tfur
Member
Its 109GB/s. Peak. MS say so in the DF article.
That is the number you should be focused on. Now, throttle that by the ddr3 transfer speed.
Then keep in mind it is at best, only for 32MB.
Its 109GB/s. Peak. MS say so in the DF article.
Actually the way I see it looks like this.
X1, peak 274GB/s average 200GB/s
PS4, peak 176GB/s average 130GB/s
Its 109GB/s. Peak. MS say so in the DF article.
Actually the way I see it looks like this.
X1, peak 274GB/s average 200GB/s
PS4, peak 176GB/s average 130GB/s
I thought the Steam Controller thread might finally put this one to bed... apparently not.
Given how often "power" has been declared to not be a major factor in console sales I have to say it's interesting how a thread about "power" is proving so huge and divisive.
Actually the way I see it looks like this.
X1, peak 274GB/s average 200GB/s
PS4, peak 176GB/s average 130GB/s
Now the only figure that can possibly be in dispute is the PS4 average figure as Sony havent been forthcoming with any test results. But MS state that
If you take the middle of that its 75% hence 130GB/s. Now I have heard guys going on about "The ps4 gets 172GB/s" which has its source in a web site where there is a quote from a director of an indi games company. No unless somehow sony have silently solved the intractable average vs peak bandwidth problem that has been in existence since the year dot, I would suggest thats a mis-quote/mis-speak etc and he was actually referring to the peak bandwidth of 176GB/s.
Maybe its 140, or if they actually managed to achieve 85% they could get 150, but even that is a long way south of 200.
This article goes into depth as to why http://archive.arstechnica.com/paedi...latency-1.html you never achieve peak for any great amount of time.
With regards to they beyond3d bullshit...hummm to be honest I think they hold their posters to pretty high standards. They certainly get well moody when you try and make a direct comparison between the x1 and the ps4!
You have absolutely no proof to back those figures up. Your post is full of conjecture and speculation.
Also, curious post history. Heh.
Uh...
Actually the way I see it looks like this.
X1, peak 274GB/s average 200GB/s
PS4, peak 176GB/s average 130GB/s
Now the only figure that can possibly be in dispute is the PS4 average figure as Sony havent been forthcoming with any test results. But MS state that
If you take the middle of that its 75% hence 130GB/s. Now I have heard guys going on about "The ps4 gets 172GB/s" which has its source in a web site where there is a quote from a director of an indi games company. No unless somehow sony have silently solved the intractable average vs peak bandwidth problem that has been in existence since the year dot, I would suggest thats a mis-quote/mis-speak etc and he was actually referring to the peak bandwidth of 176GB/s.
Maybe its 140, or if they actually managed to achieve 85% they could get 150, but even that is a long way south of 200.
This article goes into depth as to why http://archive.arstechnica.com/paedi...latency-1.html you never achieve peak for any great amount of time.
With regards to they beyond3d bullshit...hummm to be honest I think they hold their posters to pretty high standards. They certainly get well moody when you try and make a direct comparison between the x1 and the ps4!
An indie dev was getting 172GB/s out of PS4 months ago. It's going to be operating near it's theoretical peak for most devs for the entire generation.Surely the only number that isn't proved is the 130Gb/s avaerage ps4 bandwith figure, and I say as much in my post. But I also go onto say how I arrive at that number, what assumptions I have made to achieve it. I reference MS saying that you tend to run at 70-80 percent effeciency on an external interface, and the linked article explains why you dont get peak all the time.
You know this and still you say crap like this...
X1, peak 274GB/s average 200GB/s
PS4, peak 176GB/s average 130GB/s
What does the Steam controller have to do with the power difference between the PS4 and Xbone?
Yes thats because there are two interfaces to the esram, allowing reads and writes to occur simultanously. Its all in the DF article. I am not making it up!
Oh good God! That article has created a never ending nightmare.
