Ratchet & Clank technical analysis *spawn

At any rate, the original claim of needing a SSD + a major decompression block to handle PS5 games doesn't hold up to reality. At least according to Ratchet and Clank, which is working just fine on any fast HDD even on systems with 16GBs of RAM, the game is not a data intensive exercise of course, but still.

That portal sequence could very well be designed to be streamed in advance while the player is traversing ahead, especially as each portal contains just a very tiny playable area. Environment swap is nothing new, we've had Dishonored 2 and Titanfall 2 do a full swap in real time in several levels, with unique textures and assets and all, and on last gen hardware. Careful data planning and proper streaming by the developer is the key to this, SSD and decompression block may accelerate this even further, but I would argue that the CPU and GPU of the PS5 may not handle a data intensive game anyway. They are barely able to do current gen games with standard assets at sufficient quality, many times falling into the 30fps lock, or the -frankly quite miserable- 720p~1080p/60fps modes. The main benefit of the SSD and decompression block so far seems to be extremely fast loading times, nothing more.
 
At any rate, the original claim of needing a SSD + a major decompression block to handle PS5 games doesn't hold up to reality. At least according to Ratchet and Clank, which is working just fine on any fast HDD even on systems with 16GBs of RAM, the game is not a data intensive exercise of course, but still.

That portal sequence could very well be designed to be streamed in advance while the player is traversing ahead, especially as each portal contains just a very tiny playable area. Environment swap is nothing new, we've had Dishonored 2 and Titanfall 2 do a full swap in real time in several levels, with unique textures and assets and all, and on last gen hardware. Careful data planning and proper streaming by the developer is the key to this, SSD and decompression block may accelerate this even further, but I would argue that the CPU and GPU of the PS5 may not handle a data intensive game anyway. They are barely able to do current gen games with standard assets at sufficient quality, many times falling into the 30fps lock, or the -frankly quite miserable- 720p~1080p/60fps modes. The main benefit of the SSD and decompression block so far seems to be extremely fast loading times, nothing more.

What a ridiculous thing to say.
 
no 16GB is not enough if you don't have VRAM on your GPU also
not it's not loading small chunks of levels but the levels themselves, apart from the jump in the futuristic city on the flying dinosaur which is not from any level in the game, and that's why this small portion loads also faster on a HDD.
And it was always a given that with enough ram you could achieve similar results, but at a more expensive price.
 
Early on when the PS5's architecture was revealed and before Directstorage dropped, the assumption I commonly remember seeing on here was that some games would likely need 32GB to compensate and have larger install sizes due to textures being packed in a more CPU-friendly decompression format, as R&C is doing for the non-gdeflate textures. My stance at the time was "Okey dokey!". In terms of bang-per-buck, ram and storage upgrades are some of the best return for PC components, and this is even before the SSD price crash. Hell I'd imagine 16Gb is pretty rare for DDR5 systems regardless.

While 32GB has been shown to net you perhaps overall smoother frametimes with a few games over 16GB, the number of games that are actually architected to utilize more than 16GB can basically be counted on one hand today, still. If R&C for example had an option, or just shipped by default with all of its textures in .LZ4 format and thus had a ~50GB install size and required 32GB to run with high/very high textures, but in turn didn't have any impact on GPU performance when doing so, that would be a net positive based on my experience so far. Obviously Nixxes felt differently and I'd be curious to hear their reasoning.

Would the bottleneck then shift to the PCIE bus? Maybe, but interesting that R&C still shows a large performance disparity with PCI 2.0 vs 3.0 even with low textures, so not sure texture size really is the prime culprit here:


Then again I thought Diablo IV's 32GB requirement for the highest texture setting was a 'good thing' , but it doesn't work - albeit it doesn't seem like it actually loads more than 16GB, so not necessarily a fault in the approach, rather they screwed it up. Speaking of timely patches compared to Nixxes btw, I take it that still has not been fixed?

