Support NeoGAF

[SpinningBirdKick]

I hate to be the "King has no clothes" boy - but apart from this tweet - where is the benchmark ?

I mean where on here https://www.3dmark.com

I've said it before, Apisak is a prime candidate for either being part of a misinformation campaign or could easily be utilised by one.

When "journalists" accept those tweets as gospel with no further confirmation, you know you've hit paydirt to influence the next gen conversation.

How many lazy poorly researched articles have been generated from those tweets?

 

[LSWilson]

Dear Lord Cerny,

Stop being so smug and leak something to us poor forum peasants.

Regards,

#14tfTruther

I don't think Mark will let us down.

 

[sonomamashine]

Gonzalo/Flute are weak

they are realistic (nothing over the top) just discussing what it could be

New consoles are targeting 4K/VR both requiring serious graphical horsepower

yes and ... it proves nothing

3rd party devs are happy or have been wowed by the grunt of these new machines

of course compared to tablet cpu and weak gpu

Pachter has said (paraphrasing) PS5 could be $800 due to the spec it's targeting

like everything he says turn out to be reality (this is a stupid statement) like Sony&Ms will tank next gen release

PS5 is ahead on Scarlett if you believe the dev rumours

do you believe this ? i mean if you wanna lure someone you send fake stuff.

 

[SpinningBirdKick]

do you believe this ? i mean if you wanna lure someone you send fake stuff.

You know what I believe?

That the astroturfers are working crazy overtime for a reason. But what's that reason I hear you ask?

This generation the huge backlash on social media against the original XOne reveal was unprecedented and massively damaged the Xbox brand that it has still not recovered from.

I believe MS know that Scarlett is weaker than the PS5. They have backed off from the performance leader talk after all.

So there's attempts to get in front of the problems by running on going astroturfing campaigns to 1) manage down the social media expectations of the new machines; 2) trying to always keep Xbox in the conversation; 3) hype Next-gen + GamePass as a replacement for Next-gen + Killer App.

The release is a long way off in fall 2020. There's little good news to sustain brand momentum until then. After the reveals things will get even more difficult, not easier.

Those guys and girls astroturfing Xbox have their work cut out this time. It's not going to be easy.

 

[ResilientBanana]

You know what I believe?

That the astroturfers are working crazy overtime for a reason. But what's that reason I hear you ask?

This generation the huge backlash on social media against the original XOne reveal was unprecedented and massively damaged the Xbox brand that it has still not recovered from.

I believe MS know that Scarlett is weaker than the PS5. They have backed off from the performance leader talk after all.

So there's attempts to get in front of the problems by running on going astroturfing campaigns to 1) manage down the social media expectations of the new machines; 2) trying to always keep Xbox in the conversation; 3) hype Next-gen + GamePass as a replacement for Next-gen + Killer App.

The release is a long way off in fall 2020. There's little good news to sustain brand momentum until then. After the reveals things will get even more difficult, not easier.

Those guys and girls astroturfing Xbox have their work cut out this time. It's not going to easy.

How is the MS camp astroturfing while the Sony camp is not? As far as I'm concenred neither of them are saying they are the best because neither of them know.

 

[Negotiator]

Except you cant know that, but you are presenting it as a fact.

 

[rəddəM]

You know what I believe?

That the astroturfers are working crazy overtime for a reason. But what's that reason I hear you ask?

This generation the huge backlash on social media against the original XOne reveal was unprecedented and massively damaged the Xbox brand that it has still not recovered from.

I believe MS know that Scarlett is weaker than the PS5. They have backed off from the performance leader talk after all.

So there's attempts to get in front of the problems by running on going astroturfing campaigns to 1) manage down the social media expectations of the new machines; 2) trying to always keep Xbox in the conversation; 3) hype Next-gen + GamePass as a replacement for Next-gen + Killer App.

The release is a long way off in fall 2020. There's little good news to sustain brand momentum until then. After the reveals things will get even more difficult, not easier.

Those guys and girls astroturfing Xbox have their work cut out this time. It's not going to easy.

They said it eats monsters for breakfast...
And we all know who the monster in this story is... the PS5.
They're probably aiming to spec win by 0.0001TFlops and sell at a loss than repeat the less powerful more expensive fiasco again.
That's exactly why the XBox One X exists.
It's just M$ flaunting their M$'s dollars saying HELL FUCKING NO we're not going to loose for the Japs.

