• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

TRUTHFACT: MS having eSRAM yield problems on Xbox One

Status
Not open for further replies.

blade85

Neo Member
You're giving the gaming press way too much credit tho. They won't decide which console the average consumer will be buying.

Also, at this point I can't see the box selling ANYWHERE but NA, and not for the gaming parts. It's almost funny, if it wasn't so sad, how badly the Xbox One has been received in Europe. MS will face harsh problems here down the road anyway, due to Kinectspionage 2.0. The German consumer protection minister upfront demanded FULL INFORMATION about everything kinect does, which information will be recorded and how the data is used / shared. It received bad news in the mainstream media as well. And I don't have to tell you how interested europeans are in US TV, NFL, NBA and co do I?^^

And possibly in the UK, but thats entirely from the trickle down affect of the xbox360 (Had a 360 before, so should get the next one). I have no idea just how good of an "upgrade" it will be to the current cable boxes we get here as standard now days with exceptionally good controls and options.
 

mjontrix

Member
Wow, guys you can't directly compare the performance of two consoles together - it depends on how they've implemented the back-end code in the SDKs and game engines for each console. Tflops be damned, it depends on the quality of the code/algorithms implemented. Of course, there will be performance differences since there's such a large gap in Tflops performance but it won't be THAT apparent visually. I guess the main difference will be in FPS and the amount of spiking present. But even then it won't be THAT noticeable.
 

joshcryer

it's ok, you're all right now
Seems anandtech doesn't entirely disagree with my disdain for math. Agree to disagree, the math works both ways. :)

http://www.anandtech.com/show/6972/xbox-one-hardware-compared-to-playstation-4/2

You do remember this bit though:

Sony gave the PS4 50% more raw shader performance, plain and simple (768 SPs @ 800MHz vs. 1152 SPs & 800MHz). Unlike last generation, you don't need to be some sort of Jedi to extract the PS4's potential here. The Xbox One and PS4 architectures are quite similar, Sony just has more hardware under the hood. We’ll have to wait and see how this hardware delta gets exposed in games over time, but the gap is definitely there. The funny thing about game consoles is that it’s usually the lowest common denominator that determines the bulk of the experience across all platforms.

From the same article, btw, just a few pages down: http://www.anandtech.com/show/6972/xbox-one-hardware-compared-to-playstation-4/5

anandtech is just using more and less whenever they refer to each system by name. So, if the sentence (or reference) starts with PS4 they say "more" and if the sentence starts with X1 they say "less." In this case the article itself is talking about X1, so they say "33% reduction in compute power."
 
SenjustsuSage was the guy arguing for the 8GB DDR3 + esRam combo versus unified 8GB GDDR5, saying something about latency being a big advantage for the former.

So in short - I don't have any confidence in anything he is saying.

Really? He did? How much cache does he argue a relatively mildly-clocked CPU needs to stay out of the weeds?
 
You do remember this bit though:



From the same article, btw, just a few pages down: http://www.anandtech.com/show/6972/xbox-one-hardware-compared-to-playstation-4/5

anandtech is just using more and less whenever they refer to each system by name. So, if the sentence (or reference) starts with PS4 they say "more" and if the sentence starts with X1 they say "less." In this case the article itself is talking about X1, so they say "33% reduction in compute power."

In conclusion, the math works both ways, as I've said, and have been saying the entire time. Glad we could put this issue to rest lol.
 
You can call it accurate either way, because the math does indeed work out both ways. I acknowledge that going from lower to higher gives you 50%, but people seem to not be interested in acknowledging the opposite as valid in anyway, which gives you 33%. You just choose to accept one over the other, as I choose to accept one over the other.
Nonsense and bullshit. Both are perfectly accurate given correct context. There is no "accepting one over the other." To do so is idiocy.

There is no "opinion" on either being accurate. It is mathematics.
In conclusion, the math works both ways, as I've said, and have been saying the entire time. Glad we could put this issue to rest lol.
What you've been incessantly trying to do is convince everyone that one phrasing is "more accurate/correct." It is not.
 

mrklaw

MrArseFace
Yup. And that cap was probably a result of proactively addressing yield issues to insure enough supply.

Even if this is 100% true, it doesn't make it impossible for the Xbone to have games that will impress consumers enough to get them to buy the box, it just widens the performance gap of the two machines that we already knew about a bit more.

