• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Next-Gen PS5 & XSX |OT| Console tEch threaD

Status
Not open for further replies.
Time jumps purposely from 0.30 to 01.00 to inflate the total time, pure FUD.

giphy.gif


when-you-point-a-finger-at-the-moon-to-indicate-the-moon-instead-of-looking-1563639-1.jpg
Bo it's worse than that, it's every 30ms it jumps 30ms more. Yeah this should qtleast have to be taken down. This is false information.
 
Last edited:
Timecode is totally a thing. It has its origins in cinema, where it was basically used to be able to quickly find a specific frame in the timeline so it would be easier to edit and so on.
It's usually 8 digits; hours:minutes:seconds:frames / 00:00:00:00
how dare you!!!!!!!!


Bo it's worse than that, it's every 30sec it jumps 30 more. Yeah this should qtleast have to be taken down. This is false information.
go on, pull out your smartphone, put it in stopwatch mode, and time it yourself:
 

IntentionalPun

Ask me about my wife's perfect butthole
Time jumps purposely from 0.30 to 01.00 to inflate the total time, pure FUD.

giphy.gif


when-you-point-a-finger-at-the-moon-to-indicate-the-moon-instead-of-looking-1563639-1.jpg

As far as I can tell.. that's just the display of the timer, if you actually compare it to the time of the video, when it says a second has elapsed, a second has elapsed. It's just not accurately showing the fractions of a second, and just throwing numbers on the screen to simulate it.
 

IntentionalPun

Ask me about my wife's perfect butthole
I'm not sure anymore if the guy made on purpose of spread FUD or if he has really bad skills at editing lol
It's just a shitty "timer" effect.

If you actually time it, it is counting the seconds accurately as far as I can tell. The fractions of the second aren't accurately, it's just spitting some numbers out and then going back down to 0 when it reaches a second, but the seconds appear accurate as far as I can tell.

That's what TheGreatWhiteShark TheGreatWhiteShark has been going on about.

edit: As someone else pointed out, it's actually a frame timer.. that's why it goes up to 30.. 30FPS
 
Last edited:
D

Deleted member 775630

Unconfirmed Member
So why wasn't it shown? I mean, this is not a behind closed doors video that someone bootlegged out of Redmond here. It was an official MS publicity stunt, with the purpose of showing the "fast" loading and contrast it with current gen.

Why would you think they would give themselves anything but the best possible conditions to impress us, their potential clients?

So, it's either:

a) really complex to compress / decompress, etc, the data and they didn't have time/resources to do it at the time of this demo, or
b) maybe this is as good as it gets even with all the marketing BS of velocity architecture and you're hyping a dream (like the cloud last gen)

Btw I used the complexity argument because you pretty much implied it by referring to a new API that requires more code to be written to take advantage of this. From what I remember the games shown (or at least several) were MS first party games, so how hard would it be for them to write whatever code they needed to show us this amazing speed?

Also, what sort of amazing textures were these games loading that you need this much compression? State of Decay? PUBG? For real...
Because it was about old (unoptimized for VA) games, showing their load time. Not new ones, just look at what we can already do when we just run it and not do anything new to it. Doing something new might takes time, who knows, but that wasn't the point of this. So I still don't agree with your two options.
 
I gotta say, that XVA, DirectStorage and bla bla marketing from MS did nothing to build the hype. I get much more erect by hearing about Haptic Feedback, Adaptive Triggers and 3D Audio because these are tangible things I'll be able to experience later this year.

I like this, this rush to feel something new, it's a fucking new generation.
 

zaitsu

Banned
It's just a shitty "timer" effect.

If you actually time it, it is counting the seconds accurately as far as I can tell. The fractions of the second aren't accurately, it's just spitting some numbers out and then going back down to 0 when it reaches a second, but the seconds appear accurate as far as I can tell.

That's what TheGreatWhiteShark TheGreatWhiteShark has been going on about.
Who the fuck cares what he has been going about, when you can measurment loading with animation which has been made to not only hide loading but to look good from artistic point of view. Like wtf is wrong with you. Its not like this is 10s loading screen on XSX or PC.
 

