Artur Jarosik's GOLD2.7 WIP - GnGeo port work in progress (Partially optimised with AMMX2), video recorded by Tuko

http://www.apollo-core.com/knowledge.php?b=3¬e=12560

https://www.youtube.com/watch?v=_WGpP2MOKUE

Amiga 600 Vampire 2 Gold 2.7 FPU DEMO Ephidrena & Kvasigen DATASTORM 2010

https://youtu.be/BqvtjONlsD0?t=64

@edponpon

Quote:

Vampire 1200 edition, where are you?

AFAIK not before the VampireV4.

Core 2.7 reaches 30FPS on Quake 320x200

See Here

_________________
God created Paula so that Allister Brimble and Dave Whittaker could do music

Check my Amiga gameplays (ITA)!

Also, Apollo team published an interesting documentation on how V2 FPU works:



Linky

_________________
God created Paula so that Allister Brimble and Dave Whittaker could do music

Check my Amiga gameplays (ITA)!

@ErikBauer

320 x 200 30fps vampire is good but these res are very low we should be at least 640 x 480. Can we start doing some test at higher screen modes.

_________________
A1200 ACA1232 128MB Indivison MkIICr
Elbox empty Power Tower
RPi3 Emulating C64 ZX Atari PS BBC
Wii with Amiga emulation
Vampire v4 SA

Higher res get half speed AFAIK.

Keep in mind FPU core is not fully optimized, yet, and that Quake exe is compiled without any compiler that gets advantage from 080.
I guess time will tell

_________________
God created Paula so that Allister Brimble and Dave Whittaker could do music

Check my Amiga gameplays (ITA)!

@ErikBauer
I think I read that Quake only uses FPU instructions that are already implemented entirely in hardware (to the extent that any modern CPU processes instructions purely in hardware) and so is unlikely to get much faster with any FPU core improvements. Where there may be room for improvement is in the FPU code generated by the compiler. There may be room for improvement elsewhere in the code if optimisation for the 68080 including AMMX2 was implemented.

@g01df1sh

Quote:

g01df1sh wrote: @ErikBauer 320 x 200 30fps vampire is good but these res are very low we should be at least 640 x 480. Can we start doing some test at higher screen modes.

People tend to forget that 640x480 is not that easy to do.
http://old.vgamuseum.info/benchmarks/309-quake-1-software-rendering.html

640x480 software render on an AMD Athon XP 2200+ with a Voodoo 3 2000 just gets 20,2 FPS.

@WOSPUPOS4

Quote:

WOSPUPOS4 wrote: @g01df1sh Quote: g01df1sh wrote: @ErikBauer 320 x 200 30fps vampire is good but these res are very low we should be at least 640 x 480. Can we start doing some test at higher screen modes. People tend to forget that 640x480 is not that easy to do. http://old.vgamuseum.info/benchmarks/309-quake-1-software-rendering.html 640x480 software render on an AMD Athon XP 2200+ with a Voodoo 3 2000 just gets 20,2 FPS.

You wouldn't run 3D engines in higher resolutions using a software mode anyway, 640x480 and higher resolutions will look much better using OpenGL.

@WOSPUPOS4
That is because the PCI bus of that specific Voodoo holds back the Athlon. With an AGP graphics card it's much faster.

Also, if you configure the MTRR registers of a modern CPU properly, you get much higher framerates. I assume these results are from DOS, where you need to do this by a separate tool invoked before you run your application - or in AUTOEXEC.BAT. Windows and its graphics card/platform drivers tend to do this themselves anyway. This was probably the number one reason why a lot of people got the impression that even software rendering games are faster with Windows back in the days... They weren't just the system was set up properly for maximum performance.

So see the bottommost results marked as "fastvid". A Voodoo3 AGP sports over a 100fps in this case...

(Ps: I own and somewhat regularly use two AMD K6-III+ retro PC systems. Been there, done this. )

(Edit: addressed answer to the wrong person. Fixed @ nick.)

Last edited by Chain-Q on 01-Mar-2018 at 08:19 AM.
Last edited by Chain-Q on 01-Mar-2018 at 08:18 AM.

_________________
MorphOS, classic Amiga, demoscene, and stuff
"When a bridge is not enough, build a Viaduct!"
"Strip the Amiga community of speculation and we can fit every forum on a 720k floppy" (by resle)

@Chain-Q

i can make an example on x5000/40. on this machine just the swap from 6570 to 7750 make quake in software rendering increase 10 fps . Same was for Ak4 aga vs voodoo3 or a permedia2 .
Fast gpu with faster ram and better bus speed make better result in quake in software rendering mode.

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome

Core 2.7 has gone Gold!

_________________
God created Paula so that Allister Brimble and Dave Whittaker could do music

Check my Amiga gameplays (ITA)!

@ErikBauer

Congrats all! :)

_________________
Indigo 3D Lounge, my second home.
The Illusion of Choice | Am*ga

Very indepth presentation of the Vampire project by Manuel Jesus recorded by The Guru Meditation;

https://www.youtube.com/watch?v=F-egp3pWr1w

Gold 2.9 released:

RELEASE GOLD 2.9

New features

1) Better support for fast Power-OFF/ON of the AMIGA
2) Allow Software to turn off build-in MAP-ROM to access on board ROM-CHIP.
3) MAPROM mapping support for ATARI Roms
4) Memory Management support for ATARI Memory map
5) FPU speed up (now 51 MFlops at x11 according to Sysinfo)
6) FPU fully working under ATARI OS
7) Press FIRE-buttons on Power-ON will boot EMUTOS (ATARI OS)

http://www.apollo-core.com/knowledge.php?b=6¬e=14069

GOLD 2.9 - GFX Card Speed

http://www.apollo-core.com/knowledge.php?b=2¬e=14103

Vampire GOLD 2.9 running as ATARI

http://www.apollo-core.com/knowledge.php?b=2¬e=14104

Last edited by Overflow on 08-Apr-2018 at 04:37 PM.

