• Ever wanted an RSS feed of all your favorite gaming news sites? Go check out our new Gaming Headlines feed! Read more about it here.
  • We have made minor adjustments to how the search bar works on ResetEra. You can read about the changes here.

Oh no

Prophet of Truth
Member
Oct 25, 2017
653
I can't believe that there are people out there who say XSX is only brute forcing power. I guess they just intentionally ignore all the customizations MS did to the console and believe MS built a PC from newegg parts.

It's weird that the narrative is XSX = power without thought and PS5 = intelligently designed. A big one seems to be the audio chips, tons of praise for the PS5 one (and quite right too!) but it seems some folk think the XSX doesn't have one at all given how the PS5 one is touted as some huge advantage.

Both consoles have a myriad of customisations/enhancements to improve efficiency etc and I can't wait to see both in action.
 

Chettlar

Member
Oct 25, 2017
13,604
It's weird that the narrative is XSX = power without thought and PS5 = intelligently designed. A big one seems to be the audio chips, tons of praise for the PS5 one (and quite right too!) but it seems some folk think the XSX doesn't have one at all given how the PS5 one is touted as some huge advantage.

Both consoles have a myriad of customisations/enhancements to improve efficiency etc and I can't wait to see both in action.

I think the reason is because there are a few things that Sony seems to be more specialized in. So for example the whole kraken compression thing, and the custom controller, and all that, a lot of effort being made into ensuring a minimum 5.5 raw gbps transfer speed that can be processed as fast. Microsoft doesn't seem to be doing the same, although of course there's the matter of them just, not going into as much detail.

Still, I think it's a fair assessment to note (and yes the more tech head people around here have mentioned this) that the PS5 is a very nimble, specialized machine, whereas the xbox will just have that raw power advantage.

But from that you start getting people separating the differences to their extremes. People are eager for a differentiating factor, so anything that exists will get isolated and made extreme. Both consoles are doing most things the other is doing. Still, they have pretty different approaches to some things. Like power vs frequency. They are taking the exact polar opposite approach of each other.

Anyway could be wrong but I was pondering an example where like, so they made a big deal about how the PS5 streams info so fast you could stream it while a character was literally looking around. That's pretty crazy. The xbox will also clearly do this but it won't be as fast (2-4[?] vs 5.5-8). It could be that streaming info means that in a given scene the PS5 wouldn't have to hold as much data to be calculating the position and condition of at one time and be more specifically only doing what it needs right then, whereas the Xbox would hold onto more at one time, but would be able to do it just fine and wouldn't need to be dumping it out as fast. I don't know. I'm probably off base here, but I was trying to think of an example. Hopefully someone more knowledgeable could explain an instance where the PS5's design might make up for some of the loss in power. Cuz idk that's the best I have right now.
 

EvilBoris

Prophet of Truth - HDTVtest
Verified
Oct 29, 2017
16,686
Anyway could be wrong but I was pondering an example where like, so they made a big deal about how the PS5 streams info so fast you could stream it while a character was literally looking around.

Horizon does this type of thing, imagine being able to achieve that level of fidelity across more titles.
 

Trup1aya

Literally a train safety expert
Member
Oct 25, 2017
21,382
Lockhart is something that might not even get a green light anyways, depending on how the BoM shakes out and whether MS feels saucy enough to eat some costs and price it whatever Sony does with PS5.

More and more I think that Sony might have spent excess budget on on the SSD and its subsystem enough that the BoM between XSX and PS5 might not be that far apart...

If PS5 is $500 then I think MS will end up matching it with $500 for XSX. If Sony pulls off $400 somehow, then MS might employ Lockhart, XSX dual prong approach, but Lockhart even at $300 might be a tough sell to consumers.

At $250 it might have a chance for making a niche for itself as a 1080p version of XSX, but that's really hard to get to with 1TB of SSD even with say an 18CU 8 core Zen 2 SoC. You'd have to go UHD BR-less AND maybe but down the internal SSD to say 512GB...

I think a $300 Lockhart would be near impulse buy territory. It wouldnt be niche, it would be tough to keep in stock. A next gen system for the cost of a Nintendo switch. A viable first choice for those not bothered by UHD, an obvious second system for those embedded in other ecosystems.

The best 2 pronged approach MS could employ would be XSX beating ps5 on power, matching its price whilst Lockhart gobbles up the low end market.
 

EvilBoris

Prophet of Truth - HDTVtest
Verified
Oct 29, 2017
16,686
I think a $300 Lockhart would be near impulse buy territory. It wouldnt be niche, it would be tough to keep in stock. A next gen system for the cost of a Nintendo switch. A viable first choice for those not bothered by UHD, an obvious second system for those embedded in other ecosystems.

The best 2 pronged approach MS could employ would be XSX beating ps5 on power, matching its price whilst Lockhart gobbles up the low end market.

that isn't going to be achievable when the storage alone will cost half of that
 

Fatmanp

Member
Oct 27, 2017
4,438
It's weird that the narrative is XSX = power without thought and PS5 = intelligently designed. A big one seems to be the audio chips, tons of praise for the PS5 one (and quite right too!) but it seems some folk think the XSX doesn't have one at all given how the PS5 one is touted as some huge advantage.

Both consoles have a myriad of customisations/enhancements to improve efficiency etc and I can't wait to see both in action.
Agreed, if anything overclocking a chip to hell sounds more like brute force than what MS is doing.
 

Chettlar

Member
Oct 25, 2017
13,604
Horizon does this type of thing, imagine being able to achieve that level of fidelity across more titles.

Right, and it's not exactly super novel that it does that. And the CPU is still keeping track of a lot of things already anyway so that can pop in smoothly so your game doesn't like, chug suddenly when you look a certain way. Iirc it's increasingly standard practice, but yeah I think cerny was alluding to this idea. Horizon uses a smartly designed engine but yeah it's ultimately limited by the RAM pretty much. It's not streaming that info from the hard drive in real time. That's just too fast for what an HDD can do. But an SSD like the PS5's or Xbox SX's. That's different. So I think it'll be a bigger jump than we may be expecting.
 

