Why is Microsoft so far ahead of Sony with Back Compat?

[invictis]

I'm still trying to work out just how MS has blown past Sony with Back Compat.
First up we have MS bringing out the Xbox One X with a completely different RAM set up, from size, to speed, from DDR3 to GDDR5, from esram to no esram, and they got every single Xbox One game running no problem.
Then they were able to do the back compat enhancement for the X, giving full 4k output for games that were far less resolution, increased frame rates than prior. Some of the results were just stunning.
Then with this next gen they are giving BC for the full Xbox One library, as well as ton of 360 and OG Xbox games as well. Not only that but they are going to enhance OG Xbox games by adding HDR to them. Say what?
On the other side with the PS4 Pro Sony had to use a GPU that was double the size CU wise to make sure that old PS4 games would run properly (using 18 Cu's only like the OG PS4), and other than a tiny bit of frame rate increase via increased CPU clocks on the Pro, there was no enhancements.
With the PS5, it again appears that Sony had to limit themselves to 36CUs on the GPU to make BC with PS4 games viable, and still it appears that not all games will be able to be played on the PS5. And with that Sony isn't giving BC for PS3, PS2 or PS1 games.

So how are MS able to crush BC, and even add enhancements like resolution and HDR, while Sony struggles? What approach have they taken compared to Sony? Why couldn't Sony just do it the same way?

In previous gens, Sony has tried to keep BC. PS1 games played on PS2. They includes the EE of the PS2 in the PS3 initially and had BC for the PS2. But then to cut costs they ditched the BC in newer PS3 models, and since then they have fallen away.

So yeah, whats MS doing that Sony isn't, and could Sony adopt it?

 

[iroboto]

They had planned BC hardware in during the development of XBO. So it was just a matter of progression towards using that hardware to enable what they needed and just iterating forward. Sony had no way to do that with Cell and this generation. So they only started BC now. Whether they are working on classic PS1 or PS2 titles is unknown.

 

[Rootax]

And MS have a ton of software expérience of course...

 

[invictis]

And MS have a ton of software expérience of course...

So is it a case of Sony relying on hardware to get BC, while MS is doing it through software?

 

[Rootax]

So is it a case of Sony relying on hardware to get BC, while MS is doing it through software?

It can't be only one or the other, but I believe MS use some kinds of VMs (virtual machine) "layer" to run BC, and wrappers.

 

[invictis]

It can't be only one or the other, but I believe MS use some kinds of VMs (virtual machine) "layer" to run BC, and wrappers.

Is that something Sony could change over to, or it is a matter of Sony not having the knowedge and people to do it like MS do?
I realize MS are software masters, but Sony isn't a slouch in that area, so what preventing them from doing the same? Too far gone down the hardware road to come back?

 

[chris1515]

This is because Sony comes from the console side and they give lower access to hardware to devs. They need to abstract more the hardware and let a bit of performance if they want to use software for backward compatibility.

 

[invictis]

This is because Sony comes from the console side and they give lower access to hardware to devs. They need to abstract more the hardware and let a bit of performance if they want to use software for backward compatibility.

From what I understand Sony has a high level API and a low level API. Direct X also allows for high level and low level access.

 

[Arwin]

It is a combination of using weird hardware and allowing developers to use it to the fullest by exposing all of it without abstractions. Especially the Cell processor was incredibly hard for Sony.

Microsoft has used mostly PC hardware and almost always exposed everything through DirectX abstractions alone. This combination has helped them immensely, but they have also put considerable effort into making sure titles were backward compatible when the hardware differences did make this harder (most importantly with the transition from the 360’s special VRAM pool, which probably was their most exotic hardware). So they have long made it a priority as well. They also released the S model with slightly different specs between the one and the X so already have more experience with different hardware configurations to some extent there. Finally, Microsoft has started making it a priority to release their games on PC as well even for their exclusives, so that helps too.

That said, Sony has worked hard on getting their bigger titles to be rewritten to run on PC so that they could finally run on PC hardware for their steaming service PS Now (which initially atoll had to use PS3 hardware - I am not sure if some still do now). Eventually though they have rereleased some of those more important titles as basically rewritten PS4 remasters as well, and from the PS4 onwards they shifted their focus on providing an easier target for multiplatform development, making sure the hardware would be as backward compatible as possible and being able to provide more incremental steps in their hardware transitions (eg PS4 -> PS4 Pro).

They still want their developers to be able to make the most of the hardware though and so they make their hardware as backward compatible as possible. This has very likely also been an important factor for Sony choosing to use 36CUs for the PS5.

As for the rest, we don’t know the BC features of the PS5 in terms of additional features supported so we’ll have to wait and see what kind of improvements they will allow. But it is clear that Microsoft still has the advantage coming into this generation.

 

[invictis]

It is a combination of using weird hardware and allowing developers to use it to the fullest by exposing all of it without abstractions. Especially the Cell processor was incredibly hard for Sony.

