Breakthrough for the ao486 core announced - Cache

[Tieho]

I've been trying out Jagged Alliance 2 (GOG version) on these dev cores, on Win98 and DirectX 6. I think since 27 the game has worked without any graphical glitches, but the framerate is about 2 fps even in the menus. Also the music is breaking up. When you play the actual game, it seems that the voice samples play without any glitches, but still the framerate is really bad and music keeps breaking up. Is this game just too much for the core? Doesn't seem very cpu taxing to me, especially the menus. I know I could be playing this on an actual PC, but I'm just curious

[IAmParadox]

I've been trying out Jagged Alliance 2 (GOG version) on these dev cores, on Win98 and DirectX 6. I think since 27 the game has worked without any graphical glitches, but the framerate is about 2 fps even in the menus. Also the music is breaking up. When you play the actual game, it seems that the voice samples play without any glitches, but still the framerate is really bad and music keeps breaking up. Is this game just too much for the core? Doesn't seem very cpu taxing to me, especially the menus. I know I could be playing this on an actual PC, but I'm just curious

Really, dude?

System requirements
Minimum system requirements:
System:
Windows 7 or 10
Processor:
1 GHz
Memory:
512 MB RAM
Graphics:
3D graphics card compatible with DirectX 7 (compatible with DirectX 9 recommended)

AO486 doesn't have a 3D graphics card, I'm surprised it runs, at all.

[Tieho]

I've been trying out Jagged Alliance 2 (GOG version) on these dev cores, on Win98 and DirectX 6. I think since 27 the game has worked without any graphical glitches, but the framerate is about 2 fps even in the menus. Also the music is breaking up. When you play the actual game, it seems that the voice samples play without any glitches, but still the framerate is really bad and music keeps breaking up. Is this game just too much for the core? Doesn't seem very cpu taxing to me, especially the menus. I know I could be playing this on an actual PC, but I'm just curious

Really, dude?

System requirements
Minimum system requirements:
System:
Windows 7 or 10
Processor:
1 GHz
Memory:
512 MB RAM
Graphics:
3D graphics card compatible with DirectX 7 (compatible with DirectX 9 recommended)

AO486 doesn't have a 3D graphics card, I'm surprised it runs, at all.

These are the requirements from the readme.txt:
Windows 98/ME/2000/XP
Pentium 233 MHz
32 MB RAM
1 GB Hard Disk Space
4 MB Video Card
2X CD-ROM
DirectX 6.0 (or later)
DirectX Compatible Graphics Adapter
DirectX Compatible Sound Card
Windows Compatible Mouse & Keyboard

Obviously the core isn't Pentium, but just wondering why even the menus are so slow.

[Linobob]

These should be the real requirements:
https://www.systemrequirementslab.com/c ... gold/11868
It says it needs a 133 MHz CPU which is just a bit too high for the core right now...

[Cebion]

I've been trying out Jagged Alliance 2 (GOG version) on these dev cores, on Win98 and DirectX 6. I think since 27 the game has worked without any graphical glitches, but the framerate is about 2 fps even in the menus. Also the music is breaking up. When you play the actual game, it seems that the voice samples play without any glitches, but still the framerate is really bad and music keeps breaking up. Is this game just too much for the core? Doesn't seem very cpu taxing to me, especially the menus. I know I could be playing this on an actual PC, but I'm just curious

Really, dude?

System requirements
Minimum system requirements:
System:
Windows 7 or 10
Processor:
1 GHz
Memory:
512 MB RAM
Graphics:
3D graphics card compatible with DirectX 7 (compatible with DirectX 9 recommended)

AO486 doesn't have a 3D graphics card, I'm surprised it runs, at all.

Here are the real specs for Jagged Alliance 2
I don't know where you got your numbers but that may just for emulating dosbox from gog.

Minimum CPU Class Required Intel Pentium
Minimum OS Class Required Windows 95
Supported Systems/Modelsx86 (32-bit)
Minimum RAM Required 32 MB
Media Type CD-ROM, Download
Minimum CD-ROM Drive Speed Required 4X (600 KB/s)
Video Modes Supported Full screen
Video Resolutions Supported 640×480
Input Devices Supported Keyboard, Mouse

Capture.PNG (152.95 KiB) Viewed 9706 times

[Schyz]

Here are the real specs for Jagged Alliance 2
I don't know where you got your numbers but that may just for emulating dosbox from gog.

The game is from 1999, I'd say it's expected to be slow in ao486. I think anything released beyond 1996 is going to require a lot of luck to run at playable speeds.

[breiztiger]

https://www.pouet.net/prod.php?which=3540

when you try to run from floppy disk ao486 crash but run great from directory on harddrive

[IAmParadox]

And, usually when it says minimum requirements, that means, everything to the lowest setting, and, most of the time, expect a slideshow, no matter what you do.

[IAmParadox]

https://www.pouet.net/prod.php?which=3540

when you try to run from floppy disk ao486 crash but run great from directory on harddrive

This may be connected to a longtime issue with the soundcard and the floppy drive stepping on each others DMA or, memory space, I'm not exactly sure why it happens, but, this post viewtopic.php?p=6787#p6787 seems to show that the MIDI wrote to the A:\ drive's memory space, or, something.