Edit: Alex somewhat addressed this, not specifically with texture decompression GPU load but how the game could actually run if it fully utilized available ram on the PC.
 
Last edited:
Early on when the PS5's architecture was revealed and before Directstorage dropped, the assumption I commonly remember seeing on here was that some games would likely need 32GB to compensate and have larger install sizes due to textures being packed in a more CPU-friendly decompression format, as R&C is doing for the non-gdeflate textures. My stance at the time was "Okey dokey!". In terms of bang-per-buck, ram and storage upgrades are some of the best return for PC components, and this is even before the SSD price crash. Hell I'd imagine 16Gb is pretty rare for DDR5 systems regardless.

While 32GB has been shown to net you perhaps overall smoother frametimes with a few games over 16GB, the number of games that are actually architected to utilize more than 16GB can basically be counted on one hand today, still. If R&C for example had an option, or just shipped by default with all of its textures in .LZ4 format and thus had a ~50GB install size and required 32GB to run with high/very high textures, but in turn didn't have any impact on GPU performance when doing so, that would be a net positive based on my experience so far. Obviously Nixxes felt differently and I'd be curious to hear their reasoning.

Would the bottleneck then shift to the PCIE bus? Maybe, but interesting that R&C still shows a large performance disparity with PCI 2.0 vs 3.0 even with low textures, so not sure texture size really is the prime culprit here:


Then again I thought Diablo IV's 32GB requirement for the highest texture setting was a 'good thing' , but it doesn't work - albeit it doesn't seem like it actually loads more than 16GB, so not necessarily a fault in the approach, rather they screwed it up. Speaking of timely patches compared to Nixxes btw, I take it that still has not been fixed?

Have you tried deleting the DirectStorage dll from the game files to force it to not use GPU Decompression (if indeed that works)?
 
Again, Nixxes isn't some behemoth of a company that can just be working on all these Playstation games from launch. Sony has to pick and prioritize what gets ported to PC. It's basically a throughput problem and the longer you take on each item going through, the more the whole process slows down.

The alternative is to put more emphasis on the main studios doing multiplatform development instead of focusing only on Playstation during the main development cycles. But I think they're gonna be somewhat reluctant to do this because in the end, Playstation will always be their undisputed #1 priority by far.

I'm of the opinion that I'm grateful Sony is offering these games on PC at all. Some technical quibbles at launch, especially when otherwise being very playable, do not feel to me as warranting any kind of harsh bashing when they could easily just not be doing any of this at all and leaving these PS-exclusive. Especially when they've been very good about rectifying issues.

And again that is a Sony problem. They should have been investing in more studios to handle pc ports.

Right now your post is just full of making excuses for terrible Sony ports to pc gamers in which they want the full msrp of the ps5 game at launch day years later. It's appaling that you are defending them charging $60 for a game that performs worse than a game I can walk into walmart and buy for $40 bucks right now.

The last of us still has huge issues months after release but you are grateful for that ?
 
And again that is a Sony problem. They should have been investing in more studios to handle pc ports.

Right now your post is just full of making excuses for terrible Sony ports to pc gamers in which they want the full msrp of the ps5 game at launch day years later. It's appaling that you are defending them charging $60 for a game that performs worse than a game I can walk into walmart and buy for $40 bucks right now.

The last of us still has huge issues months after release but you are grateful for that ?

TLOU and Horizon:Zero Dawn at launch, yes - pretty awful. But on the whole, Sony's PC ports have been quite far from 'terrible' imo. Like many ports recently, there's been a host of bugs at launch depending upon the title, but most seem to be addressed in relatively short order. I also will cut the porting studios some slack in that porting first-party games that were hand crafted to take advantage of a closed platform is a different matter than porting a title designed from the outset to be multiplatform - we have plenty of examples of those games shipping with severe bugs on the PC, and little being done to improve them months after release. Sony has shown worlds more commitment than say, EA in actually giving a shit. R&C isn't in the same world as TLOU, indicated by some of its more egregious issues being fixed in a few days. I have little doubt in a couple of weeks most of the remaining niggling issues will be remedied, the very high texture settings may never be as optimized as I would like, but this is nothing like TLOU.