IntentionalPun

Ask me about my wife's perfect butthole
Who the fuck cares what he has been going about, when you can measurment loading with animation which has been made to not only hide loading but to look good from artistic point of view. Like wtf is wrong with you. Its not like this is 10s loading screen on XSX or PC.
Hey don't shoot the messenger here.

I think loading an entire word in a couple seconds is amazing.

But that timer appears accurate.. you can compare it against the video time.

*shrug*
 
Last edited:
Last edited:
I am just basing my speculation on what we have seen.
It is more of a denial to say that on astrobot loading times are big because the developers wanted us to look at the ....tunnel (I can provide quotation on that)
or that insomniac while being the frontrunner for ps5 SSD, intentionally bloats up the loading times because ...artistic reasons (I can also provide quotation for this too)
Do you know in which hardware astrobot was running? Or are you assuming it was on PS5? Honest question here.
 
So, guys, not trying to add fuel to the fire here, but I've seen quite some comments about the Ratchet & Clank loading times video and how the timer jumps up, so I think we should clarify that the timer is okay, it's using TC (timecode). TC is used in video productions, it has its origins in cinema and it is usually used to pinpoint a frame in the timeline for easier editing and so on.

It usually has 8 digits; hours:minutes:seconds:frames / 00:00:00:00



If you play the video at 0.25x speed, you'll clearly see the frame counter going up to 29 and resetting to 0 just in time to hit one second, and since we know the gameplay shown is 30fps, I think it's accurate.

Now, whatever we may think about the loading time being fast or not, that's another discussion (in my opinion, it's blazing fast), but I just wanted to clarify that.
 
Just opened a video editor and checked, time seems about right, but the time keeps going until he lands, which is stupid, it doesn't take ~6 seconds to load the level, it takes just over 2.
I'd say that the counter on that video starts way too late.
If you see my own video (look at top of this page), which I cut from the latest gamescom video,
we are out of gameplay the moment ratchet climbs those stairs.
all the gliding down, monster and jump into void is not gameplay.
 
Last edited:
So, guys, not trying to add fuel to the fire here, but I've seen quite some comments about the Ratchet & Clank loading times video and how the timer jumps up, so I think we should clarify that the timer is okay, it's using TC (timecode). TC is used in video productions, it has its origins in cinema and it is usually used to pinpoint a frame in the timeline for easier editing and so on.

It usually has 8 digits; hours:minutes:seconds:frames / 00:00:00:00



If you play the video at 0.25x speed, you'll clearly see the frame counter going up to 29 and resetting to 0 just in time to hit one second, and since we know the gameplay shown is 30fps, I think it's accurate.

Now, whatever we may think about the loading time being fast or not, that's another discussion (in my opinion, it's blazing fast), but I just wanted to clarify that.

HOW DARE YOU, AGAIN!!!
;]
 

IntentionalPun

Ask me about my wife's perfect butthole
I'd say that the counter on that video starts way too late.
If you see my own video, which I cut from the latest gamescom video,
we are out of gameplay the moment ratchet climbs those stairs.
all the gliding down, monster and jump into void is not gameplay.
You have no clue how much of it is actually used for "loading" though vs. how much is a stylistic choice.

PS5 certainly wouldn't need that much time for asset loading... if Ratchet is doing a bunch of calculations it might need some extra time.. but between compression, shared assets, and the insane I/O... you wouldn't need any more than a couple seconds ever to load anything on PS5 asset wise.

Entire RAM for a game is probably at most 14GB... subtract out shared assets (let's say 2GB).. and let's be super conservative on compression (since not all assets are textures) and say compression shrinks another 2 GB...

You are left with 10GB... and that can be loaded in under 2 seconds on PS5.

Asset loading ISNT the only factor of a "loading time" though.. so who really knows? You certainly don't.
 
Last edited:

YoodlePro

Member
would probably make a much more intelligent conversation than what we are having here right now with some folks
woooooow.

