Open Lara engine on the ATARI Jaguar

[JagChris]

It's absolutely impressive how much overhead is involved with the OpenLara browser version.

[Atlantis]

Open Lara, on Jaguar? Hope so; we Will have nother jag game to include in the ”Beat off to ” thread wink wink 

[Cyprian]

11 hours ago, leech said:

Funny idea of the day; PiStorm for the Jaguar.  Just throwing that out there. 

+1

[Lostdragon]

On 2/15/2022 at 5:58 PM, Stephen said:

I guess just trying to temper expectations.  After 27 years of constant "Jag can run Quake if only ###" arguments it gets beyond tiring.

It goes beyond even that I often feel. 

 

 

It's been firmly established that the likes of John Carmack (Doom), Lee Briggs (WTR), Peter Johnson (Power Drive Rally), let alone say the coders of Battlesphere, could of gotten further  incremental performance out of the Jaguar, with further optimisation on existing engines/coding for the hardware from scratch/coding the hardware in a different manner, had they gone onto produce more titles, used experiences gained from titles they had written... 

 

 

We also know from others in the industry, the Tramiel's pushed for texture mapping etc, even though they knew the damage it would do to a games performance, they felt the machine had to he seen to be able to compete on a level playing field, next to the 3DO,Saturn etc. 

 

Madness, but dem was the times. 

 

But there's a vast difference between we never saw the hardware pushed to it's absolute limits, due to it's short commercial lifespan etc and... 

 

 

It running games clearly designed with Pentium PC's and consoles with dedicated 3D hardware in mind. 

 

 

At times i am not even sure what the community really wants anymore, fun, original titles for the Jaguar or 3D just for the sake of 3D.

 

 

Zero 5 was a respectable demonstration of the Jaguar doing polygon 3D, but i wish as much work had gone into the gameplay, it's too hard to be enjoyable. 

Edited February 17, 2022 by Lostdragon

[leech]

On 2/17/2022 at 11:34 AM, Lostdragon said:

At times i am not even sure what the community really wants anymore, fun, original titles for the Jaguar or 3D just for the sake of 3D.

I feel a lot of this is down to the same reason the FujiNet exists, or we want accelerators for the Atari TT030 / Falcon, etc.  To see what the machines were really REALLY capable of.  The Jaguar is in that weird position where it didn't have a long lifespan, and so had a small software library, some of which were very good (AvP, T2k, etc) and some that were... well not good.  And there's a very large gap between playable framerates and not on the games.  Showing that some developers either didn't have the time, talent, or well the care to actually optimize.  So the upper limit of the Jag (as far as the feelings of the fans) has never been hit.  So while unique 2D games are awesome, for the most part they could potentially have a port for lesser machines like the Sega Genesis.  The 3D games (for whatever magical limit people think the Jag may be capable of) seem more interesting from the 'let's see what this old hardware can do!' perspective.

 

It also is an awesome 'Challenge' platform, as there are definitely hardware bugs, and coding for custom chips always seems like something a Demo group should work on, as somehow those coders figure out some sort of magic to make some of the coolest things!  Kind of surprised the Jag doesn't have a decent demo group, though I guess getting the software to the users has always been a problem, maybe with the GameDrive, that may change.

[Gunther]

On 2/17/2022 at 6:34 PM, Lostdragon said:

It goes beyond even that I often feel. 

 

 

It's been firmly established that the likes of John Carmack (Doom), Lee Briggs (WTR), Peter Johnson (Power Drive Rally), let alone say the coders of Battlesphere, could of gotten further  incremental performance out of the Jaguar, with further optimisation on existing engines/coding for the hardware from scratch/coding the hardware in a different manner, had they gone onto produce more titles, used experiences gained from titles they had written... 

 

 

We also know from others in the industry, the Tramiel's pushed for texture mapping etc, even though they knew the damage it would do to a games performance, they felt the machine had to he seen to be able to compete on a level playing field, next to the 3DO,Saturn etc. 

 

Madness, but dem was the times. 

 

But there's a vast difference between we never saw the hardware pushed to it's absolute limits, due to it's short commercial lifespan etc and... 

 

 

It running games clearly designed with Pentium PC's and consoles with dedicated 3D hardware in mind. 

