PS5 - The Future of Gaming | The Power of V

[kac77]

Where's the math that makes it obvious? Yes texture files may potentially be stored in large contiguous archives but games will randomly access small bits of data all across the file at runtime. You don't sequentially read the entire file into memory in order to access specific bytes.

If random access peaks at 64 MB / per sec how long would it take to load a level? Definitely more than a few minutes considering some games are already at 8GB per map, some larger than that. Therefore it's literally impossible to make sweeping suggestions that it's all random access when obviously it's not. The math under the assertion would mean only 3.8 gigs is loaded within a minute on a SSD. So obviously the data is packed whereby it's not all random. I can't remember when playing any game on a SSD where it took more than 2 minutes to load. But that's what it would be if it was all random access.

There is no streaming of data off of drives at present not that I've seen. I think Rage tried it. I'll have to research it but I think that's the only title that I know of that tried that. Other than that no, when you load a map all of it is sitting in VRAM for the most part, unless there's not enough RAM on the card. In which case you'll get stutter because it has to request the data, which is why drive speed matters.

Nobody is arguing that sequential reads aren't used or aren't important.

That's literally what you said. You said, " They also don’t know where the sequential speed will be useful. " But now you're saying no one is arguing that when it's been said mutliple times here.

The question is what specific use case benefits from 5.5GB/s off the SSD. As Anandtech points out even the PS5 SSD's speed is a joke compared to RAM so it's not a substitute for that. Sony obviously thinks it was worth the extra effort and cost vs an off the shelf SSD but they haven't really told us why.

Anand is a magazine. They have been wrong before. Many times. So that's not really a point I would make. As stated before SSD's are not meant to replace RAM but they are imperative in order to load the data into memory quick enough. At 5.5 GB/s you could load an entire map into memory easily within 2 - 3 seconds. That's big no matter who says it isn't because well the math doesn't lie. If all of the data is random access then it's going to take forever. That's why the graph that was posted was so telling. It literally proves the opposite of the point being made. That's why I didn't say anything initially it's best to just sit back and watch people backtrack and say "I didn't say that," or whatever.

 

[MangoSeed]

Therefore it's literally impossible to make sweeping suggestions that it's all random access when obviously it's not.

Nobody said this.

That's literally what you said. You said, " They also don’t know where the sequential speed will be useful. " But now you're saying no one is arguing that when it's been said mutliple times here.

Or this.

Since you seem more interested in arguing against imaginary boogeymen instead of having an actual technical debate I’ll leave you to it.

 

[MangoSeed]

At 5.5 GB/s you could load an entire map into memory easily within 2 - 3 seconds.

Let me try a different approach since you’re struggling to see the point. How do you know that sequential read speed is the primary determinant for loading an entire map in 2-3 seconds? What is the shape of the workload that makes that true considering the queue depths and in flight reads required to saturate an NVMe controller?

 

[kac77]

Nobody said this.



Or this.

Since you seem more interested in arguing against imaginary boogeymen instead of having an actual technical debate I’ll leave you to it.

Um I quoted your post. So that's what the quotes mean. But OK no one said it. Whatever... didn't like the math huh? LOL So I guess where back to SSD's matter and 5.5 GB per sec isn't anything to sneeze at. Thanks.

 

[kac77]

Let me try a different approach since you’re struggling to see the point. How do you know that sequential read speed is the primary determinant for loading an entire map in 2-3 seconds? What is the shape of the workload that makes that true considering the queue depths and in flight reads required to saturate an NVMe controller?

Becuase if random access peaks at 64 MB per sec then it would stand to reason it's not all random. Meanwhile you're arguing you never said it but arguing here about "how do I know?" . Make up your mind please. Just be consistant that's all I ask.

 

[MangoSeed]

Becuase if random access peaks at 64 MB per sec then it would stand to reason it's not all random. Meanwhile you're arguing you never said it but arguing here about "how do I know?" . Make up your mind please. Just be consistant that's all I ask.

Sorry I can't be consistent with your imagination. Literally nobody in this thread said all storage access is random so you're wasting your time arguing with yourself.