Microsoft has used mostly PC hardware and almost always exposed everything through DirectX abstractions alone. This combination has helped them immensely, but they have also put considerable effort into making sure titles were backward compatible when the hardware differences did make this harder (most importantly with the transition from the 360’s special VRAM pool, which probably was their most exotic hardware). So they have long made it a priority as well. They also released the S model with slightly different specs between the one and the X so already have more experience with different hardware configurations to some extent there. Finally, Microsoft has started making it a priority to release their games on PC as well even for their exclusives, so that helps too.

That said, Sony has worked hard on getting their bigger titles to be rewritten to run on PC so that they could finally run on PC hardware for their steaming service PS Now (which initially atoll had to use PS3 hardware - I am not sure if some still do now). Eventually though they have rereleased some of those more important titles as basically rewritten PS4 remasters as well, and from the PS4 onwards they shifted their focus on providing an easier target for multiplatform development, making sure the hardware would be as backward compatible as possible and being able to provide more incremental steps in their hardware transitions (eg PS4 -> PS4 Pro).

They still want their developers to be able to make the most of the hardware though and so they make their hardware as backward compatible as possible. This has very likely also been an important factor for Sony choosing to use 36CUs for the PS5.

As for the rest, we don’t know the BC features of the PS5 in terms of additional features supported so we’ll have to wait and see what kind of improvements they will allow. But it is clear that Microsoft still has the advantage coming into this generation.

So I would assume that the PS5 would have the power to run PS1 and 2 games through an emulator? Why wouldn't they just do that? Seems an easy solution.

 

[chris1515]

It is a combination of using weird hardware and allowing developers to use it to the fullest by exposing all of it without abstractions. Especially the Cell processor was incredibly hard for Sony.

Microsoft has used mostly PC hardware and almost always exposed everything through DirectX abstractions alone. This combination has helped them immensely, but they have also put considerable effort into making sure titles were backward compatible when the hardware differences did make this harder (most importantly with the transition from the 360’s special VRAM pool, which probably was their most exotic hardware). So they have long made it a priority as well. They also released the S model with slightly different specs between the one and the X so already have more experience with different hardware configurations to some extent there. Finally, Microsoft has started making it a priority to release their games on PC as well even for their exclusives, so that helps too.

That said, Sony has worked hard on getting their bigger titles to be rewritten to run on PC so that they could finally run on PC hardware for their steaming service PS Now (which initially atoll had to use PS3 hardware - I am not sure if some still do now). Eventually though they have rereleased some of those more important titles as basically rewritten PS4 remasters as well, and from the PS4 onwards they shifted their focus on providing an easier target for multiplatform development, making sure the hardware would be as backward compatible as possible and being able to provide more incremental steps in their hardware transitions (eg PS4 -> PS4 Pro).

They still want their developers to be able to make the most of the hardware though and so they make their hardware as backward compatible as possible. This has very likely also been an important factor for Sony choosing to use 36CUs for the PS5.

As for the rest, we don’t know the BC features of the PS5 in terms of additional features supported so we’ll have to wait and see what kind of improvements they will allow. But it is clear that Microsoft still has the advantage coming into this generation.

PSnow still use motherboard with CELL CPU.

 

[TS_JBG]

So I would assume that the PS5 would have the power to run PS1 and 2 games through an emulator? Why wouldn't they just do that? Seems an easy solution.

Because that's a commercial decision more than a technical one.
The question isn't "can we run ps1/2 games on our new hardware" but more "How much will we ear if we invest into an emualtor, and is it worth the cost of relicensing ?"

 

[Arwin]

So I would assume that the PS5 would have the power to run PS1 and 2 games through an emulator? Why wouldn't they just do that? Seems an easy solution.

PS1 games only came on CDROM that the PS4 drive physically cannot read. I assume the same for PS5. So they would have to be rereleased on PSN or you play them through PSNow.

These games still work on any PS3 of course.

I don’t know what the situation will be with PS5 there but I’m not sure how many PS1-2 games people would still want to play on that. I would perhaps play a few PS2 games like Buzz if that were possible.

 

[Proelite]

MS had the time, money, culture and willpower.

They also had Dave Cutler.

 

[Shifty Geezer]

So is it a case of Sony relying on hardware to get BC, while MS is doing it through software?

If you watched "The Road to PS5" presentation, you'll see Cerny mention that in the past, they included the old console hardware. That's too costly, so going forwards you either design your new hardware to be compatible with the old (PS5 designed to run PS4 code) or you use emulation (PS1 and PS2 emulation on PS4), or various shades in-between with some hardware features to support efficient software emulation.

Alternatively, your software is abstracted away from the hardware so the same executable runs on different hardware. You see this on PC and mobile devices, and more so on something like HTML5 code. The same iOS app will run on old and new iPads because the hardware interprets the software and maps it onto the hardware.

One advantage with software abstraction is scalability. Put a PS1 chip in a PS5 and it'll run PS1 games in their original manner, but run an emulator and you can get high-def upscaling and the like. But if the software isn't expecting that, you are forcing it and may break something.