It's not worth telling juniors to go to hell or making ban bets over. Nor is it immediately going to make PS4 games have double the resolution/framerate of their Xbone counterparts.

a downgrade of the bus speed is one thing, a downclock of the actual GPU is something else.
 

Neo C.

Member
Not certain of this. Given the current 'long generation', I suspect that a worldwide launch delay longer than a couple weeks will cause terrible losses. MS would be better served to simply abandon Japan and accept defeat across Europe (including UK) to concentrate where its console's particular features would work best, namely the USA.

If the yield problem is deeper than suggested and the downclocking is only a bandaid, a delay of a few months will probably cost less than a repeat of the RROD disaster (now with downclocking!).

I'm very interested to see how MS will decide in the very near future.
 

joshcryer

it's ok, you're all right now
In conclusion, the math works both ways, as I've said, and have been saying the entire time. Glad we could put this issue to rest lol.

Hey, how about we agree to do it andandtech's way?

If we refer to PS4 first, we say "more."

If we refer to X1 first, in our sentence, we say "less."

Then fanboys on both sides can agree that whichever system they are referring to first they will use each others preferred terminology. And we can all be friends. :)
 

Septimius

Junior Member
Yup. And that cap was probably a result of proactively addressing yield issues to insure enough supply.

Even if this is 100% true, it doesn't make it impossible for the Xbone to have games that will impress consumers enough to get them to buy the box, it just widens the performance gap of the two machines that we already knew about a bit more.

It's not worth telling juniors to go to hell or making ban bets over. Nor is it immediately going to make PS4 games have double the resolution/framerate of their Xbone counterparts.

You mean even if it's 0% less than true.
 

twobear

sputum-flecked apoplexy
What the hell kind of engineering nightmare have they created? Weaker, more expensive and more difficult to produce than the PS4. Is there a precedent for this kind of thing? I guess the GeForce FX line back in the day?
 
What the hell kind of engineering nightmare have they created? Weaker, more expensive and more difficult to produce than the PS4. Is there a precedent for this kind of thing? I guess the GeForce FX line back in the day?

Seems like microsoft greenlighted this project way to late.
 
Hmmmm

Currently we know that 10% of GPU resources are dedicated to the OS for the Xbox One. Even with a GPU down clock I can't see MS reducing the number of GPU FLOPS dedicated to the OS so I think its reasonable to expect that the # of FLOPS dedicated to the OS would stay fixed regardless of a down clock.

10% of 1.229TFLOPS = 0.123TFLOPS dedicated to the OS.

A 100-200MHz reduction in the GPU clock speed would bring it down to 0.922-1.075TFLOPS.

Subtract that from 0.123TFLOPS and you got 0.799-0.952TFLOPS left for games.
I doubt that's how it works. MS didn't reserve a set number of "flops". That's impossible. They most likely reserved a portion of a CU, or a whole CU unit. Giving XBone 11 out of 12 CUs for games.

With a down clock, all CUs would be affected, so the OS portion of the "flops" would be affected, as in decrease, since your taking 10% of an overall smaller number.

The OS would have a slightly less powerful CU. Would most likely be pretty minor.
 

joshcryer

it's ok, you're all right now
Could someone explain what a "yield problem" is? I'm lost on the subject.

When you make a chip you are doing things at a very small scale. So sometimes it's not perfect and you have to throw away a chip. Now if you're doing a lot of things at a very small scale (like having 5 billion transistors) there's more chance of something not being perfect and you having to throw it away.

Basically people are arguing that because Microsoft opted to have a much larger chip with a lot more transistors (little logic gate thingies) they're having to throw away more chips than they wanted.

The "wafers" that the chips are made on are really expensive and have to go through a very clean environment and conditions, so if you wind up throwing away a chip you lose money (and that also delays production).

Let say you bake 100 cakes.
You expected 60 cakes to be good enough.
But it seems only 30 cakes are good enough.

You have a problem :p

^- or that :p
 

Nikodemos

Member
If the yield problem is deeper than suggested and the downclocking is only a bandaid, a delay of a few months will probably cost less than a repeat of the RROD disaster (now with downclocking!).
Early production yield issues were a near certainty with such a complex chip. Though my take is that it would be better for MS to simply accept restricted initial supply than to downgrade the console's capabilities (aiming for process maturation leading to better yields). Having the early buzz about their console be "It's really weak and it sucks!" is worse than not having enough consoles to sell everywhere.
 