As I stated in my first post the question is where 5.5GB/s is advantageous vs commodity SSD speeds. In all the data on game load times that we have to date it's random access times that separate SSDs from the pack, not sequential speeds. The obvious conclusion there is that there are diminishing returns to even faster sequential reads. How you're unable to grasp that simple point is beyond me.

 

[kac77]

In all the data on game load times that we have to date it's random access times that separate SSDs from the pack, not sequential speeds.

So you're not saying this or you are? I'm still waiting for you to answer the question that if it's all random then why doesn't it take 2 minutes or more to load onto VRAM from an SSD. I guess trolling is where we are now. OK have at it I guess.

 

[MangoSeed]

LOL So I guess where back to SSD's matter and 5.5 GB per sec isn't anything to sneeze at. Thanks.

Did someone say SSDs don't matter? You have an uncanny knack for arguing with shadows.

 

[MangoSeed]

So you're not saying this or you aren't? I'm still waiting for you to answer the question that if it's all random then why doesn't it take 2 minutes or more to load onto an SSD. I guess trolling is where we are now. OK have at it I guess.

I'm going to assume that you can appreciate that real workloads are a combination of random accesses and sequential reads. Now assuming you understand that hopefully you also understand that actual performance may be dependent more on random accesses once you reach some threshold for sequential speed. The evidence we have (from SSD reviews) points to the conclusion that most SSDs are already fast enough and level loading performance is primarily determined by random access.

You seem to be taking Sony's word at face value that more sequential speed is better when we have evidence to the contrary. Don't know if I can break it down any simpler than that.

 

[kac77]

I'm going to assume that you can appreciate that real workloads are a combination of random accesses and sequential reads.

I said this posts ago. I literally said you can't decouple random from sequentional but then this was said, " In all the data on game load times that we have to date it's random access times that separate SSDs from the pack, not sequential speeds. " When literally both matter but OK.

Now assuming you understand that hopefully you also understand that actual performance may be dependent more on random accesses once you reach some threshold for sequential speed. The evidence we have (from SSD reviews) points to the conclusion that most SSDs are already fast enough and level loading performance is primarily determined by random access.

If you think it is then you have to answer the math question. There's no way around it. Also, if you think it's all little files then open up a game and look you won't see them. They are in one file per map and many times less than that...for a reason. This is all controllable by the developer by the way.

You seem to be taking Sony's word at face value that more sequential speed is better when we have evidence to the contrary. Don't know if I can break it down any simpler than that.

Not taking it at face value but I'm not about to discount basic arithmetic. That would make me a fanboi which I am not. I plan to buy all of the consoles and game on my PC as well so I really don't care who does what. But I do care about senseless acusations that bounce all over the place for fanboyism sake that's ridiculous.

 

[MangoSeed]

I said this posts ago. I literally said you can't decouple random from sequentional but then this was said, " In all the data on game load times that we have to date it's random access times that separate SSDs from the pack, not sequential speeds. " When literally both matter but OK.

Yes both matter in absolute terms. That's not up for debate. The question is whether there is any practical benefit to Sony's approach vs other SSDs. PCIe 4.0 SSD reviews so far say no.

Not taking it at face value but I'm not about to discount basic arithmetic. That would make me a fanboi which I am not. I plan to buy all of the consoles and game on my PC as well so I really don't care who does what. But I do care about senseless acusations that bounce all over the place for fanboyism sake that's ridiculous.

Your math is based on theoretical 4K random access throughput tests. Obviously real world workloads don't consist of purely 4K random reads. You believe Sony that real workloads will benefit from even faster sequential speeds. The evidence says otherwise. Sony may very well have good justification for their approach. We just haven't seen it yet.

What accusation is that btw?

 

[MangoSeed]

If you think it is then you have to answer the math question. There's no way around it. Also, if you think it's all little files then open up a game and look you won't see them. They are in one file per map and many times less than that...for a reason. This is all controllable by the developer by the way.

Yes I am familiar with how large files are read. You certainly don't sequentially read the whole file into memory. You use random seeks to the bytes that you want from disk.

 

[kac77]

Yes both matter in absolute terms. That's not up for debate. The question is whether there is any practical benefit to Sony's approach vs other SSDs. PCIe 4.0 SSD reviews so far say no.

