[Mefisutoferesu]

OK, in a most likely vain attempt to save this thread from Vysez's big locking stick... how do you suppose PS3 will provide BC?

I'm thinking PS1 will be entirely software emulation. As some have said, considering how compatible 3rd party PS1 emus are, I imagine Sony could do a rather excellent job. Especially if they've been working on it since the PS2 launch in preparation of PS3.

As to PS2... I'm thinking the EE should be relatively easy to emulate on CELL, PPE is the main core and VU1 and VU0 are just pushed onto SPE 0 and 1 (or does the number start at 1??). Yes, it's complicated, I'm not saying it's easy, but it seems like a natural progression, no? That said I think the GS will probably be full blown added to the RSX, it's only some 35 million, transistors and what with the phantom transistors on the G70 and the possible close relation, it seems probable.

That said, I'm actually kinda hoping they just put the eDram on the RSX and emulate the rest. That way it should be relatively easy to add AA and AF to PS2 games, or so I hope, and I'm hoping a LOT.

Also, on an interesting note if the RSx turns out to be very generic as many would assume, do you think part of the reason is future compatability?? Make it easy for PS4 to support PS3 and PS2 and PS1... oh goodness!!

[pcostabel]

Originally Posted by Mefisutoferesu

OK, in a most likely vain attempt to save this thread from Vysez's big locking stick... how do you suppose PS3 will provide BC?

I'm thinking PS1 will be entirely software emulation. As some have said, considering how compatible 3rd party PS1 emus are, I imagine Sony could do a rather excellent job. Especially if they've been working on it since the PS2 launch in preparation of PS3.

As to PS2... I'm thinking the EE should be relatively easy to emulate on CELL, PPE is the main core and VU1 and VU0 are just pushed onto SPE 0 and 1 (or does the number start at 1??). Yes, it's complicated, I'm not saying it's easy, but it seems like a natural progression, no? That said I think the GS will probably be full blown added to the RSX, it's only some 35 million, transistors and what with the phantom transistors on the G70 and the possible close relation, it seems probable.

Hmm, I rather think they will take an approach similar to PS2 emulation. Emulating a MIPS on the PPE could be feasible, but quite slow, especially considering all the timing issues with the VIF/GIF etc. OTOH, the gs is a very simple rasterizer: the RSX will have no problem emulating it and adding enhancements like AA and higher res, just like PS1 games are enhanched when emulated on PS2. It is easier for Sony to add a stripped down EE in hardware than deal with the nightmare of a software emulation.

[Mefisutoferesu]

That makes a good bit of sense, but is it viable to do CPUs in Hardware and GPUs in software? I imagine that would add to the pricing no? More over if Sony want's to maintain BC on down through the PS generations it's gonna be a problem no? Or will it just happen in PS4 when emulation will be a snap even if it's dog slow? Anyway, how much would having the additional CPUs add to the PS3s cost? I can't imagine it's prohibitive, but I doubt its free.

[Almasy]

I´m only talking out of my ass here, but I´d say we probably will only see hardware from two generations in a closed box. Say, PS5 will only have PS4 hardware to help emulation, the rest of the systems will be emulated through software (Afterall, it should be powerfull enough). I´d say it´s much more likely to see GS emulated through the RSX and just toss the MIPS core in there.

I don´t know how it´d run, and I think the vector units could be emulated quite well through the SPU´s, but I don´t know how much work it´d take to emulate the MIPS core on the PPC core.

[Farid]

Originally Posted by Mefisutoferesu

OK, in a most likely vain attempt to save this thread from Vysez's big locking stick...

Nice try, but the thread is going to be hit no matter what.

But since I don't want to stop this new discussion, I just split the thread into a new, and I hope meaningful, discussion.

So, here we go, this thread is about speculating about what methods the PS3 will use to provide BC.

[Arty]

Originally Posted by Vysez

So, here we go, this thread is about speculating about what methods the PS3 will use to provide BC.

