Support NeoGAF

[MarkMe2525]

So glad that you can all justify Spencers lies about cloud compute and physics being done by server farms and how the cloud would make Xbox One the most powerful console ever....

The past seven gameless years are now all worth it thanks to the spec sheet.... just another few gameless years ahead till the next spec sheet!

Thanks Phil!!!

P.S. Please more drm, subscriptions and microtransactions.. /S

(Probably went too hard here.... I'm gonna get some icecream and come back later...)

I knew you have already stated you might of went too hard. I just have to say, I have owned an Xbox one since launch and have over 350 games for it. I also own a PS4 with 5 select exclusives that I wanted to play. I just don't know where the gameless argument comes from because it's just not true.

 

[Dory16]

Cerny never said a 36CU with higher clock would beat a 52CU with lower clock.

The comparison he made was for when the resulting teraflops are equal, using higher clock to get there is more advantageous than adding more CU.

It's true that he was comparing 2 different CU configurations amounting to the same number of teraflops.
When he says that's there's 36% more rasterisation in the highly clocked one though, he's implying that it performs better than a 36% more powerful GPU and every objective observer thought about the XSX.
That's what the article is getting at. It's deconstructing Cerny's well perceived indirect but intentional assertions.

 

[Bryank75]

It's true that he was comparing 2 different CU configurations amounting to the same number of teraflops.
When he says that's there's 36% more rasterisation in the highly clocked one though, he's implying that it performs better than a 36% more powerful GPU and every objective observer thought about the XSX.
That's what the article is getting at. It's deconstructing Cerny's well perceived indirect but intentional assertions.

Consider who is saying this too...a guy that works on a PC only strategy game. Strategy games are usually not graphically intense and he has no experience or insight into PlayStation or consoles in general....

It only benefits him to stand on the MS side of things..... where he has a stake.

 

[Dory16]

Consider who is saying this too...a guy that works on a PC only strategy game. Strategy games are usually not graphically intense and he has no experience or insight into PlayStation or consoles in general....

It only benefits him to stand on the MS side of things..... where he has a stake.

Fair point. But the benefit of technical arguments is that they can be judged on their merit. There have been several benchmark tests done, including by digital foundry that disprove Cerny's theory. We don't have to just take it at face value because Cerny said it. Once RDNA 2 chips are available, the same tests will be repeated to take scaling into account but the laws of physics tend to be stubborn.

 

[welsay01]

It's true that he was comparing 2 different CU configurations amounting to the same number of teraflops.
When he says that's there's 36% more rasterisation in the highly clocked one though, he's implying that it performs better than a 36% more powerful GPU and every objective observer thought about the XSX.
That's what the article is getting at. It's deconstructing Cerny's well perceived indirect but intentional assertions.

No. Your original thread title was about Cerny being dishonest when really the writer of the article is either misinformed or dishonest himself.

I'd be interested in seeing someone create a poll here to find out if most people really think that the higher clock of the PS5 completely closes the gap between it and the XSX or only lessens it. From my short time on the forum I can only recall seeing the latter.

 

[Bryank75]

Fair point. But the benefit of technical arguments is that they can be judged on their merit. There have been several benchmark tests done, including by digital foundry that disprove Cerny's theory. We don't have to just take it at face value because Cerny said it. Once RDNA 2 chips are available, the same tests will be repeated to take scaling into account but the laws of physics tend to be stubborn.

However they were all tested in the PC paradigm..... where games are scaled to the card. Not in the console paradigm where games are coded precisely for to the metal and frequency.

Isn't it possible that there is more performance to be eked out of the overclocking in a console?

 

[martino]

However they were all tested in the PC paradigm..... where games are scaled to the card. Not in the console paradigm where games are coded precisely for to the metal and frequency.

Isn't it possible that there is more performance to be eked out of the overclocking in a console?

in the end in both case and since more than one gen now this is settings tweaking for most.

 

[Dory16]

No. Your original thread title was about Cerny being dishonest when really the writer of the article is either misinformed or dishonest himself.

I'd be interested in seeing someone create a poll here to find out if most people really think that the higher clock of the PS5 completely closes the gap between it and the XSX or only lessens it. From my short time on the forum I can only recall seeing the latter.

Cerny clearly says in his presentation that it is too challenging to keep many CUs constantly fed with meainingful work (as if parallel computing was somehow a bad idea and we should all return to single cores).

At the back of that, We had a developer from Crytek (never mind posters on this forum) saying that the Ps5 is the better of the 2 consoles precisely for the reason that this article is refuting. Because of the inefficiency of large GPUs.

At the very least Cerny is overselling his solution and I trust that he knows much better than what he proclaimed. The article is pertinent and objective on that basis.

 

[Dory16]

However they were all tested in the PC paradigm..... where games are scaled to the card. Not in the console paradigm where games are coded precisely for to the metal and frequency.

Isn't it possible that there is more performance to be eked out of the overclocking in a console?

If it is possible to the extent that Cerny claimed in his presentation, we still have to see it in real games. PC graphic cards did not confirm that and we will soon find out if consoles do. Many experts are sceptical, as you saw in the article.

