Where? What was it?
Baseless Next Generation Rumors with no Technical Merits [post E3 2019, pre GDC 2020] [XBSX, PS5]
- Thread starter BRiT
- Start date
- Status
Where? What was it?
That particular test case seemed to be the result of a typo in the config file used for collecting test data.
The subsequent test results had Arden tests:
My deduction is that the python script then erroneously searched in Arden project and added it to Oberon regression test results.
Silent_Buddha
Legend
Considering how it struggles to even match XBO-X, maybe those are "R600 FLOPS"?
Regards,
SB
Why not? It'd be on an outdated architecture lacking modern features that'd make it a superior product. It'd be akin to PS3 coming with G70 instead of G80 - had PS3 had Tesla architecture, it'd have been a beast and cleaned the floor with XB360.
There's an argument that BC is important, but large userbase of PS4 isn't it. Wii had a large userbase, and Wii U's BC killed it. PS2 had a large userbase and PS3 basically gave up fresh. Older consoles like the 16 bitters didn't care too much for BC and didn't suffer. The argument is more that the market and expectations are changing.
I think that's a little selling Sony short. MS has hardware abstraction and VMs, so it's easier for them, for sure, although they'll still face potential compatibility issue with different GPU operations. They don't have flawless XBox emulation on XBO, for reference. Sony have their own libraries that devs use - they aren't writing machine code. There may be the possibility of simply recompiling the things for the new architecture and offering them as free downloads, which Sony could have planned for in advance. We don't really know what BC Sony ideas Sony had in place. Requiring hardware BC means not only tying themselves to GCN for another generation, but also what about going forwards? If future BC is important for Sony from now through their trademarked PlayStation 6..10, they need to be thinking about that now, which can't rely on using a GPU design from 20 years earlier.
I suppose the concern, as Iroboto points out, is things like Apex Legends and Fortnite just running without the devs needing to port. By and large though, BC hasn't stopped people upgrading.
Edit: Also, if next-gen is going to be the first of the 'locked in' generations, a notably weaker console may see those with no particular platform affiliation jump to MS for the more powerful box playing their cross-plat games in better quality, and then stick with MS forever more. If Sony want to lock everyone in, they have to give as few reasons as possible to swap. A less powerful setup with architectural disadvantages will see a bigger performance delta than XBO vs PS4 and that'll be enough to push some over.
Last edited:
Yea on the MP side of the equation. Keeping the player base together and large is the “new” business model. It’s already so challenging to get to a critical mass, it’s very difficult to ask developers to toss their very successful title and make a new one and hope the traction continues. Sometimes it does, sometimes you get halo 5.
being able to upgrade and play all those titles you still enjoy playing is pretty freeing. You won’t feel like you need to build an all new library just so that you can enjoy your new console. But be able to enjoy the games you already like but now even more enhanced.
Other business strategies is to let the players play the previous titles in an on going series to build up hype before release. This is much more effective with BC available for instance.
IMO; if Sony hypothetically missed some targets because they prioritized resources to ensuring BC over say having more power; I have no doubt that was the right move.
I won’t consider a PS5 without BC. I would very much want to play the PS4 titles I missed.
Last edited:
D
Deleted member 7537
Guest
Their electronics division is still their biggest one, but that includes a lot of different products. Gaming alone is like 25% of their revenue.
CES is not the place. You want an event with fans like PS Meeting with a lot of fans screaming "yeeeeahhh" and "ooooooooooh".
I know what you mean! I was a PC and Nintendo gamer for a decade until I bought the PS3 at launch, with full hardware BC. So I got to discover all the PS2 games I missed, the whole series of R&C, Sly, Jak and Daxter, ICO, etc... (so many greatest hits in bargain bins). BC isn't just for existing gamers to upgrade, it gives a whole bunch of great exclusives at launch for those who never had that console. You start the generation showered with great games while everyone else whines about the launch day line up.
Last edited:
I'm likely to upgrade my PC as the new consoles come out, so if I buy a console later it'll probably be a PS5. In that case, there's a bunch of PS4 games that I could play that I haven't had access to since I don't own a PS4. Conversely, if I wasn't buying a PC, I'd prob get Xbox series X because I have the One and I'd be able to keep all of my games. BC is a great feature.
Betanumerical
Veteran
That particular test case seemed to be the result of a typo in the config file used for collecting test data.
The subsequent test results had Arden tests:
My deduction is that the python script then erroneously searched in Arden project and added it to Oberon regression test results.
its clear imo that they are just reusing scripts and messed it up somewhere. I think the better question to ask here is why the raytracing on Oberon is not being tested when we /know/ that the PS5 has hardware raytracing in the GPU?.
Regression testing for PS4/4Pro modes that don't have any RT?
Betanumerical
Veteran
I could buy that, but RT is missing in the native mode as well for Oberon / Ariel when it isn't for Sparkman / Arden.
Hardware could contain RT. Driver and APIs may not register it as being available at the time of the test.
Betanumerical
Veteran
Could be, but I would be surprised if the driver was working for Sparkman / Arden and not Ariel / Oberon when everything indicates that Sony was further a long. Everything from the Github leaks indicates to me that the files where bulk copied of somewhere else in July this year, this makes me feel like a lot of the files / folders may be 'stale' and not up to date.
I also thought there was a good chance for a 2019 launch from a hardware perspective but if you look at there software there was no chance of a 2019 launch.