If these numbers were true it would be like saying the PS4 has 256x more memory than the Xbone (32MB vs 8192MB). Stop using a tiny cache as the be-all, end-all.
According to OldSchoolNerd's maths (and microsoft), x360 has a 256 GB/s eDRAM
Creative math taught in Redmond.According to OldSchoolNerd's maths (and microsoft), x360 has a 256 GB/s eDRAM + 22,4 GB/s GDDR3, adding up to 280GB/sec! Bandwidth monster !
Cancel nextgen
Hasn't there been confirmation it's more of a scratch pad than actual cache?
The way you "see" it isn't the way it actually works. It would behoove you to check in with reality every once in a while.Actually the way I see it looks like this.
That's because you are cherry picking info that confirms your bias. Just because MS "Measured" 200GB/s in one game does not imply it is averaging 200GB/s or that figure holds up for more than a few operations or in more than one engine.
It doesn't really matter the exact number, the fact that all the data is accessible by the amount is what matters. It has enough for 18CUs according to AMDs own released line of Southern Island cards. How much is needed by 12CUs, 200GB/s? No, more like 100GB/s.
No system RAM ever achieves peak continually. The DDR3 in the x1 is 68GB/s but they only get 50-55Gb/s average out of it. The ESRAM achieves peak more readily because it has a dedicated read bus and a dedicated write bus, but even the write bus has bubbles in it and can only get 7/8ths of its max.So the GDDR5 in the PS4 doesn't achieve peak, but you have no issue believing Rangers that the eSRAM's 32MB will somehow quadruple the systems aggregate bandwidth?
Then you are a fool, one of them (my bet is Rangers) or new to B3D. There is no vetting or posters, the only thing they do is try to keep things on topic, by that doesn't keep idiots from posting. So many people there have pretended to be insiders here and got banned, but happily post there without consequence. Hell eastman posts here, but never tried the "insider" shtick that he does there. Most of the B3D faithful are non-technical MS fans. eastman, Tap In, Rangers, RudeCurve, Blackjedi, AlphaWolf, etc. etc. There are about five members total who's opinion is worth a damn.
Because the eSram is on die and extremely low latency. They are able to push those CUs up to a higher utilisation.
X1, peak 274GB/s average 200GB/s
Oh good God! That article has created a never ending nightmare.
Where did they say one game?. They really wouldn't have been very through if they only used one game! Its an average utilisation figure. Over an entire second (ie producing 30 or 60 frames) it averaged 200GB/s. Thats not a few operations thats an entire seconds worth of data! If one engine can do it, I suggest others could too.
Where did they say one game?. They really wouldn't have been very through if they only used one game! Its an average utilisation figure. Over an entire second (ie producing 30 or 60 frames) it averaged 200GB/s. Thats not a few operations thats an entire seconds worth of data! If one engine can do it, I suggest others could too.
Yes but clearly something is going on. MS have measured the GPU/CPU combined getting through 200GB/s. I think the most the CPU is allowed is 20Gb/s, so the GPU is getting though min 180GB/s. Its doing something with that data!
No system RAM ever achieves peak continually. The DDR3 in the x1 is 68GB/s but they only get 50-55Gb/s average out of it. The ESRAM achieves peak more readily because it has a dedicated read bus and a dedicated write bus, but even the write bus has bubbles in it and can only get 7/8ths of its max.
As I have said before the size of the ESRAM is not relevent as it is big enough. This is evidenced by the fact that MS have measured it running at 150GB/s, running real code. They are not doing that for fun, its being extremely heavily utilised for a reason. That reason being its there, its fast, and its big enough.
Thanks for the personal attack, not really necessary in a technical discussion.
But back to the topic. A few posts up I have quoted a few of the guys there to support what I am saying. Shifty Geezer, Gubbi, Silent_Buddha. Do you repect any of them?