Abstraction gives you a degree of future-proofing, allowing software to move hardware, at the cost of performance overhead. HTML 5 will run on every device - Android phone, iPad, Windows, console (if it has a good enough browser) - but it won't be efficient enough to allow for a RDR AAA title. And a platform exclusive exploiting the system architecture, poking values into registers at specific clock cycles to eke out a bit more performance, will wow people on what the game is able to achieve but be locked to that hardware and then very hard to carry across.

It seems as time goes on, the value of hitting the hardware decreases and the value in software portability increases. There's a long discussion here from 2011 on the value of BC and a poll on how much lost power through abstraction people would want and a poll on use. I think a lot of opinions have shifted, including my own.

 

[goonergaz]

My views on BC changed with Sony. Initially I thought it was really important, especially day one to help bridge the gap in games available. However, like Sony suggest from its data, BC was rarely used - it ends up being a ‘nice to have’ feature that minimal people really ‘need’.

Then we come to next gen, suddenly BC is really important because of the ecosystem and many games now are much longer and even infinite. People will want to take their games with them.

MS have had more reasons to ply the BC functionality. They have needed to regain trust after the ‘always online’ etc fiasco and they also wanted to give more VFM as the console lacked in power and exclusives. (IMHO)

So that’s probably why MS are so far ahead and Sony behind, as well as the technical reasons given.

 

[SumoSaki]

Because Sony is complacent.

 

[invictis]

PS1 games only came on CDROM that the PS4 drive physically cannot read. I assume the same for PS5. So they would have to be rereleased on PSN or you play them through PSNow.

These games still work on any PS3 of course.

I don’t know what the situation will be with PS5 there but I’m not sure how many PS1-2 games people would still want to play on that. I would perhaps play a few PS2 games like Buzz if that were possible.

Yeah I remember now that Sony took out the ability to read CDRom. Ok, so that answers PS1.

 

[invictis]

My views on BC changed with Sony. Initially I thought it was really important, especially day one to help bridge the gap in games available. However, like Sony suggest from its data, BC was rarely used - it ends up being a ‘nice to have’ feature that minimal people really ‘need’.

Then we come to next gen, suddenly BC is really important because of the ecosystem and many games now are much longer and even infinite. People will want to take their games with them.

MS have had more reasons to ply the BC functionality. They have needed to regain trust after the ‘always online’ etc fiasco and they also wanted to give more VFM as the console lacked in power and exclusives. (IMHO)

So that’s probably why MS are so far ahead and Sony behind, as well as the technical reasons given.

Personally I cant go back and play PS1 games, they just look so bad to what we are used to playing now. But some 360.games still look ok, and deff the PS4 and Xbox One games still look really good, that I can still see myself playing them next gen as well, so I now see BC as important as well.
But seeing MS enhancing alot of games with resolution and HDR is crazy good, and kinda gives the game a new lease of life.
I agree that MS needed to go over and above to regain their reputation with gamers, and since Phil came on board they have done just that.
And the Xbox One X is an amazing console.

 

[invictis]

It is a combination of using weird hardware and allowing developers to use it to the fullest by exposing all of it without abstractions. Especially the Cell processor was incredibly hard for Sony.

Microsoft has used mostly PC hardware and almost always exposed everything through DirectX abstractions alone. This combination has helped them immensely, but they have also put considerable effort into making sure titles were backward compatible when the hardware differences did make this harder (most importantly with the transition from the 360’s special VRAM pool, which probably was their most exotic hardware). So they have long made it a priority as well. They also released the S model with slightly different specs between the one and the X so already have more experience with different hardware configurations to some extent there. Finally, Microsoft has started making it a priority to release their games on PC as well even for their exclusives, so that helps too.

That said, Sony has worked hard on getting their bigger titles to be rewritten to run on PC so that they could finally run on PC hardware for their steaming service PS Now (which initially atoll had to use PS3 hardware - I am not sure if some still do now). Eventually though they have rereleased some of those more important titles as basically rewritten PS4 remasters as well, and from the PS4 onwards they shifted their focus on providing an easier target for multiplatform development, making sure the hardware would be as backward compatible as possible and being able to provide more incremental steps in their hardware transitions (eg PS4 -> PS4 Pro).

They still want their developers to be able to make the most of the hardware though and so they make their hardware as backward compatible as possible. This has very likely also been an important factor for Sony choosing to use 36CUs for the PS5.

As for the rest, we don’t know the BC features of the PS5 in terms of additional features supported so we’ll have to wait and see what kind of improvements they will allow. But it is clear that Microsoft still has the advantage coming into this generation.

If we just take the Xbox One X and PS4 Pro this generation. To get PS4 games to run on the Pro, Sony had to use a GPU set up that allowed them to turn off half the 36CU GPU so that it mirrored the PS4s 18CU GPU. MS on the other hand didn't need to do that, they went with a 40CU GPU that didnt in anyway double the OG Ones CU count, and adding to that they went with a totally different RAM set up that is nothing like the OG One. DDR3 to GDDR 5, 8GB to 12GB, esram to no esram. And every game ran spot on.
At the time I didnt really think about how much of an achievement that was, but watching how Sony has struggled and it has effected the CU count of the PS5 to some degree, its shown just how big it was. It also makes me wonder why Sony cant replicate it.