• 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.

Next-Gen PS5 & XSX |OT| Console tEch threaD

Status
Not open for further replies.

JLB

Banned
It has partnership with UEFA Champions League for the whole competitions since years back. I think since PS1, not only that competition, you see Sony and Playstation all over the place. See this picture it's in 1997 in the Champions League. The team is Juventus, the biggest Italian team and one of the biggest in Europe.

Juventus+1997.jpg

one of the biggest lol
keep dreaming
 

darkinstinct

...lacks reading comprehension.
What are the chances Sony show something about PS5 at PAX East, assuming they don't cancel it due to coronavirus? By the way, Ryan said 'hundreds of consumer events across the globe'.
Hundreds of events just means that after they have announced it they will do a truck tour or something similar at retailers.
 

DJ12

Member
I think if the Github leaks are still relevant but incomplete:

PS4: 18 CU
PS4 Pro: 36 CU
PS5: 54 CU

18+18+18= 54 CU :pie_thinking:
But it's 18 x 2 aka butterfly design. so that doesn't really work as ain't no way in hell PS5 is 72 CUs.

More likley, it's 56, and CUs are just disabled in BC modes as mentioned by a few Sony patents.
 
Last edited:

darkinstinct

...lacks reading comprehension.
If Sony wants to repeat the butterfly design they used to allow native BC between PS4 base/Pro, that would be the most probable CU number(three banks of 18).
I keep reading this but it makes little sense. Wouldn't a repeat of that mirrored design (who calls it butterfly?) suggest 72 CU? They deactivated half of the CUs for PS4. So now they would deactivate half of them for PS4 Pro and half of half of them for PS4. And they could run 72 CU (basically Navi 23 with 8 CU deactivated) at 1.4 GHz for 12.9 TF.
 

Bo_Hazem

Banned
But it's 18 x 2 aka butterfly design. so that doesn't really work as ain't no way in hell PS5 is 72 CUs.

More likley, it's 56, and CUs are just disabled in BC modes as mentioned by a few Sony patents.

18x3 is 54, x4 is 72. If they go that route then it'll be very predictable that PS5 Pro is gonna be 72 of 18x4.

Only speculations, anyway. I think the 2000Mhz tests are either endurance/stress tests or to give the developer some headroom to tweak a game before optimizing it at like 1.7-1.5 Ghz.🤷‍♂️
 
Last edited:

DaGwaphics

Member
Xbox can mandate whatever they want as the platform holder, it’s their platform.

As clearly quoted by JAMMA, what they have said thus far points to games releasing 2 years into XSX still being playable on their current gen systems, nothing has been said to draw distinction between 1sr, 2nd or 3rd party.

Not pleased in the slightest about this messaging, and I welcome some evidence to the contrary or at least a statement that is clear enough to make that interpretation.

They've referenced their first-party studios in every statement regarding cross-gen support, it doesn't get any clearer than that. As a rule, third-party games are cross-gen for a year or two on all platforms every generation. It's hard to believe this is a confusing concept for those that follow the industry.
 

01011001

Banned
Xbox can mandate whatever they want as the platform holder, it’s their platform.

As clearly quoted by JAMMA, what they have said thus far points to games releasing 2 years into XSX still being playable on their current gen systems, nothing has been said to draw distinction between 1sr, 2nd or 3rd party.

Not pleased in the slightest about this messaging, and I welcome some evidence to the contrary or at least a statement that is clear enough to make that interpretation.

I love how you hang on their words when it comes to the exact wording about what games are supposedly meant... but that detail where it was 2 years from now (which was said near last years E3) then gets translated by you toi 2 years into the SX's lifecylcle...

just stop already. it's obviously only 1st party titles that will be included in this and even if they weren't 90% of all games that will release until 2021 will be on both by default
 

Neo Blaster

Member
I love how you hang on their words when it comes to the exact wording about what games are supposedly meant... but that detail where it was 2 years from now (which was said near last years E3) then gets translated by you toi 2 years into the SX's lifecylcle...

just stop already. it's obviously only 1st party titles that will be included in this and even if they weren't 90% of all games that will release until 2021 will be on both by default
But every start of generation third parties are just the ones normally making crossgens anyway, even if MS didn't mandate that. It's safer for them till the new install base is big enough.
 