 

[CrustyBritches]

I don't think Apisak is pulling the scores from the regular 3DMark Results search. On top of that, that 12k physics score is what you'd see from a CPU like a R5 1400 4c/8t @ 3.8GHz. A 20K+ score would be more like a ~24-25K GPU score with a ~19-20K CPU score. Lastly, Apisak said "5700XT + 3700X = Next-gen console simulator".

 

[Racer!]

Thats not how AMD custom design works at all. RDNA2 is almost certainly very similar GPU design with few bits and pieces + RT, which Sony and MS can modify as they see fit.

They have AMD roadmap, and as it was case with last gen consoles and mid gen refreshes, they can implement things not in current AMD lineup even though they may be based on current GPU.

It is funny how guys who have been decoding chips for years, before they end up in tech site, think its clearly PS5 chip, yet guys on this forum say "Nah...they will be using RDNA2 therefore that chip is not PS5".

With that line of thought, everything can be anything. Im trying to follow something resembling logic here. Radeon Rays IS RDNA2, as I understand AMD slides. Flute and Gonzalo are based on Navi Lite/RDNA according to yourself.

We are not suppose to prove that Gonzalo and Flute are not PS5/Scarlett. To do that we must know what it actually is, and thats not going to happen before AMD tells us. With what Im seeing so far, this could be anything, but almost certainly not PS5/Scarlett.

Sony and Microsoft are almost certain to go for 7nm+. To manufacture a power/performance sensitive product from 2021 onwards, without the benefits of euv would be pure stupidity. It will be cheaper as it requires far fewer steps and is in high volume manufacturing as of now. Everything will be euv going forward, future nodes depend upon it. Also, RDNA2 is confirmed to be using it.

 

[R600]

I don't think Apisak is pulling the scores from the regular 3DMark Results search. On top of that, that 12k physics score is what you'd see from a CPU like a R5 1400 4c/8t @ 3.8GHz. A 20K+ score would be more like a ~24-25K GPU score with a ~19-20K CPU score. Lastly, Apisak said "5700XT + 3700X = Next-gen console simulator".

Apisak already said only overall score is visible, graphics one is hidden.

 

[luca_29_bg]

2080 super is 27k :/

 

[SlimySnake]

apparently there is a lot of thermal headroom for a 64CU Navi

is this the entire desktop setup or just the GPU?

its crazy how the wattage jumps drastically to 258w by only increasing the clocks by 150W. Who knows, maybe you gain 50-60w by reducing clocks down to 1.8 ghz.

 

[xool]

is this the entire desktop setup or just the GPU?

its crazy how the wattage jumps drastically to 258w by only increasing the clocks by 150W. Who knows, maybe you gain 50-60w by reducing clocks down to 1.8 ghz.

(says GPU power at the top)

Power roughly scales linear with frequency .. except generally you also need to increase voltage to operate at a higher frequency, and power consumption scales with voltage squared .. this makes worse case of power use proportional to frequency cubed (approximation)..

Maybe ~1.65GHz might get you down to 140W

[edit the base 5700XT will be running at 1.85GHz (top speed) for the 186W 'score' .. right ??]

 

[Mass Shift]

How is the MS camp astroturfing while the Sony camp is not? As far as I'm concenred neither of them are saying they are the best because neither of them know.

That's just the plain truth. I mean there's alot of assumptions being made against baseless rumors and non-confirmations. When you chase any of these articles back to their sources you usually have nothing burgers.

An APU gets spotted and it's, Oh, that's the PS5 or that one is for Scarlett. There's far too much projection and little to nothing relevant to back any of it up.

I believe MS know that Scarlett is weaker than the PS5. They have backed off from the performance leader talk after all.

Honestly though, if you don't know with a certainty what your competition has hiding under the hood, it would make little sense to make such a claim of hardware superiority this early.

And I don't sense that same lack of confidence from MS as it was in 2013. They came out boldly and declared "hardware accelerated Ray tracing, 4K and 120fps". In 2013 MS was only willing to share how many billions of transistors Durango was compromised of.

It was ridiculous, everyone saw straight through the obfuscation. MS didn't want to talk specs at all. This time you get the real sense that they actually do have something significant to reveal and want to put a real showcase of games behind it. I get that from Sony as well.

In many ways by skipping E3 Sony took away the necessity for MS to over expose themselves. It was met with criticism, but if MS had done what everyone expected they would have revealed their long game far too early. Sure I wanted to see Fable but it will be far sweeter to see Scarlett running it next year. And if we don't see something from Gears soon I'm going to also assume that it's being delayed until next year.

