English Amiga Board


Go Back   English Amiga Board > Main > Retrogaming General Discussion

 
 
Thread Tools
Old 27 March 2024, 17:47   #21
Minuous
Coder/webmaster/gamer
 
Minuous's Avatar
 
Join Date: Oct 2001
Location: Canberra/Australia
Posts: 2,641
I was thinking more at the hardware level.

At the software level, Coffin is OS3.9-based, ApolloOS is AROS-based.

A similar point can be made that using AROS is not going to reveal much about what happens on AmigaOS, as it is quite different.
Minuous is offline  
Old 29 March 2024, 13:25   #22
VladR
Registered User
 
Join Date: Dec 2019
Location: North Dakota
Posts: 741
Quote:
Originally Posted by Minuous View Post
I was thinking more at the hardware level.

At the software level, Coffin is OS3.9-based, ApolloOS is AROS-based.

A similar point can be made that using AROS is not going to reveal much about what happens on AmigaOS, as it is quite different.
Yeah, the reality of dealing with multiple Amiga OS is an unfortunate side effect. From quick browsing of apollo forums, I can already see that getting the engine to run on all 3 OS will be a significant undertaking.

I will have to sign up for the multiboot option and start testing all 3 OSs.

I figure the uninitialized timer.device is just a first of many such OS-specific issues.

I will need to implement persistent / crash-safe logging (e.g. RamDisk or similar) for sure...
VladR is offline  
Old 29 March 2024, 13:30   #23
VladR
Registered User
 
Join Date: Dec 2019
Location: North Dakota
Posts: 741
The benchmark scene I am working on is heavily inspired by section that starts at 7:03 here:
[ Show youtube player ]

I'm realizing that Sega Model 1 is a really good reference point for my future visual style : sharp High-Poly flatshading with far draw distance.
No shimmering textures, just elegant clear-cut shapes.

This Benchmark should answer the question of how far can this scene complexity be pushed with 040/060, as I already have a pretty good idea with my Vampire.

No idea on PiStorm. Yet. In theory, emu68 should shred Vampire to smithereens, but this will be the real test (how many frames more rendered compared to some abstract MIPS figure)
VladR is offline  
Old 29 March 2024, 19:53   #24
copse
Registered User
 
Join Date: Jul 2009
Location: Lala Land
Posts: 522
That flat-shaded look is great. The popping and view distance are the cost of doing business and not terrible.
copse is offline  
Old 30 March 2024, 12:03   #25
VladR
Registered User
 
Join Date: Dec 2019
Location: North Dakota
Posts: 741
Quote:
Originally Posted by copse View Post
That flat-shaded look is great. The popping and view distance are the cost of doing business and not terrible.
True, but unlike the fixed arcade/console HW, on Amiga we have the advantage of huge performance differences between various configs, so there is at least an option of having full draw distance (if your HW has the power).

The draw distance is a simple parameter in my engine, so I'll keep it tweakable by user.
LOD requires manual process of creating different art assets (basically ~tripling the manual work for 3 levels), so that's less likely to be present.
Due to the amount of work, I'd rather create 2-3 different environments (than merely lower LOD assets) in same time.

The only thing that remains to be answered is the ratio between SysInfo's MIPS figure and actual framerate.

Meaning, does 1,600 MIPS of PI4b actually translate into 10x framerate of Vampire's 160 MIPS ? Or is it 10% more or exactly how much in between those two extremes?

That's what this benchmark is about - getting one real-world number that will allow us to compare the meaningful performance of each CPU (only under RTG, of course).

I should be finished with the StarWars environment today, if all goes well...
VladR is offline  
Old 30 March 2024, 18:44   #26
d4rk3lf
Registered User
 
d4rk3lf's Avatar
 
Join Date: Jul 2015
Location: Novi Sad, Serbia
Posts: 1,649
Thumbs up for the decision to use flat shaded poly's, instead of low res textures.
It always looked way better visually to me, then any textured game, until 2000.