Could you please modify the title of this thread to reflect that.

[Reznor007]

Originally Posted by Almasy

I´m only talking out of my ass here, but I´d say we probably will only see hardware from two generations in a closed box. Say, PS5 will only have PS4 hardware to help emulation, the rest of the systems will be emulated through software (Afterall, it should be powerfull enough). I´d say it´s much more likely to see GS emulated through the RSX and just toss the MIPS core in there.

I don´t know how it´d run, and I think the vector units could be emulated quite well through the SPU´s, but I don´t know how much work it´d take to emulate the MIPS core on the PPC core.

Based on current PC PS2 emulators using dynamic recompilation CPU cores, I'd say it would be tight, considering that splitting the emulation of the MIPS and VU's over the PPE and SPE's wouldn't really be practical. Ask any of the MAME developers about using dual core Athlon's/P4's and they will say any gains you get would be taken back from time to sync everything. Also, it isn't really possible to split the emulation of the single MIPS core over multiple processors, and that's where a good portion of the time is spent. I also don't know if an individual SPE has enough power to emulate a VU though...it would certainly outperform one, but emulating is completely different from just running its software.

Here's a post about someone running Kingdom Hearts on the most compatible PS2 emulator available, getting 25 minutes of gameplay in 5 hours of time. http://www.ngemu.com/index.php?action=post&id=1909

Semi-related, MAME emulating San Francisco Rush is extremely slow. The system is only a 200MHz R5000 and a 33MHz TMS32031 for sound, but even without emulating the Voodoo graphics hardware the game only runs about 10-15% speed on the fastest CPU's around. This is using dynarec CPU cores for the MIPS also.

[Bohdy]

The MIPS core could be feasibly emulated on the PPU I think, as it is not terribly fast with a fairly low instructional thoroughput (afaik). The VU code could be dynamically recompiled and easily handled by the SPE's as they are SIMD units anyway, and thereby lend themselves to paralleling.

What is more salient to me (at least) is how MS is managing to emulate the Xbox in real time. Sure they have licensed the NV2A hardware for the GPU side of things, but emulating the 700mhz P3 must be a nightmare with multiple cores providing seemingly little benefit. My best guess as to how they are doing it given the info availible is through some form of static recompilation. I have done some research on the topic in the past, but it is a rather infrequently used method of emulation despite the huge speed gains. Often each program's execution path needs to be individually traced to make sure that conditional jump targets are all recompiled, which would explain the "emulation profiles".

[Farid]

Originally Posted by serenity

Could you please modify the title of this thread to reflect that.

Done!

I changed it the first time, but I forgot to move pcostabel's post from the other, ill-fated, thread. So, I when I move it, the mege function changed the thread title again.

[lip2lip]

I am hoping that 512 kB cache on main core will be enough to hold instruction set conversion tables, this is probably only thing that could stop software emulation, but may not be an issue because of high dedicated ram bandwidth. because of high frequency of spes, they should be able to emulate timing of vu0, vu1. I am surprised no one mentioned, usually biggest burner in chip emulation is branch prediction logic; but as I understand it, branch prediction in new chips is long enough to not notice any differences.

[jvd]

I believe its all going to be emulation . I don't see them sticking in any bit of the ps2 in there . Its 6or 7 years old and I'm sure for whatever they could think of using it for (the ee for sound) could better be used by a dedicated newly designed chip .

[one]

(Since I was late to the original thread...) This is from PS Meeting in the last month. It's from day 1

[BlueTsunami]

Originally Posted by one

(Since I was late to the original thread...) This is from PS Meeting in the last month. It's from day 1

Well...theres the picture. From day 1.

Also about the emulation. I believe Sony will probably put some type of lecacy PS2 hardware in the PS3 (what exactly I don't know) to help with the emulation of PS2 games. As far as PSX..I believe the PS3 is more than capable to have the emulation software run purely on software (without the help of specific hardware for the PSX).