NippleViking

Member
May 2, 2018
4,491
I think a $300 Lockhart would be near impulse buy territory. It wouldnt be niche, it would be tough to keep in stock. A next gen system for the cost of a Nintendo switch. A viable first choice for those not bothered by UHD, an obvious second system for those embedded in other ecosystems.
Yep. Or more importantly it's for people (particularly parents) who simply want whatever's the cheapest point of entry to next-gen zeitgeists.
that isn't going to be achievable when the storage alone will cost half of that
I agree that a Lockhart for $300 isn't viable at launch, but 2 years down the track - when the generation finally acquires momentum - is where you need to be looking. Year 1 will be predominantly enthusiasts and early adopters anyway, who will almost assuredly gravitate towards SX.
 

EvilBoris

Prophet of Truth - HDTVtest
Verified
Oct 29, 2017
16,686
Right, and it's not exactly super novel that it does that. And the CPU is still keeping track of a lot of things already anyway so that can pop in smoothly so your game doesn't like, chug suddenly when you look a certain way. Iirc it's increasingly standard practice, but yeah I think cerny was alluding to this idea. Horizon uses a smartly designed engine but yeah it's ultimately limited by the RAM pretty much. It's not streaming that info from the hard drive in real time. That's just too fast for what an HDD can do. But an SSD like the PS5's or Xbox SX's. That's different. So I think it'll be a bigger jump than we may be expecting.

Yeah I think so, let's hope this stuff is genuinely valuable and simple to use, a number of the customisations in the PS4 barely got touched and I'm sure it's the same for the Xbox One.

I did read somewhere that the customisations for the XS are almost like a multiplier for the RAM, giving the numerical equivalent of 30GB.
That's pretty bananas when you think about it like that, what kind of leap could we expect?
 

Chettlar

Member
Oct 25, 2017
13,604
Agreed, if anything overclocking a chip to hell sounds more like brute force than what MS is doing.

Well they're not exactly overclocking it to hell. The way this boost clock isn't quite like your standard boost clock. Digital Foundry's video explains it a bit better.

When people mean brute force they mean raw power doing something smarter with less power could do. Boost clock to respond to load and keep power consistent isn't brute forcing. That's...not really what that means. For example, you'd need a like 25TFLOP GPU to bruteforce what the xbox series X will do, if you didn't have that dedicated raytracing hardware on there, very generally speaking. So you could brute force it with higher power, or design it more smartly. So brute forcing is the appropriate word for xbox vs PS5 here, but it's still an oversimplification.

Even if this was a straightforward overclock, trying to keep up with something else in terms of raw power is not "brute forcing." Brute forcing is something you do to a given task, not a descriptor of something's traits. You don't brute force power. Brute force is the act of you using power to do a task.
 

Prine

Attempted to circumvent ban with alt account
Banned
Oct 25, 2017
15,724
It's weird that the narrative is XSX = power without thought and PS5 = intelligently designed. A big one seems to be the audio chips, tons of praise for the PS5 one (and quite right too!) but it seems some folk think the XSX doesn't have one at all given how the PS5 one is touted as some huge advantage.

Both consoles have a myriad of customisations/enhancements to improve efficiency etc and I can't wait to see both in action.
I've said this before, the forum disproportionately leans towards Sony, we have multiple polls that show this, so whilst PS5 is well designed, we get a large portion of people automatically assuming if that's true the XSX can't be. If enough of these people post in a thread falsely saying XSX is "off the shelf" then that obviously sets an incorrect narrative that people who don't know much about the XSX adopt unconsciously and unfortunately a then stigma forms.
 

big_z

Member
Nov 2, 2017
7,797
I think the reason is because there are a few things that Sony seems to be more specialized in. So for example the whole kraken compression thing, and the custom controller, and all that, a lot of effort being made into ensuring a minimum 5.5 raw gbps transfer speed that can be processed as fast. Microsoft doesn't seem to be doing the same, although of course there's the matter of them just, not going into as much detail.

Microsoft is supposed to talk about their storage solution more in the future. Not sure if it will be viewable by public or just devs. Richard Geldreich, who specializes in this stuff says Microsoft is using something that can best kraken and they're doing it in a smarter more efficient way where you only take the parts of the Buffalo needed, instead of moving the whole thing. We know Microsoft's speeds are sustained, we only know where Sony's tops out. In the end real world results probably won't be too different between them as both companies are trying to do the same thing but with a different approach.
 

nib95

Contains No Misinformation on Philly Cheesesteaks
Banned
Oct 28, 2017
18,498
It's weird that the narrative is XSX = power without thought and PS5 = intelligently designed. A big one seems to be the audio chips, tons of praise for the PS5 one (and quite right too!) but it seems some folk think the XSX doesn't have one at all given how the PS5 one is touted as some huge advantage.

Both consoles have a myriad of customisations/enhancements to improve efficiency etc and I can't wait to see both in action.

I don't think it's necessarily the audio chip stuff, I think it's because the PS5 has many innovative and/or much bigger unknowns, that show they're thinking quite far outside the box (whether that's good or bad remains to be seen).

For example, the fact that they're running such high clocks (unseen frequencies really). Or that their boosted speeds aren't really actually boosted per se, nor tied to thermal dynamics (eg throttled based on temperature like other boost modes in other products) but instead to a set power load (something that afaik hasn't been done before).

Then there's the custom SSD, which has technology/performance akin to Saturate PCIe 4.0 SSD tech that isn't even available yet, and might not even be able till just after the PS5 launches.

Add to that we don't really even know why they've gone so hard on this SSD tech, nor exactly what sort of dividends it will or won't have, or how it might shape game design, notably with their exclusives. But already some are theorising about potential benefits it could have, even as far as things like saving system images or offloading OS functions from memory (caching it aside) to provide access to far more of the GDDR6 ram for actual game use etc.

Right now there are quite a lot of unknowns, and lots of details and information yet to be shared from both Microsoft and Sony, so I guess we'll just have to wait and see. Personally I think both systems sound really competent, and I'm genuinely excited for both.
 

Theorry

Member
Oct 27, 2017
61,057
Its funny to read the first posts of this thread. "What a beast" etc. Couple of days later it suddenly doesnt matter and things are "off the shelf"
 

Chettlar

Member
Oct 25, 2017
13,604
Yeah I think so, let's hope this stuff is genuinely valuable and simple to use, a number of the customisations in the PS4 barely got touched and I'm sure it's the same for the Xbox One.

I did read somewhere that the customisations for the XS are almost like a multiplier for the RAM, giving the numerical equivalent of 30GB.
That's pretty bananas when you think about it like that, what kind of leap could we expect?