That would be one hell of a console launch. I know there will be cross-gen but next-gen enhanced Halo, Fable and Gears along with whatever these new studios and acquisitions are cooking up would be very enticing motivations.

MS isn't backing off, their HOLDING back.

 

[Fake]

Someone really voted at 5TF? Show yourself you monster.

 

[ResilientBanana]

That's just the plain truth. I mean there's alot of assumptions being made against baseless rumors and non-confirmations. When you chase any of these articles back to their sources you usually have nothing burgers.

An APU gets spotted and it's, Oh, that's the PS5 or that one is for Scarlett. There's far too much projection and little to nothing relevant to back any of it up.

I just have a feeling that AMD/MS/Sony are simply not stupid enough to test their hardware on a website that people have access to.

 

[LordOfChaos]

Someone really voted at 5TF? Show yourself you monster.

You can click them

pawel86ck , just trolling?

 

[Fake]

Mod of War please, ban that monster.

 

[Mod of War]

Mod of War please, ban that monster.

 

[R600]

I just want to add that there are reasons why people think Gonzalo/Flute is PS5 and not Scarlett.

1) First Gonzalo leak showed codename that fit Sony's console and not MS. It certainly is gaming APU, but its much more likely to be Sony then MS, duo to naming convnetion.

2) Gonzalo PCI ID matched Ariel PCI ID (that means GPU part), and Ariel was rumored to be PS5 before anything else leaked.

3) Flute showed 16GB of RAM on 256 bit bus with 18Gbps chips. This means max BW of system is 576GB/s (if full clocks) or min of 512GB/s (if downclocked). Benchmark showed 528GB/s.

4) 256bit bus with 16 GDDR6 chips mean this APU CANNOT be Scarlett because Scarlett showed 10 chips (14Gbps) on 320bit bus

IMO both of these consoles should have similar BW and similar TF, because there is sweet spot for TF/BW, as was case in last gen and mid gen refresh. This is why Matt said both are not showing their hand yet, because they could end up very close to one another.

In theory, MS has more headroom because 320bit bus is wider and they use slower memory. They can even go for 16Gbps memory in retail and get 672GB/s, while Sony can at most get 576GB/s.

In the end it seems MS wanted to save a bit on memory while Sony wanted smaller die. I do believe Reddit PCB leak was from someone near manufacturing site that got hold of PCB used for dev kits and 316mm² should mean 40CUs/RT hw + Zen2 with 1/4 of cache and 256bit bus fit like a glow. Because PCB leak was first one to specify 18Gbps chips that are now confirmed by Flute benchmark, but these chips where and still arent even in full production. Best you can get in mass product is 16Gbps from Samsung (these can be OCd to 19Gbps so there is space for 18Gbps obv).

For MS, I dont believe die size is as big as people predict, and I think its closer to 330mm² then 380mm², and with 320bit bus they will add 16mm² on their die compared to Sony by default so even with smaller die size, Sony should have similar amount of die for compute tasks.

What to make of it? My personal opinion is Sony picked sweet spot. Got just enough BW, processing power and die size to fit it all nicely while MS went with wider bus and perhaps more CUs to be able to counter Sony, depending on their performances.

I dont see Sony needing to counter MS, but can see it vice versa. Sony losing 10% power battle with smaller die size means nothing to them or customers, but MS is done if they have bigger die and lower performance.

So I think this is why Matt on Era said his educated guess is that MS will have a bit more omph, because from everything we know, their design (wider bus, slower memory and bigger die?) gives them headroom to squeez a bit more out of their design then Sony. Although I like Sonys way more tbh...Really would love to see 320mm² die with 9TF and RT hw, Zen2 and narrower bus with fastest available memory to counter wider bus design.

 

[NickFire]

That's just the plain truth. I mean there's alot of assumptions being made against baseless rumors and non-confirmations. When you chase any of these articles back to their sources you usually have nothing burgers.

An APU gets spotted and it's, Oh, that's the PS5 or that one is for Scarlett. There's far too much projection and little to nothing relevant to back any of it up.



Honestly though, if you don't know with a certainty what your competition has hiding under the hood, it would make little sense to make such a claim of hardware superiority this early.

And I don't sense that same lack of confidence from MS as it was in 2013. They came out boldly and declared "hardware accelerated Ray tracing, 4K and 120fps". In 2013 MS was only willing to share how many billions of transistors Durango was compromised of.