Last edited:

OsirisBlack

Banned
54 is my guess as well, just because there is no way they are going to come in with 36 CUs like the PS4 Pro, it makes no sense! If they did 54 CUs and clocked them at 1800Mhz it puts it at 12.44 TFlops, or even if they went down to 1700Mhz it puts it at 11.75 TFlops which even that would be an amazing jump over what we currently have. Putting it at 1700Mhz and calculating in the RDNA > GCN efficiency it would put the PS5 at roughly ~8 times faster than the base PS4 console, which is a pretty good jump.
Hmm, this could be a thing.

Also an update on the C-virus and Chinese production. I have a friend that owns a hair company who’s products come from (you guessed it) China. Word is it’s much worse than the media is portraying and most companies that are still operating are doing so against the wishes of the state. COVID-19 is not slowing and there was a recent (today) containment and quarantine of people returning to work. Also other countries are quietly quarantining and checking people that are returning from all parts of Asia. https://www.nytimes.com/interactive/2020/world/asia/china-wuhan-coronavirus-maps.html

interesting read. I’m not into mass panic so I can see why it would be understated In the media. During my tenure overseas in our military conflicts a lot of the shit we went through and saw on the news was just a blatant lie. I will never forget them reporting that we only lost 7 soldiers for the first few mo that we were in Iraq.
 
Last edited:
But it's 18 x 2 aka butterfly design. so that doesn't really work as ain't no way in hell PS5 is 72 CUs.

More likley, it's 56, and CUs are just disabled in BC modes as mentioned by a few Sony patents.

This is the most logical explanation if the Github leak is incomplete. If AMD were stress testing the chip specifically for backwards compatibility then they’d probably disable that cluster and the ray tracing components.
 

SlimySnake

Flashless at the Golden Globes
18x3 is 54, x4 is 72. If they go that route then it'll be very predictable that PS5 Pro is gonna be 72 of 18x4.

Only speculations, anyway. I think the 2000Mhz tests are either endurance/stress tests or to give the developer some headroom to tweak a game before optimizing it at like 1.7-1.5 Ghz.🤷‍♂️
I wouldnt be surprised if MS is at 72 already.

the die size phil tweeted about is 405mm2. The cutdown zen 2 CPU should be around 40-50mm2. The memory controllers and other stuff etc. shouldnt take up more than 20mm2. That leaves us with 340mm2 for the GPU. One Dual CU contains two CUs and is around 5mm2. That should be enough for a 251mm2 navi 10 with 40 CUs plus an extra 76 CUs. Assuming, 15mm2 for ray tracing and n7p's 7% extra density, we are probably looking at something in the high 60s at the very least.

Another thing to keep in mind is that the rtx 2080 has only 2944 SPs or 46 CUs worth of stream processors, and yet its a 11.4 tflops gpu at 1.95ghz game clocks. So if Sony is indeed going with a slightly smaller gpu, they should be able hit 12 tflops with just 48 CUs IF 12 tflops was their target and they really wanted a small 271mm2 gpu. i wouldnt be surprised if their APU is even bigger than Microsoft's. that giant devkit isnt cooling a 36CU gpu running at 2.0 ghz.
 

ANIMAL1975

Member
You have to wonder which version of events are unfolding in China. Is it the official numbers or the much scarier version you see talked about on twitter and whatnot. I usually write off most social media mass hysteria but I cant really say I trust the Chinese gov't. If its really bad I imagine all sorts of stuff will be affected.
After Sony saying that will not attend on the MWC in Barcelona, and instead will be streaming to the event

Sony has been closely monitoring the evolving situation following the novel coronavirus outbreak, which was declared a global emergency by the World Health Organization on January 30th, 2020.
"As we place the utmost importance on the safety and wellbeing of our customers, partners, media and employees, we have taken the difficult decision to withdraw from exhibiting and participating at MWC 2020 in Barcelona, Spain.