Looking forward to see some videos running benchmarks on different configs.

Good luck Vlad!
d4rk3lf is offline  
Old 30 March 2024, 21:43   #27
VladR
Registered User
 
Join Date: Dec 2019
Location: North Dakota
Posts: 741
Quote:
Originally Posted by d4rk3lf View Post
Thumbs up for the decision to use flat shaded poly's, instead of low res textures.
It always looked way better visually to me, then any textured game, until 2000.

Looking forward to see some videos running benchmarks on different configs.

Good luck Vlad!
Thanks. I've always been a fan of flatshaded look, it just somehow never got used a lot. Instead, like you said, we got grainy, uber-pixelated mess for half a decade - till we got 3DFX...

I spent few hours browsing apollo forums in last few days and it looks like Vampire V4 has now dedicated portion of FPGA for Maggie (which does texturing in HW, including bilinear filtering).

But even from a very limited testing I've done so far on V4, I would prefer high-poly flatshaded scenes at 24-bit.
Regardless, aiming for RTG will make it run on V4 (and all other RTG configs). Aiming for Maggie would only make it run on V4...
VladR is offline  
Old 30 March 2024, 22:10   #28
copse
Registered User
 
Join Date: Jul 2009
Location: Lala Land
Posts: 522
I think where we have seen it used occasionally, it has stood out. Although that may be my own retro preferences. So it's nice to see modern projects that are bringing it back, like Thunder Helix for instance.

[ Show youtube player ]
copse is offline  
Old 30 March 2024, 22:10   #29
d4rk3lf
Registered User
 
d4rk3lf's Avatar
 
Join Date: Jul 2015
Location: Novi Sad, Serbia
Posts: 1,649
@VladR
Well.. I am not a coder...
...but maybe to consider this...

Frontier game worked even on lowest Amiga's, and in my opinion, had a very decent looking, and detailed flat shading look.
Sure, on OCS/ECS-es with 68k 7Mhz, and 1MB Ram, when you land on planet, you get to a pretty much slide show, but it's still impressive.

I repeat.. I am not coder... so take everything I say with a reserve, but I don't see why minimum configuration for a very cool, flat shaded poly (pretty detailed), fast paced racer, couldn't be done with a fast 030 with 4 or 8MB Ram, with high framerate (even with this lowest specs)...

Oh wait.. I just remember...
There was a GEM of the game... a GEM...
That somehow managed very smooth racers on 68k 7Mhz lowest Amigas, using flat shaded polygons.
No second Prize:
[ Show youtube player ]

I have played it on my Amiga 500, and couldn't believe how smooth it was... never saw anything like it.. whoever code this was genius.

Btw.. I have nothing against Vampires.. but if you want to sell your game, if you go, something like 030 goal, you should have much much wider audience.
Maybe some specific Vampire version with higher res.. or something...

Last edited by d4rk3lf; 30 March 2024 at 22:16.
d4rk3lf is offline  
Old 31 March 2024, 04:32   #30
VladR
Registered User
 
Join Date: Dec 2019
Location: North Dakota
Posts: 741
Quote:
Originally Posted by d4rk3lf View Post
@VladR
Frontier game worked even on lowest Amiga's, and in my opinion, had a very decent looking, and detailed flat shading look.
Sure, on OCS/ECS-es with 68k 7Mhz, and 1MB Ram, when you land on planet, you get to a pretty much slide show, but it's still impressive.
Yes, but that framerate is not going to fly in 2024. That era is long gone, unfortunately. Also, it's a very different engine compared to RTG-based Chunky framebuffer. You need to do a lot of experimenting with how much can Blitter do in parallel with the available DMA slots of CPU and that's a lot of work - just read couple threads in ASM section of this forum about it - it's a combinatorial explosion of approaches to try and see which one works the best.

In future, if this works out, I would like to extend my engine to support sub-030 configs, yes. But not right now when everybody is getting Pi, emu68 and a host of other accelerators.