B3D is an excellent source for laughs these days. Glad to see juniors linking such excellent comedy here. Saves me the work of actually sifting through B3D's bullshit to get to the fun stuff
On that note, we need to pack shit up. No amount of developer commentary citing personal experiences developing on both platforms will ever quell the uprising of "junior"s spouting nonsense and spreading FUD.
Sorry, juniors, I'll take a dev's word over yours all day every day - even if its news I don't want to hear.
Not that this isn't very entertaining, but you're aware that junior bans are permanent, right? Just so we're clear?
Yes thats because there are two interfaces to the esram, allowing reads and writes to occur simultanously. Its all in the DF article. I am not making it up!
it's ok these reputation management firms have plenty of employees.
According to the DF article the ESRAM is restricted to reading or writing on each bus at a time and it means 102GB is it's peak transfer rate. According to the DF article by using instruction tetris which lets them do some extra writes or reads while certain other instructions happen they can achieve more than that. They did 133GB while doing alpha transparency blending, a operation that generally inflates bandwidth numbers. Then someone took an absurd step and projected out this very limited trick into the general case and assuming they could do this every 8/9 clock ticks to get 198GB/s.
This was before the upclock so 109GB is after the up clock and 198GB became 208GB. But the last number is very very funny math. MS has been repeating that funny math. It's the equivalent to claiming a stock Chrysler Caravan can do 0-60 in 2.46 seconds because they dropped one from a crane and clocked it. It has almost no bearing to it's real world performance.
So... what's next? CU performance scaling, Audio chip, Hidden Gpu, Secret sauce or NDA?
You cant blame MS for doing it though can you? Those guys have spent years designing and building something they are proud of and probably a bit too deeply attatched to, then the internet says its crap...when they know it isn't. They want to get their side of the story out. Which is fair enough I think.
To my mind it can be used to challenge the notion that the ps4 is going to be all over the x1 ... something I am trying to do here! For my sins....
But anyway all this has brought on quite a considerable thirst and I'm off for a few pints of Guiness. Laters.
I was hoping this was just a delusional fan, but there's so much willfully disingenuous BS here this man HAS to be getting paid for this. There's no way this is an honest misinterpretation of figures at this point.
You cant blame MS for doing it though can you? Those guys have spent years designing and building something they are proud of and probably a bit too deeply attatched to, then the internet says its crap...when they know it isn't. They want to get their side of the story out. Which is fair enough I think.
To my mind it can be used to challenge the notion that the ps4 is going to be all over the x1 ... something I am trying to do here! For my sins....
But anyway all this has brought on quite a considerable thirst and I'm off for a few pints of Guiness. Laters.
Where did they say one game?. They really wouldn't have been very through if they only used one game! Its an average utilisation figure. Over an entire second (ie producing 30 or 60 frames) it averaged 200GB/s. Thats not a few operations thats an entire seconds worth of data! If one engine can do it, I suggest others could too.
But back to the topic. A few posts up I have quoted a few of the guys there to support what I am saying. Shifty Geezer, Gubbi, Silent_Buddha. Do you repect any of them?
Judging by your arguments, I would say you're already a few pints in!
MS was caught with their pants down. Plain and simple. They underestimated a damaged competitor, made some poor functionality and technology decisions, and now will pay the price.
X1 won't fail, but I would wager it will fail in relation to what the x360 has achieved.
So... what's next? CU performance scaling, Audio chip, Hidden Gpu, Secret sauce or NDA?
Surprised his posts don't start with "hey guys" and end with "thanks Microsoft!"Amazing how they keep showing up. This might be one of the best threads ever for moderation purposes.
Seeing "we" in such a blatant FUD post... delusional fanboy or terribly transparent shill? YOU DECIDE!
$100 price gap alone will guarantee that much. The power gap is really only relevant to the core and hardcore. Casual fans aren't going to be doing side by side comparisons for anything except a price tag.
Microsoft shill hydra confirmed!Grimløck;84023573 said:They don't die, they multiply. Whenever one gets bished, another steps up to take the mantle, desperately trying to dance the same tired dance.