[Arty]

Originally Posted by one

(Since I was late to the original thread...) This is from PS Meeting in the last month. It's from day 1

How does that contribute to this thread

How do you think BC is possible on PS3, that is the question.

[MasaC]

<a href="http://ps3media.ign.com/ps3/image/article/636/636028/ps-meeting-2005-fun-with-slides-part-iii-20050722101609165.jpg" target="_blank">Backward compatibility from day one</a> will be achieved through a combination of hardware and software <a href="http://hardware.gamespot.com/Story-ST-15015-2097-4-4-x" target="_blank">according to Ken Kutaragi</a>.

The details of how this will work have yet to be announced but if PS2 b/c is anything to go by, there will be some PS2 hardware inside the PS3. On the PS2 there is a PS1-cpu that doubles as an i/o-chip that handles the USB- and firewire-ports.

It's possible that there will be the EE+GS-chip that's used in the PSTwo to double as an i/o-chip on the PS3, perhaps shrunk to a smaller process. If this is the case I'd expect PS1-emulation will be done entirely in software on the cell.

Maybe the PC-leagcy logic on PS3-variant of the G70 will be replaced with PS2-legacy logic. I don't know if this is possible at all, though.

[Phil]

Originally Posted by pcostabel

Hmm, I rather think they will take an approach similar to PS2 emulation. Emulating a MIPS on the PPE could be feasible, but quite slow, especially considering all the timing issues with the VIF/GIF etc. OTOH, the gs is a very simple rasterizer: the RSX will have no problem emulating it and adding enhancements like AA and higher res, just like PS1 games are enhanched when emulated on PS2. It is easier for Sony to add a stripped down EE in hardware than deal with the nightmare of a software emulation.

That sounds very feasable IMO. The only thing I'm wondering is how RSX is supposed to be emulating the eDRAM and its high bandwidth... wouldn't CELL technically have more (internal) bandwidth available that would help with the emulation of the eDRAM memory?


BTW; One, best of thanks for providing that picture. That should certainly put the other discussion to rest for good.

[Shifty Geezer]

The combined EE+GS definitely won't appear on a smaller process as it's already at 90nm and if Sony could use 65nm, they'd be using that for Cell! I doubt EE+GS will feature at all. The hardware aspect will be, I guess, some hardware emulation that maps some PS2 functions onto the PS3 hardware. The EE+VUs will probably be emulated on Cell, and GS emulated on PSX. Though as the Cell can seemingly handle in software graphics beyond PS2's capabilities maybe the bulk of emulation is in the Cell? The biggest concern has to be the 48 GB/s GS eDRAM and how they could tackle the fillrate, but as has been mentioned by others smarter than me, use of compression could alleviate brute-force BW requirements.

[london-boy]

Originally Posted by Shifty Geezer

The combined EE+GS definitely won't appear on a smaller process as it's already at 90nm and if Sony could use 65nm, they'd be using that for Cell!

Now, i'm only saying this, not expecting it to happen, but just for the sake of argument, don't you think a tiny little chip with about 40M trannies would be MUCH easier to fab on a new process like 65nm than Cell?
If anything, they might want to try their new 65nm process on something "easy" like the EE+GS, not on a big and complicated chip like Cell. So it would make sense if EE+GS were on 65nm and Cell stayed at 90nm for the time being.
Having said that, i find it very unlikely that they'll put EE+GS in there.

[Shifty Geezer]

Perhaps you are right, and they'll move PS2 production over to 65nm as well to make that more profitable, while keeping Cell at 90nm. Would it not be considerably more expensive though (several dollars per unit) to include PS2 in PS3 than emulate with some RSX extras for aid? Even with a process shrink. How much does the existing EE+GS chip cost, and how much would a process shrink save on that?

[Squeak]

Originally Posted by Phil

That sounds very feasible IMO. The only thing I'm wondering is how RSX is supposed to be emulating the eDRAM and its high bandwidth...

