Playstation 5 [PS5] [Release November 12 2020]

loot boxes to get more hairstyles maybe

6sQlSGE.gif
 
I also doubt its a slim, but it could be something that helps Somy with certain things. Cost down across a bunch of chips, basically a smaller revision.
Also the PS3 removed hardware BC components before EU launch believe. So its not unheard of doing costs revisions within the first year.
Wether that is N6 or something else who knows. But my interpretation of TSCM leaker for N6 was that it would go into mass production next year and increase yield vs N7.
Which would mean more per wafer, correct? Also might have some power and heat envelopes that helps other aspects, maybe cheaper ram or heatsink etc.
Then again, I have zero knowledge in this area, so :D

AFAIK, N6 reduces the power consumption but there's no perf. gains on the node. Maybe there's a density increase, but I don't know enough on it.

Had forgotten about the PS3 getting that revision, but see that also fits into what I mentioned before about not necessarily seeing systems get revisions that soon unless they were addressing notable issues. Tho in PS3's case the "notable issue" was mainly production costs being ridiculously high; they couldn't axe the Blu-Ray drive, integrate the CPU/GPU into a single package or even change up the RAM and since BC wasn't as big a thing at that time it made sense to reduce costs by getting rid of the legacy hardware.

If N6 goes into mass production next year, I still think 2021 is too early. 2022 might be a better fit, though. I'm curious if they could leverage N6 for a hypothetical PS5 Pro, but not like something which is a clear boost like the Pro was over PS4 base. More like squeezing out a tad more performance from the older power consumption profile, and increasing storage space. Maybe bundle in PSVR with it too as another SKU, encourage wider adoption for VR going forward while PSVR2 can still be kept standalone.

loot boxes to get more hairstyles maybe

6sQlSGE.gif

This is in-game, right? If so, extremely impressive. Pretty much CG quality, at least on the level of some of Capcom's recent RE CG films.
 
Thats what I thiught too. Some people need to take the chill pill :LOL:

Sorry, I mistook your joke for snobbery.

Likewise, what gives with the PS5 Pro rumors? They don't sound realistic to me, at all tbh. That is going to be a VERY expensive Pro model since you are essentially doubling the size of the APU (not exactly, but close to it). At the very least, it's paying more for silicon; even if that's on the N6 process the cost savings are essentially negated and then some because of the extra silicon for doubling the GPU. Plus, they'd have to rework the memory controllers, probably add more memory or bandwidth (likely include not only a good amount of Infinity Cache but also increase the GDDR6 bandwidth a tad since that is still needing sharing with the CPU, audio, and SSD I/O), bump up the cooling, increase storage space size (that's actually probably the big complaint that seems to be had with PS5)...all of that increases costs.

Personally, I'd love a PS5 Pro Duo a few years down the line, when 3nm is economically feasible: a doubled PS5, consisting of 2 PS5 APU's as chiplets connected to ~2-4GB of HBM as a cross-chiplet cache.

It would certainly be expensive, but I think that's the purpose of the higher tier model. The base PS5's APU isn't particularly large, and will be tiny on 3nm. It'll also be child's play to power and cool on that node. 14gbps GDDR6 won't be all that expensive in a few years' time either.

So the cheap, lower end will be taken care of by the base console. Why also aim for the also cheap, but slightly less low end of the market? In an era of $1000 consumer GPU's?

Release a $700 PS5 Pro Duo in 2025 and spend the next 5 years getting to a point where a more refined design can be sold at $500 as the PS6.

And, I'm also just thinking but, doesn't going with a 72 CU design go against Cerny's own narrow & fast philosophy? Their BC is also hardware-based so that will lock them in to a 72 CU GPU (at the very least) for PS6, and node shrinks are only getting more expensive, not less. Sounds like it'd be a bit of a financial nightmare for Sony regardless of how profitable PS5 ends up being. That's why I think the PS5 Pro stuff is really just in relation to something server-related, it makes the most logistical and financial sense.

The PS5's narrow and fast philosophy went against the PS4's wide and slow philosophy. 36CU's was the smallest Sony could go while ensuring near flawless to substantially enhanced PS4 Pro BC. Any smaller would impede replacing the PS4 Pro, any larger would create a more costly chip.

Would a 72CU chip even be much larger than that of the XSX? Or if Sony were to go with a chiplet design, they'd have a single chip design to bin according to whether one goes in the base model or two go in the Pro model. Might that be enough to offset some of the cost of more expensive 3nm manufacturing?

Also, you'd think Sony have to be thinking of how a seemingly really strong PS5 Pro affects future perception for a PS6; people seem to already feel a sense of diminishing returns going from the PS4 Pro (and for MS, One X) to the PS5 (and again for MS, Series X). 10th-gen consoles won't be able to rely on SSDs as a new innovation, nor probably VR, either. So some 20.5 TF, 72 CU PS5 Pro in whatever year between 2021-2024 it'd release, that just locks Sony in to a relatively big (and expensive) GPU setup for PS6 unless they completely redesign how they handle BC (or goodness forbid, just do away with it, which I don't think anyone wants). It also means they'd have to go to some ridiculous means to build a PS6 that appears worthy of being next-gen in terms of technological grunt, have to include yet even more cache, memory capacity and bandwidth (all increasing costs), etc.

I'm not so sure. I think this go around, with a solid GPU architecture, a solid CPU architecture, and solid IO, there's scope for rolling generations. An adequately powerful PS5 Pro could be a fairly limited PS6.

With rolling generations and a substantially powerful PS5 Pro, the PS6 doesn't need to be much more powerful. If the PS5 Pro is 20.5 TF's, the PS6 could be 72CU's clocked at 2.8GHz for 25.8TF's. The same principle can apply across the board.