"The Sony press conference will now instead take place at the scheduled time of 8:30am (CET) on February 24, 2020 as a video via our official Xperia YouTube channel to share our exciting product news.

"Sony would like to thank everyone for their understanding and ongoing support during these challenging times."

I'm starting to think that probably there was a Playstation Meeting in the planning of the PS5 reveal (where there's smoke...) but, due to the recent events, was also canceled and they're now preparing a streamed Stat of Play reveal like HeisenbergFX4 HeisenbergFX4 hinted some time ago! (dude iu killing me)
Same strategy as MWC...
 
After Sony saying that will not attend on the MWC in Barcelona, and instead will be streaming to the event



I'm starting to think that probably there was a Playstation Meeting in the planning of the PS5 reveal (where there's smoke...) but, due to the recent events, was also canceled and they're now preparing a streamed Stat of Play reveal like HeisenbergFX4 HeisenbergFX4 hinted some time ago! (dude iu killing me)
Same strategy as MWC...
HeisenbergFX4 more recently mentioned they are planning another Wired article.
 

HeisenbergFX4

Gold Member
I'm starting to think that probably there was a Playstation Meeting in the planning of the PS5 reveal (where there's smoke...) but, due to the recent events, was also canceled and they're now preparing a streamed Stat of Play reveal like HeisenbergFX4 HeisenbergFX4 hinted some time ago! (dude iu killing me)
Same strategy as MWC...

I hate that IU has become so mediocre but so happy I was able to attend the IU/Purdue game with Coach Knight returning. Have been following IU for as long as I can remember back in the late 70's.

As far as the State of Play goes was told that almost a month ago or so that Sony was not doing a Feb reveal and he expected that SoP instead.

He doesnt expect a full reveal until closer to E3 but that was his guesses from hearing whispers but I would imagine only a very small handful of people know this.

And he did say he sees one more Wired article.

Again though even in his full reveal he doubts they talk about the total teraflops.

We shall see whats coming.
 

Disco_

Member
That is the worrying part but I honestly suspect the Oberon being benchmarked in those steppings are either early silicon or the same silicon reflective of the final chip but with either bugs preventing a cluster of CUs to turn on, or a cluster turned off to mimic the PS4 Pro's GPU tests for PS4 compatibility (where a chunk of the CUs were disabled).

That's why I think it's more probable there are around 48CUs on the chip active, and therefore could provide around 10.4 TF Navi - 11 TF Navi within the sweetspot range (1.7GHz - 1.8GHz). Otherwise, running a 36CU chip at even 2GHz is going to come off as a very poor solution because of the amount of extra power generated in order to hit those numbers, it'd be incredibly insufficient (you would likely see only up to 2.05GHz at most)

Desktop GPU sweet spot=/=APU sweet spot especially given that whatever they've learned for rdna2/7nm+ COULD be implemented for consoles at 7nm. We know rdna1 refresh is coming in 2020 and it's probably safe to assume efficiency will be improved. Even a slight improvement of 10% can move that sweet spot quite a bit.

54 is my guess as well,
IIRC 54 requires 3 SEs in which case you'd end up with a die that's bigger than a 56cu/2 se setup. 54 would be bigger, more expensive, and have less CUs. Does that sound as something Cerny would do?
 
Last edited:

Gavin Stevens

Formerly 'o'dium'
Slightly off topic but... guess which dumb tit just replaced his worn out shitty old analog sticks with new ones on his PS4, only for his controller to now not turn on, at all...

Every connection is made, every ribbon in, I even managed to get those 2 shits of spring in for the triggers... but nope, no dice, won’t turn on, won’t charge...

Urg...

On the plus side, the pad has two very nice new analog sticks 🤣

To be fair, it was a launch controller, and the rubber has only just started to peel, which is... fine. Battery was horse shit anyway. Will buy a new battery off eBay and see if that helps, otherwise it’s An eBay spares jobbie.

(Not overly mad, I didn’t really need much of an excuse to get a new white one haha! Mine was so old it didn’t even have the light strip!)
 
Last edited:

PSYGN

Member
Slightly off topic but... guess which dumb tit just replaced his worn out shitty old analog sticks with new ones on his PS4, only for his controller to now not turn on, at all...