No review benchmarks SSDs with a game other than loading times. Please link I would would be interested.

Your math is based on theoretical 4K random access throughput tests.

No it's not. The math is literally pulled from a post from an actual benchmark. It's not theoretical it's actual.

Obviously real world workloads don't consist of purely 4K random reads.

Yup totally agree. Said that days ago.

You believe Sony that real workloads will benefit from even faster sequential speeds. The evidence says otherwise. Sony may very well have good justification for their approach. We just haven't seen it yet.

What accusation is that btw?

What evidence? No one has done it before in a console. So don't know where you are pulling that from.

 

[kac77]

Yes I am familiar with how large files are read. You certainly don't sequentially read the whole file into memory. You use random seeks to the bytes that you want from disk.

Between files yes, which are large not small jpegs in a folder. But they are copied onto your drive at once by a single or sometimes more process. The blocks used will be sequential in nature literally to improve performance. NTFS does this. ZFS does this. EXT3/4 does this. All file systems do this. We aren't on Fat32 anymore where we don't need to defrag once a week because of this. But you are free to argue otherwise I guess.

 

[MangoSeed]

No review benchmarks SSDs with a game other than loading times. Please link I would would be interested.

Yes, I'm referring to load time benchmarks as that's the only data we have. E.g. https://www.techspot.com/review/1893-pcie-4-vs-pcie-3-ssd/

No it's not. The math is literally pulled from a post from an actual benchmark. It's not theoretical it's actual.

There's a benchmark of real gaming workloads that measures random access throughput? Would love to see that.

What evidence? No one has done it before in a console. So don't know where you are pulling that from.

That's right, nobody has done it on a console. We just have Sony's word to go on.

 

[kac77]

Yes, I'm referring to load time benchmarks as that's the only data we have. E.g. https://www.techspot.com/review/1893-pcie-4-vs-pcie-3-ssd/

That's testing PCI3 vs PCI4. That's not much to go on and it's game loading not specific levels so while it's insightful it doesn't really help one way or the other. But thank you for providing it. I always like a good review.

.

There's a benchmark of real gaming workloads that measures random access throughput? Would love to see that.

Do we need to redefine theorhetical vs synthetic? What was posted was synthetic but definitely not theoretical. Theoretical is what the drive manufacturer releases as optimium performance. Synthetic is a benchmark that is meant to showcase specific workloads on real hardware. What was posted earlier and debated i might add was synthetic and used within arguments. If that post which you were agreeing with wasn't good enough for me to use then really no one should use it.

That's right, nobody has done it on a console. We just have Sony's word to go on.

Exactly. But that's not what's happening. Are you completely discounting what Sony is saying?

 

[MangoSeed]

Exactly. But that's not what's happening. Are you completely discounting what Sony is saying?

They haven't said much so far except that they're pushing the envelope on sequential speeds. They haven't explained or demonstrated why that should be a priority for any practical use case. So not much to respond to. The skepticism is based on the fact that games today do not benefit from significantly higher sequential read speeds than what's available on a commodity Sata SSD. The reason being bottlenecks in other components of the system.

 

[Nobu]

They haven't said much so far except that they're pushing the envelope on sequential speeds. They haven't explained or demonstrated why that should be a priority for any practical use case. So not much to respond to. The skepticism is based on the fact that games today do not benefit from significantly higher sequential read speeds than what's available on a commodity Sata SSD. The reason being bottlenecks in other components of the system.

I think sony has removed some of those bottlenecks, which is why they say it cannot be done on PC and why they say it makes such a big difference.

Now, I don't think it's impossible for PC, but it'd require a significant change in how storage and memory are positioned in systems.

 

[next-Jin]

I think sony has removed some of those bottlenecks, which is why they say it cannot be done on PC and why they say it makes such a big difference.

Now, I don't think it's impossible for PC, but it'd require a significant change in how storage and memory are positioned in systems.

I mean that’s my thought process, if none of this mattered there’s no reason to waste the resources and cost to push reads so high and completely rework the I/O. Historically they could have just used a user replaceable NVMe drive like they’ve done with spinners since PS3 and call it a day beefing up their GPU instead.