So putting aside the fact that you only say one thing over and over, paraphrased so it's not treated as spam, you also have the audacity of insinuating that speaking to yourself would be a "much more intelligent conversation", even though you so "intelligently" avoid all counter arguments coming your way.

Do everyone a favour and gtfo of here and stop smelling your own farts.
 

Mr Moose

Member
I'd say that the counter on that video starts way too late.
If you see my own video (loot at top of this page), which I cut from the latest gamescom video,
we are out of gameplay the moment ratchet climbs those stairs.
all the gliding down, monster and jump into void is not gameplay.
I used the video from here:
From the moment he enters the portal to the moment he exits is the loading the new level.
I got roughly 2.26 seconds.
 
Because it was about old (unoptimized for VA) games, showing their load time. Not new ones, just look at what we can already do when we just run it and not do anything new to it. Doing something new might takes time, who knows, but that wasn't the point of this. So I still don't agree with your two options.
I don't know why you're being so "generous" in your assessment of this demo Microsoft themselves put out. They control their own narrative and it's on them to prove their shit works.

They made a whole video whose point was to demonstrate this and until we saw the PS5 solution everyone was like "ok, fine". So now you're coming up with excuses for them, somehow implying poor little Microsoft somehow was done wrong, and it would be so much better if only...

Man, there's no if only. They presented this, proudly, as a demonstration of how great their loading speeds are. Clearly the competition is miles better than them, period.

I agree with your point above thought "... who knows...". Exactly. So don't make up excuses, go from what is factual. If tomorrow MS shows some title loading new worlds as quickly or faster then R&C then great. We can then give kudos and say well done. Not now.
 

geordiemp

Member
xbone games do not use the double compression of the new xboxes Geordie.
only xbox series x/s games will.
so it's ~double the time.
also, this procedure is not a read (load). its write image of current game(save), then load image of selected game.

I thought you were more knowledgeable than what you write here.

And one of the games was 1 GB.....also compression is not double for Zlib, the Bcpack like oodle compresses the data when making the textures, so they are not loaded faster. Bcpack reorders and Mp4 effect on the image before converting to BCn

The point is as I understand the technology

1. SFS is a LOD management system and tries to load less, if LOD is late its blended in by hardware (goosen quote) so its not as obviously ugly as pop in. But its still late - was halo pop in blended ?

2. Everything else is software to overcome the really slow xbox IO system, see Linus tech test of 27 Gbs SSD on W10 it runs faster in Linux and is slowed down allot in windows

3. VA is a software solution to reduce the IO overhead of standared MS IO system, more Linux like. Hardware is the LOD blender and the zlib decompression although MS were cagey on details at Hotchips...

It will still be slow but will remove SOME of the bottlenecks to allow reasonable SSD which is what was shown in the demos.

The rest is allot of PR. I am happy to be proven wrong
 
Last edited:

IntentionalPun

Ask me about my wife's perfect butthole
xbone games do not use the double compression of the new xboxes Geordie.
only xbox series x/s games will.
so it's ~double the time.
also, this procedure is not a read (load). its write image of current game(save), then load image of selected game.

I thought you were more knowledgeable than what you write here.
XBone games.. ran on a machine with only ~5GB of RAM available to devs...

XSX games should be using more than double that really.

Also compression is for textures ins't it? 3D models are a significant portion of RAM usage. You can't just take a compression ratio and cut the entire asset load in half.

Really what should be pointed out is that this is a full "resume" of a game.. and a game doens't need to fill the entirety of RAM to load a level. That's what asset streaming systems handle. But the amount that needs to be loaded in order for the game to start playing is probably similar to the full RAM dump of an XBone game.. but that depends on the game.
 
Last edited:
I don't know why you're being so "generous" in your assessment of this demo Microsoft themselves put out. They control their own narrative and it's on them to prove their shit works.

They made a whole video whose point was to demonstrate this and until we saw the PS5 solution everyone was like "ok, fine". So now you're coming up with excuses for them, somehow implying poor little Microsoft somehow was done wrong, and it would be so much better if only...

