'we got da balance'

You have the mic, Shills, but please no backtracking. I've just had my rug cleaned.


What's funny is that he acted like he busted me for taking credit of somebody else's post but he also forgot to mention that everything else posted in the xbox forum literally comes from neogaf but yet people post new threads as if they found the article or wrote it themselves including him.

The denial is so strong
#27Foxx3kPosted 9/22/2013 9:15:47 AM
It's really not shocking to me that fanboys are confusing respect with endorsement.
---
[LanParty nF4 Ultra-D] [AMD64 3700+ San Diego] [2x 1gb Corsair XMS 3-3-3-8] [2x 250gb Barracuda] [Soundblaster Audigy 2 ZS] [X850XTPE]
#28airdanielPosted 9/22/2013 9:20:59 AM
Looks promising. Thanks for the info.
---
GT: Rancub
#29RollnThunder213Posted 9/22/2013 10:17:22 AM(edited)
SoulTrapper posted...
Interesting article, I do have one problem with it:

Tying into that previous block, Digital Foundry's own analysis came to the same conclusion that MS presented: that more CUs does not linearly scale to more power. They had a mock-PS4 PC (with roughly the same specs) and a mock-XBone PC, and despite the PS4 having 50% more CUs, it only yielded a 24% advantage. While this is in no way saying that the GPU in XBone is superior to PS4's, it means that if MS is smart with optimization and Sony isn't, any performance gap in the area of GPUs would be largely mitigated by the time final drivers roll to devkits.

So pretty much all of the talk about the GPU is theoretical BS assuming that Sony doesn't know how to optimize.
If they do, the PS4 will possibly be more than 50% better.


The point is that despite have 50% more CUs it doesn't just result in a 50% advantage for the PS4's GPU. The Xbox One has 14 CUs in the hardware, so if that was the end all be all they would allow developers to use them, they instead found better gains with the upclock than the increase in CUs. The benchmarks done by DF only come out to only 24% advantage compared to the 50% more CUs. He said if Microsoft's optimization is good and Sony's isn't, that there would 0% advantage, not "MORE than 50%". You're heading the wrong direction. Of course if both optimization efforts are successful then it's variable a bit more, but still not "50% better"

The only way to know is to wait until both consoles are out and full benchmark tests can be done on the finished hardware if you're that interested.

sworder posted...
As far as the topic, this only proves that MS should have gone with GDDR5. Easy 176GB/s > kinda complex ~150 GB/s

Reading this, it's clear PS4 will still be more powerful and simpler to develop for but MS has made sure it won't be too noticeable


You're mixing up your bandwidths in MB or GB.

RAM in PS4 176MB/s

RAM in Xbox One 68MB/s for main, 150GB/s for the 32MB of ESRAM (in read/write of actual code)

Both systems are just going to be using resources inefficiently and using brute force to play games at launch. The system's little differences aren't going to show until a few years from now when developers totally understand how to bring out the advantages of each console's hardware intricacies.
---
"No control beyond these skies"
GT: RollnThunder213
#30METALINGUS5150Posted 9/22/2013 10:21:12 AM
RollnThunder213 posted...
SoulTrapper posted...
Interesting article, I do have one problem with it:

Tying into that previous block, Digital Foundry's own analysis came to the same conclusion that MS presented: that more CUs does not linearly scale to more power. They had a mock-PS4 PC (with roughly the same specs) and a mock-XBone PC, and despite the PS4 having 50% more CUs, it only yielded a 24% advantage. While this is in no way saying that the GPU in XBone is superior to PS4's, it means that if MS is smart with optimization and Sony isn't, any performance gap in the area of GPUs would be largely mitigated by the time final drivers roll to devkits.

So pretty much all of the talk about the GPU is theoretical BS assuming that Sony doesn't know how to optimize.
If they do, the PS4 will possibly be more than 50% better.


The point is that despite have 50% more CUs it doesn't just result in a 50% advantage for the PS4's GPU. The Xbox One has 14 CUs in the hardware, so if that was the end all be all they would allow developers to use them, they instead found better gains with the upclock than the increase in CUs. The benchmarks done by DF only come out to only 24% advantage compared to the 50% more CUs. He said if Microsoft's optimization is good and Sony's isn't, that there would 0% advantage, not "MORE than 50%". You're heading the wrong direction. Of course if both optimization efforts are successful then it's variable a bit more, but still not "50% better"

The only way to know is to wait until both consoles are out and full benchmark tests can be done on the finished hardware if you're that interested.

sworder posted...
As far as the topic, this only proves that MS should have gone with GDDR5. Easy 176GB/s > kinda complex ~150 GB/s

Reading this, it's clear PS4 will still be more powerful and simpler to develop for but MS has made sure it won't be too noticeable


You're mixing up your bandwidths in MB or GB.

RAM in PS4 176MB/s

RAM in Xbox One 68MB/s for main, 150GB/s for the 32MB of ESRAM (in read/write of actual code)

Both systems are just going to be using resources inefficiently and using brute force to play games at launch. The system's little differences aren't going to show until a few years from now when developers totally understand how to bring out the advantages of each console's hardware intricacies.


No, in the PS4 it's 176 GB/s
and in the Xbone, it's 68 GB/s for main.