Current Generation Hardware Speculation with a Technical Spin [post GDC 2020] [XBSX, PS5]

Status
Not open for further replies.
There were plenty slower drives than the stock. When discussing the PS5's SSD in 'The Road to PS5' Mark Cerny compared it to PS4, stating the stock HDD performance ranged between 50 and 100MB/s. In their GDC Spider-Man post-mortem, Insomniac said that had to go with 20MB/s because some much slower drives were used by gamers to "upgrade" their PS4/Pro.



I'm also not aware that PS4 checked/rejected drives. If it does, the reasons for doing so, for guaranteeing a minimum level of performance, would be incredibly difficult to do on PS4 given clean (new/empty) drives offer optimum performance but you really want to know what performance will be like with a heavily used drive where your data may be scattered from the outer to inner edges and not necessarily contiguously. It would take literally hours to create that worse-case data environment on a HDD for testing. On PS5 minimum performance testing is much simpler, you don't have drive geometry and seeks to fixate on so it's really only the sustained read and write performance which you can test in a few minutes.

It'll be interesting to see what Sony does but I absolutely not believe they will allow PS5 games to run from NVMe drives that have an overall slower performance profile than the internal solid state drive. I'm sure developers also do not want to be having to deal with that variable performance headache for another generation, it adds work to development and creates another support issue, i.e. users complaining about stutters and so on. Then you get the delightful job of diagnosing their crappy drive. No thanks. :nope:

Sony has to limit operation to SSDs that can run at the stated-to-death “super speed” of the onboard SSD.

If not, games will simply not work as intended when installed on the second SSD and it would be a complete mess.

Either that or we’ll be able to simply store games in that second SSD but unable to play straight from there, and every time we want to play them, they’ll have to move onto the onboard SSD, which at the speeds they run at wouldn’t really take long.
 
The official specs says so:
Code:
Dimensions:
PS5:
    Approx. 390mm x 104mm x 260mm (width x height x depth)(excludes largest projection, excludes Base)
PS5 Digital Edition:
    Approx. 390mm x 92mm x 260mm (width x height x depth)(excludes largest projection, excludes Base)

From https://blog.playstation.com/2020/0...499-for-ps5-with-ultra-hd-blu-ray-disc-drive/
Wait...so it might not fit in my new cabinet I spent the weekend setting up with temp controlled fans!?

WTH is the full size ?
 
Why exclude the base? Can the PS5 be readily placed without it?

It can be placed with the stand in different locations based on orientation, that might make an absolute hight and width harder although they could give sizes per orientation.
 
Really? I was under the impression that it included the protrusions.


Everytime I see this, it amazes me that the PS5 is almost as large as the guy's torso.

That said, I love how quiet it is. I'm glad both companies are really focusing on the sound profile of their consoles. I've been building silent PCs for my own use for over a decade now, so in the unlikely event I ever decide to get a console again, at least I won't have to worry about either being annoyingly loud.

Regards,
SB
 
Why did they make it look so stupid? The design is horrible.

Personally I actually like the aesthetic; disliked it at first but over time it's grown on me. Would say it invokes some of the feeling of high-end gaming system designs from the early '90s. Not necessarily that design language but the ambition of the aesthetic.

If you know of some system designs like the SEGA/Victor Wondermega (imho probably the best-looking console of all time) , PC-FX (which is what Series X reminds me most of design-wise, just more plain/utilitarian), etc., or even the Japanese SFC and Saturn...I guess something like that.
 
There were plenty slower drives than the stock. When discussing the PS5's SSD in 'The Road to PS5' Mark Cerny compared it to PS4, stating the stock HDD performance ranged between 50 and 100MB/s. In their GDC Spider-Man post-mortem, Insomniac said that had to go with 20MB/s because some much slower drives were used by gamers to "upgrade" their PS4/Pro.
I didn't mean to say it would be impossible to find slower drives, though I hadn't put much thought into how to best find a drive slower than a 5400rpm 3gbs laptop drive.

On PS5 minimum performance testing is much simpler, you don't have drive geometry and seeks to fixate on so it's really only the sustained read and write performance which you can test in a few minutes.
Off the top of my head, either Sony's test lab or the PS5 would need to catch some things like drives dropping write performance based on whether the SLC cache has been exhausted, which can halve sustained write performance in various places. Then there's empty/full differences in performance, which have been benchmarked ranging from almost none to 20-60%. The same possible range occurs with empty/full average latency, and there are some stark examples of cheap drives with 99th percentile latency spikes in the range of milliseconds for full drives.
Sony's approach, going by Cerny's presentation, seems to be profiling candidate SSDs against the games they have in development. Hopefully that's sufficiently exhaustive, representative enough of what games could be doing years from now, and not biased towards non-fragmented clean/empty test drives.

I think being thorough would take more time per drive if done in a test lab, and possibly beyond the scope of what the console could do at installation time.
I don't think Sony can rescind approval if it turns out drives that seemed fine enough in 2021 start showing problems several years later.