That being said, yes I agree that Sony's prices for these games on PC, especially 2 years later when they're either been half off for a year + or literally a freebie PSN+ monthly add on are egregious. Like R&C is $80 Cndn now. Come on man. You just are not going to get that investment from the PC community for a 10-hour platformer shooter at that price, no matter how graphically impressive it might be. Be real.
 
Last edited:
Platform warring will not be tolerated
The Master Race Crowd pleasing decision was made by using PS5s performance mode to show off high end PCs better IQ and general settings...
So easy to see through the real motivation behind this video..
Laughable...
I wish DF would still be a place with no bias towards a certain platform and their quality output has suffered since a while ...
Btw... With VRR and 120hz mode that performance mode on PS5 runs often at around 80fps...
While the 40fps mode retains IQ and settings from 30fps mode..
 
Regarding the section on HDD, again, some brilliant info around how the game performs on first run after boot vs cached to system RAM with an HDD. This could explain some of the videos out there showing HDD performance that is very reasonable.

Yeah, that's why in my HDD videos I specifically rebooted after each test. I think some videos out there are just showing the benefits of caching.

It's also very interesting that Nixxes have not implemented a pre-caching system for this date and simply rely on the Windows file cache.

Yeah, and pulling from Windows cache may also not be most efficient method either as opposed to if the game's process was actually utilizing it. I want to see 32GB taken advantage of more games in general! I honestly thought we would by now.
 
The Master Race Crowd pleasing decision was made by using PS5s performance mode to show off high end PCs better IQ and general settings...
So easy to see through the real motivation behind this video..
Laughable...
I wish DF would still be a place with no bias towards a certain platform and their quality output has suffered since a while ...
Btw... With VRR and 120hz mode that performance mode on PS5 runs often at around 80fps...
While the 40fps mode retains IQ and settings from 30fps mode..

It's a review of a PC game for PC gamers. PC gamers don't care about playing games at 30 or 40fps for the most part so the comparison isn't particularly relevant. Even before the PC version released DF have stated that Performance RT mode is the best way to play on the PS5 and despite all that, there certainly were multiple instances where Alex compared to the PS5's fidelity mode. Not only that but he clearly called out the PC versions deficiencies vs the PS5 version throughout the video so it seems to me your claims of bias are pretty selective.
 
The Master Race Crowd pleasing decision was made by using PS5s performance mode to show off high end PCs better IQ and general settings...
So easy to see through the real motivation behind this video..
Laughable...
I wish DF would still be a place with no bias towards a certain platform and their quality output has suffered since a while ...
Btw... With VRR and 120hz mode that performance mode on PS5 runs often at around 80fps...
While the 40fps mode retains IQ and settings from 30fps mode..
Laughable comment. It's that way because the RT Performance mode is by far the way to play on PS5. Nobody sane would choose the other modes.
 
Laughable comment. It's that way because the RT Performance mode is by far the way to play on PS5. Nobody sane would choose the other modes.

Regardless Alex did use the Quality mode for comparisons several times, and the majority of his PC shots involved DLSS too. Not his fault the PC just has access to a superior reconstruction method atm.

Really just put this kid on ignore people, you're not going to have any sort of 'discussion' with them, at no point will they be "Ah ok, I see your perspective now". They're here for a specific reason.
 
Laughable comment. It's that way because the RT Performance mode is by far the way to play on PS5. Nobody sane would choose the other modes.
Oh is that so? 40fps mode is best compromise between IQ and responsivness and fps...

Performance Mode was definitely the best choice if you want to show PC gains more pronounced - wich you might do if you a master race zealot.

But it is dishonest to say the least...