Man, there's no if only. They presented this, proudly, as a demonstration of how great their loading speeds are. Clearly the competition is miles better than them, period.

I agree with your point above thought "... who knows...". Exactly. So don't make up excuses, go from what is factual. If tomorrow MS shows some title loading new worlds as quickly or faster then R&C then great. We can then give kudos and say well done. Not now.
I already wrote to explain to geordiemp geordiemp that this quick resume demo
a) is not just loading. its saving the current game image so it can resume at present state, then loading the selected game image to its last saved stage. notice: writing to SDD is slower than reading
b) current xbone games do not use the double compression tech that series s/x have. so its basically ~double the time for this procedure.

forza 7 is ~100gb and state of decay is ~20, yet they take the same time for this procedure, as it is basically a memory dump along with game details. I've read somewhere that each such image is ~9gb
 
Bo, you're wrong. The counter is timecode. It measures seconds and video frames. Every second of video is appx 25 frames. it isn't skipping anything.

Edit: Reading back through this thread I think a few of you owe ol' Sharky an apology. Geez, you guys are super quick to jump to conclusions.

It's amazing (not in a good way).


TheGreatWhiteShark TheGreatWhiteShark
Okay I apologize as I can see that it is a measurement of frames per second which is not even what this discussion was about. We are talking about loading times. So I apologize for letting myself even get into this because it wasn't even going to help this argument. Why use fps to measure loading times? I don't even really care. He was talking to Bo in seconds for loading and used a video showing a timer for frames. Moving on.
 

geordiemp

Member
I already wrote to explain to geordiemp geordiemp that this quick resume demo
a) is not just loading. its saving the current game image so it can resume at present state, then loading the selected game image to its last saved stage. notice: writing to SDD is slower than reading
b) current xbone games do not use the double compression tech that series s/x have. so its basically ~double the time for this procedure.

forza 7 is ~100gb and state of decay is ~20, yet they take the same time for this procedure, as it is basically a memory dump along with game details. I've read somewhere that each such image is ~9gb

No think of BCpack OR oodle BCn as saving an image as a JPEG for simplicity. If that JPeg is 1 GB for example, that is your load size. It does not matter if the original file before JPEG was 10 GB. Does not matter. Its still loading 1 GB.

It stil loading a 1 GB file, and its used as a JPEG (or BCn) - if Zlib or kraken uncompress adds 10 % so it becomes 1.1 GB, so what ?

Loading is not twice as fast. What we see is their best loading (so far).

Both systems use RDO and a compression to create the BCn files and decompressed using Zlib or kraken. The uncompressed by Zlib is still RDO compressed Bcn.
 
Last edited:
D

Deleted member 775630

Unconfirmed Member
I don't know why you're being so "generous" in your assessment of this demo Microsoft themselves put out. They control their own narrative and it's on them to prove their shit works.

They made a whole video whose point was to demonstrate this and until we saw the PS5 solution everyone was like "ok, fine". So now you're coming up with excuses for them, somehow implying poor little Microsoft somehow was done wrong, and it would be so much better if only...

Man, there's no if only. They presented this, proudly, as a demonstration of how great their loading speeds are. Clearly the competition is miles better than them, period.

I agree with your point above thought "... who knows...". Exactly. So don't make up excuses, go from what is factual. If tomorrow MS shows some title loading new worlds as quickly or faster then R&C then great. We can then give kudos and say well done. Not now.
Well you have your interpretation of that video, and I have mine. Since the start I saw that and thought, OK so this is when you do nothing special. Because indeed, we've seen PS5 speeds, we know the theoretical numbers of both. So there's no reason for XSX to be this slow, unless it's not properly optimized to use the system. Which is why, like I said, I see this as a "install and run, did nothing special to it".
 

zaitsu

Banned
I'd say that the counter on that video starts way too late.
If you see my own video (loot at top of this page), which I cut from the latest gamescom video,
we are out of gameplay the moment ratchet climbs those stairs.
all the gliding down, monster and jump into void is not gameplay.
It will be funny to see your reaction when R&C would launch from main menu to playable game under time needed to travel through this portal :)
Maybe then youll undesrstand something like art direction and choose of lenght of this animation.
 