It was ridiculous, everyone saw straight through the obfuscation. MS didn't want to talk specs at all. This time you get the real sense that they actually do have something significant to reveal and want to put a real showcase of games behind it. I get that from Sony as well.

In many ways by skipping E3 Sony took away the necessity for MS to over expose themselves. It was met with criticism, but if MS had done what everyone expected they would have revealed their long game far too early. Sure I wanted to see Fable but it will be far sweeter to see Scarlett running it next year. And if we don't see something from Gears soon I'm going to also assume that it's being delayed until next year.

That would be one hell of a console launch. I know there will be cross-gen but next-gen enhanced Halo, Fable and Gears along with whatever these new studios and acquisitions are cooking up would be very enticing motivations.

MS isn't backing off, their HOLDING back.

I agree with the notion that playing too many cards would let Sony change a few things to come out as the more powerful console later. But I think you are re-writing history (a bit) about Sony taking away the necessity to over expose themselves. I recall that after Sony said no to E3, MS was pushing the message regularly that they would be there and bringing a ton to show off. Any necessity to show more than they did was entirely self-inflicted. They could have avoided all backlash had they not said (and likely leaked) enough to get everyone super excited for a true console reveal. And now it kind of seems they are paying a price for it, because the chatter has not been very good the last month.

 

[SlimySnake]

(says GPU power at the top)

Power roughly scales linear with frequency .. except generally you also need to increase voltage to operate at a higher frequency, and power consumption scales with voltage squared .. this makes worse case of power use proportional to frequency cubed (approximation)..

Maybe ~1.65GHz might get you down to 140W

[edit the base 5700XT will be running at 1.85GHz (top speed) for the 186W 'score' .. right ??]

1.95 ghz.

 

[bitbydeath]

APU = CPU + GPU on a single die, tha f...

Ray Tracing is built onto the PS5's GPU.

 

[Lort]

And I'm confused too , because I have never said anything like that. In fact I dont consider 8'th gen consoles perfectly balanced systems, just the opposite because of very weak (tablet like) CPU. "R600" was suggesting all games out there can be maxed out at around 150fps on modern CPU's, so I have posted CPU benchmarks that shows even the best CPU's out there can dip below 60fps in current games.

If you need even the best CPU to run a game at 60 fps .. then all you need is a CPU on console that can run 30 fps... and they do .. therefore one could argue they have a good enough CPU.

 

[ethomaz]

The forever waiting game continues...

Ninjas are killing all leakers.

 

[sonomamashine]

Ray Tracing is built onto the PS5's GPU.

We know wtf ...

 

[R600]

Flute - platform
Gonzalo - SOC

 

[xool]

I just have a feeling that AMD/MS/Sony are simply not stupid enough to test their hardware on a website that people have access to.

Starting to wonder about Apisak -according to them Sony tested the PS4 final version APU on 3dmark



A story that DF repeated on Eurogamer :

The next example is the closest we'll see to ironclad evidence that the 3DMark database does host preliminary tests for Sony console hardware. A processor with a 1.6GHz clock speed appears to have gone through the 3DMark test with the product code DG1001FGF84HT some years ago. Take a close look at the shot below and you'll see that a DG1000FGF84HT code appears on the processor of a final, retail PlayStation 4.

Sounds convincing - but this doesn't seem to have been news back pre-PS4 release .. is APISAK rewriting history ? And good luck finding those benches on 3dmark too ..

.. the alternative is deliberate creation of a (made up) paper trail ..

Probably just enough info to keep "PS5" rumors bubbling (publicity)

[honestly someone should tweet DF and ask if they've checked they haven't been fed a line ..]

 

[bitbydeath]

We know wtf ...

I'm not sure which part you are not getting.
The GPU houses Ray Tracing.
Navi10 GPU does not support Ray Tracing.
Navi20 GPU supports Ray Tracing
5700/5700XT utilises Navi10 which is why it does not support Ray Tracing.

You need Navi20 which is used in the 5900.
Gonzalo / Flute does not have these components and therefore does not support Ray Tracing. Meaning Gonzalo / Flute aren't next-gen as it is incompatible with what we already know about next-gen.

 

[R600]

Starting to wonder about Apisak -according to them Sony tested the PS4 final version APU on 3dmark



A story that DF repeated on Eurogamer :



Sounds convincing - but this doesn't seem to have been news back pre-PS4 release .. is APISAK rewriting history ? And good luck finding those benches on 3dmark too ..

.. the alternative is deliberate creation of a (made up) paper trail ..