Dude it's genuinely so much more exciting than last gen for me. I remember being really annoyed hearing CPUs weren't going to be that good back in 2013. We're getting a bigger leap than the tflops numbers suggest for a number of reasons, and that's for BOTH consoles, and that's why I'm so excited.

Like, consoles not only getting SSDs, but top of the line? One console so top of the line it's no on shelves right now? VIRTUAL RAM ON A CONSOLE??

I genuinely didn't expect such a huge jump, especially with how expensive some things are right now. It's seriously exciting right now.

PC gamers have been preaching that an SSD is the single biggest isolated upgrade you can do to your PC, for years now. Having it now be the baseline is so great, and it's great to see neither Sony or Microsoft are doing the bare minimum.

Seriously Cerny was smart to mention Jak 2 to demonstrate how long we've been dealing with slow hard drives and how it affects the literal design of games we play. Maybe now with things like virtual ram other memory problems will be way more opened up too. I cannot wait for a couple years from now when we start seeing some stuff we didn't realize we took for granted just didn't work.

Microsoft is supposed to talk about their storage solution more in the future. Not sure if it will be viewable by public or just devs. Richard Geldreich, who specializes in this stuff says Microsoft is using something that can best kraken and they're doing it in a smarter more efficient way where you only take the parts of the Buffalo needed, instead of moving the whole thing. We know Microsoft's speeds are sustained, we only know where Sony's tops out. In the end real world results probably won't be too different between them as both companies are trying to do the same thing but with a different approach.

Oh for real? Duuuuuude. I love the sound of that. I wish Microsoft would literally do a GDC type presentation too. I freaking loved the Cerny talk. It was like, really accessible if you were a laymen but interested party like myself, and covered everything I wanted to know about the tech in the PS5 pretty much. So I'd love something similar for xbox too.

Is xbox also doing similar to PS5 in terms of like having an extra chip on the die specifically for facilitating data transfer at those high speeds so it can take some load off the CPU?

Fuck, if you're into the tech side of this, this is so exciting. I remember last gen being like, excited because man consoles were truly showing how much they were struggling to keep up and just, really wanted to do certain things they couldn't, or could but performance would tank so bad and textures couldn't catch up and yada yada. With XB1/PS4 it felt like they finally could do those things, in proper HD, with proper performance for the most part (I think we forget how many last gen games had truly, truly abysmal performance. It's a lot). But that's kinda it ya know? Like there's a limit to the excitement there because it was more, hey we're already doing most of this stuff but now we can like, actually do it and that's it. It's not compromised and we have breathing room. But I don't feel like gameplay or other things like, fundamentally changed in terms of possibilities. This generation it feels like there's a bigger leap, and fundamentally we've removed the storage bottleneck, we've added raytracing, which is completely new, and we're getting sound hardware that isn't just half a CPU core anymore. Stuff that might seriously have us rethink how a lot of games are designed, and just thinking about things I know of off hand that are due to technical limitations about to be relieved, man it's exciting.

Basically I just kinda felt like this generation was a continuation or part 2 of the last generation. And this generation feels like a serious leap, although it may be a few years before we really get that separation.
 
Last edited:
Jan 4, 2018
119
Horizon does this type of thing, imagine being able to achieve that level of fidelity across more titles.
Fairly certain that's wrong. Horizon does frustum culling, which pretty much every game does (but there's a famous gif showing it for Horizon in particular). This means the game only renders what the player is looking at, but that doesn't mean it only stores in memory what the player is looking at. The stuff behind the player is still in memory even though it's not being rendered because the HDD would take too long to stream it when the player quickly turns around. On PS5 and Series X with their super fast SSDs, games like Horizon would not only be able to not render all that stuff, but also not store it in memory.

TL;DR: Not rendering stuff behind you doesn't mean the stuff behind you isn't in memory. It's still in memory for games designed for slow HDDs like Horizon, but it won't be on games designed for fast SSDs.
 
Last edited:

Chettlar

Member
Oct 25, 2017
13,604
Fairly certain that's wrong. Horizon does frustum culling, which pretty much every game does (but there's a famous gif showing it for Horizon in particular). This means the game only renders what the player is looking it, but that doesn't mean it only stores in memory what the player is looking at. The stuff behind the player is still in memory even though it's not being rendered because the HDD would take too long to stream it when the player quickly turns around. On PS5 and Series X with their super fast SSDs, games like Horizon would not only be able to not render all that stuff, but also not store it in memory.

TL;DR: Not rendering stuff behind you doesn't mean the stuff behind you isn't in memory. It's still in memory for games designed for slow HDDs like Horizon, but it won't be on games designed for fast SSDs.

Right that's what I was trying to explain, and I was wondering like, if you bridge those concepts. Idk I might be off base. I hope I'm not misleading anyone. I'm just speculating like, well what if, you could have certain data streamed into the RAM really quickly so you can offload some work off the CPU or whatever. Idk. So like, maybe having such a faster SSD could help the PS5 in this regard, to where there is some work it doesn't have to do as much because it's not dealing with as much data at a given time, or something. I dunno man. lol.
 

EvilBoris

Prophet of Truth - HDTVtest
Verified
Oct 29, 2017
16,686
Fairly certain that's wrong. Horizon does frustum culling, which pretty much every game does (but there's a famous gif showing it for Horizon in particular). This means the game only renders what the player is looking it, but that doesn't mean it only stores in memory what the player is looking at. The stuff behind the player is still in memory even though it's not being rendered because the HDD would take too long to stream it when the player quickly turns around. On PS5 and Series X with their super fast SSDs, games like Horizon would not only be able to not render all that stuff, but also not store it in memory.

TL;DR: Not rendering stuff behind you doesn't mean the stuff behind you isn't in memory. It's still in memory for games designed for slow HDDs like Horizon, but it won't be on games designed for fast SSDs.

Gotcha, I assumed that famous gif was it in action.
 

Prine

Attempted to circumvent ban with alt account
Banned
Oct 25, 2017
15,724
I think the reason is because there are a few things that Sony seems to be more specialized in. So for example the whole kraken compression thing, and the custom controller, and all that, a lot of effort being made into ensuring a minimum 5.5 raw gbps transfer speed that can be processed as fast. Microsoft doesn't seem to be doing the same, although of course there's the matter of them just, not going into as much detail.

Still, I think it's a fair assessment to note (and yes the more tech head people around here have mentioned this) that the PS5 is a very nimble, specialized machine, whereas the xbox will just have that raw power advantage.