Personally, I'm loving the PS4 Pro enhanced games I'm playing on my PS5, as dynamic resolutions don't appear to dip and shimmer really at all, and framerates are a dream (especially FFXV's "lite" mode.)

Moore's law is slowing down, game development time is increasing. I really think the generations can afford to last longer - something like 10 years rather than 6 - and adequately powerful mid-gen consoles are the way to achieve that IMO.
 
This is in-game, right? If so, extremely impressive. Pretty much CG quality, at least on the level of some of Capcom's recent RE CG films.

yes it's ingame, or at least replay camera.
it looks really good indeed for a football game. But it's still not at the level of that shrek 2 gif from 2004.
 
AFAIK, N6 reduces the power consumption but there's no perf. gains on the node. Maybe there's a density increase, but I don't know enough on it.

There was this a while back:
https://www.anandtech.com/show/1422...echnology-7-nm-with-higher-transistor-density
TSMC states that their N6 fabrication technology offers 18% higher logic density when compared to the company’s N7 process (1st Gen 7 nm, DUV-only), yet offers the same performance and power consumption.

So it doesn't look like a great win.
 
All while according to the TSMC leaker Sony started to produce their SoCs very late and at a lower volume than the SeriesX's.

There was this talk about 6nm PS5 via a TSMC leaker..

Aquariuszi is not a TSMC worker. He works in an IC packaging and verification company.

What he said is in his company there were fewer PS5 SOCs under verificaion. He didn't say TSMC
produce less PS5 SOCs.
 
I also doubt its a slim, but it could be something that helps Somy with certain things. Cost down across a bunch of chips, basically a smaller revision.
Also the PS3 removed hardware BC components before EU launch believe. So its not unheard of doing costs revisions within the first year.
Wether that is N6 or something else who knows. But my interpretation of TSCM leaker for N6 was that it would go into mass production next year and increase yield vs N7.
Which would mean more per wafer, correct? Also might have some power and heat envelopes that helps other aspects, maybe cheaper ram or heatsink etc.
Then again, I have zero knowledge in this area, so :D

Launch PS3s had ps2 hardware. (At Europe)

At least the 60Gb model had it(not sure about the 40gb one, or the 80gb)

But some definetly had it, I have one of those too.

40gb didnt have card readers and some other items I believe
 
It really depends how one defines moore's law. In layman terms progress measured in pure fps has slowed dramatically down from 20 years ago. 1080ti is 3.5 years old and latest fastest gpu's are barely 100% faster in average. 60fps vs. 30fps. In reality 1080ti is now 4 years old chip. It's likely in 4 years perf maybe double again. Improvements are slowing down despite the power draw going up, chips getting bigger etc.

Hope for really big gains in performance are purely in innovation outside of adding more of the same. Innovation like figuring out design with lower power draw to be able to use more transistors, dedicated units for accelerating specific things, ray tracing being new thing can see optimizations in implementation, dlss like solutions maybe becoming good enough to provide more perf etc.

From my POV good engineering makes it looks like moore's law is alive. It would look very different if old architecture was just ported to 7nm or 5nm

This is what anandtech has to say about tmsc roadmap and promises. Note power improvements and perf don't stack up. Adding all those transistors into chips in 5nm would eat some serious power unless something smart is done in design phase.

Compared to it’s N5 node, N3 promises to improve performance by 10-15% at the same power levels, or reduce power by 25-30% at the same transistor speeds. Furthermore, TSMC promises a logic area density improvement of 1.7x, meaning that we’ll see a 0.58x scaling factor between N5 and N3 logic

upload_2020-12-3_15-46-23.png

 
Last edited:
If N6 goes into mass production next year, I still think 2021 is too early. 2022 might be a better fit, though. I'm curious if they could leverage N6 for a hypothetical PS5 Pro, but not like something which is a clear boost like the Pro was over PS4 base.
A die shrink for a Lite/Silm version has been a tradition for volume & yield improvements. Not a surprise; maybe 2022 given alleged AMD roadmaps for their own products (a good projection of AMD’s process tech roadmap).

As for a hypothetical PS5 Pro, the margin of improvements for N7 to N6 does not seem enough to pull off a 2x improvements in GPU (PS4 to PS4 Pro). Especially when N6 was touted to offer only a 16% higher logic density, and the lack of iso comparison figures seem to indicate a limited to no power saving.

So I would guess the next major SoC upgrade would be on 5nm, and no earlier than fall 2022 (judging by the alleged 6nm Rembrandt APU launching in 2022). This is unless a chiplet config (that still resembles the same system architecture with unified memory) is on the table, while bumping the SoC power budget above the PS5 design target is acceptable (which is unlikely).
 
Last edited:
I also doubt its a slim, but it could be something that helps Somy with certain things. Cost down across a bunch of chips, basically a smaller revision.
Also the PS3 removed hardware BC components before EU launch believe. So its not unheard of doing costs revisions within the first year.
Wether that is N6 or something else who knows. But my interpretation of TSCM leaker for N6 was that it would go into mass production next year and increase yield vs N7.
Which would mean more per wafer, correct? Also might have some power and heat envelopes that helps other aspects, maybe cheaper ram or heatsink etc.
Then again, I have zero knowledge in this area, so :D
It is not unheard of but Sony had to reduce costs and make the price attractive fast because the PS3 was going downhill and punching holes in Sony's pockets.
Which is the complete opposite of PS5. They cut off all the "extras" that were barely going to be used.
 
The Dualsense features on Worms Rumble is freaking amazing, unlike big games like Fortnite and Rainbow Six
Rainbow Six is mixed bag for me but Fortnite is by far the worse implantation I've seen so far
Anyone with a PS5 should try Worms as it's free with PSPlus, every gun has different trigger resistance and they feel very satisfying to use
 
Back
Top