Probably just enough info to keep "PS5" rumors bubbling (publicity)

[honestly someone should tweet DF and ask if they've checked they haven't been fed a line ..]

Apisak is from Taiwan. I have a suspicion he has someone working really close with these things to get the data he gets. We'll see but Gonzalo is only 1 of hundreds of products he posted.

Funnily enough, he didnt even say it was PS5, it was Komachi that connected it to Ariel and then entire thing went crazy.

 

[R600]

I'm not sure which part you are not getting.
The GPU houses Ray Tracing.
Navi10 GPU does not support Ray Tracing.
Navi20 GPU supports Ray Tracing
5700/5700XT utilises Navi10 which is why it does not support Ray Tracing.

You need Navi20 which is used in the 5900.
Gonzalo / Flute does not have these components and therefore does not support Ray Tracing. Meaning Gonzalo / Flute aren't next-gen as it is incompatible with what we already know about next-gen.

Aaaaand you are wrong.

Polaris does not have FP16. Vega has FP16.

Since PS4 Pro has FP16, it cant be Polaris based.

Except it is.

 

[sonomamashine]

I'm not sure which part you are not getting.
The GPU houses Ray Tracing.
Navi10 GPU does not support Ray Tracing.
Navi20 GPU supports Ray Tracing
5700/5700XT utilises Navi10 which is why it does not support Ray Tracing.

You need Navi20 which is used in the 5900.
Gonzalo / Flute does not have these components and therefore does not support Ray Tracing. Meaning Gonzalo / Flute aren't next-gen as it is incompatible with what we already know about next-gen.

the part that AMD made some customizations for Sony & Ms, and that the GPU line has nothing to do with custom APU's .

 

[ResilientBanana]

Starting to wonder about Apisak -according to them Sony tested the PS4 final version APU on 3dmark



A story that DF repeated on Eurogamer :



Sounds convincing - but this doesn't seem to have been news back pre-PS4 release .. is APISAK rewriting history ? And good luck finding those benches on 3dmark too ..

.. the alternative is deliberate creation of a (made up) paper trail ..

Probably just enough info to keep "PS5" rumors bubbling (publicity)

[honestly someone should tweet DF and ask if they've checked they haven't been fed a line ..]

Not only that, but can a PS4 even run Windows? I'm sure there is a possibility but it seems unlikely that they would.

 

[bitbydeath]

Aaaaand you are wrong.

Polaris does not have FP16. Vega has FP16.

Since PS4 Pro has FP16, it cant be Polaris based.

Except it is.

That came out before the hardware was released. Sony doesn't need to custom for a part that will already exist.

 

[llien]

Radeon Rays IS RDNA2, as I understand AMD slides.

But RR is oldrr than Navi and older than RTX.

AMD Has Its Own Ray-Tracing Technology

AMD announced widened support for Radeon Rays, its GPU accelerated ray-tracing developer software.

www.tomshardware.com

 

[ResilientBanana]

I'm not sure which part you are not getting.
The GPU houses Ray Tracing.
Navi10 GPU does not support Ray Tracing.
Navi20 GPU supports Ray Tracing
5700/5700XT utilises Navi10 which is why it does not support Ray Tracing.

You need Navi20 which is used in the 5900.
Gonzalo / Flute does not have these components and therefore does not support Ray Tracing. Meaning Gonzalo / Flute aren't next-gen as it is incompatible with what we already know about next-gen.

But at the same time people think the Zen 2 for consoles will be EUV so why couldn't RT be a console only modification to the 5700?

 

[Mass Shift]

I just have a feeling that AMD/MS/Sony are simply not stupid enough to test their hardware on a website that people have access to.

Indeed

 

[DeepEnigma]

I just have a feeling that AMD/MS/Sony are simply not stupid enough to test their hardware on a website that people have access to.

My thoughts as well.

 

[sonomamashine]

Indeed

My thoughts as well.

The response is :

Starting to wonder about Apisak -according to them Sony tested the PS4 final version APU on 3dmark



A story that DF repeated on Eurogamer :



Sounds convincing - but this doesn't seem to have been news back pre-PS4 release .. is APISAK rewriting history ? And good luck finding those benches on 3dmark too ..

.. the alternative is deliberate creation of a (made up) paper trail ..

Probably just enough info to keep "PS5" rumors bubbling (publicity)

[honestly someone should tweet DF and ask if they've checked they haven't been fed a line ..]