If RSXs framebuffer compression is good enough to give the equivalent of 48Gb/s bandwidth, what does that tell us about the 32Gb/s EDRAM in Xenos? Is it silicon well spend, or would it have been better to but some (less area consuming) bandwidth saving logic on the main die?

[Phil]

^^ Yeah, I did think about compression, but what about latency

[Shifty Geezer]

Originally Posted by Squeak

what does that tell us about the 32Gb/s EDRAM in Xenos?

1.) That's BW TO the eDRAM. Internally, for the work it does, it's got 256 GB/s. PS2's 48 GB/s BW isn't enough for what Xenos has to do next-gen. 2.) That's so totally off topic . I'll pretend I didn't reply and I'm really talking about PS3's BC solution, which I believe uses magic, perhaps small gremlins and fairies.

[Squeak]

Originally Posted by Shifty Geezer

1.) That's BW TO the eDRAM. Internally, for the work it does, it's got 256 GB/s. PS2's 48 GB/s BW isn't enough for what Xenos has to do next-gen. 2.)

But that's "only" for AA and stencil. AFAICS 10Mb is just enough for Z and backbuffer, so If you want to do alpha blending or render to texture, you have to go through system memory or bus to VPU.

[Shifty Geezer]

If you want to continue this discussion you'd best take it to another thread.

[Fafalada]

Originally Posted by Pcostabel

OTOH, the gs is a very simple rasterizer: the RSX will have no problem emulating it and adding enhancements like AA and higher res

I disagree, simple though GS may be, the way it gets used in games is not simple to emulate at all. Unless RSX comes with certain GS extensions (in regards to address modes), this is very much in question whether it's possible to do fast enough.
As for AA/resolution enhancements - unless Sony follows up on MS idea of game profiles, forget about it, it's just not going to happen.

Originally Posted by Squeak

If RSXs framebuffer compression is good enough to give the equivalent of 48Gb/s bandwidth, what does that tell us about the 32Gb/s EDRAM in Xenos? Is it silicon well spend, or would it have been better to but some (less area consuming) bandwidth saving logic on the main die?

You don't need to think of it in such brute force manner - SDTV framebuffers are relatively small and the better optimized the game, the more closely it follows page buffer coherency.

IMO something around 128KB renderbuffer cache would trivially serve bandwith requirements of typical GS rendering patterns 90% of the time, even with much slower external memory then what RSX has available. And regular texture bandwith is a non-issue alltogether.

However, IMO there's two real problems with GS emulation.
One is that you need to emulate its addressing schemes in detail - majority of effects in PS2 games are coded around specific behaviour of memory accesses and addressing, you'll not get away with any kind of HLE for that.
The other problem is the few cases of games that ping-pong between texture&render buffer a lot (sometimes on every rendered primitive, using same memory addresses for texture and render buffer) - which will kill bandwith on any GPU without unified eDram (meaning Xenos like setup would be useless to assist this also). The only way I see around this issue is hoping that the emulation of the rest of rendering would be faster then real GS by a large enough amount to compensate for the slower rendering whenever this kind of rendering packets are encountered over course of a frame.

Anyway, one thing I'm certain about is that if they ended up sticking a full GS chip in there, it will NOT be part of RSX die.

Originally Posted by Bohdy

What is more salient to me (at least) is how MS is managing to emulate the Xbox in real time. Sure they have licensed the NV2A hardware for the GPU side of things, but emulating the 700mhz P3 must be a nightmare with multiple cores providing seemingly little benefit.

Are they actually going to use NV2A hardware in there? I was under impression that it'll be some sort of highlevel emulation, so basically whenever a game does lowlevel stuff that isn't quite within DX spec, they'll code it into one of those game profiles (hence the initial low compatibility rate).

The P3 emulation is an interesting question though, if they were really sticking with static recompiles then you could well be looking at separate "emulation profile" for each game - which at least to me would sound like there'll never be anything close to full library compatibility.