 

[Bryank75]

If it is possible to the extent that Cerny claimed in his presentation, we still have to see it in real games. PC graphic cards did not confirm that and we will soon find out if consoles do. Many experts are sceptical, as you saw in the article.

Games already get huge optimization by being coded to the specifics of the console....as you say, we will have to see.

I doubt there will be much difference between the actual performance of the two boxes on 3rd party games when all is said and done.

 

[Dory16]

Games already get huge optimization by being coded to the specifics of the console....as you say, we will have to see.

I doubt there will be much difference between the actual performance of the two boxes on 3rd party games when all is said and done.

There will be some fabulous looking games on both consoles whether they are 3rd or 1st party, like every generation.
I think the reason why power matters to gamers is the increased confidence they feel about their purchase, knowing that every beautiful game that they see on the other console could theoritically made to look at least at good if not better on theirs, even if the gap is not meaningful. Those things are expensive and I guess people want the absolute most for their money, it's human.

 

[welsay01]

Cerny clearly says in his presentation that it is too challenging to keep many CUs constantly fed with meainingful work (as if parallel computing was somehow a bad idea and we should all return to single cores).

At the back of that, We had a developer from Crytek (never mind posters on this forum) saying that the Ps5 is the better of the 2 consoles precisely for the reason that this article is refuting. Because of the inefficiency of large GPUs.

At the very least Cerny is overselling his solution and I trust that he knows much better than what he proclaimed. The article is pertinent and objective on that basis.

What Cerny said is true within the scope of the scenario he mentions. He never said anything about the XSX. Assuming implications from what he says is injecting your own bias.

Same with that Cytek developer and any poster making such assumptions. Just because they misinterpreted does not make Cerny dishonest.

 

[Dory16]

What Cerny said is true within the scope of the scenario he mentions. He never said anything about the XSX. Assuming implications from what he says is injecting your own bias.

Same with that Cytek developer and any poster making such assumptions. Just because they misinterpreted does not make Cerny dishonest.

Well I respectfully disagree in the sense that I find naive to assume that Cerny is unaware that he is operating in a context of intense business competition with Xbox. Even if he personally doesn't, his employer does and everything he said that day I'm sure wasn't heard for the first time by Sony's bosses. I know you don't believe that there haven't been discussions at Sony about how his presentation would make the PS5 look against the XSX as well as whether the message was in the best interest of console sales.
It wasn't an internal seminar at Sony's interactive headquaters, it was the first public statement by Sony about their product. Not only did they know that everyone was watching, they knew that everyone was holding the XSX spec sheet while listening. Cerny is not an innocent littel nerd just making GPUs and oblivious to his surroundings. Overselling in that context actually goes beyond just Cerny's personally responsibility. This is a corporations's business strategy in action.

 

[welsay01]

Well I respectfully disagree in the sense that I find naive to assume that Cerny is unaware that he is operating in a context of intense business competition with Xbox. Even if he personally doesn't, his employer does and everything he said that day I'm sure wasn't heard for the first time by Sony's bosses. I know you don't believe that there haven't been discussions at Sony about how his presentation would make the PS5 look against the XSX as well as whether the message was in the best interest of console sales.
It wasn't an internal seminar at Sony's interactive headquaters, it was the first public statement by Sony about their product. Not only did they know that everyone was watching, they knew that everyone was holding the XSX spec sheet while listening. Cerny is not an innocent littel nerd just making GPUs and oblivious to his surroundings. Overselling in that context actually goes beyond just Cerny's personally responsibility. This is a corporations's business strategy in action.

Of course he doesn't operate in a bubble, but the fact of the matter is he never said what the article is claiming he said.

If he's so dishonest why would he even mention any negatives at all like memory being 33% further away?

Could we also say Microsoft is dishonest or overselling locked frequency and high CU count or RAM bw split? Why didn't they mention any of their solution's downsides? Are people naive enough to believe they have a flawless solution?

 

[Dory16]

Of course he doesn't operate in a bubble, but the fact of the matter is he never said what the article is claiming he said.

If he's so dishonest why would he even mention any negatives at all like memory being 33% further away?

Could we also say Microsoft is dishonest or overselling locked frequency and high CU count or RAM bw split? Why didn't they mention any of their solution's downsides? Are people naive enough to believe they have a flawless solution?

I think Cerny just did his job as a Sony employee. When Xbox engineers said to Digital Foundry in 2013 that GDDR5 would have put them in an uncomfortable design position for the Xbox One, they got rightfully laughed at by every reader and countless people on this forum. When Cerny states that unprecedented overclocking is the best performance solution going forward because that's how you keep very performant GPUs cool (I'm paraphrasing), it should draw scepticism as well.
I concede that the article is wrongfully saying that Cerny directly compared the PS5 to the XSX. He did not. His listeners did and perceived his innuendos. Without Cerny's presentation I am not sure that the Crytek guy goes on record to downplay large GPUs and then retract it.

 

[Piku_Ringo]

So glad that you can all justify Spencers lies about cloud compute and physics being done by server farms and how the cloud would make Xbox One the most powerful console ever....