But from that you start getting people separating the differences to their extremes. People are eager for a differentiating factor, so anything that exists will get isolated and made extreme. Both consoles are doing most things the other is doing. Still, they have pretty different approaches to some things. Like power vs frequency. They are taking the exact polar opposite approach of each other.

Anyway could be wrong but I was pondering an example where like, so they made a big deal about how the PS5 streams info so fast you could stream it while a character was literally looking around. That's pretty crazy. The xbox will also clearly do this but it won't be as fast (2-4[?] vs 5.5-8). It could be that streaming info means that in a given scene the PS5 wouldn't have to hold as much data to be calculating the position and condition of at one time and be more specifically only doing what it needs right then, whereas the Xbox would hold onto more at one time, but would be able to do it just fine and wouldn't need to be dumping it out as fast. I don't know. I'm probably off base here, but I was trying to think of an example. Hopefully someone more knowledgeable could explain an instance where the PS5's design might make up for some of the loss in power. Cuz idk that's the best I have right now.

Just want to highlight this part, and I guess you touched upon it, with MS being cautiously quiet, perhaps it's due to not giving away their competitive edge here, but MS have thier custom implementation of BCPack which is supposed to be even more impressive then Kraken, this seems to be lost with the current narrative amongst other things. But I agree MS are being coy with thier implementation refinements and Velocity Architecture at the moment.
 

Lukas Taves

Banned
Oct 28, 2017
5,713
Brazil
I don't think it's necessarily the audio chip stuff, I think it's because the PS5 has many innovative and/or much bigger unknowns, that show they're thinking quite far outside the box (whether that's good or bad remains to be seen).

For example, the fact that they're running such high clocks (unseen frequencies really). Or that their boosted speeds aren't really actually boosted per se, nor tied to thermal dynamics (eg throttled based on temperature like other boost modes in other products) but instead to a set power load (something that afaik hasn't been done before).

Then there's the custom SSD, which has technology/performance akin to Saturate PCIe 4.0 SSD tech that isn't even available yet, and might not even be able till just after the PS5 launches.

Add to that we don't really even know why they've gone so hard on this SSD tech, nor exactly what sort of dividends it will or won't have, or how it might shape game design, notably with their exclusives. But already some are theorising about potential benefits it could have, even as far as things like saving system images or offloading OS functions from memory (caching it aside) to provide access to far more of the GDDR6 ram for actual game use etc.

Right now there are quite a lot of unknowns, and lots of details and information yet to be shared from both Microsoft and Sony, so I guess we'll just have to wait and see. Personally I think both systems sound really competent, and I'm genuinely excited for both.
What you are saying seems like Ps5 is the one brute forcing.

MS went with a fast ssd, but designed a solution to improve texture compression and reduce the amount of data that is being loaded. Sony brute force by adding a super fast ssd.

Ms changed the design of the console in order to make room for the thermal envolope of running a fairly large chip at high frequencies. Sony brute force the clocks on a small chip to the point throttling is needed because if they were if a fixed (and lower) clock solution the size of their chip would be a big performance delta. Let's not forget that even so there will be a performance delta as SX has a gpu, cpu and bandwidth advantage.

Then there's the audio chip that is getting completely unreasonable praise when it's they catching up. And not even at launch because at launch it will only support headphones when Windows Sonic already works with home theater systems too. The innovation that they are doing with audio which is creating a service to map you head and ears, is unrelated to ps5 hardware.

Oh and there's the BC. Nothing elegant about that either. It's also definitely brute forcing to have the cpu and gpu to cripple themselves to previous consoles. And that is the culprit of them going with a small gpu which them made them brute force the clocks to get more performance.

Cerny said they literally can't go any higher than 2.23 GHz because the chip literally stops working. How is that elegant in any meaning of the word?
 
Jan 4, 2018
119
i am pretty sure rdna at 4 tf is greater then gcn at 6. Wasn't it around 30 percent gains in efficiency, or am I not remembering it right?

also Lockhart will be rdna 2 which is most likely an improvement over rdna 1.


bottom line is Lockhart will be more powerful then xbox one x in gpu department. I wouldn't worry yourself over that
According to Digital Foundry's testing, the gain from Polaris to Navi is about 27%. So Series S's 4 TF would be equivalent to 5.08 TF, which is pretty far off from One X's 6 TF. RDNA 2 could get it closer, but I still think it'll probably fall short. Could be wrong though.

I'm certainly not worried about it. The only reason you'd want it to be equal or greater to Xbox One X's GPU is to play Xbox One games at 4K. Otherwise, that comparison doesn't matter. Its job is to play next gen Xbox games at around 1080p or 1440p, and 4 TF is enough to do that considering Series X is at 12 TF and targeting 4K.
 
Last edited:

nib95

Contains No Misinformation on Philly Cheesesteaks
Banned
Oct 28, 2017
18,498
What you are saying seems like Ps5 is the one brute forcing.

MS went with a fast ssd, but designed a solution to improve texture compression and reduce the amount of data that is being loaded. Sony brute force by adding a super fast ssd.

Ms changed the design of the console in order to make room for the thermal envolope of running a fairly large chip at high frequencies. Sony brute force the clocks on a small chip to the point throttling is needed because if they were if a fixed (and lower) clock solution the size of their chip would be a big performance delta. Let's not forget that even so there will be a performance delta as SX has a gpu, cpu and bandwidth advantage.

Then there's the audio chip that is getting completely unreasonable praise when it's they catching up. And not even at launch because at launch it will only support headphones when Windows Sonic already works with home theater systems too. The innovation that they are doing with audio which is creating a service to map you head and ears, is unrelated to ps5 hardware.

Bit of a silly post given the PS5 also has custom IO and compression stuff to help out on that front (hence it isn't just brute force), and that we haven't actually even seen the design of the PS5 nor its cooling solution to even be able to criticise it comparative to the Xbox Series X's.

On top of that, I think it's far too early to say the Series X has the bandwidth advantage (given the 10GB/6GB difference comparative to the PS5). It really depends on how or how much of the PS5's ram is used (and whether the SSD speed can be utilised to free up more usable ram), and also whether devs end up wanting to use more than 10GB on the XSX for GPU or graphics, especially as the OS footprint comes down (which is usually the case as a generation progresses).

Cerny said they literally can't go any higher than 2.23 GHz because the chip literally stops working. How is that elegant in any meaning of the word?