Septimius

Junior Member
It took me a while to process this before I spit coffee on my keyboard, asshole. :)

<3

Let say you bake 100 cakes.
You expected 60 cakes to be good enough.
But it sees only 30 cakes are good enough.

You have a problem :p

So the APU-die is created within too low margins, so too many aren't useable? And a solution to that is down-clocking, since heat is the issue? Or otherwise more chips can be used?
 
I've heard the exact same thing as Thuway.

GPU is getting DC from 2 different sources. One that told me several weeks ago and the other that reconfirmed yesterday.
 

Binabik15

Member
Are those 30 cakes 50% less powerful than some other cakes?

Those cakes have sufficient secret sauce sprinkled on them, the others do not.

Man, what happened to "Devs have more confidence in MS, are unclear about PS4" and "It'll be a wash"?

^OooooohoooOh.

IF BruceLeeRoy and Thuway are saying this, I'm on board.
 
Apple came out of nowhere and scared them shitless, thats what happened, and scared MS is stupid MS.
They're a company in turmoil, scrabbling around for a niche that Apple or Google can't knock them out of, the presumed chance to get in every american home with a TV must seem like the greatest thing imaginable for them.
The core means nothing. Wii proved that going for the mass-market can get you the glory.

Can get.

Perfectly stated, but the core does mean something. They buy games. Casuals do not. Casuals won't return now that they play free games on their phones and tablets.
 
Not really. MS went with cheap ddr 3 ram and a larger die for edram. As the die size decreases with each micron drop they are able to fit more chips on a wafer .


While sony's die will also get smaller their ram GDDR wont go down in price as quickly since its mostly used in video cards and the ps4 itself will be the main driver of cost reduction. Lets also not forget that they are using a cutting edge density.

Sony is most likely going to be stuck with the same amount of gddr chips for the majority of this generation.
I've posted cites that the PS4 would be using Stacked DDR3. This turned out to be wrong in the short term because it's not ready. Everyone is expecting Sony to move to Stacked DDR3 at the first refresh and that will reduce power and cost.

Everything about the PS4 screams AMD 2014 design released a year early with compromises.

As to the Xboxone rumor, my take is ESRAM is having yield problems. My understanding is that the TDP power limits are calculated using efficient code with as little memory move as possible. If your code is larger than can be contained in ESRAM for instance and you assumed when designing the Silicon that it could fit, you would be over TDP budget it it can't. You have two choices when you discover the issue, rewrite your code or downclock. Both options are configurable using software and can be updated later. I.E. you can downclock now and when you fix the software issue you can configure the clock higher with a firmware update. There is a rush to release and Microsoft may elect to fix the issue later.
 

Raide

Member
Those cakes have sufficient secret sauce sprinkled on them, the others do not.

Man, what happened to "Devs have more confidence in MS, are unclear about PS4" and "It'll be a wash"?

Damn. I am torn between these cakes. I think they should retire and rethink the cake production. Don't want to jump into the next-generation half baked now.
 

joshcryer

it's ok, you're all right now
So the APU-die is created within too low margins, so too many aren't useable? And a solution to that is down-clocking, since heat is the issue? Or otherwise more chips can be used?

Not the cake guy here, but yes! That's exactly it.

Remember how you could "unlock a core" on some AMD cpus? The reason you could do that was because they were unstable at higher temperatures. At least initially (turns out there were entire batches that were fully unlockable stable). So the heat is causing imperfect transistors / logic gates to behave badly and cause problems.
 
Those cakes have sufficient secret sauce sprinkled on them, the others do not.

Man, what happened to "Devs have more confidence in MS, are unclear about PS4" and "It'll be a wash"?

^OooooohoooOh.

IF BruceLeeRoy and Thuway are saying this, I'm on board.

Only poster I saw say that was Arthur Gies
 
This entire project is starting to look like a colossal failure from a design standpoint. If this is true, Sony will end up with a 100% more powerful console with the same silicon budget. MS made a terrible mistake with DDR3.
 
I've heard the exact same thing as Thuway.

GPU is getting DC from 2 different sources. One that told me several weeks ago and the other that reconfirmed yesterday.
Really hoping they don't go through with this in the final units. Really hoping this isnt true, and I don't even plan on buying an XBone. This will hurt multi platform games and therefor PS4 as well. Most developers will program to the lowest common denominator. I see this being REALLY bad for multiplatform games, which is what 75% of people play.