 

 

At times i am not even sure what the community really wants anymore, fun, original titles for the Jaguar or 3D just for the sake of 3D.

 

 

Zero 5 was a respectable demonstration of the Jaguar doing polygon 3D, but i wish as much work had gone into the gameplay, it's too hard to be enjoyable. 

Janky 3D Jag games are simply more interesting to consumers.

Even people that have never played on one, nor ever intend to, have also found this interesting.

It's upsetting for you, but thats the way it is.

 

Edited February 19, 2022 by Gunther

[agradeneu]

4 hours ago, Gunther said:

Janky 3D Jag games are simply more interesting to consumers.

Even people that have never played on one, nor ever intend to, have also found this interesting.

It's upsetting for you, but thats the way it is.

 

I think the interest for 2D games pushing the hardware is higher.

 

[agradeneu]

 

wrong thread lol

Edited February 19, 2022 by agradeneu

[PeterG]

On 2/17/2022 at 7:34 PM, Lostdragon said:

At times i am not even sure what the community really wants anymore, fun, original titles for the Jaguar or 3D just for the sake of 3D.

I think what is important here is that people have the feeling that they haven't seen the 3D games the Jaguar might be able to deliver or let's say the perceived potential to.
But it depends on what is expected. You want to habe Tomb Raider as it was on the Playstation? It is not going to happen. But we get a glimpse of how it might look...

 

On the other hand the first time I played Iron Soldier I was blown away. Sure it wasn't the prettiest game. The buildings were some blocks and everything aside from the marvelous helicopters and maybe also the tanks looked kind of bland. But did it feel like being in a huge mech blasting and stomping everything to the ground? Yes. So in that regard the 3D did everything it needed to do at a stable framerate in an enjoyable game (and the second one was even improved in some parts).

I don't know who said it in one of the threads here but the beauty of it is that you don't have to choose one over the other and then we can have awesome 2D and 3D games.

 

[JagChris]

On 2/7/2022 at 4:31 PM, Chilly Willy said:

 

Given the speed on the GBA, an N64 version should run rather nicely, even with software rendering.

 

A 32X version using one SH2 and plain C should be slow, but not a slideshow - it's basically how my yeti3d demo on the 32X works. If you made the same kind of optimizations to the 32X version as the GBA version, it should should be at least as fast, and faster if you can pull some dual processor stunts like Vic did for D32XR. The main thing going for the 32X version is you can have all the game logic and everything running on the SH2 rather than the 68000. The SH2 has a mature and stable C/C++ compiler, unlike the JRISC, so acting like the main SH2 is the only processor, and all the MD stuff is just a support chip to be initialized and forgotten about is a valid way of programming on it. So I wouldn't be surprised if an initial build on the 32X ran better than the current build on the Jaguar. In the end, the Jaguar version should be better, but it will take more work to get there.

 

 

It looks like he's actually begun on the 32X version. This should be interesting to watch.

[Chilly Willy]

I ought to point him toward the D32XR github repo. It has the latest support code for unified int handling as well as how to do int driven DMA and timing. He seems to be working from some of my older support code, which isn't too surprising.

 

[Chilly Willy]

I've gotten in contact with him, and pointed him towards the newer code. He seems rather nice. He also provided me with an initial build to try: works pretty well - no sound yet, around 4 FPS on real hardware with no optimizations (plain fixed point C++ codebase). Runs at 8 to 10 FPS on Fusion 3.64 (emulators always run 32X games fast as they neglect things like bus delays and cpu contention). This guy deserves a medal for his work on OpenLara. ?

[swapd0]

 

I think that this is about the performance that could I get on the Jaguar.

 

I wonder how it moves on a more complex scene, like the two bridges later in the first level.

 

Edited March 9, 2022 by swapd0

[Bill Loguidice]

3 hours ago, swapd0 said:

 

 

I think that this is about the performance that could I get on the Jaguar.

 

I wonder how it moves on a more complex scene, like the two bridges later in the first level.

 

That runs a heck of a lot better on the 32X than I expected at this stage. Incredibly impressive. 

[42bs]

3 hours ago, swapd0 said:

I think that this is about the performance that could I get on the Jaguar.

How much GPU code is in your current port?