Such an OCS engine is a fantastic weekend project for somebody with a stable job who can tinker with it for a year during weekends (or fulltime for few weeks when in between the jobs). I will gladly let someone else serve that niche (and there's few people on this site doing this very thing and have been for few years).


Quote:
Originally Posted by d4rk3lf View Post
I repeat.. I am not coder... so take everything I say with a reserve, but I don't see why minimum configuration for a very cool, flat shaded poly (pretty detailed), fast paced racer, couldn't be done with a fast 030 with 4 or 8MB Ram, with high framerate (even with this lowest specs)...
Of course it can be done with fast 030. But most likely not using RTG. It'd have to use C2P conversion.

I believe I can write a reasonably performing C2P in a week or two, as I already have some experience with 6 Bitplanes in EHB mode (for static GFX - executable render compo). But, as with all things Amiga, I really need at least 3 different versions of C2P based on the config. So, it's probably going to take a month to burn just on C2P.

But, if it was just for 030, then I am pretty confident I can churn out a reasonably competitive and heavily optimized (and many times rewritten) C2P ASM routine in 10-14 days.


Quote:
Originally Posted by d4rk3lf View Post
Oh wait.. I just remember...
There was a GEM of the game... a GEM...
That somehow managed very smooth racers on 68k 7Mhz lowest Amigas, using flat shaded polygons.
No second Prize:
[ Show youtube player ]

I have played it on my Amiga 500, and couldn't believe how smooth it was... never saw anything like it.. whoever code this was genius.
Yes, I am aware of NSP. To get that level of performance out of OCS would require at the very least 3 months full-time (after the game is done), just running benchmarks on various approaches with Blitter and seeing which performs the best. And there's still possibility I might not even get there after burning through that time, so this is an incredibly risky approach for first-gen game.

Quote:
Originally Posted by d4rk3lf View Post
Btw.. I have nothing against Vampires.. but if you want to sell your game, if you go, something like 030 goal, you should have much much wider audience.
I hear you and fully agree.
The benchmarks will be the final arbiter. I honestly have no clue how fast 50MHz 030 in RTG is. I've seen few YT vids of 50 MHz 030 and it was borderline from what I recall - but it was a different kind of engine, so I need to see data from my engine to be able to see if it's viable or not in RTG (obviously, bypassing RTG with a C2P approach would make the 030 a totally viable config - no disagreement there but I don't have that yet unfortunately).

Quote:
Originally Posted by d4rk3lf View Post
Maybe some specific Vampire version with higher res.. or something...
Higher Res is already supported. I've run it on V4 in all ~15 resolutions from 320x200 all the way up to 1920x1080 (It's doing about 3fps in 1920x1080, which is not that bad, actually).
I'm going to leave resolution choice up to user.
V4 is currently running in 24-bit color and will have a separate method of procedural colorization/fog.
15/16-bit support is next (technically, I have half of the pipeline working in 16-bit already), and 8-bit will be the last.
VladR is offline  
Old 31 March 2024, 05:04   #31
VladR
Registered User
 
Join Date: Dec 2019
Location: North Dakota
Posts: 741
Quote:
Originally Posted by copse View Post
I think where we have seen it used occasionally, it has stood out. Although that may be my own retro preferences. So it's nice to see modern projects that are bringing it back, like Thunder Helix for instance.