Nice video showing Lightwave rendering and workflow using A2000 equipped with Vampire.

Ofcourse, Lightwave isnt optimized for Apollo Core features, so the performance leaves quite a bit to be desired considering the un-utilized power;

https://www.youtube.com/watch?v=YXUtCF9mMJM

Apollo Team Activity Report (June 2018)

http://www.apollo-core.com/knowledge.php?b=1¬e=15593

Quote:

June Activity Report Exciting weeks lately in the Apollo Team as quite a lot of things happened in a very small timeframe on both hardware and software sides. Summer will for sure be hot in the Vampire and Amiga community. Sit down, relax and enjoy a glass of blood with us because we have several topics to cover this month. Vampire V4 If your remember last Activity Report from May, we had to admit we were struggling on a bug we weren�t able to locate the source and this for weeks and weeks. During the night of 2nd of June, ceaich and BigGun finally found the reason and fixed it, definitively. Hurray ! Some team members were during that particular evening attending the OufParty 5 in Martigny (Switzerland). When they got the information, they instantly left the dinnertable, leaving creamy raclette and fresh white wine alone to teleport themself in front of their Vampire V4 developer boards to test that freshly baked core : for the first time ever since Neuss, the Vampire V4 worked perfectly on every Amiga 500 mainboard tested. I won�t comment what came next, mostly because I�m missing part of it and because it was mainly made from alcohol and cigars (which might explain why I�m missing those parts). So from now, our next step is now to send Vampire V4 cards to all testers in team to go forward on finalizing the product before releasing it on the market. Those final steps will require some time but we can now see the light out of the tunnel. Software-side I wanted to put software as second topic of this activity report because we also got some nice development in the Apollo Team cavern : - Bax is working as always as a shaolin monk on the driver side. New SDNet and ExpNetV2 should bring new features like multicasting/SANA-II (letting you use network within ShapeShifter), while improving performance and reliability. - Nonarkitten is working on GnGeo and is squeezing every single FPS she can out from it - Inflammasome is working on improving his STonAmiga software Card Registration As stated in the apollo-core forum, we have announced we are closing card registration website at end of June. This had to end to serve its purpose and I will try to explain why. As you might know, we have been aware that some cloned cards were in the wild that�s the reason why the registration system came to battle against such black market. Registering a card on our website allowed us to double check that we had corresponding serial number in our database and add it to our whitelist. Only « downside » of a non-registered card was the user was getting a black and white screen on Digital Output (and only beginning from GOLD2.7 cores) but could still use its card. Cards coming from majsta were all already in our database, ence were « pre-whitelisted », cards coming from kippa weren�t all in that database. As of today, users are all getting cards flashed with GOLD2.5 core, which doesn�t implement the whitelist system and let them already use their card out of the box in full color. This give us some time to cook new and updated GOLD2.9 cores including those serials. This system isn�t ideal as there is still a period for users to wait for an updated whitelist to flash their card and get full colors but it is the only solution to make sure cloned cards won�t offer same level of features as a legit card. We are not only protecting our IP but also customer investments here. I would personnaly like to thank all people who played the game and registered their card. It was a pain not only for customers but also for team as we would have prefered to spend time on other things than building a database of serials, cooking whitelist, updating cores, etc. We have also read some funny comments about GDPR regarding our registration system. It�s obvious that some more knowledge regarding what GDPR really is might be required there. We have a Privacy Policy stated on our website, directly accessible within the form, a confirmation dialog in the registration form asking for approval and we only ask for datas we require to bring users a good level of support. Datas we are processing aren�t « sensitive » (unlike medical datas, religion and sexual preference, etc.) and people pushing us on that kind of discussion field are in fact spreading FUD (but we are used to it). I�ll stop there otherwise some might think that a famous amiga hardware producer wrote that paragraph (which would require use of the usual criminal damage, slander, defamation, insults, conspiracy and some more blah blah words). Today we think we have erradicated any will to build and sell more cloned cards. The registration served its purpose. Some users, might still not have registered their card but we will support them in that process even if registration system is closed. Just get in touch with me when needed. No worry. GOLD3 BigGun announced GOLD3 for Vampire cards for this summer. We are all pretty confident that our AGA will be really super and every day core is being improved with new fixes to let us play more and more games. It�s actually difficult to describe how good it feels when an AGA games runs perfectly on an OCS/ECS Amiga but each us might remember in the days how frustrating it was not being able to run them on our « Denised » miggies. This time is soon over.

Getting online from macOS8.1 with coffin r53 - Surfing with Icab.

https://www.youtube.com/watch?v=esvf4xpQa6M


X86 Power !

http://www.apollo-core.com/knowledge.php?b=2¬e=16335



GOLD 2.11 Coming

Quote:

GOLD 2.11 is in preparation. GOLD 2.11 will provide some improvements A) Elude Demos run better B) Partial protection for Bad Blitter programming code Which might fix some games. We like to thank Daniel for his brainstorming and thank Kiero for giving us insight in his Demo source.

Enforce 0.2 On the V4

https://www.youtube.com/watch?v=9NWNTNXZxUw&t

And for those that want this option, it seems like kolla is testing/researching a stable manner of disabling the FPU;

http://eab.abime.net/showthread.php?t=93488&page=2

@Overflow

80386DX performance in PCTask (?), nice!