It'll be interesting to see what Sony does but I absolutely not believe they will allow PS5 games to run from NVMe drives that have an overall slower performance profile than the internal solid state drive. I'm sure developers also do not want to be having to deal with that variable performance headache for another generation, it adds work to development and creates another support issue, i.e. users complaining about stutters and so on. Then you get the delightful job of diagnosing their crappy drive. No thanks. :nope:

I think Sony would be trying to avoid being so selective that it rules out most of the 7.0 GB/s PCIe 4.0 drives, especially early on when there is only a few of them, while not relaxing standards to the point that games have to account for a worst-case performance that negates a significant portion of the custom SSD's benefits.

This assumes a generally performant custom SSD without its own performance cliffs, although perhaps we should expect some. If they're mostly in-line with other drives, the impact's not as bad as it would be if the expansion and built-in drives tank in different performance regimes--forcing games to assume worse performance throughout. It has more channels than many drives, but it's also a DRAM-less controller of custom design.

How can Sony stop users from putting in dodgy drives? They can publish a list of acceptable models and adopt a 'buyer beware' policy for drives not in it, or is it practical to add such a list to firmware updates?

Sony has to limit operation to SSDs that can run at the stated-to-death “super speed” of the onboard SSD.

If not, games will simply not work as intended when installed on the second SSD and it would be a complete mess.

Either that or we’ll be able to simply store games in that second SSD but unable to play straight from there, and every time we want to play them, they’ll have to move onto the onboard SSD, which at the speeds they run at wouldn’t really take long.

Cerny's presentation seemed to indicate a concern in testing drives to make sure they weren't too alien in performance profile compared to the PS5's own drive. A requirement that games swap onto the main drive would seem like they'd be less worried about that, since that would be a straightforward swap prior to the game running.

Future SSDs with significantly better controllers and/or NAND than Sony's drive could lead to the reverse being preferred. Without good numbers, we can't be sure that the most recent PCIe 4.0 drives aren't already better than a design from a team of undisclosed experience with sometimes finicky SSD implementations.
 
Without good numbers, we can't be sure that the most recent PCIe 4.0 drives aren't already better than a design from a team of undisclosed experience with sometimes finicky SSD implementations.

A user here reported faster load times on his nvme gen 3 ssd pc in the optimized (as claimed before the test being done) for ps5 DMC game.
 
A user here reported faster load times on his nvme gen 3 ssd pc in the optimized (as claimed before the test being done) for ps5 DMC game.

Not true, PS5 runs faster. He said load as fast but its runs a bit better on PS5. 3.2 s on NVME 970 EVO 3.5 GB/s PCIE3 with a 3900X against 2.1 seconds on PS5. If you compare the speed it match with the difference in SSD speed 5.5 GB/s against 3.5 GB/s.

ThoroughGleamingAuk-size_restricted.gif


I doubt they use kraken and optimized decompression or oodle texture. 7 GB/s PCIE 4 SSD will probably run faster on this games.

they added a button prompt if people want to read the tips. Too fast.
 
Last edited:
A user here reported faster load times on his nvme gen 3 ssd pc in the optimized (as claimed before the test being done) for ps5 DMC game.
It's a single data point, although I think there can be a case made that it may not be definitive since we're not necessarily benchmarking just the SSD when looking at a load screen, and it may not be fully representative on a cross-gen game.


I doubt they use kraken and optimized decompression or oodle texture. 7 GB/s PCIE 4 SSD will probably run faster on this games.
Do you mean you doubt the PC is using kraken, or the PS5? On the PS5, the lossless compression is meant to be transparent to software, so it probably means zlib or kraken must be in use.
 
It's a single data point, although I think there can be a case made that it may not be definitive since we're not necessarily benchmarking just the SSD when looking at a load screen, and it may not be fully representative on a cross-gen game.



Do you mean you doubt the PC is using kraken, or the PS5? On the PS5, the lossless compression is meant to be transparent to software, so it probably means zlib or kraken must be in use.

No you need to pack your game with zlib or kraken, this is not automatic on PS5. Some other games will load much faster and are probably design around the fast SSD than this special edition of a current gen Ember labs CTO told the game Kena load from OS menu to gameplay in 2 seconds. Devs of NBA 2k1 talk about 2s loading and R&C will probably load faster.

And futur first party games will probably load a bit faster than this when they will adopt oodle texture.
 
Last edited:
  • Like
Reactions: snc
No you need to pack your game with zlib or kraken, this is not automatic on PS5. Some other games will load much faster and are probably design around the fast SSD than this special edition of a current gen Ember labs CTO told the game Kena load from OS menu to gameplay in 2 seconds. Devs of NBA 2k1 talk about 2 s loading and R&C will probably load faster.
The game in question was modified in other ways to implement PS5 features. Kraken or zlib would be part of the last stage of generating package. What barrier would a lossless compression pass present versus adding RT and game modes?
 
Status
Not open for further replies.
Back
Top