PS4 Pro Official Specifications (Codename NEO)

Status
Not open for further replies.
[emoji106]
Interesting tech, if it's producing results that are not visually discernible from true 4k, it'd be great to see the additional resources that would go towards 4k applied to other upgrades.
 
imo, for the time being as long as we agree not to use Ultra/Extreme as our benchmark settings, I think 4K is achievable. I think over time you're going to see creative use of the hardware in which you'll begin to see the settings scale upwards; if they can't get native, they may be able to get there over time. 4Pro price/performance will certainly impress over time.
 
Digital Foundry said that Skyrim on the PS4P will render in 4K natively in their Skyrim Remastered article over the weekend.

Yeah that's why I was specific to say cutting edge games ;) Remasters of PS3 games even with lots of visual improvements should quite reasonably be expected to run at 4K native.
 
Interesting tech, if it's producing results that are not visually discernible from true 4k, it'd be great to see the additional resources that would go towards 4k applied to other upgrades.

The overall interpretation I get from a number of accounts is that it is possible to tell in a side-by-side comparison, although this is muddling various impressions.
The Tomb Raider implementation was described as being softer, although overall still better than 1080p.
The high-res mode for the Mass Effect was considered higher-quality, with the admission that its visual design made it harder to see if there were artifacts.
(caveat: Various artifacts the authors were assuming to come from checkerboard rendering or reprojection in general were considered difficult to see at a distance.)

The choice to go with checkerboard isn't without certain trade-offs, per the Rainbow Six description of it. There are certain choices taken to minimize the obviousness of the reprojected pixels. I can see tracking the primitive ID can help better estimations for what to re-project, although I'm not sure it would change something like the decision to restrict how rapidly various effects could alternate between light and dark in order to avoid a more visually obvious sign of reprojection.
 
It's also been described as having a dedicated processor/chip/special sauce for the checkerboard, while other GPUs have to do all the processing themselves .
 
If I can ask the question from another angle, why is checkerboard used for the majority of PS4P games while it was previously a very rare occurrence? So far almost any PS4P games that didn't reach 4K native are using Sony's checkerboard implementation. Very rarely used on PC and current consoles, and obviously it's not really expected to be used on scorpio.

Was the central problem the artifacts or the performance cost?
 
But those same "equivalent" PC's that they built to represent the PS4P should also be entirely capable of producing a checkerboard based 4k output if developers choose to implement it (which is another question entirely).

The real question is what quality or performance disadvantages would they suffer on account of not having the ID buffer available.

It's also been described as having a dedicated processor/chip/special sauce for the checkerboard, while other GPUs have to do all the processing themselves .
Can GPUs create a triangle ID buffer and what sort of cost is that?

The overall interpretation I get from a number of accounts is that it is possible to tell in a side-by-side comparison, although this is muddling various impressions.

The choice to go with checkerboard isn't without certain trade-offs, per the Rainbow Six description of it.
One of the problems is we're using a catch-all term that is more specific than needs be, and we thus overlook other non-checkboard reconstruction techniques that devs can be using.

We really should change to using 'reconstruction' and only mention checkboard or other methods by name when expressly talking about those methods.
 
Can GPUs create a triangle ID buffer and what sort of cost is that?

Cerny made it sound like it is possible but at a significant cost. Direct quote
It's all hardware based, written at the same time as the Z buffer, with no pixel shader invocation required and it operates at the same resolution as the Z buffer. For the first time, objects and their coordinates in world-space can be tracked, even individual triangles can be identified. Modern GPUs don't have this access to the triangle count without a huge impact on performance.
 
Can GPUs create a triangle ID buffer and what sort of cost is that?
Not necessarily as transparently as the parallel write path for the PS4 Pro's buffer, but some form of tracking a primitive or triangle ID is part of the process for the following: http://www.confettispecialfx.com/gdce-2016-the-filtered-and-culled-visibility-buffer-2/ .

Even without a dedicated path, the magnitude of some of those wins may bear out even on architectures without Sony's hardware tweak.

One of the problems is we're using a catch-all term that is more specific than needs be, and we thus overlook other non-checkboard reconstruction techniques that devs can be using.

We really should change to using 'reconstruction' and only mention checkboard or other methods by name when expressly talking about those methods.