It doesn’t make sense if there were no direct benefit to gameplay and design. It wasn’t about no load times either, Cerny mentioned that briefly but went into far more details on other aspects.

 

[Marees]

I mean that’s my thought process, if none of this mattered there’s no reason to waste the resources and cost to push reads so high and completely rework the I/O. Historically they could have just used a user replaceable NVMe drive like they’ve done with spinners since PS3 and call it a day beefing up their GPU instead.

It doesn’t make sense if there were no direct benefit to gameplay and design. It wasn’t about no load times either, Cerny mentioned that briefly but went into far more details on other aspects.

From my understanding of the Anandtech article they make it very clear that Sony has worked to remove the bottlenecks

1. Decompression hardware (to replicate this in a PC you would need atleast 16 core processor)
2. DMA controller (to replicate this in PC you would need high system RAM & GPU VRAM such as 32gb RAM, & 12+GB vram)
3. Multi channel access to SSD, you can do serial & random i/o in parallel without impacting each other ( to replicate this you would need a SSD faster than anything available today in the market)

 

[Revdarian]

The benefits of the lowered latency between asking for the data and actually getting it in ram ready to use is that your buffers can now be all that much smaller, this gives you a bigger effective amount of ram, you don't have to preload the data of the next 2 blocks in a city in every hexagon around the player, you can instead load what you are watching and a small extra angle worth of data in every direction.

The benefits of the ID based file system stored in SRAM and the way the controller automatically groups up small reads into a bigger sized read is a much higher throughput for the reads of what usually were multiple small files, giving you a more uniform read performance (SRAM was chosen because of the much smaller latency VS DRAM), aka small 4k random queue depth numbers are not that harsh for their controller.

In any case the patents are there to be read, GDC videos showing how game design has been constrained by the storage also exist and developers with actual experience and not just armchair knowledge has shared their praises explaining a bit what can be possible that couldn't in the past, for the users it won't be initially obvious but for them this is only the beginning, they do need to exercise their imagination after years assuming that the constraints were unavoidable.


Tbh I wanted the multiverse effect of Ratchet and Clank to appear in a Doctor Strange game but hey, this is nice too. (I will expect more film like cuts jumping instantly from one scene to a wildly different one to appear in future narrative games, the prospect of GTA 6 doing this with multiple protagonists is interesting).

 

[MangoSeed]

The benefits of the lowered latency between asking for the data and actually getting it in ram ready to use is that your buffers can now be all that much smaller, this gives you a bigger effective amount of ram, you don't have to preload the data of the next 2 blocks in a city in every hexagon around the player, you can instead load what you are watching and a small extra angle worth of data in every direction.

The biggest difference in game design will be due to having SSDs as standard on both consoles. However an SSD is still dog slow compared to RAM whether it peaks at 2.5GB/s or 5.5GB/s. So you still need to use compression and smart software to stream data from disk well before it's needed.

I'm looking forward to seeing what unique experiences the PS5 can deliver but I would be very surprised if they show anything that's not also possible on XSX.

 

[Revdarian]

Well of course that it's slower than current ram, so you still have a slight buffer, but the size of the buffer is orders of magnitude smaller than with hdd.
As for the compression, of course that's why both have it, for the efficiency. About the possibility, most things should be possible as long as you scale back the assets accordingly, similarly what xsx does could be done on the ps5 as long as you scale the resolution accordingly, these two consoles are absurdly close VS generations past.

Small edit: I would like to point out that currently there's never enough bandwidth, even on vram color compression is a thing to help you better utilize the resources and is theorized that it is one of the things that help Nvidia be more power efficient than AMD.

 

[Marees]

"PS5 and Xbox Series X games will feel different, better and more visceral"
— EA Chief Studios Officer Laura Miele

"dynamic weather systems will impact the athletes, their equipment and the playing field. Games are going to feel different, better and more visceral."