Except that's not at all what Cerny said at all. Infact what he implied was that they could have actually pushed the clockspeed even higher but capped it in order to keep the on chip logic running efficiently. To quote....

"Running a GPU at 2Ghz was looking like an unreachable target with the old fixed frequency strategy. With this new paradigm we're able to run way over that, infact we have to cap the GPU frequency at 2.23Ghz so that we can guarantee the on chip logic operates properly."
 
Last edited:

big_z

Member
Nov 2, 2017
7,797
Is xbox also doing similar to PS5 in terms of like having an extra chip on the die specifically for facilitating data transfer at those high speeds so it can take some load off the CPU?

I believe there is no cpu load, might have even been mentioned in the digital foundry video. Microsoft's solution removes bottlenecks that Sony is designing around. Should be interesting to see in more detail how their setup works.
 

Chettlar

Member
Oct 25, 2017
13,604
Gotcha, I assumed that famous gif was it in action.

Well it couldn't be because the PS4 doesn't have an SSD, and even if it did, it's bottlenecked by (oh man this is going to really annoy me again now that I'm thinking about it) SATA freaking 2. Because I guess SATA 3 costs too much. I dunno lol.

But yeah that gif is what the GPU is rendering at a given time. The CPU is keeping track of where everything is still and the data is still there to access. I mean, that's what RAM is for. Random Access Memory. So the CPU can just pull that info from the RAM and throw it to the GPU to render in real time.

With SSDs, like the other poster said, you're no longer limited by what you can hold in your 8, well now 16GB. Or well, like 6 and now 13. So now not only will that info be rendered as needed, but it won't even all have to be in the RAM all at one time, meaning scenes can be wayyyyyyyyy more complex because now I mean like, basically the whole SSD is your console's oyster, lol.

Just want to highlight this part, and I guess you touched upon it, with MS being cautiously quiet, perhaps it's due to not giving away their competitive edge here, but MS have thier custom implementation of BCPack which is supposed to be even more impressive then Kraken, this seems to be lost with the current narrative amongst a lot of other things.

Yeah see I just completely have not heard of that.

I think both companies should have come out with a nice little blog posts with a link to a video, instead of a livestream. So both cover what they can do, but also hype up their fanbase or whatever. I dunno. I liked Cerny's talk so I just want more of those.

Ms changed the design of the console in order to make room for the thermal envolope of running a fairly large chip at high frequencies. Sony brute force the clocks on a small chip to the point throttling is needed because if they were if a fixed (and lower) clock solution the size of their chip would be a big performance delta. Let's not forget that even so there will be a performance delta as SX has a gpu, cpu and bandwidth advantage.

ehhhhh I feel like calling lowering clock speeds "brute forcing" a bit of a stretch.

Brute forcing means doing the simple, powerful solution to complete a task, when another solution could do the same job with less power or energy. So in other words, more efficiently. I guess in a certain perspective you could say Sony is brute forcing Sata speeds, but I mean, idk I feel like it's more appropriate to say Microsoft is trying to find ways to shortcut.

I feel like typically people use the phrase "brute force" to imply doing something inefficiently with lots of power. Having a fast drive with a few bottlenecks as possible...isn't really being inefficient. It's just, the normal way of doing it. By contrast look at the raytracing hardware in xbox. Microsoft said it's basically the equivalent of 13 teraflops of power, just as one chip, to just handle the raytracing. That's if you bruteforced it. Key point here, is you would be bruteforcing it because it's hardware not built for raytracing, so trying to do what something else built for it can do, but less efficiently, would be brute forcing. Note that the raytracing hardware can do exactly what the 13tflops could do. Exactly. The 13tflops isn't more powerful. It's the same, but less efficient. That's the key point here.

Like, what you're saying is kinda like saying Sony is brute forcing their 16GB of ram, where microsoft isn't. But that just isn't true. Sony's single 16GB pool of fast ram is faster than Microsoft's. Microsoft's split pool is not necessarily more efficient. It is physically just less powerful. Being more powerful isn't brute forcing.

Likewise Sony's drive is just more powerful. It is not brute forcing anything. It physically is more capable than Microsoft's drive. If Microsoft's drive did everything Sony's did, but Sony's used more power and was supposedly faster despite similar work, then that would be brute forcing.

Brute forcing is about inefficiently doing the same work as something more efficient with less cost. Something being more powerful to make up for something else being less powerful is not brute forcing. In fact that's...literally what efficiency is.

Imagine a racetrack down a winding road like Fujimi Kaido. A fast car goes down the road quickly. A faster car goes down it more quickly. You might go faster than the faster car if you clip a few corners, (or with electronics, cut out some work you can get away with). You're more limited and might still lose the longer stretches when they happen, but that might be okay for your purposes and your balancing act. A brute force approach just barrels down the hill, ignoring the roads, and crashing through trees. Like a tank. Really powerful, not very efficient, but it gets there around the same time as the fast car going down the road normally.

So I feel like right now you're calling the car that is faster than a slower car that clips some corners, "brute forcing." And that's just not accurate I feel. Brute forcing means expending something inefficiently when you could do something smarter. Just having faster hardware do something to help something else is the literal opposite of that.

Cerny said they literally can't go any higher than 2.23 GHz because the chip literally stops working. How is that elegant in any meaning of the word?

A crescent wrench can only go so big, but it's more elegant than a hammer. Designing exactly what you need to within the costs you have is like, the definition of elegant. You have to look at the big picture. Sony's design probably could work for a bigger machine. But they designed it how a machine that costs as much as it does would work if you took the approach they chose.

Also, where are you getting the "literally stops working." Just curious.
 
Last edited:

AzorAhai

Member
Oct 29, 2017
6,617
I don't know much about power but it seems like a beast. I'm used to getting my 3rd parties on PlayStation but could these specs really make a difference for the games ? Or are we getting to a point where graphics are so advanced that such a gap will not be obvious ?
 

Chettlar

Member
Oct 25, 2017
13,604
I don't know much about power but it seems like a beast. I'm used to getting my 3rd parties on PlayStation but could these specs really make a difference for the games ? Or are we getting to a point where graphics are so advanced that such a gap will not be obvious ?

Xbox games will look better. How much better really depends. I don't think anyone exactly knows just yet.
 