Imagine the outrage ... Showing in a PC Video about best PCs settings at least also the best visual version of PS5s Ratchet & Clank as well ...
Cheap as fuck and not worth coming from a big outlet like DF ..
I mean - yeah off course everybody has his favorite platform - nothing wrong with that - but not offering the "competing" platform in your video at least the best visual performance as well is beyond cheap actually...
It is truly something else
Just wow..
 
Last edited:
Amazing video which answered so many big questions about the game. How Alex managed to produce it so quickly, especially given the patch that landed just today, I have no idea, impressive stuff!

My absolute highlight was the load speed test. Alex confirmed, or rather ruled out GPU speed (so decompression speed is not an issue here - certain posters take note) along with memory speed, drive speed (which we already knew) and surprisingly CPU speed (above a certain level at least as one would expect). So perhaps starting to look more likely now that the game engine really does cap this, or is at least code limited in some way. I guess it does make sense as these transitions still need to work in a way that makes sense on future hardware, so allowing them to run completely unlimited would be a bad idea. So what speed to you limit them too? Maybe Nixxes asked Sony lol....

I also really loved the section about the background programs and how that effects framerate with both vram and system ram usage. Very illuminating.

Clearly the game performs better on PS5 than it does on the 2070S as we should really expect, but the gap is smaller than I expected it to be based on this video. 8GB is obviously limiting things but it's not a bad showing for the old GPU overall, although DRS does muddy the comparison somewhat.

I definitely appreciated the additional bugs picked up at the start. Nixxes seem to be very responsive at fixing things like this so the more they are brought to their attention the better. This is a game I'd really like to see perfected on the PC. Unfortunate that it will have to be post launch, but such is the way of the world atm.

Regarding the section on HDD, again, some brilliant info around how the game performs on first run after boot vs cached to system RAM with an HDD. This could explain some of the videos out there showing HDD performance that is very reasonable. It's also very interesting that Nixxes have not implemented a pre-caching system for this date and simply rely on the Windows file cache.
I'm not sure it's expected when RT is in use. And yeah without knowing what the DRS bounds are its really hard to make performance comparisons. We would need some scenes with resolutions matched.
 
Please no condescending comments like that here, this is not neogaf, everybody has the right to choose what suits them better, i enjoyed the 40fps mode and had no problem with it.
Especially since DF other Members always ALWAYS say that if a 40fps mode is available and stable that it is the best option , the best mix out of visuals fps and responsiveness. Alex sits next to them nodes his head as if he has to.
What he realy thinks is shown here...
And yeah he said he does not have a capture card for such modes - ok.

But then if work schedule for a corporation of Alex and the member with such a card (Rich) is not possible why not simply adress it in the video?
Just a mention of a (at least visually )better mode of PS5 but why it is not shown.
And to pretend that it is all only about PC is such a bad faith argument as well.
We all know that as soon it is PS5 footage and PC footage the secret videotitle is "Game xyz - PC vs PS5"
 
These comparisons never bring out the best in camps. One side always wanting hardware superiority in every inch of their particular chosen gaming platform. One should celebrate that the PC can match or excel gaming consoles in most instances, but we should also appreciate that gaming consoles at times don't require brute strength or super expensive hardware in certain instances to perform the same task... and typically in a much more efficient and less costly manner.

Maybe, I'm too old to care about these fanboy PC/console wars... but it's a little bit tiring seeing the nonsense following these comparisons, rather than having a fruitful conversation without the smartass digs coming from both sides.
 
I'm not sure it's expected when RT is in use. And yeah without knowing what the DRS bounds are its really hard to make performance comparisons. We would need some scenes with resolutions matched.

RT performance on PS5 is 1080p-1440p apparently, as per DF's PS5 video. But yeah, in raster for a highly optimized native PS5 game, I would expect you'd see above 2070S performance - but you would expect it would gain that back with RT. There's not the performance gap like there is with TLOU, but the PS5 looks to outpace a 2080 when using RT here which is rarely, if ever seen.
 
Back
Top