"During my first demo when I experienced the instant load times and massive worlds, I knew immediately this was going to set a new bar in terms of player expectations. When players see how accessible and immersive this generation of technology is, there will be no turning back."

https://www.gamesindustry.biz/artic...-will-feel-different-better-and-more-visceral

 

[DukenukemX]

From my understanding of the Anandtech article they make it very clear that Sony has worked to remove the bottlenecks

1. Decompression hardware (to replicate this in a PC you would need atleast 16 core processor)
2. DMA controller (to replicate this in PC you would need high system RAM & GPU VRAM such as 32gb RAM, & 12+GB vram)
3. Multi channel access to SSD, you can do serial & random i/o in parallel without impacting each other ( to replicate this you would need a SSD faster than anything available today in the market)

Without knowing anything about the PS5's hardware, these things you claim might as well be a work of fiction.
1. Why 16 cores specifically? Have you tested this or seen tests for this?
2. How did you conclude you needed system ram plus GPU VRAM to equal the DMA controller on the PS5?
3. Again, what makes you need faster NVME drives to replicate this?

If I had to take a guess what the PS5 SSD is like, then I believe that Sony solders the NAND chips to the motherboard and uses this DMA controller as a sorta SSD controller. That would explain the claims that Sony is making but you guys do realize that's a mistake right? 825GB of SSD storage is not a lot in 2020, and games built for the PS5 will be much larger. If you can't upgrade the SSD storage then you will be deleting games off the SSD and downloading them very often. Also, NAND chips do eventually fail and if the NAND chips are soldered to the motherboard then when they do fail then you need another PS5. From the sounds of the Xbox Series X, you can still replace the SSD for an upgrade or as a method to repair the console. Maybe Sony might allow external storage to transfer games to and from the SSD but that still sounds like it's going to take over an hour just to download and install a game.

 

[kac77]

Without knowing anything about the PS5's hardware, these things you claim might as well be a work of fiction.
1. Why 16 cores specifically? Have you tested this or seen tests for this?
2. How did you conclude you needed system ram plus GPU VRAM to equal the DMA controller on the PS5?
3. Again, what makes you need modern NVME drives have an issue with this?

If I had to take a guess what the PS5 SSD is like, then I believe that Sony solders the NAND chips to the motherboard and uses this DMA controller as a sorta SSD controller. That would explain the claims that Sony is making but you guys do realize that's a mistake right? 825GB of SSD storage is not a lot in 2020, and games built for the PS5 will be much larger. If you can't upgrade the SSD storage then you will be deleting games off the SSD and downloading them very often. Also, NAND chips do eventually fail and if the NAND chips are soldered to the motherboard then when they do fail then you need another PS5. From the sounds of the Xbox Series X, you can still replace the SSD for an upgrade or as a method to repair the console. Maybe Sony might allow external storage to transfer games to and from the SSD but that still sounds like it's going to take over an hour just to download and install a game.

Same limitations exist for PC. I have a 1TB SSD then a larger HDD for storage and games that's don't require that level of performance, or once I finish a game . Every month or two I spend time moving my steam library over to the HDD.

I don't see how the console is really any different.

 

[Aurelius]

"PS5 and Xbox Series X games will feel different, better and more visceral"
— EA Chief Studios Officer Laura Miele



https://www.gamesindustry.biz/artic...-will-feel-different-better-and-more-visceral

She's overselling it, but there is a degree of truth in that these consoles will remove a number of barriers. It's been ages since you could play a game that just loaded within a few seconds (if it wasn't instantaneous). And there will probably be moments where you'll realize that an experience was technically impossible (or at least, difficult) on earlier consoles, if not many PCs.

 

[Armenius]

Same limitations exist for PC. I have a 1TB SSD then a larger HDD for storage and games that's don't require that level of performance, our once I finish a game . Every month or two I spend time moving my steam library over to the HDD.

I don't see how the console is really any different.

The difference is you won't be able to run games made for PS5 off of slower external storage, period. Microsoft has already said as much with regards to the Series X. If you want to play games you have to move it from storage to the internal SSD first.

 

[kac77]

The difference is you won't be able to run games made for PS5 off of slower external storage, period. Microsoft has already said as much with regards to the Series X. If you want to play games you have to move it from storage to the internal SSD first.