ThatNerdGUI

Prophet of Truth
Member
Mar 19, 2020
4,551
On top of that, I think it's far too early to say the Series X has the bandwidth advantage (given the 10GB/6GB difference comparative to the PS5). It really depends on how or how much of the PS5's ram is used (and whether the SSD speed can be utilised to free up more usable ram), and also whether devs end up wanting to use more than 10GB on the XSX for GPU or graphics, especially as the OS footprint comes down (which is usually the case as a generation progresses).

The amount of RAM available has nothing to do with bandwidth. Any bandwidth restrictions would be due to the bus width. And no, the SSD won't change that either.
 

gremlinz1982

Member
Aug 11, 2018
5,331
Bit of a silly post given the PS5 also has custom IO and compression stuff to help out on that front (hence it isn't just brute force), and that we haven't actually even seen the design of the PS5 nor its cooling solution to even be able to criticise it comparative to the Xbox Series X's.

On top of that, I think it's far too early to say the Series X has the bandwidth advantage (given the 10GB/6GB difference comparative to the PS5). It really depends on how or how much of the PS5's ram is used (and whether the SSD speed can be utilised to free up more usable ram), and also whether devs end up wanting to use more than 10GB on the XSX for GPU or graphics, especially as the OS footprint comes down (which is usually the case as a generation progresses).
You have a CPU and OS to run. What Microsoft have done is have OS and CPU related tasks running on the slower pool or RAM and the GPU working off faster RAM.
You are going to allocate some memory to have those tasks met, and by extension, bandwidth. I also do not know how many tasks one is going to hive off to SSD when it is much slower, the same SSD loads stuff onto RAM, and you do are running stripped down OS mainly related to gaming tasks.
 

Alexandros

Member
Oct 26, 2017
17,815
I don't know much about power but it seems like a beast. I'm used to getting my 3rd parties on PlayStation but could these specs really make a difference for the games ? Or are we getting to a point where graphics are so advanced that such a gap will not be obvious ?

They will make a difference and I think it will be obvious. That doesn't mean that the difference will be huge but I believe it's a safe bet that multiplatform games will look and run better on Microsoft's console.
 

big_z

Member
Nov 2, 2017
7,797
I don't know much about power but it seems like a beast. I'm used to getting my 3rd parties on PlayStation but could these specs really make a difference for the games ? Or are we getting to a point where graphics are so advanced that such a gap will not be obvious ?

We're still a long way off before power becomes pointless. It will be noticeable when compared side by side, possibly more so in certain cases but ps5 isn't going to feel like a generation behind by any means.
 

nib95

Contains No Misinformation on Philly Cheesesteaks
Banned
Oct 28, 2017
18,498
The amount of RAM available has nothing to do with bandwidth. Any bandwidth restrictions would be due to the bus width. And no, the SSD won't change that either.

The amount of ram does have to do with bandwidth in this case, because only 10GB's of the XSX's ram is actually quicker than the PS5's.

And if as NXGamer and a few others theorised as a possibility, the OS on the PS5 can be cached aside facilitated by the ultra fast SSD (a second or two delay on the OS stuff isn't the end of the world), then that frees up an additional 2.5GB or whatever PS5's OS footprint will be, for graphics use. And lest we forget the entirety of the PS5's ram has the same bandwidth.

Ordinarily I wouldn't give as much credence to such theories, but then Sony's principle software engineer for the PS5 had this to say about NXGamers video, and I can't think of many other details he could be referring to that DF missed.




Even then though, as much as Microsoft mentioned itse "typically easy" for devs to fill up 3.5GB of the slightly slower memory with audio, CPU data etc, there's still no guarantee devs won't want to use more than 10GB ram for GPU/graphics related stuff, nor that the OS footprint over time won't get smaller. I guess we'll just have to wait and see how it pans out.
 
Last edited:
Feb 1, 2018
5,242
Europe
On top of that, I think it's far too early to say the Series X has the bandwidth advantage (given the 10GB/6GB difference comparative to the PS5). It really depends on how or how much of the PS5's ram is used (and whether the SSD speed can be utilised to free up more usable ram), and also whether devs end up wanting to use more than 10GB on the XSX for GPU or graphics, especially as the OS footprint comes down (which is usually the case as a generation progresses).

TBH almost all of those "theories" are pipe dreams. Nice to speculate about them, but fact is that XSX has a clear bandwidth advantage.
 

Lukas Taves

Banned
Oct 28, 2017
5,713
Brazil
Bit of a silly post given the PS5 also has custom IO and compression stuff to help out on that front (hence it isn't just brute force), and that we haven't actually even seen the design of the PS5 nor its cooling solution to even be able to criticise it comparative to the Xbox Series X's.
Compression block is a given I was talking about having support for a compression format aimed at textures, and the sampler feedback feature that extends tiled resources all the way to the ssd loading only what is needed.

Not making any comparisons of what is going to be better in practice but it's clear which one is brute forcing the problem and most likely at a price.

On top of that, I think it's far too early to say the Series X has the bandwidth advantage (given the 10GB/6GB difference comparative to the PS5). It really depends on how or how much of the PS5's ram is used (and whether the SSD speed can be utilised to free up more usable ram), and also whether devs end up wanting to use more than 10GB on the XSX for GPU or graphics, especially as the OS footprint comes down (which is usually the case as a generation progresses).
It's not. SX have a bandwidth advantage and a sizeable one at that. And unlike esram it has a very large pool of that faster ram. Also as they explained the gpu has full access to that bandwidth without contention from the cpu, the contention happens only at the slower pool. So in practice it will actually be an even bigger delta because when the ps5 processor uses the ram it reduces the bandwidth the gpu can use disproportionately (the same thing that happens with ps4) only that it happens in the entire pool.



Except that's not at all what Cerny said at all. Infact what he implied was that they could have actually pushed the clockspeed even higher but capped it in order to keep the on chip logic running efficiently. To quote....

"Running a GPU at 2Ghz was looking like an unreachable target with the old fixed frequency strategy. With this new paradigm we're able to run way over that, infact we have to cap the GPU frequency at 2.23Ghz so that we can guarantee the on chip logic operates properly."
He did. He said that with that pulsating pattern they could go even higher but had to cap at 2.23 because past that point the chip literally stops working.
 

solis74

Member
Jun 11, 2018
43,011
Compression block is a given I was talking about having support for a compression format aimed at textures, and the sampler feedback feature that extends tiled resources all the way to the ssd loading only what is needed.

Not making any comparisons of what is going to be better in practice but it's clear which one is brute forcing the problem and most likely at a price.