Artifacting when there is a significant disparity between the current and previous frame(s) from the same primitive is not without precedent. A game can opt to limit how quickly such strobing happens, or it can show up as some kind of banding or interlacing--like it did for Killzone SF. I'm presuming that the DF authors were keeping an eye out for issues like that with the particle effects in the PS4 Pro games.
 
Does anyone know if the 4K method being developed by Insomniac (which I understand builds on work from epic) for Spiderman is using the ID buffer?
 
Again, and again and again ... I don't know why that myth is so strong ...
SATA 3 won't improve the performance in any meaningfull way. Even for a SSD Sata 2 is more than enough. Only for really really high performance SSDs, the interface would make a difference. This is only in non-realworld benchmarks. Or if you copy gigabytes of data at once. Than it would make a little difference.
For everything else, SATA 2 is more than enough. You wouldn't even notice if the consoles had a sata 1 interface in it, just because the HDDs (with 5200 spins) are not fast enough.
Also we saw that the PS4 hasn't that much better loading speeds if you insert an SSD. It is a bit better, yes, but there is another limiting factor for consoles HDD performance, not yet known.
It might be just a limitation of some chips that checks if the data is valid and came from a secure place (copy protection).
The external drive on the xbox one is just faster, because the external drive has no OS that's loaded from it (so it is game-exclusive) and it is basically the drive itself that is faster than the buildin HDD, most times.

Until we observe an SSD in a PS4 Pro compared to PS4 we won't know the real world impact...but I'm guessing it will have some impact on loading times.

But on a purely technical point...pretty sure SATA 2 interface limits any relatively new SSD by about half compared to SATA 3 interface.

Xbox One external drive is using USB 3 interface which 5Gb/s compared to SATA 2 interface which is 3Gb/s.
 
Yeah, even entry-level SSDs can reach upwards of 400MB/s. I just put an entry-level Plextor MLC 128GB SSD in a friend's system that only had SATA2 ports and it was definitely capped at SATA2 speeds.

I'm sure it will make a considerable difference in load times.
 
Last edited:
To be honest, I have my finger ready on the Cancel Pre-order button until the final day. Right now I still see very little reason to actually go ahead. Especially after reading that FFXV will not actually support the Pro at least initially. That was a huge reason for me to get the bloody thing.
 
Are you sure you can resist the exquisite HDR and high res patch for UC4 and finally be able to play it and see it in what the creators have intended it to;)?

I've played through UC4 maybe 3-4 times already and got the plat so i'm not really looking forward playing the game again. Will probably have a Pro by the time the SP dlc comes out which is what i'm looking forward to. I'd buy a Pro just to play the best version of Gravity Rush 2 but that got delayed so i don't really have a reason to own one so it's better to wait and see if there are any problems to sort out in the first batch :)

I also haven't played Bloodborne in a loooong time so if there's a patch close to the release of the Pro i'm game.
 
To be honest, I have my finger ready on the Cancel Pre-order button until the final day. Right now I still see very little reason to actually go ahead. Especially after reading that FFXV will not actually support the Pro at least initially. That was a huge reason for me to get the bloody thing.
=(
If money is tight and you can't risk not getting what you want, I'd cancel. It's been a recent mantra of mine. It's harder if there are pre-order sales (aka, Amazon Prime E3 games pre-order discounts) but if you hold out, you can save yourself some money. Maybe even enough for black friday if you can wait that long.
 
Money isn't tight, I just think that between PSVR and the Pro, I'm spending a liiiiiiiittle too much on gaming, which I wouldn't mind really if it actually made a lot of sense. But as it stands, without a 4K TV, it still doesn't make a huge amount of sense. We'll see, I have 9 days to be convinced.
 
Money isn't tight, I just think that between PSVR and the Pro, I'm spending a liiiiiiiittle too much on gaming, which I wouldn't mind really if it actually made a lot of sense. But as it stands, without a 4K TV, it still doesn't make a huge amount of sense. We'll see, I have 9 days to be convinced.
If you have a PS4 already and don't have a 4k (and thus not HDR either) TV then sounds like waiting would make sense. At least until there's a bunch of enhanced titles?
 
Status
Not open for further replies.
Back
Top