That really not that bad. Like I literally do this all of the time. Here's the scenario. I want to play Rise of the Tomb Raider. It's on my HDD. However, I know that if I don't want to experience crazy load times it's best if I move it. So that's what I do. It's like 5 minutes of my time. I'll get up go take a poo and when I come back I'm ready to rock.

 

[SeymourGore]

It's like 5 minutes of my time. I'll get up go take a poo and when I come back I'm ready to rock.

I miss the days when I could finish making poop in 5 minutes.

 

[IdiotInCharge]

That really not that bad. Like I literally do this all of the time. Here's the scenario. I want to play Rise of the Tomb Raider. It's on my HDD. However, I know that if I don't want to experience crazy load times it's best if I move it. So that's what I do. It's like 5 minutes of my time. I'll get up go take a poo and when I come back I'm ready to rock.

Now ask console folks to figure this out.

Even if Sony provides a means of expanding 'playable' storage, it's going to have one hell of a markup due to the customizations they're making in order to 'jump' ahead in performance while keeping the cost of the base unit down.

 

[kac77]

Now ask console folks to figure this out.

Even if Sony provides a means of expanding 'playable' storage, it's going to have one hell of a markup due to the customizations they're making in order to 'jump' ahead in performance while keeping the cost of the base unit down.

Well if Apple can sell 16GB of memory for $2000 someone will go for it.

 

[DukenukemX]

Same limitations exist for PC. I have a 1TB SSD then a larger HDD for storage and games that's don't require that level of performance, or once I finish a game . Every month or two I spend time moving my steam library over to the HDD.

I don't see how the console is really any different.

Yea but that's games under 100GB in size. With the hardware consoles are getting then I can expect games to reach 200GB in size. This will be a big problem. Also, NAND flash memory fails and if the SDD is soldered to the motherboard then you'll have to buy another PS5. If you want larger storage then you'll have to buy another more expensive PS5.

 

[T4rd]

Yea but that's games under 100GB in size. With the hardware consoles are getting then I can expect games to reach 200GB in size. This will be a big problem. Also, NAND flash memory fails and if the SDD is soldered to the motherboard then you'll have to buy another PS5. If you want larger storage then you'll have to buy another more expensive PS5.

Cerney cited in his brief back in April that games may actually get smaller since they don't have to put several duplicate assets in the game to make it easier for the HDD to seek/load them. If anything, larger textures may cancel that out and make the games about the same size or maybe slightly larger in general. 200 GB games being common at all seems highly speculative, I'd say maybe up to 100 GB at the most for the foreseeable future, with a few exceptions over that like CoD and RDR2.

Also, the storage won't be soldered on and will be replaceable/expandable with commercial drives approved (not licensed or proprietary) by Sony's standards.

https://www.polygon.com/2020/3/18/21185298/ps5-ssd-storage-expansion-xbox-series-x

 

[next-Jin]

Cerney cited in his brief back in April that games may actually get smaller since they don't have to put several duplicate assets in the game to make it easier for the HDD to seek/load them. If anything, larger textures may cancel that out and make the games about the same size or maybe slightly larger in general. 200 GB games being common at all seems highly speculative, I'd say maybe up to 100 GB at the most for the foreseeable future, with a few exceptions over that like CoD and RDR2.

Also, the storage won't be soldered on and will be replaceable/expandable with commercial drives approved (not licensed or proprietary) by Sony's standards.

https://www.polygon.com/2020/3/18/21185298/ps5-ssd-storage-expansion-xbox-series-x

To clarify the primary SSD will be soldered, expandable storage can be a normal M.2.

 

[DukenukemX]

Cerney cited in his brief back in April that games may actually get smaller since they don't have to put several duplicate assets in the game to make it easier for the HDD to seek/load them. If anything, larger textures may cancel that out and make the games about the same size or maybe slightly larger in general.

I've heard this before but what games actually do this? Without showing a specific example I'm gonna say it's rare to find a game that just duplicates files to boost loading time. Good chance any game that does this is due to laziness from the developer as instead of using the same asset from the same source, it would be easier to just package it together with the same wad file or whatever. Also, it would be rather hard to direct where the files are placed on a hard disk, which is why I think it's bullshit. Game sizes won't get smaller.