Why would he ruin hes reputation

 

[LordOfChaos]

The response is :

Why would he ruin hes reputation

3600 followers

reputation

¯\_(ツ)_/¯


They probably gained a lot just on this interest, and most will forget to unfollow if it's false lol

 

[vpance]

THEBE chip, brings back memories of the Sweetvar leak in 2012. I remember he and his friend got sued or something.

 

[SonGoku]

Gonzalo is not 5700 or XT. Its chip based on Navi 10, with any modifications that Sony or MS may request.

Except in semi custom you dont. You can pick and chose, as they already have in past, to put Vega features 2yrs away into Polaris based GPU.

RT is no simple add on like RPM (FP16*2). The whole GPU pipeline must be designed around RT for optimal efficiency and the design of the WGPs (CUs) must change to include necessary bits
RDNA2 brings all of the architecture changes necessary for RT hw support, the GPU core must be different.

The Navi 10 core its not RT compliant and they cant just duck tape rt cores to it.

 

[sonomamashine]

3600 followers

reputation

¯\_(ツ)_/¯


They probably gained a lot just on this interest, and most will forget to unfollow if it's false lol

This guy leaks are solid and often quoted by websites and leakers (yt)

 

[R600]

That came out before the hardware was released. Sony doesn't need to custom for a part that will already exist.

Sony or MS dont care one bit when AMD releases their GPUs. They care about IP. They see bits they can use and pick and chose what to use in their semi custom APUs. Thats why Pro had Vega features before they got out even though it was Polaris based, and that GPU didnt have it.

 

[R600]

RT is no simple add on like RPM (FP16*2). The whole GPU pipeline must be designed around RT for optimal efficiency and the design of the WGPs (CUs) must change to include necessary bits
RDNA2 brings all of the architecture changes necessary for RT hw support, the GPU core must be different.

The Navi 10 core its not RT compliant and they cant just duck tape rt cores to it.

"Navi 10 core is nor RT compliant and they cant just duckt tape RT cores to it".

Before we even know how AMD RT works, we have people claiming this.

Its completely reasonable that RDNA2.0 is upgrade in same way GCN upgrades where before, therefore its likely keeping pretty mucj 1:1 core design with some low level upgrades and RT. It may not be the case, but MS nor Sony shot for RDNA2.0 when they started with console design 3 years ago. They had tech and Navi core at their disposal to pick and chose what they want in SOC.

 

[xool]

THEBE chip, brings back memories of the Sweetvar leak in 2012. I remember he and his friend got sued or something.

This is pure gold .. how it starts :

Alright, I'm in Toronto and was talking to my buddy who works at the AMD headquarters here.

Thing is - he wasn't lying, knew everything

 

[R600]

This is pure gold .. how it starts :



Thing is - he wasn't lying, knew everything

Yea those where crazy times

This one is great as well. 2yrs ahead of release...

So, apparently, a lot of game developers have recently received the new PS4 and - Pastebin.com

Pastebin.com is the number one paste tool since 2002. Pastebin is a website where you can store text online for a set period of time.

pastebin.com

When PS4 was 4GB and MS was surely going for 8GB. He obviously messed up a bit about 1.2TF being CPU, but rumor was on point in many ways.

 

[SonGoku]

they are realistic (nothing over the top)

For 1440p its fine, for next gen 4K gaming or approximates its under powered

codename that fit Sony's console

Based on a single digit, thats far from conclusive evidence and can be faked to.

Ariel was rumored to be PS5 before anything else leaked.

The single source for this is Komachi speculating its PS5

18Gbps chips

Not proven, there are other explanations.

Before we even know how AMD RT works, we have people claiming this

Before we even know how AMD RT works, we have people claiming this.

Its completely reasonable that RDNA2.0 is upgrade in same way GCN upgrades where before, therefore its likely keeping pretty mucj 1:1 core design with some low level upgrades and RT.

To get any acceptable RT performance the whole GPU pipeline must be designed around it. AMD patent also points towards this
hw RT support is no inconsequential feature upgrade, there is no simple way to do this without redesigning the GPU pipeline.

 

[Bryank75]

 

[bitbydeath]

Sony or MS dont care one bit when AMD releases their GPUs. They care about IP. They see bits they can use and pick and chose what to use in their semi custom APUs. Thats why Pro had Vega features before they got out even though it was Polaris based, and that GPU didnt have it.

Why would they completely overhaul Navi10 to be Navi20 when they could just use Navi20?

What you are suggesting makes no sense.