[ Show youtube player ]
That's a phenomenal vid, thank you !
Keep 'em coming, if you spot some other ones
VladR is offline  
Old 31 March 2024, 05:10   #32
VladR
Registered User
 
Join Date: Dec 2019
Location: North Dakota
Posts: 741
Quote:
Originally Posted by d4rk3lf View Post
Looking forward to see some videos running benchmarks on different configs.
That's going to have to be done by people with those configs, but my thinking is that if I replace some dry synthetic benchmark with a Star Wars scene from Sega Model I, then it should hopefully entice people to download and run it

I've finished the basic shape of the corridor mesh segment today - see attachment. It's all modelled in Notepad++

Tomorrow I'll work on more details on the floor and the signature red pipes so that then the camera can be scripted accordingly.
Attached Thumbnails
Click image for larger version

Name:	Benchmark03.gif
Views:	66
Size:	45.0 KB
ID:	81937  
VladR is offline  
Old 01 April 2024, 03:32   #33
VladR
Registered User
 
Join Date: Dec 2019
Location: North Dakota
Posts: 741
I spent whole day reading Amiga docs and googling but can't seem to find anything inherently wrong with my timer.device code. The only way is to debug it on V4 as it works fine under WinUAE, but my SD extension cable will arrive in about 10 days and I am not willing to keep plugging SD card back and forth till the slot is damaged.

I did, however, implement a simple VBL IRQ FrameCounter, so I have at least a rudimentary idea how it performs in different resolutions (within the precision of +- 1/120s).

Code:
  Vampire V4 (No AMMX/Maggie, Just CPU)

-----------------
PolyCount: 1,020 tris
Color Depth: 32-bit
Total Frames rendered:1,000
Method: Brute-Force (no LOD)

-----------------
 320x200: 77 fps
 640x200: 63 fps
 480x270: 62 fps
 640x360: 53 fps
 640x400: 39 fps
 720x576: 33 fps
 848x480: 26 fps
 800x600: 22 fps
 960x540: 20 fps
1280x720: 13 fps
1440x900: 7.4 fps
1920x1080:4.8 fps
16-bit modes should raise the fps a bit and 8-bit even further, obviously.


It's quite surprising seeing ~5 fps full redraw at FullHD at 16.7 Mil colors. It would actually be useable in many genres (outside of racing and FPS) - certainly turn-based 3D games could work well in 10 fps (with half screen covered in HUD and redrawing just other half) - I certainly played many games at that framerate on PC...
VladR is offline  
Old 03 April 2024, 19:42   #34
VladR
Registered User
 
Join Date: Dec 2019
Location: North Dakota
Posts: 741
The micro-SD card cable extender (to save the SD card slot on V4) arrived, but since the Vampire has the two other connectors (HDMI and power) literally right next to each other, I have these options:
1. No power and no HDMI, but SD extender fits in
2. Power + HDMI On, but no SD Extender

I didn't want to wait another 2 weeks to see if the right-angle connectors would work (and spend $50 on it), so I bit the bullet and got the networking working from my PC.

Now I can upload builds right from my PC.


Of course, compared to Atari Jaguar's Skunkboard, the upload process is very clunky and cannot yet be automated on Windows - it is -literally- drag and drop!
Plus ApolloExplorer doesn't like files over 250 KB (it just hangs, does not finish uploading it), so I had to spend half day cleaning up all the data so the EXE fits under 250 KB, but at least it -kinda/sorta- works with screeching teeth...



What would be the 100% working way to upload files from PC to V4 ? FTP ? Samba ? Preferably something that can run from command-line.


Any ideas ?
VladR is offline  
 


Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools

Similar Threads
Thread Thread Starter Forum Replies Last Post
10 Years of Amiga Gaming Part 10: 1995 kad3t Retrogaming General Discussion 43 29 September 2022 18:20
Amiga 2000 keyboard accepts Amiga 500 key/spring/plungers? rjd324 support.Hardware 1 30 May 2020 14:10
Amiga joystick on pc for xyz time, but this time with adapter? :) Srksi support.WinUAE 3 24 May 2018 03:57
Uaeunp (24-09-10) refuses to acknowledge certain formats. MethodGit support.WinUAE 12 05 November 2010 00:21
F/S: Vidi Amiga 24-bit real time colour digitiser John64 MarketPlace 4 06 June 2009 18:47

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +2. The time now is 14:59.

Top

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.
Page generated in 0.18368 seconds with 16 queries