The past seven gameless years are now all worth it thanks to the spec sheet.... just another few gameless years ahead till the next spec sheet!

Thanks Phil!!!

P.S. Please more drm, subscriptions and microtransactions.. /S

(Probably went too hard here.... I'm gonna get some icecream and come back later...)

Who the hell are you to disrespect God Spencer like that?!

 

[welsay01]

I think Cerny just did his job as a Sony employee. When Xbox engineers said to Digital Foundry in 2013 that GDDR5 would have put them in an uncomfortable design position for the Xbox One, they got rightfully laughed at by every reader and countless people on this forum. When Cerny states that unprecedented overclocking is the best performance solution going forward because that's how you keep very performant GPUs cool (I'm paraphrasing), it should draw scepticism as well.
I concede that the article is wrongfully saying that Cerny directly compared the PS5 to the XSX. He did not. His listeners did and perceived his innuendos. Without Cerny's presentation I am not sure that the Crytek guy goes on record to downplay large GPUs and then retract it.

Of course we should be skeptical of what we're being told. But it's another thing to automatically assume that people are being dishonest or putting words in their mouth.

And again I haven't seen anyone saying PS5's clocks make it more powerful than XSX. I've only seen people saying it narrows the gap, but there's still a gap. This is why it'd be very interesting to see a poll because we're both perceiving things so different.

I wasn't on this forum during the XBO reveal, so I don't know how the reaction was, but their listed benefits for going with DDR3 with eSRAM were sound, however they also did not admit the downside of the setup, which is that it'd be more of a challenge to use.

I'm not sure where you gathered that Cerny said overclocking was the best way to keep a GPU cool though, but if you mean the whole variable clocks thing, again, if he's so dishonest, why would he mention the downside of it having to downclock in certain instances at all?

 

[Dory16]

I clearly stated that I was paraphrasing Cerny. He said that cooling and quietness is what led them to fixed power budget/variable clocks. When cooling is your priority I don't think that you come up with the highest frequencies ever seen in consoles even if they are variable. Unless of course you are trying to catch up with a competitor. And then you HAVE to make a case for high clocks. Which is contested in the article. I think it's a completely healthy discourse.

I already explained that this is bigger than Cerny the individual. He's the spokesperson of a corporate strategy. My original title was dishonest OR misguided. I'm happy to stick with "overselling" because there's no doubt he (basically Sony) was.

 

[welsay01]

I clearly stated that I was paraphrasing Cerny. He said that cooling and quietness is what led them to fixed power budget/variable clocks. When cooling is your priority I don't think that you come up with the highest frequencies ever seen in consoles even if they are variable. Unless of course you are trying to catch up with a competitor. And then you HAVE to make a case for high clocks. Which is contested in the article. I think it's a completely healthy discourse.

I already explained that this is bigger than Cerny the individual. He's the spokesperson of a corporate strategy. My original title was dishonest OR misguided. I'm happy to stick with "overselling" because there's no doubt he (basically Sony) was.

So overselling yet still pointing out some disadvantages? I think I'd accept someone calling him a poor salesman before I'd accept someone saying he's overselling.

I can't recall the last time I saw a salesperson point out any flaw of their product outright. Seems like a poor tactic if you're trying to oversell people on something.

That would be like the d2d vacuum salesperson saying "bagless vacuums are great; you never have to spend a dime on expensive disposable bags! Sure all the dust comes back out into your face when you empty it, but that's a minor inconvenience." Well he sure oversold that Dyson to me.

 

[Dory16]

So overselling yet still pointing out some disadvantages? I think I'd accept someone calling him a poor salesman before I'd accept someone saying he's overselling.

I can't recall the last time I saw a salesperson point out any flaw of their product outright. Seems like a poor tactic if you're trying to oversell people on something.

That would be like the d2d vacuum salesperson saying "bagless vacuums are great; you never have to spend a dime on expensive disposable bags! Sure all the dust comes back out into your face when you empty it, but that's a minor inconvenience." Well he sure oversold that Dyson to me.

Sorry but the fact that he stated the obvious about parts of the system that people care less about when he could, does not change the overselling of high frequencies in a GPU. The SSD is another matter that the article aims to mitigate but I'll wait until first party games come out to judge whether Sony was overselling that part.

 

[slade]

It's true that he was comparing 2 different CU configurations amounting to the same number of teraflops.
When he says that's there's 36% more rasterisation in the highly clocked one though, he's implying that it performs better than a 36% more powerful GPU and every objective observer thought about the XSX.
That's what the article is getting at. It's deconstructing Cerny's well perceived indirect but intentional assertions.

Wasn't the comparison between the PS4 pros 36 CU's to the PS5's 36 CU's. And that while the number of CU's was the same, because of architecture improvements the PS5 CU's were better.

 

[Dory16]

Wasn't the comparison between the PS4 pros 36 CU's to the PS5's 36 CU's. And that while the number of CU's was the same, because of architecture improvements the PS5 CU's were better.

Not quite. he compared 2 abstract configurations, one with more CUS but lower clocks than the other.

 

[sinnergy]

It is what it is, doesn't mean it's bad .