200 GB games being common at all seems highly speculative, I'd say maybe up to 100 GB at the most for the foreseeable future, with a few exceptions over that like CoD and RDR2.

Obviously that depends on the game but now that consoles can do 1440p to 4k and have Ray-Tracing, we'll start see much larger games. It'll probably be a few years from now before games average closer to 200GB but it'll happen.

Also, the storage won't be soldered on and will be replaceable/expandable with commercial drives approved (not licensed or proprietary) by Sony's standards.

https://www.polygon.com/2020/3/18/21185298/ps5-ssd-storage-expansion-xbox-series-x

Then Sony's claims for faster loading and etc are bunk. M.2 NVME's or SSD's have a clear speed limit and the PS5 won't be making these load faster just because they're in a PS5. Unless Sony bought some special SSD's that are faster than ones on the market today.

 

[T4rd]

I've heard this before but what games actually do this? Without showing a specific example I'm gonna say it's rare to find a game that just duplicates files to boost loading time. Good chance any game that does this is due to laziness from the developer as instead of using the same asset from the same source, it would be easier to just package it together with the same wad file or whatever. Also, it would be rather hard to direct where the files are placed on a hard disk, which is why I think it's bullshit. Game sizes won't get smaller.

Obviously that depends on the game but now that consoles can do 1440p to 4k and have Ray-Tracing, we'll start see much larger games. It'll probably be a few years from now before games average closer to 200GB but it'll happen.


Then Sony's claims for faster loading and etc are bunk. M.2 NVME's or SSD's have a clear speed limit and the PS5 won't be making these load faster just because they're in a PS5. Unless Sony bought some special SSD's that are faster than ones on the market today.

If you just go watch Cerny's presentation, you'll see specific examples and explanations as to why all of these assumptions are wrong.

 

[next-Jin]

If you just go watch Cerny's presentation, you'll see specific examples and explanations as to why all of these assumptions are wrong.

He doesn’t even need to watch that, the freaking Ars Technica war series has several videos talking about exactly why its mandatory to have duplicate files on HDs (or in some cases like Myst) CD/DVDs.

Or Twitter, Developer Reddit, etc.

He is on my ignore list for this reason. You can lay this stuff out and give him links, yet still argues. It’s not like he has critical responses on the topic, he has no response period. He won’t acknowledge your point and bring it back up in a few posts.

 

[DukenukemX]

If you just go watch Cerny's presentation, you'll see specific examples and explanations as to why all of these assumptions are wrong.

You mean the PS5 lead system architect Mark Cerny? That Cerny? I might as well ask a representative from the coal industry for proof of clean coal.

He doesn’t even need to watch that, the freaking Ars Technica war series has several videos talking about exactly why its mandatory to have duplicate files on HDs (or in some cases like Myst) CD/DVDs.

I can understand CD/DVDs because you stamp them out, but files on a HDD are not stamped out. They don't know where the files will be placed just as much as you do. What if most of the data is near the inner portion of the drive? What if the files fragment? I can think for myself ya know.

He is on my ignore list for this reason. You can lay this stuff out and give him links, yet still argues. It’s not like he has critical responses on the topic, he has no response period. He won’t acknowledge your point and bring it back up in a few posts.

Do me a favor and continue to ignore me instead of responding to me indirectly. You layed shit out in front of me. Any links? Any videos? Anything that compares to loading times on PC?

 

[T4rd]

He doesn’t even need to watch that, the freaking Ars Technica war series has several videos talking about exactly why its mandatory to have duplicate files on HDs (or in some cases like Myst) CD/DVDs.

Or Twitter, Developer Reddit, etc.

He is on my ignore list for this reason. You can lay this stuff out and give him links, yet still argues. It’s not like he has critical responses on the topic, he has no response period. He won’t acknowledge your point and bring it back up in a few posts.

Oh, I know all to well his MO and pretty much called it on the first page in hopes to keep his shitposting out of the thread. I prolly should ignore him as well by this point, but he's fun to poke sometimes when I'm bored too. He's like one of the women on the View trying to talk about anything political; it's somewhat entertaining even if it's completely ignorant and annoying.