Plus compititon is good for the industry, and XBone needs every performance gain it can get right now to stay competitive with Sonys console, this would be a big step in the opposite direction.
 

joshcryer

it's ok, you're all right now
Everything about the PS4 screams AMD 2014 design released a year early with compromises.

hUMA's going to change everything. /hackers

(But yeah I've been saying this, and I hope AMD succeeds. PC architecture needs a nice boost.)

MS made a terrible mistake with DDR3.

Alternatively, Sony got damn lucky with GDDR5. If the supply chain wasn't there they would've been SOL.
 
Early production yield issues were a near certainty with such a complex chip. Though my take is that it would be better for MS to simply accept restricted initial supply than to downgrade the console's capabilities (aiming for process maturation leading to better yields). Having the early buzz about their console be "It's really weak and it sucks!" is worse than not having enough consoles to sell everywhere.

Not if your goal was always to beat Google and Apple to the punch addressing the TV audience at-large, rather than competing head-to-head with Sony for core gamers. This has always been about establishing a beachhead vs. Google and Apple--the sooner, the bigger, the better, esp. in terms of marketshare--and they're still clearly more powerful than anything those guys have brought to market.

Like I've said in the past many times, while Microsoft was making core gamers happy and selling "just" 80M units in the process, Apple and Google sold ONE BILLION of their devices that people are using as "second screens," which are competing for consumers' attention in the TV room right now. Microsoft's shareholders see that as the biggest threat to everything Microsoft has done thus far, spending the last 15 years and billions of dollars establishing the brand in the living room.

And you know what? As much as we may not like it, those shareholders are probably right. Microsoft isn't in the business of selling 80-100M units when the other guys are selling over a billion combined. That doesn't interest them. The "core gamer" space has proven too small.

The only question now is, will it work, or has the "second screen" and the slow death of live TV killed the "set top box" model MS is perusing already?
 
This entire project is starting to look like a colossal failure from a design standpoint. If this is true, Sony will end up with a 100% more powerful console with the same silicon budget. MS made a terrible mistake with DDR3.
Yep. The whole thing looks like a big mistake and it started from the planning and design stages. The whole foundation is flawed. You can pretty much nail the blame on the "media all in ONE console" vision. They prioritized this and it cost them big time. All this bullshit that's distracted from what this thing should of focused on, GAMES. With out all this TV bullshit, they would of never mandated 8GB of RAM back in 2010, forcing them to come up with this ridiculous esRAM solution.

I still don't understand why they choose eSRAM over edram? Why? Isn't edram much higher bandwidth and less die space?
 
Damn. I am torn between these cakes. I think they should retire and rethink the cake production. Don't want to jump into the next-generation half baked now.

:)

More accurate is this story:

Say you have 10 eggs. The best poached eggs are cooked at 150ºF in the pot.

You cook the 10 eggs in the pot but only 8 of them cook as perfect poached eggs, 2 of them overcook because they are too hot inside (discard these for a salad or something).

So you turn down the pot temperature to 135ºF and you cook another 10 eggs and none are spoiled. These ain't 10 perfect poached eggs but at this point you don't give a shit cos you're hungry and you have 10 consumable poached eggs.

Amirite?
 
Not if your goal was always to beat Google and Apple to the punch addressing the TV audience at-large, rather than competing head-to-head with Sony for core gamers. This has always been about establishing a beachhead vs. Google and Apple--the sooner, the bigger, the better, esp. in terms of marketshare--and they're clearly more powerful than anything those guys have brought to market.

Like I've said in the past many times, while Microsoft was making core gamers happy and selling "just" 80M units in the process, Apple and Google sold ONE BILLION of their devices that people are using as "second screens," which are competing for consumers' attention in the TV room right now. Microsoft's shareholders see that as the biggest threat to everything Microsoft has done thus far, spending the last 15 years and billions of dollars establishing the brand in the living room. Microsoft isn't in the business of selling 80-100M units when the other guys are selling over a billion combined. That doesn't interest them. The "core gamer" space has proven too small.

Wont really matter when a stronger box can just as easily do all that stuff. And the stronger box will grant you more hype. So far the rumors are not really helping microsoft.
Weaker console higher price. I would bet google and apple could release a box as cheap as $99 to compete as a cable box running ios or android.

I really hope microsoft's 1 billion investment in games will pay off.
Only way they can probably increase hype under games.
 
Status
Not open for further replies.
Top Bottom