[+cubanismo]

5 hours ago, swapd0 said:

I think that this is about the performance that could I get on the Jaguar.

That looks like higher FPS than Supercross to my eyes. Ship it.

[phoboz]

The SH-2 RISC processors of the 32x are probably on par with the RISC processors of the Jaguar, but the 68k CPU is just way to weak for this (otherwise we should already have seen a Genesis version of open lara). The only feasable option to get something playable on the Jaguar is to execute the 3D calculations on the RISC processor(s), and to use the Blitter for drawing polygons.

 

There is a major challange to do this compared to the 32x way: That is because the off-the-shelf SH-2 processors of the 32x have a gcc compiler for C/C++ code (e.g. the language that open lara is written in), while coding for the custom made RISC processors of the Jaguar means re-writing much of the code in pure assembly. I am not saying anything can't be run on the 68k, but certainly not all the heavy calculations should be there.

 

It will be very interesting to see if this can ever be realised, otherwise I do not expect to see much performance increase on the Jaguar version?

 

By the way, is there a Dremcast version of this?

Dreamcast has the even more powerful SH-4 processor and higher resolution graphics than the PSX. I would expect that an enhanced versions is possible there.

Edited March 9, 2022 by phoboz

[sirlynxalot]

57 minutes ago, phoboz said:

By the way, is there a Dremcast version of this?

Dreamcast has the even more powerful SH-4 processor and higher resolution graphics than the PSX. I would expect that an enhanced versions is possible there.

 

Yes, here's information on that.

https://github.com/histat/OpenLara-dc

 

[alucardX]

1 hour ago, phoboz said:

The SH-2 RISC processors of the 32x are probably on par with the RISC processors of the Jaguar, but the 68k CPU is just way to weak for this (otherwise we should already have seen a Genesis version of open lara). The only feasable option to get something playable on the Jaguar is to execute the 3D calculations on the RISC processor(s), and to use the Blitter for drawing polygons.

 

There is a major challange to do this compared to the 32x way: That is because the off-the-shelf SH-2 processors of the 32x have a gcc compiler for C/C++ code (e.g. the language that open lara is written in), while coding for the custom made RISC processors of the Jaguar means re-writing much of the code in pure assembly. I am not saying anything can't be run on the 68k, but certainly not all the heavy calculations should be there.

 

It will be very interesting to see if this can ever be realised, otherwise I do not expect to see much performance increase on the Jaguar version?

 

By the way, is there a Dremcast version of this?

Dreamcast has the even more powerful SH-4 processor and higher resolution graphics than the PSX. I would expect that an enhanced versions is possible there.

Great summary here phoboz. Someone like @Atari_Owl with his 3d programming background and understanding of the Jaguar hardware as well as experience writing assembly for it could be a good candidate for this, if he were so inclined.

[+cubanismo]

1 hour ago, phoboz said:

That is because the off-the-shelf SH-2 processors of the 32x have a gcc compiler for C/C++ code

There's a gcc compiler for the Jaguar RISC processors too (See jaguar/bin/dos/agpu/2.6 for the copy in my SDK, as well as jaguar/docs/docs/gcc.txt for the documentation). It's just very, very old, only supports C, and has a few... issues. Would have been nice if source were available, which legally should have been required given the GPL, but that's not helping anything at this point.

 

However, it is something you could use to kick-start your assembly efforts. Mock up your algorithm in C, generate a bunch of assembly using the RISC gcc, manually scan through it for the known bugs (It doesn't handle the RISC processor scoreboard bugs for example, and I think there is an undocumented bug where it doesn't handle jumps properly in various cases), and manually tweak the *really* perf-sensitive parts. I've been meaning to go through a round of work like this to validate that workflow, but... life and other projects always intervene.

[swapd0]

4 hours ago, 42bs said:

How much GPU code is in your current port?

0%

[42bs]

Just now, swapd0 said:

0%

Ok, then there is plenty room for optimization ?

[alucardX]

1 minute ago, 42bs said:

Ok, then there is plenty room for optimization ?

100% room to be exact!

[Judas-Prime]

4 hours ago, alucardX said:

100% room to be exact!

At the very least Up to %100

[Jagosaurus]

On 3/9/2022 at 6:53 PM, Judas-Prime said:

At the very least Up to %100