It's not. SX have a bandwidth advantage and a sizeable one at that. And unlike esram it has a very large pool of that faster ram. Also as they explained the gpu has full access to that bandwidth without contention from the cpu, the contention happens only at the slower pool. So in practice it will actually be an even bigger delta because when the ps5 processor uses the ram it reduces the bandwidth the gpu can use disproportionately (the same thing that happens with ps4) only that it happens in the entire pool.




He did. He said that with that pulsating pattern they could go even higher but had to cap at 2.23 because past that point the chip literally stops working.

cool info.
 

nib95

Contains No Misinformation on Philly Cheesesteaks
Banned
Oct 28, 2017
18,498
Compression block is a given I was talking about having support for a compression format aimed at textures, and the sampler feedback feature that extends tiled resources all the way to the ssd loading only what is needed.

Not making any comparisons of what is going to be better in practice but it's clear which one is brute forcing the problem and most likely at a price.


It's not. SX have a bandwidth advantage and a sizeable one at that. And unlike esram it has a very large pool of that faster ram. Also as they explained the gpu has full access to that bandwidth without contention from the cpu, the contention happens only at the slower pool. So in practice it will actually be an even bigger delta because when the ps5 processor uses the ram it reduces the bandwidth the gpu can use disproportionately (the same thing that happens with ps4) only that it happens in the entire pool.


He did. He said that with that pulsating pattern they could go even higher but had to cap at 2.23 because past that point the chip literally stops working.

But we don't know the compression details of the PS5 besides that it uses Kraken. The assumption is RDO BCx+Kraken, but we don't know if Sony has something custom or not. Either way, based on the figures, it does seem Microsoft has a slight compression advantage.

Regarding the ram bandwidth, it was the slower portion that I was referring to about potentially balancing things out depending on how much ram is available to use on the PS5 vs the XSX, notably for graphics/GPU (eg if devs wanted to use more than 10GB).

I also wouldn't consider it a particularly sizeable bandwidth advantage either. For context, the PS4's 8GB's of ram was 158% faster than the XO's, whilst the PS5's ram is 34% faster than 6GB's of the Series X's, and 10GB of the Series X's ram is 25% faster than the PS5's. The bandwidth delta is no where near what it was this gen .

Lastly, do you have a link to Cerny saying the chip literally stops working after 2.23 GHz? Because that isn't what was said in the quote I posted directly from the presentation. Though I suppose it's semantics on whether not guaranteeing on chip logic operates 100% properly constitutes as not working, so maybe that's where our consideration of his quote differs.
 
Last edited:

Lukas Taves

Banned
Oct 28, 2017
5,713
Brazil
ehhhhh I feel like calling lowering clock speeds "brute forcing" a bit of a stretch.
I agree, I'm calling pushing the clocks so high to the point they can't sustain it brute forcing, not the throttling itself.

Brute forcing means doing the simple, powerful solution to complete a task, when another solution could do the same job with less power or energy. So in other words, more efficiently. I guess in a certain perspective you could say Sony is brute forcing Sata speeds, but I mean, idk I feel like it's more appropriate to say Microsoft is trying to find ways to shortcut.
Exactly why I'm calling the overclock as brute forcing. We know from GH that they started pushing for higher clocks only relatively recent as of last year, likely reacting to SX specs. So we know that initially they haven't designed it like this, it was brute forced to try and close the gap.

I feel like typically people use the phrase "brute force" to imply doing something inefficiently with lots of power. Having a fast drive with a few bottlenecks as possible...isn't really being inefficient. It's just, the normal way of doing it. By contrast look at the raytracing hardware in xbox. Microsoft said it's basically the equivalent of 13 teraflops of power, just as one chip, to just handle the raytracing. That's if you bruteforced it. Key point here, is you would be bruteforcing it because it's hardware not built for raytracing, so trying to do what something else built for it can do, but less efficiently, would be brute forcing. Note that the raytracing hardware can do exactly what the 13tflops could do. Exactly. The 13tflops isn't more powerful. It's the same, but less efficient. That's the key point here.
Exactly, I'm just pointing out how ridiculous it is to say SX design is brute force or non elegant, or that somehow Ps5's design will allow for better utilization of resources. None of this are true and if anything Sony had to bruteforce some aspects of their design to match what Ms was doing.

Like, what you're saying is kinda like saying Sony is brute forcing their 16GB of ram, where microsoft isn't. But that just isn't true. Sony's single 16GB pool of fast ram is faster than Microsoft's. Microsoft's split pool is not necessarily more efficient. It is physically just less powerful. Being more powerful isn't brute forcing.
I don't think I said sony brute forced the memory. And no that is not an advantage for PS5. Ms have a clear bandwidth advantage over PS5, and they also solved the memory contention problem with their design without adding costs by having a separate memory pool, it was quite clever as well.

Likewise Sony's drive is just more powerful. It is not brute forcing anything. It physically is more capable than Microsoft's drive. If Microsoft's drive did everything Sony's did, but Sony's used more power and was supposedly faster despite similar work, then that would be brute forcing.

Brute forcing is about inefficiently doing the same work as something more efficient with less cost. Something being more powerful to make up for something else being less powerful is not brute forcing. In fact that's...literally what efficiency is.
And wouldn't say that potentially achieving the same practical results with a much cheaper SSD would be efficient? Or that losing the cost advantage of a smaller chip by increasing the thermals is efficiency?

We had reports that sony was designing a box for $400, with both the overclocking and the costly SSD do you still think that they are on target to hit that without any subsiding? We know now that Ms' and Sony's bom are actually a lot closer than we were expecting before, and with Ms having a significantly bigger processor and significantly faster memory. Does it sound efficient that ultimately Sony's decisions may have costed them the pricing advantage?


Also, where are you getting the "literally stops working." Just curious.
From Cerny himself. He literally said that during the presentation.
 

Bunzy

Banned
Nov 1, 2018
2,205
According to Digital Foundry's testing, the gain from Polaris to Navi is about 27%. So Series S's 4 TF would be equivalent to 5.08 TF, which is pretty far off from One X's 6 TF. RDNA 2 could get it closer, but I still think it'll probably fall short. Could be wrong though.

I'm certainly not worried about it. The only reason you'd want it to be equal or greater to Xbox One X's GPU is to play Xbox One games at 4K. Otherwise, that comparison doesn't matter. Its job is to play next gen Xbox games at around 1080p or 1440p, and 4 TF is enough to do that considering Series X is at 12 TF and targeting 4K.

