To maintain the BW/FLOPs of 5700 XT, with 50 GB/s for CPU and a 10% inefficiency overhead, Anaconda would need 16.5 Gbps chips on a 320-bit bus. Sporty, but doable. Roughly 660 GB/s.
-
Ever wanted an RSS feed of all your favorite gaming news sites? Go check out our new Gaming Headlines feed! Read more about it here.
Next-gen PS5 and next Xbox speculation launch thread |OT7| - nm
- Thread starter Mecha Meister
- Start date
- Status
I would like herds of machines that roam dynamically rather than a specific area. So when you encounter one it's almost like a legendary Pokémon 😃
gonzalo leak in april showed a qualifying model which DF indicated means it was pretty much ready for release. its entirely possible they had it made for devkits to be shipped out.
It's been long reported that SOny is working with AMD on Navi. They are likely doing development on Navi and RT together. Why wouldnt Sony have a prototype version for devkits?
by april, the 5700xt chips shouldve been in full production since it got released two months later. if the ps5 was 10 tflops, they couldve just shipped a devkit with teh 5700xt instead of a vega gpu. neither had ray tracing but if what you are saying is true and the ray traced cards werent ready in april, then why go with vega and not a rdna card?
Any rumors about the potential xbox unveiling?
The terrain should affect the beasts too, so you can lure them into less favorable terrains, and such.
Yeah that would be fun and interesting .
Think certain robot can't go certain places is part of the game design but i hope they allow them to roam much further from there home ground .
Scarlet, Lockhart. Definitely pointing towards a Valentine's day reveal.
It's not like most gamers have anything to do on that date anyway. (I'm obviously kidding)
That's using RDNA1 tflops / bandwidth ratio. And you can always optimize the bandwidth accesses on a console.
Ergo: they won't use that much bandwidth, even at 12tflops.
To assume RDNA 2.0 is different is just that - an assumption.
I have no idea what "optimize bandwidth accesses" means on a console. You think they can do things different than PC to a measurable improvement? There are slides out there that specifically show inefficiency on PS4 as CPU accesses rise, which runs counter to your argument. Hence the booking of an overhead.
Lol
You enjoy this don't you? 😄
Woooh, I didn't know this. Thanks for the insight. Makes sense tho.
It's gonna make the surprise all the more sweet.
Because everything is based on theories and leaks that are still not being told in a whole.
Like, all we know about Gonzalo is that it's possible that it does 2Ghz. It doesn't say anything about the final clock that Sony finally lands on. We also don't know the CU count.
how are we going to tailspin back to 8 TF with this line of reasoning?!? Stay on message,
IT WAS GCN FLOPS ALL ALONG AUSTIN
I asked you before in the thread that when you claimed MS has "leapfrogged" Sony with the 12TF Scarlett and it was some new/revision to the target spec what your source was for that.
I'm still waiting.
If 12TF has always been the target spec for Scarlett (and we have literally no reason to assume otherwise) then the previous PS5 performance advantage hasn't changed.
Seems a lot of people on this thread have been pushing a narrative they would like to be true (MS somehow last minute massively revising target performance by 20%+ and leapfrogging Sony) rather than one that actually has any evidence whatsoever to back it up.
Can anyone please give me some kind of source that the 12TF is a new/revision to target spec?
Last edited:
Guys I think we are focusing on the wrong metric. TF is one thing but the clock speed as DukeBlueBall said - will have a significant impact on performance and pixel quality.
Could a higher clocked GPU with slightly lower Teraflops perform better than a GPU with lower clocks but a higher Teraflop count?
And than there's Sony's MYSTERIOUS SSD tech to help mitigate bandwith issues.
Could a higher clocked GPU with slightly lower Teraflops perform better than a GPU with lower clocks but a higher Teraflop count?
And than there's Sony's MYSTERIOUS SSD tech to help mitigate bandwith issues.
maybe I don't remember correctly but the perfornance advantage was measured on the dev kits performance.
On the same arch ? It would be hard to find example but it's possible theoreticaly.
I'm pretty sure that Klee posted that the game his friend is developing performs better on the PS5 devkit than the Xbox devkit
I have to ask - why? Would you be upset if they just pre-rendered everything? I really don't understand the fascination with non-interactive graphics being a benchmark... in video games.
It's so jarring when cutscenes run at 30fps but gameplay runs at 60fps. Total Immersion killer. So I have to disagree.
Here is the thing.
At this point, there is to much counter-information to state anything.
They want Xbox to be more powerful that's why, even though it won't matter anyways. Lol
Are you actually assuming Tom Warren is a MS employee now or something ?
Let's try to link all the leaks: PS5 > xb >10.7Tflops, PS5 is 10.24TF.
Possible scenario:
1. Current PS5 devkits are indeed 10.24TF with 40CUs and 2GHz. Using a 5700XT@2GHz.
2. Final silicon is about 300mm2, in 7nm+, still under test. So the target would be higher than 10.24TF
by a lot (52CUs @ 2GHz, for example).
3. SONY could just use off-shelf 5700XT to confuse people. The additional cost to make a 5700XT devkit
is not high. And it won't affect the development very much as long as they can deliver more powerful
final devkits in time.
well, a console that runs its GPU at much more than 2.0GHz it will be quite the achievement
Sounds hard to get the platinum. Less random more specific ness and closeness please devs!
I laid out very logical points in other posts in this thread that take all sources of information into account in a reasonable fashion, and you just seem to want to pick a fight or something I guess. We will all see what these things actually hit when fully revealed, but I will admit that I would be pretty surprised if PS5 exceeds 12 TF at retail (or if Scarlett does either).