Every connection is made, every ribbon in, I even managed to get those 2 shits of spring in for the triggers... but nope, no dice, won’t turn on, won’t charge...

Urg...

On the plus side, the pad has two very nice new analog sticks 🤣

To be fair, it was a launch controller, and the rubber has only just started to peel, which is... fine. Battery was horse shit anyway. Will buy a new one off eBay and see if that helps, otherwise it’s An eBay spares jobbie.

(Not overly mad, I didn’t realt need much of an excuse to get a new white one haha! Mine was so old it didn’t even have the light strip!)

Did you plug in the battery ribbon and the lightbar ribbon? Unless you did some shoddy soldering job or pulled the ribbons out forcibly at some weird angle and scraped the contacts at the ends I don't think it would break.
 
Last edited:

Gavin Stevens

Formerly 'o'dium'
Did you plug in the battery ribbon and the lightbar ribbon?

Yup, both in, both secure. Took it apart several times and checked it all. The battery has been shit for ages, like, an hours charge out of it on a good day. So it’s likely just a coincidence that it failed as I took it out, maybe I shorted it or something.
 

B_Boss

Member
Already posted?


I am currently working as game tester in a 3rd party studio we are now working on Devkit 2
PS5
1.12.6Tflops RDNA 1.5
2. AMD ZEN 2 @3.6Ghz
3. 18GB GDDR6+4GB ddr4
4. SSD@5.5GB/S 500GB
5. Dedicated cores for RT and 3D Audio
6. Bandwidth 576GB/S
DS5 on Devkit 3 has a alot of new "tweaks" (my friend reported me working at WWS). No information on Price, Design, reveal and BC

XSX~.
1. 11.8Tflops Rdna 1.5
2. AMD Zen 2@3.7Ghz
3. 16GB GDDR6+4GB ddr4
4. SSD@3.8 GB/s 1TB
5. Dedicated RT cores (<PS5)
6. Bandwidth 596GB/S
XSX is almost final can expect GPU Tflops to be around 12.1 or more and some changes in Ram.

As always, until officially proven, take with a grain of salt.
 
Last edited:

SlimySnake

Flashless at the Golden Globes
Already posted?
I am currently working as game tester in a 3rd party studio we are now working on Devkit 2
PS5
1.12.6Tflops RDNA 1.5
2. AMD ZEN 2 @3.6Ghz
3. 18GB GDDR6+4GB ddr4
4. SSD@5.5GB/S 500GB
5. Dedicated cores for RT and 3D Audio
6. Bandwidth 576GB/S
DS5 on Devkit 3 has a alot of new "tweaks" (my friend reported me working at WWS). No information on Price, Design, reveal and BC

XSX~.
1. 11.8Tflops Rdna 1.5
2. AMD Zen 2@3.7Ghz
3. 16GB GDDR6+4GB ddr4
4. SSD@3.8 GB/s 1TB
5. Dedicated RT cores (<PS5)
6. Bandwidth 596GB/S
XSX is almost final can expect GPU Tflops to be around 12.1 or more and some changes in Ram.


As always, until officially proven, take with a grain of salt.
So PS5 has yet to see its final form.

Devkit 3 to have HBM confirmed.

i am surprised at the high cpu clocks. i expected them to be at 3.2 ghz since that seems to be the sweet spot for perf vs power usage. 3.6 and 3.7 suggests they are targeting 7nm+ since its supposed to give 10% in power efficiency.
 
IIRC 54 requires 3 SEs in which case you'd end up with a die that's bigger than a 56cu/2 se setup. 54 would be bigger, more expensive, and have less CUs. Does that sound as something Cerny would do?

Maybe if he is feeling frisky I guess he might! :messenger_tears_of_joy:

But in all honesty it was just a complete guess on my part, I am guessing it will be a 30 WGP/60 CU/3 SE setup with 1 WGP/2 CU disabled on each SE for yield leaving it at 27 WGP/54 CU/3 SE for the setup, again it is just a total pull a rabbit out of my ass guess!
 
Last edited:
Status
Not open for further replies.
Top Bottom