I don't think that they would of even entertained the idea of releasing the last of us 2 and ghost of tsushima after PS5.
D
Deleted member 11852
Guest
I agree but stupider things and have happened and this reconciles the 9Tf and 12Tf numbers, Scarlet being twice as powerful as Xbox One X and sources saying the top higher-tier consoles are pretty close in performance.
Maybe I missed something, but why is Oberon accepted at PS5's GPU? I assume you've seen the full info by now. Why would the author have access to this AMD/Sony project - how does that make sense? Oberon lacking RT hardware is a hurdle for me because if this is so then this means the best case scenario is Mark Cerny was being incredibly disingenuous in his interview with Wired. I can understand Sony PR people doing this, this is their job, but Cerny is not a Sony employee and the man has a reputation in the industry.
Not only this but the economics of selling the box at any loss has to dovetail with the long-term plan for recouping spent costs. When Sony and Microsoft have sold a cheaper box they've both sold well (and more software sales = licensing = profit) and when they've sold expensive boxes they've both done worse relative to cheaper boxes. For decades the economics of the console market has been predicated partly on accessories (e.g. controllers, memory cards) but mostly on profits from software sales and licensing but Microsoft look have have given up a slice of that with Game Pass, which I simply don't fathom how they are running at a profit given they need to compensate publishers for lost sales and they are also presumably surrendering licensing/publishing fees.
Come on man, you're better than this. The 'special math' is laid in AMD's RDNA white paper. Page 21. You need to consider all the facts, not just the ones you like.
This is the aspect that I would be least surprised at. Microsoft's competence from decades of Windows and Xbox hardware drivers and APIs is in backwards and forwards compatibility through software abstraction whereas Sony have traditionally employed brute-force approaches to backwards compatibility; witness chunks of PS1 hardware included in PS2, chunks of PS2 hardware being included in early PS3 models and the design of PS4 Pro - as BRiT says, a "frankenstein GPU"
I can well believe Sony would chose to go with sub-optimal technology choices to secure backwards compatibility to reduce market bleed, i.e. folks not wanting to lose their game libraries. Why this assumes AMD don't have a viable solution to Sony's backwards compatibility strategy, which I assume they do because they're AMD.
Last edited by a moderator:
Betanumerical
Veteran
Oberon has three modes.
Native - 2Ghz - 36 CU
BC1 - 911mhz - 36 CU
BC2 - 800mhz - 18 CU
These modes match the PS4 and PS4 Pro exactly.
Makes it pretty clear to me it was, at least at some point related to the PS5. The lack of raytracing tests makes me confused as the GPU 100% has it.
D
Deleted member 11852
Guest
Related to PS5 I get, but a lot of folks are real keen to accept it as PS5's GPU and therefore representative of retail PS5 specifications. It does not seem likely AMD would have final retail silicon 17-18 months before launch. It also begs the question why PS5 need a base PS4 B/C profile rather than a default PS4 Pro profile. Very little of this makes sense but like I said a few pages back, folks will believe what they want to believe.
Betanumerical
Veteran
I can believe the BC2 mode as not every game works with all 36CUs of the PRO, hence why the PRO itself has a base PS4 mode. There are some oddities in the testing data though that make no sense to me.
1st Oddity.
The bandwidth per CU from L0 appears to be 2x that of GFX10 and Arden/Sparkman(this could be old or wrong for Arden/Sparkman) in Oberon. This is even specifically noted in the document.
2nd Oddity.
The bandwidth of Oberon changes between some of the tests. Its 448GB/s most of the time but other times it appears to achieve bandwidth to the memory of ~530GB/s without compression.
I honestly don't know what's going on here but it feels to me like something is off.
Edit:.
Third Oddity.
The Wave size for Oberon(64) seems to be 2x the size of the Wave size for Sparkman / Arden(32), this only appears to happen sometimes though?. This matches the wave size for MI100 (another GPU in the leak).
Last edited:
D
Deleted member 11852
Guest
Fair enough, I didn't know this. This further reinforces my view that Sony aren't great at the backwards compatibility thing, which is I suspect for the simple reason that traditionally they have not had to concern themselves with playing today's games tomorrow when developing hardware and, more importantly, their APIs. For Microsoft and their developers, I think this is just inherent to their design philosophy - it's decades of habit that is part of their culture.
Oddities like these produce more questions than answers.
Last edited by a moderator:
For MS, it's not decades of habit but a conscious decision made with XBO's design. The XBox was the box designed to run Direct X, little more than a PC in architecture. Emulation should therefore be 'easy' like on PC, but it isn't and XBO's selection of OXB games is as weak as the PS2 library on PS4. 360 emulation is achieved through bloody good emulation and ongoing work to make it happen. Choices for XBO to use VMs etc. will serve them well for BC, absolutely, but it's not a 'corporate DNA' type thing.
The problem for Sony seems to be the GPU architecture changing from GCN to RDNA, and working fundamentally differently. In theory, PS4 games could be rebuilt for RDNA if Sony implement the libraries on it.
Hmmm. What exactly in Navi10 is GCN that'll help compatibility? It's using RDNA compute units, so that part is already completely different.
The problem for Sony seems to be the GPU architecture changing from GCN to RDNA, and working fundamentally differently. In theory, PS4 games could be rebuilt for RDNA if Sony implement the libraries on it.
Hmmm. What exactly in Navi10 is GCN that'll help compatibility? It's using RDNA compute units, so that part is already completely different.
- Status
Similar threads
- Locked