That's not really how it's works, doing math like that doesn't work because it's not apples to apples
 

solis74

Member
Jun 11, 2018
43,011
sounds awesome!

DirectStorage
DirectStorage is an all new I/O system designed specifically for gaming to unleash the full performance of the SSD and hardware decompression. It is one of the components that comprise the Xbox Velocity Architecture. Modern games perform asset streaming in the background to continuously load the next parts of the world while you play, and DirectStorage can reduce the CPU overhead for these I/O operations from multiple cores to taking just a small fraction of a single core; thereby freeing considerable CPU power for the game to spend on areas like better physics or more NPCs in a scene. This newest member of the DirectX family is being introduced with Xbox Series X and we plan to bring it to Windows as well.


GPU Work Creation
Xbox Series X adds hardware, firmware and shader compiler support for GPU work creation that provides powerful capabilities for the GPU to efficiently handle new workloads without any CPU assistance. This provides more flexibility and performance for developers to deliver their graphics visions.


Hardware Decompression
Hardware decompression is a dedicated hardware component introduced with Xbox Series X to allow games to consume as little space as possible on the SSD while eliminating all CPU overhead typically associated with run-time decompression. It reduces the software overhead of decompression when operating at full SSD performance from more than three CPU cores to zero – thereby freeing considerable CPU power for the game to spend on areas like better gameplay and improved framerates. Hardware decompression is one of the components of the Xbox Velocity Architecture.
 
Jan 4, 2018
119
That's not really how it's works, doing math like that doesn't work because it's not apples to apples
How so? That 27% just came from testing games on two different GPUs with the same amount of TF (product of shader cores and frequency) but different architectures, and then determining how much faster the new architecture is by itself. The end result is a multiplier you can apply to TF to see how fast the GPU will be even across architectures.
 

Lukas Taves

Banned
Oct 28, 2017
5,713
Brazil
But we don't know the compression details of the PS5 besides that it uses Kraken. The assumption is RDO BCx+Kraken, but we don't know if Sony has something custom or not. Either way, based on the figures, it does seem Microsoft has a slight compression advantage.

Regarding the ram bandwidth, it was the slower portion that I was referring to about potentially balancing things out depending on how much ram is available to use on the PS5 vs the XSX, notably for graphics/GPU (eg if devs wanted to use more than 10GB).

I also wouldn't consider it a particularly sizeable bandwidth advantage either. For context, the PS4's 8GB's of ram was 158% faster than the XO's, whilst the PS5's ram is 34% faster than 6GB's of the Series X's, and 10GB of the Series X's ram is 25% faster than the PS5's. The bandwidth delta is no where near what it was this gen .

Lastly, do you have a link to Cerny saying the chip literally stops working after 2.23 GHz? Because that isn't what was said in the quote I posted. Though I suppose it's semantics on whether not guaranteeing on chip logic operates 100% properly constitutes as not working, so maybe that's where our consideration of his quote differs.
Ps4 ram wasn't 158% faster than Xbone because there was esram. That actual problem with the console wasn't raw speed but rather that it was so small 32mb.

For SX you could say the problem is similar in theory, only now you have 10GB of ram instead of just 32MB.

You are also not considering the system reservation on PS5 so instead of 6 it actually just has about 3 (3.5GB) of memory that is faster than the 3.5 pool on SX. However, as I said, there is memory contention on that pool.

As sony shows on PS4, that is an issue:

jpg


As you can see the problem is that as cpu uses bandwidth the total CPU+GPU bandwidth reduces. This still happens on Ps5, and likely to an even higher extent now that the cpu is significantly more powerful, and not on SX thanks to that elegant and efficient memory design Ms came up with.

I don't have a link other than the presentation. I watched it fully at once so I can't tell you exactly when he tells that as well, but he certainly does. He says something that going higher than that would make the inner logic of the chip to stop working.
 
Nov 2, 2017
2,275
For example, the fact that they're running such high clocks (unseen frequencies really). Or that their boosted speeds aren't really actually boosted per se, nor tied to thermal dynamics (eg throttled based on temperature like other boost modes in other products) but instead to a set power load (something that afaik hasn't been done before).
Nvidia cards have multiple throttles including a power throttle. You can choose at what limit you want to throttle it: for example 80% of TDP or even 110%. That was the way to mine cryptocurrency in the day: you'd power limit the card hard and while the clocks & performance dropped the drop in power was much larger so it was way more efficient.

Power and heat are correlated though and usually you hit temperature limits earlier than power limits, atleast on PC where it doesn't matter how much power a GPU draws.
 

Deleted member 20297

User requested account closure
Banned
Oct 28, 2017
6,943
Ps4 ram wasn't 158% faster than Xbone because there was esram. That actual problem with the console wasn't raw speed but rather that it was so small 32mb.

For SX you could say the problem is similar in theory, only now you have 10GB of ram instead of just 32MB.

You are also not considering the system reservation on PS5 so instead of 6 it actually just has about 3 (3.5GB) of memory that is faster than the 3.5 pool on SX. However, as I said, there is memory contention on that pool.
As sony shows on PS4, that is an issue:

jpg


As you can see the problem is that as cpu uses bandwidth the total CPU+GPU bandwidth reduces. This still happens on Ps5, and likely to an even higher extent now that the cpu is significantly more powerful, and not on SX thanks to that elegant and efficient memory design Ms came up with.

I don't have a link other than the presentation. I watched it fully at once so I can't tell you exactly when he tells that as well, but he certainly does. He says something that going higher than that would make the inner logic of the chip to stop working.

I think that is a perfect example for what I discussed back then in 2013 - theoretical max is just that, theoretical. Especially on PS4 this was misleading as the design with the Onion and Garlic busses led to a worse than linear bandwidth distribution when CPU demanded bandwidth and the graph also shows that even in the best case the theoretical max not achievable.
 

etta

Banned
Oct 27, 2017
3,512
I just noticed there's no optical out port on the back of the Xbox!
That's a huge disappointment, and I'm surprised no one seems to mention it.
Sad face

Horizon does this type of thing, imagine being able to achieve that level of fidelity across more titles.
And in much, MUCH, grander fashion! If Horizon did that with 100 MBps read speed, imagine how much more is possible with 2.4-5 GBps!
That's the one aspect I'm super excited for the PS5, what they can do with all that bandwidth in open world games.
 
Last edited: