Xbox Series X [XBSX] [Release November 10 2020]


Ok this is interesting. Not sure if there are other examples in existence that we dont know of because they werent revealed.
But this is the first example I am hearing where the Series S is becoming a problem for Series X.
 

Ok this is interesting. Not sure if there are other examples in existence that we dont know of because they werent revealed.
But this is the first example I am hearing where the Series S is becoming a problem for Series X.
I'm sure this story came up about a month ago and the speculation was that it was likely Series S's reduced memory footprint (8Gb high-speed RAM, 2gb slower RAM) that was probably causing the issue. Other games have simply paired back native rendering resolution, asset quality and frame rates to make run on the smaller console.

Halo Infinite was supposed to supports 4 player split screen but that was yanked fairly late in development, although I'm sure there was some 'trick' to getting it work.
 
Halo Infinite was supposed to supports 4 player split screen but that was yanked fairly late in development, although I'm sure there was some 'trick' to getting it work.
Considering halo:i was cross gen don't think XSS would be the cause of anything as ran on XO. From what I remember seemed to play ok on XSS when you could hack it on.

Need to read article but your right the most obvious issues was always speculated to be memory. Need to send ATG to help them out, get SFS into a game for once 😉
 
Considering halo:i was cross gen don't think XSS would be the cause of anything as ran on XO. From what I remember seemed to play ok on XSS when you could hack it on.
I'm sure the 'trick' to get split-screen working in Infinite worked on both current and previous generation consoles.
 
LOL. A lot of these S sucks stories are just sour grapes from Devs. How come PC devs can support 1033 different configs, but Xbox devs can't support 2? Give me a break.
 
Last edited:
LOL. A lot of these S sucks stories are just sour grapes from Devs. How come PC devs can support 1033 different configs, but Xbox devs can't support 2? Give me a break.
Nobody is saying " S sucks" apart from you. So.. uh... well done? :runaway:
 
Development methodology needs to change.
You don't try to optimise in the last 2 months to get to 60fps. You need to keep it in mind and develop towards it during the whole development.
It's why we see so many badly performing or buggy games. It's not the hardware.

Regardless of what people think of XSS it's a similar thing. You don't just try to get it working well at the end.
Vision on the XSX & PS5 optimise as you go along on the XSS. Most of the time those optimizations will carry over to the other consoles and make for a better game on there also.
In most cases it probably wouldn't be much of an issue.
 
Development methodology needs to change.
You don't try to optimise in the last 2 months to get to 60fps. You need to keep it in mind and develop towards it during the whole development.
It's why we see so many badly performing or buggy games. It's not the hardware.

Regardless of what people think of XSS it's a similar thing. You don't just try to get it working well at the end.
Vision on the XSX & PS5 optimise as you go along on the XSS. Most of the time those optimizations will carry over to the other consoles and make for a better game on there also.
In most cases it probably wouldn't be much of an issue.
Yeah, it's kinda ridiculous this is happening. The Series S is really solid hardware and it's been around nearing 3 years, so it should've been factored in.

That being said, if splitscreen's a consistent issue for developers, just remove splitscreen from the feature parity checklist. I think that's a fair differentiating feature between high and low end models of a console.
 
Development methodology needs to change.
You don't try to optimise in the last 2 months to get to 60fps. You need to keep it in mind and develop towards it during the whole development.
It's why we see so many badly performing or buggy games. It's not the hardware.

Regardless of what people think of XSS it's a similar thing. You don't just try to get it working well at the end.
Vision on the XSX & PS5 optimise as you go along on the XSS. Most of the time those optimizations will carry over to the other consoles and make for a better game on there also.
In most cases it probably wouldn't be much of an issue.

I'd likely say its smarter to just develop a game around the series s and then optimize around series x/ps5 with higher resolution / textures and more effects. The series s is capable of all the same hardware features and effects of the series x.

I actually think the series x and ps5 releases would benefit a lot more from that.
 
I'd likely say its smarter to just develop a game around the series s and then optimize around series x/ps5 with higher resolution / textures and more effects. The series s is capable of all the same hardware features and effects of the series x.