[ToothbrushThreepwood]

Pentiums are also a whole different animal compared to 486 architecture: dual-pipelines, much stronger FPU (we don’t even have one at all ) and more means than you can’t expect nearly the same performance MHz-for-MHz with a 486. Especially in the later parts of the 90ies, when most games relied on the tricks of the pentium to do advanced graphics. Quake, for example, was optimized for pentium on the assembly level. This awesome video illustrates this with a P60 outperforming 486ers of much higher the clock speeds (in Quake, but not in the other benchmarks): https://youtu.be/NLrKxWL73Mw

[Schyz]

https://www.pouet.net/prod.php?which=3540

when you try to run from floppy disk ao486 crash but run great from directory on harddrive

This may be connected to a longtime issue with the soundcard and the floppy drive stepping on each others DMA or, memory space, I'm not exactly sure why it happens, but, this post viewtopic.php?p=6787#p6787 seems to show that the MIDI wrote to the A:\ drive's memory, or, something.

I can confirm the behaviour, if using a memory manager (EMM386/QUEMM) there seems to be a region in the upper memory area that messes with the floppy drive. Probably that region wasn't used back in the day but ao386 decided to use it to adrress the FDD, so memory managers use it to loadhigh TSRs or map EMS, expecting it to be available. To avoid corruption, what I do is to disable EMM386 in config.sys and reboot to copy the files (or install) from FDD to HDD when needed.

Pentiums are also a whole different animal compared to 486 architecture: dual-pipelines, much stronger FPU (we don’t even have one at all ) and more means than you can’t expect nearly the same performance MHz-for-MHz with a 486. Especially in the later parts of the 90ies, when most games relied on the tricks of the pentium to do advanced graphics. Quake, for example, was optimized for pentium on the assembly level. This awesome video illustrates this with a P60 outperforming 486ers of much higher the clock speeds (in Quake, but not in the other benchmarks): https://youtu.be/NLrKxWL73Mw

Also, Pentium platforms usually benefit from using PCI, which is a big improvement over ISA.

[IAmParadox]

Some late 486 boards had PCI... hint...

[tontonkaloun]

I'd like to say it's not too difficult to compile yourself, but got a 'Can't convert time-limited SOF into POF, RBF, etc.' message in the log.

I could be wrong, but I think that is due to using the "Lite" version of Quartus.

you have a compiled file to test it?

I have compiled the DEV branch.

It would probably be be better to wait for official release from the people who know what they are doing, but if you can't...

EDIT:
Joystick: add 4-buttons and gravis pro.
ao486_Cache28_be08f83.zip


EDIT:
Joystick: support for variable cpu clock and some fixes.
ao486_Cache29_8ad57c2.zip


EDIT:
L2 Cache: 2 times size, LRU exchange strategy, reduced dirtyflag resource usage
ao486_Cache31_087914e.zip


EDIT:
Bypass aligned reads directly into CPU
ao486_Cache32_aa65f9e.zip


EDIT:
Fixed shared memory
ao486_Cache33_7431d4b.zip


EDIT:
Uart: fifo-off compatibility, implement mark/space parity.
ao486_Cache34_ece74ab.zip

Hello,

The latest version (cache34) doesn't seem to work !!
It freezes at startup

Should I revert to an earlier version of the main Mister file? (I see it is older in your archive) ?

[Apolonius]

I have an error after load SoftMPU, it can't detect serial interface and I have tried on ports 1 to 4, anyone have a tip?.

Regards.

[abbub]

Yep. Same here. cache34 freezes on boot, bringing up the OSD locks the system.

[lroby74]

Hi,
it's same thing for me, while cache33 is wonderfull! Doom @ 30,11fps and with music ingame

Hello,

The latest version (cache34) doesn't seem to work !!
It freezes at startup

[kathleen]

Yep. Same here. cache34 freezes on boot, bringing up the OSD locks the system.

Cache34 works on my side if I use the Main Mister file provided in the Cache24 archive, with the other one, it freezes at the boot also.

[tontonkaloun]

I suspected it but I was too lazy to do it in this heat!


thank you for your feedback

[IAmParadox]

I would stick to 33, the changes in 34 that most likely cause the hangs have to do with disk i/o and could possibly mess up your HDDs without the proper changes in the new Mister file. I'd wait until a fix is added to the dev repo.

The Mister executable, included with cache33 is the same one for most of the cache releases, up until cache34.

[tontonkaloun]

I would stick to 33, the changes in 34 that most likely cause the hangs have to do with disk i/o and could possibly mess up your HDDs without the proper changes in the new Mister file. I'd wait until a fix is added to the dev repo.

The Mister executable, included with cache33 is the same one for most of the cache releases, up until cache34.

I think like you, I will wait to see which version of the main Mister will be released in the next updates and I will also stay at 33.