Last edited:
No think of BCpack OR oodle BCn as saving an image as a JPEG for simplicity. If that JPeg is 1 GB for example, that is your load size. It does not matter if the original file before JPEG was 10 GB. Does not matter. Its still loading 1 GB.

It stil loading a 1 GB file, and its used as a JPEG (or BCn)

Loading is not twice as fast.
once again, I really though you were more knowledgeable than what you write here.
maybe you read stuff only regarding the playstation, thats why
 

geordiemp

Member
once again, I really though you were more knowledgeable than what you write here.
maybe you read stuff only regarding the playstation, thats why

I am, its you that dont understand.

Lets use images as a simpler analogy....close enough best TIFFs and JPEG as an example

Original file = 4
JPEG = 2
Zipped Jpeg = 1.8

Loads 1.8
Unzips Jpeg = 2
 
Last edited:
You dont understand Bcn, thats fine, it is complex Rate distortion and compression is not simple and people often combine the compression of RDO with what Kraken and Zlib do.

Hence your 2 x as your adding RDO compression of Bcn + Zlib.

The XSX decompression is only Zlib.

Do you want me to explain more ? This doubling is-understanding is also what confuses Ps5 loading speeds.
once again you describe the quick resume procedure as a load, I have already written twice to you that this procedure entails first a full image save of the current game in its present moment, then loading of the selected game in its last saved image moment.
watch that video again and notice that forza 7 for example loads mid-game while lapping.
if you boot forza 7 from scratch, until you reach the track hotlapping you need to pass 5 or 6 different loading points.
thats why its called "quick resume", and you still keep referencing it as a "load". basically its a 9BG image save, then a 9GB image load.

I dont know what else to tell you. You should read more about these stuff before appearing so certain.
 

geordiemp

Member
once again you describe the quick resume procedure as a load, I have already written twice to you that this procedure entails first a full image save of the current game in its present moment, then loading of the selected game in its last saved image moment.
watch that video again and notice that forza 7 for example loads mid-game while lapping.
if you boot forza 7 from scratch, until you reach the track hotlapping you need to pass 5 or 6 different loading points.
thats why its called "quick resume", and you still keep referencing it as a "load". basically its a 9BG image save, then a 9GB image load.

I dont know what else to tell you. You should read more about these stuff before appearing so certain.

Lets use images as a simpler analogy....close enough best TIFFs and JPEG as an example

Original file = 4
JPEG = 2
Zipped Jpeg = 1.8

Loads 1.8
Unzips Jpeg = 2

Yes, you can say hey it loaded 4 file size, but it dod not, it loaded a lossy JPEG of 2.

Also cave was 1 GB

XSX BCpack RDO + Zlib = Ps5 oodle RDO + Kraken (roughly same technique)
 
Last edited:
Lets use images as a simpler analogy....close enough best TIFFs and JPEG as an example

Original file = 4
JPEG = 2
Zipped Jpeg = 1.8

Loads 1.8
Unzips Jpeg = 2

Yes, you can say hey it loaded 4 file size, but it dod not, it loaded a lossy JPEG of 2.

Also cave was 1 GB

XSX BCpack RDO + Zlib = Ps5 oodle RDO + Kraken (roughly same technique)
no, lets describe it for what it is!
 
I genuinely appreciate the answer and the trouble you went to to gather up that info. It's interesting to fill in the gaps.

Is that all it is though? A group of people in a discord group, some of whom were a bit naughty on Era? It's hardly al-qaeda is it? I mean, that kind of group reporting can happen, and probably has happened, here.

I've seen the DF guy getting hounded for being part of that discord, even though none of his responses, at least those that I've read, seem overly problematic to me.

He's laughed at some of the more extreme fans, sure, I imagine that he's had to put up with a lot though, much of it coming from a particular group of 'you know who' fans, it's probably good for his soul to let off a little steam from time to time.

There seems to be some 'guilt by association' involved.