I actually think the series x and ps5 releases would benefit a lot more from that.
There's a lot more to graphics and the makeup of a game than just 'slider options' like texture resolutions and whatnot. You build ambitious next-gen games by pushing the core foundations of the visuals, not just cranking settings. This requires targeting PS5/XSX as a baseline. You then 'make it work' for Series S by whatever cuts are needed. It makes more sense to leave Series S as the afterthought given it is literally the console built for people who dont care as much about the technical side of things and just want to be able to play new games as cheaply as possible.

And the Series S is not actually 'really solid hardware'. The cuts made to the memory setup in particular are damn near crippling. Building games around 10GB of RAM(just two more than last gen) would be catastrophic for the notion of actually getting a true generational leap.
 
There's a lot more to graphics and the makeup of a game than just 'slider options' like texture resolutions and whatnot. You build ambitious next-gen games by pushing the core foundations of the visuals, not just cranking settings. This requires targeting PS5/XSX as a baseline. You then 'make it work' for Series S by whatever cuts are needed. It makes more sense to leave Series S as the afterthought given it is literally the console built for people who dont care as much about the technical side of things and just want to be able to play new games as cheaply as possible.

And the Series S is not actually 'really solid hardware'. The cuts made to the memory setup in particular are damn near crippling. Building games around 10GB of RAM(just two more than last gen) would be catastrophic for the notion of actually getting a true generational leap.

Like I said , The series S is capable of everything that the series x and ps5 are capable of. Design a game around that taking advantage of all the hardware features for the system and then move on from there.

Sony made spiderman based on the ps4 and moved up from there to offer ray tracing and other features the system offered. The leap between the ps4 and ps5 is a huge gulf vs the jump from the s to the x.

All optimizations done for the series s will benefit the x and ps5. So working on the series s version first would be the smarter way.

As for ram amounts even the 16 gigs of the series x and ps5 are crippling.

Ps1 had 3 MB of ram
Ps2 had 32MB of ram
Ps3 had 512 gigs
Ps4 had 8 gigs
PS5 has 16 gigs

Xbox 64mb
Xbox 360 512
Xbox one 8gigs
Xbox sereis s 10 gigs
Xbox series x 16 gigs


Regardless of the series s or x/ps5 this is by far the smallest leap in ram amounts while last generation was the biggest
 
And the Series S is not actually 'really solid hardware'. The cuts made to the memory setup in particular are damn near crippling. Building games around 10GB of RAM(just two more than last gen) would be catastrophic for the notion of actually getting a true generational leap.

I like the Series S - but I do think MS should have put in 2 more GB of ram and pushed the clocks on the GPU a little higher. The bandwidth is reasonably high relative to the power of the GPU, and even having double the memory area across two of the eight 16-bit channels (which includes the OS area) isn't likely to be big deal IMHO.

Blimey, in terms of BW the Series S has even the very newest and highest end APU configured with it's highest spec ram beat by a mile. The very top end APU I can see on AMD's site atm is the Ryzen 9 7940HS, with a 12 CU Zen 3 GPU running at up to 2.8 ghz (other places say 2.9), and the fastest listed memory is LPDDR5x 7500. The CPU is crazy fast too. So you're looking at an APU there with something very similar to a ~4.1TF RDNA 2 part (at peak speed) with a total system BW of about half of the Series S. There will be other factors of course, but overall Series S BW seems to be decent for what it is. Even PS5 only has double the BW despite its 10.3 TF GPU and sick depth fillrate.

As for RAM quantity, it's certainly not ideal but there's a bit more to it than just comparing total system memory with last gen. PS4 and X1 had 5 GB available for games. Series S apparently started with 8 GB for games and that's now up to '8.x' with "several hundreds of additional MB" available for games if devs switch certain OS features off (iirc). That was first mentioned in a dev kit update last year. So while Series S only has 2GB more ram than X1, it actually has 3GB + potentially more that last gen available for games. It does seem MS are at least listening.

I suspect that MS might store some kind of translation table for a part of the SSD in reserved ram, so that may limit the amount that they can free up for games. I hope they give the option to devs to free up everything they can though, and just keep the bare minimum like Xbox Live and screenshot capture running in the background. They could always release an refreshed S down the line with another two GB to add everything back in again. Key thing is probably to raise the baseline as much as possible I reckon.
 