[abbub]

Cool, cool. cache33 works, though is it a known issue that the UART is broken / doesn't work? I have mtcp dos networking up and running under the current (05/10/2020) release under PPP mode, but it doesn't seem to work on the newer versions with the cache. I assume this is related to the switch from the commercial uart implementation to the free implementation?

[Schyz]

Some late 486 boards had PCI... hint...

Does ao486 core use a PCI bus?

[bbond007]

Cache34 works on my side if I use the Main Mister file provided in the Cache24 archive, with the other one, it freezes at the boot also.

Sorry, I somehow included an old MiSTer file and not the most current one I had just built... fixed....

[Caldor]

I don't know where you got your numbers but that may just for emulating dosbox from gog.

Minimum CPU Class Required Intel Pentium
Minimum OS Class Required Windows 95
Supported Systems/Modelsx86 (32-bit)
Minimum RAM Required 32 MB
Media Type CD-ROM, Download
Minimum CD-ROM Drive Speed Required 4X (600 KB/s)
Video Modes Supported Full screen
Video Resolutions Supported 640×480
Input Devices Supported Keyboard, Mouse

Generally GOG displays the requirements for running the game using their DOSBox or ScummVM setup and not what would be required to run the game natively in DOS. But 133mhz Pentium as minimum requirements... it is pretty amazing that it runs at all. Several games wont run due to their FPU requirement. With this core, just doing stuff in SVGA is pretty hard on the CPU and such it seems. Especially 3D games.

[rhester72]

First time trying to bring a new (read: git tip) build up off the ground - using boot0.rom from git dev branch and boot1.rom both from git dev and from the ET4000, in all cases I get the following (from an otherwise-working vhd and unchanged config from the latest official ao486 release):

- Proper video BIOS ID string (either 2MB from Tseng Labs or the Bochs one)
- Proper indication of primary and secondary HDD
- Hang

It also seems to be spin-waiting or something, because UI response becomes EXTREMELY sluggish.

Switching back to the last official boot0.rom results in nothing but a black screen (but normal UI behavior). Resetting settings resulted in my MiSTer diving off the deep end and stopping all video output until power cycle.

Anything obvious I'm missing (a setting that needs to be changed, perhaps)?

[IAmParadox]

First time trying to bring a new (read: git tip) build up off the ground - using boot0.rom from git dev branch and boot1.rom both from git dev and from the ET4000, in all cases I get the following (from an otherwise-working vhd and unchanged config from the latest official ao486 release):

- Proper video BIOS ID string (either 2MB from Tseng Labs or the Bochs one)
- Proper indication of primary and secondary HDD
- Hang

It also seems to be spin-waiting or something, because UI response becomes EXTREMELY sluggish.

Switching back to the last official boot0.rom results in nothing but a black screen (but normal UI behavior). Resetting settings resulted in my MiSTer diving off the deep end and stopping all video output until power cycle.

Anything obvious I'm missing (a setting that needs to be changed, perhaps)?

Try a lower clock speed, if you are trying to start at 100Mhz, and, a lower speed fixes it, you may have the same problem I had with my de 10 board, which I have returned, and am waiting on a new one to arrive.

If it isn't this, then, maybe you are using the old Mister executable included with the cache34 download... like everyone else, until the uploader fixes it.

[IAmParadox]

Cache34 works on my side if I use the Main Mister file provided in the Cache24 archive, with the other one, it freezes at the boot also.

Sorry, I somehow included an old MiSTer file and not the most current one I had just built... fixed....

I just redownloaded 34, and, it's the same.

Can you make all the versions you've compiled, available, somewhere? I'd like to get the ones I never downloaded, for science reasons.

[rhester72]

Interestingly, it was the (lack of an) updated MiSTer file that was preventing me from booting - I didn't know it was needed.

What does it change?

[dshadoff]

Try a lower clock speed, if you are trying to start at 100Mhz, and, a lower speed fixes it, you may have the same problem I had with my de 10 board, which I have returned, and am waiting on a new one to arrive.

If it isn't this, then, maybe you are using the old Mister executable included with the cache34 download... like everyone else, until the uploader fixes it.

This core should not be expected to run at high clock speeds (but it might sometimes).
Although it's constantly changing, the timing closure was around 50MHz a day or so ago - nowhere near 100MHz (though it works to varying degrees at speeds better than the closure speed). If your board doesn't run a deep-development not-even-beta-yet core at the top speed listed on a control panel, don't despair. Maybe the core will change; maybe the control panel will change.

They are trying to reduce complexity on critical paths to allow higher speeds when a good fit takes place, but don't expect that it runs at high speeds (though that is the goal).

Wait for the next release for any sense of expected speeds.

[rhester72]

Point taken.

As of dev tip right now, Norton 8 says overall performance (both CPU and disk) is at 486SX/33 levels, which feels about right to me having also tested (a whole bunch of) id software. If it doesn't improve any further at all, it's still a REMARKABLE effort and I'm getting serious waves of nostalgia from it. Unbelievable effort! Who'd have ever thought MiSTer representing so well at 90s PC gaming?