This isn't true though. There really hasn't been much contradictory information at all coming from the reliable sources posted in this thread so far aside from the whole Lockhart is/is not cancelled thing.
All of the reliable sources have actually been rather spot on so far.
It's the pastebin/fake insider nonsense that has derailed this thread, as well as people trying to boost unreliable people as something more than spewers of garbage that has hurt this thread the most. As I said earlier, some people are intentionally misleading others by pretending to know what they are talking about, which is having a knock on effect as people are reading their garbage and going into other threads and perpetuating it. That also has an effect here as someone comes in and says "I heard X!" And then we spend two pages either discussing the new garbage or debunking it by explaining why it's wrong
So based off of insiders we CAN state that the consoles are close to each other and the PS5 is said to be stronger, but that's not confirmed yet. They are also both stronger than 10.24 teraflops.
Procedurally generated legendary machines that have different stats and strats - start out fully kitted in armor, longer range attacks, etc. Just imagine a Frost Claw that can cloak itself like a Stalker.
And that is the problem, one day is one thing the other is another
Unless multiple sources are saying the same thing, it is hard to believe in anyone, even insiders with good track record.
So is MS still supposed to have two skus? I haven't been in here in awhile
There are likely some edge cases, but I haven't seen enough on RDNA's architectural details to come up with a list offhand. In general, though, it's not enough to make a big difference for the wide range of rendering circumstances produced during the course of generating a frame for your average game.
The SSD and ray tracing stories for both Microsoft and Sony remain intriguing unknowns. I think it's a safe bet that neither party has something fast enough on the SSD front to make up for latency or bandwidth deficiencies in their chosen RAM configuration. Whatever we get will be slow enough by comparison that it acts as a distinct storage tier, it'll just be a nice change of pace that it's not slower than RAM by as many orders of magnitude as classic hard drives.
It sucks, but you have to ignore the garbage and look for reliable sources who have dropped reliable news in the past such as Klee, Matt and Tomwarren!
MS does have two SKUs, Lockheart and Anaconda, as confirmed by tomwarren in a fantastic recent article.
Even if they are the exact same, there will still be visual differences due to differences in APIs, tools, etc. The great work that the team at Digital Foundry does is not going away any time soon.
Are you being obtuse are purpose? In relation to the discussion on TF gains from Polaris to RDNA.
Digital Foundry found far less than what is being thrown about in this thread currently, your number calculations included.
Lockhart has thrown a wrench into this. There will be multiple "-gates" next-gen
To be fair the actual summary of what has been said by "leaks" or "insiders" is nothing more than the following, which of course makes a somewhat surprising multiply sourced 12 TF retail power target for Xbox Scarlett interesting here when it would put the PS5 at needing to best that 12 TF number at retail if Xbox hits their target (yes these are targets for both of them until they make the final call on starting mass production). And of course the Xbox Scarlett dev kits have been further behind in getting out to more developers until very recently too apparently. I don't know who ends up on top by what sounds like a small amount eventually either way, but it really isn't accurate to portray things as a bunch of clear on record sources having PS5 more powerful either given the actual comments and record here.
- Andrew Reiner tweeted about PS5 being more powerful than Xbox Scarlett at E3 based on "dev chatter" and apparent target specs known to some at the time, but he never indicated whether it was a slight difference or anything before bailing out of the conversation
- Colin Moriarty said something about PS5 being definitively more powerful than Xbox Scarlett at the same E3 time frame, and information from Klee and Jason from Kotaku since definitely has pointed toward very close power which made this "source" not on the mark it appears
- Klee has a friend that is developing a 2021 or later next gen game that says the PS5 is slightly more powerful than the Xbox Scarlett (think OG Xbox One versus Xbox One S per his own comparison) based on just what he knows from the dev kits that they have had and whatever other specs info they were provided by Sony and Microsoft
- Jason from Kotaku most recently after the 12 TF Xbox Scarlett information came out said that he has heard that they are both VERY powerful and that developers said it was too early to be able to say for sure which would be more powerful (this recent wording from him seems to get lost in the discussion here)
- Matt as a mod on here has consistently said that they would be very close in power also and that it won't really matter either direction, but he hasn't clearly gone on record with which he believes is the stronger console even after the 12 TF Scarlett information
I can't speak for him but them being pre rendered means my gear can't show in the cuts scenes .
Sometimes i watch them over with different out fits or take pics .
Spidey was a game i did that a fair amount with and there few others.
If everyone was surprised by the news that Lockhart wasn't dead, isn't it possible and perhaps reasonable given the proximity of the news, that people were also surprised to learn that Scarlett is targeting 12TF?
Before the WC article, other than a disregarded article from JeuxVideo, not a single insider or prominent news organization were talking about 12 TF targets for Scarlett. Now I see some discussing 12TF as if of course it was.
I have something to say :-) This video is misleading in context of this thread (comparing navi to current gen console like ps4). They use R9 280X (which is fine couse cu's number is same as in rx570) and conclude with "From Tahiti to Navi, a notional 32 CU part based on the new architecture would deliver a 62 per cent improvement in compute power.". It's fine comparison but problem is that ps4 gpu equivalent radeon 7850 has 1.25x better teraflop/performance ratio than r9 280x so navi to ps4 is closer to 1.3x teraflop/perf advantage. (btw 9.216tf * 1.3x = 11.98tf ;d)
A discussion would be, an estimate based on nothing is nothing worth discussing.
Honestly unless you're providing numbers I don't think the information is helpful or reliable at this stage. It doesn't sound like 3rd parties had real kits, more akin to the 7970s they had for the xbox one. We're better off trying to finding PS5's APU
Thank god for that, those guy are exceptional with their videos.
- Status