Last edited:
Like I said , The series S is capable of everything that the series x and ps5 are capable of. Design a game around that taking advantage of all the hardware features for the system and then move on from there.

Sony made spiderman based on the ps4 and moved up from there to offer ray tracing and other features the system offered. The leap between the ps4 and ps5 is a huge gulf vs the jump from the s to the x.

All optimizations done for the series s will benefit the x and ps5. So working on the series s version first would be the smarter way.

As for ram amounts even the 16 gigs of the series x and ps5 are crippling.

Ps1 had 3 MB of ram
Ps2 had 32MB of ram
Ps3 had 512 gigs
Ps4 had 8 gigs
PS5 has 16 gigs

Xbox 64mb
Xbox 360 512
Xbox one 8gigs
Xbox sereis s 10 gigs
Xbox series x 16 gigs


Regardless of the series s or x/ps5 this is by far the smallest leap in ram amounts while last generation was the biggest
Using a clearly cross-gen game like Spiderman is a very good example of why this isn't a good approach.
It just seems completely ridiculous to base your entire development foundation around a weak machine that's only catered to people who dont care much about technical aspects, and then just cranking 'slider settings' up a bit for the PS5/XSX versions. That's such a terrible direction to go to and would result in a super underwhelming generation. Yes, games would all perform very well, but that's not why people buy a new generation console. People want games that are a very clear generational leap.

And no, the 16GB isn't crippling because the whole point of the move to fast SSD's is to ensure much greater efficiency of RAM usage. But even with this setup, 10GB is still pitiful. That's still a very hard cap in terms of what you can do within any given frame.
 
Just to respond to these particular points.
It makes more sense to leave Series S as the afterthought
I disagree, it shouldn't be treated as an afterthought. Doing that will lead to broken games and a quality of product that didn't need to be that bad on it. By keeping it in mind during development process will likely lead to games that will also perform and look better on lower end pc's.
Building games around 10GB of RAM(just two more than last gen)
They forecasted the use of virtual texturing in the engine and with the use of ssd to get away with less memory bump(for the target resolution)
We've not really seen games needing to make heavy use of it yet due to cross gen.
I will say that I'm on the record believing more memory and a bump in gpu performance would've really made a nice difference though.
 
I will say that I'm on the record believing more memory and a bump in gpu performance would've really made a nice difference though.

That would have been great and I don't disagree, unfortunately economics certainly disagree.

If you have more memory you need a way to fill that memory. HDDs were already struggling to fill 8 GB of memory in less time than an ice age (slight exaggeration :p).

That means you need to move to an SSD at the very least. But an SSD is going to be more costly than an HDD.

Complicating matters is that GDDR memory isn't getting cheaper all that quickly. Hell, GDDR memory may not even be declining in price for the forseeable future.

So pick one, SSD or more memory. I think the compromise was relative good, a small bump in memory (8 GB to 16 GB) and the addition of a much faster storage solution.

However, to get the most out of it means to rearchitect your engine to make better use of the new tech that is meant to assist with streaming (the custom hardware in each machine). Unfortunately, as we've seen from the vast majority of developers who are faced with limited budgets and an escalating cost of development, it's cheaper to continue to architect your games the same way they were architected in past generations. Because it's cheaper than rearchitecting your entire rendering engine and storage interface.

So, we end up with a very small (in terms of console "generation" increase) bump in memory amount with a large bump in storage speed that's mostly not being used for anything other than faster traditional asset loading for most developers.

And even if the I/O was fully exploited, it's unlikely we'd get what some expect from a "generational" increase in graphics IQ.

That said, I'd argue that for this generation of consoles, Nanite (high geometric density) and Lumen (higher quality lighting) have the potential to deliver a generational leap in graphics when used together. Nanite without Lumen or Lumen without Nanite would still be impressive but not nearly as much as both combined. It's why RT in the vast majority of games is so disappointing, it just highlights how low the geometric density and quality of current games is. OTOH if the geometric density and quality were higher but without some form of ray casted light, it'd just accentuate how bad the lighting is.

RT needs highly detailed geometry in the same way that highly detailed geometry needs RT.

Regards,
SB
 
Last edited:
Back
Top