I use this place alongside a lot of other people. Some of those people post nonsense and are utterly fact-resistant. I'm sure you've had to deal with it to.

The thing is, people from outside this forum probably lump us all together in the exact same way that Sony fans have decided that anyone in that discord group is part of some sweeping conspiracy.

It seems to me, if your examples are typical, that a small number of people have overreacted to the discord group and then made loud and confident assertions that evil was afoot. It seems those assertions have been taken at face value by nearly everyone else (me included).

There really doesn't seem to be a lot to it though, just more fanboy turfwars. :)

Again, thanks for the info, it's been really helpful.
Wow, I feel like I just sat through a therapy session - lots of good in what you said but don't make excuses for any of those guys. They are more brand loyal than game loyal. Which is why they are glad for Sony games coming to PC. Although they have no idea that Sony First Parties are a gaitway drug.
 
I already wrote to explain to geordiemp geordiemp that this quick resume demo
a) is not just loading. its saving the current game image so it can resume at present state, then loading the selected game image to its last saved stage. notice: writing to SDD is slower than reading
b) current xbone games do not use the double compression tech that series s/x have. so its basically ~double the time for this procedure.

forza 7 is ~100gb and state of decay is ~20, yet they take the same time for this procedure, as it is basically a memory dump along with game details. I've read somewhere that each such image is ~9gb
on these points:

a) I'm going to assume MS engineers are savy enough to partition some OS reserves for this floating game state. So (again assuming) variables are being tracked and instantaneously stored to memory in such a way that you don't need to initiate a special new process every time you i) jump to the taskbar and ii) click start on a different game. If that's the case, the whole "saving the current game" isn't a metric you can use as excuse for the loading to take longer time.

b) Why do I care about the compression at this stage? What needs to be compressed there when we're looking at some basic textured games that run on Xbox 1 like state of decay and pubg? This is not like you're having some ultra high quality 4k or 8k textures for all kinds of elements in the game. Yeah, sure, in next gen games having kraken and other compression algorithms will be really important to load all these textures to memory without running out of space. This shouldn't be a problem on the games they used to demo this feature.

On an apples to apples comparison, Spider Man was loading in 0.8 seconds on the behind closed doors demo Sony used. I'm pretty sure if MS had anything close to this, they would have shouted it out from the rooftops ages ago.
 
Well you have your interpretation of that video, and I have mine. Since the start I saw that and thought, OK so this is when you do nothing special. Because indeed, we've seen PS5 speeds, we know the theoretical numbers of both. So there's no reason for XSX to be this slow, unless it's not properly optimized to use the system. Which is why, like I said, I see this as a "install and run, did nothing special to it".
So you're just confirming Microsoft are absolutely incompetent at selling their own product.

That's me giving the most generous interpretation possible.
 

banjo5150

Member
Hey don't shoot the messenger here.

I think loading an entire word in a couple seconds is amazing.

But that timer appears accurate.. you can compare it against the video time.

*shrug*
I see no problem with it being a second or two, I agree that is amazing. I just wonder is it possible to be just about seamless if they wanted it to be? I remember Cerny saying in Road to the PS5 talk about loads can be so fast its quite jarring. So could these 2 sec loads basically be them slowing it down on purpose for a transition?
 
D

Deleted member 775630

Unconfirmed Member
So you're just confirming Microsoft are absolutely incompetent at selling their own product.

That's me giving the most generous interpretation possible.
Nope, stop putting words in my mouth. You are the one that is giving a twist to their loading time video. It's a video from April/May where they are showing a current-gen game loading. That's it. I'm not being generous, just stating facts.

You on the other hand are deliberately being negative. You already understand the situation, you know that there are optimisations to be made that they didn't make, and you keep trying to put something negative on it. You started by saying that this is representative for VA, that there weren't improvements to be made. I tell you what they can do, then it's crappy marketing, and on and on and on.

It's loading times. PS5 will be faster. If PS5 can load something within a second, XSX will load it in 2-3 seconds. That's it.
 
Status
Not open for further replies.
Top Bottom