New VAMPIRE-GFX Driver Beats CV64 and Picasso-IV | |
---|
|
---|
| | Gunnar von Boehn (Apollo Team Member) Posts 6263 13 Sep 2017 07:44
| The Apollo-team did great work on the new VAMPIRE-GFX driver. The new GFX driver is AMMX enabled and improves GFX-performance significantly. AMIGA GFX performance is now clearly lead by SAGA - VAMPIRE. Here you can see a comparison to other AMIGA GFX cards. CLICK HERE Lets thank Henryk and Philippe for this great work!
| |
| | Obetto Sannala
Posts 61 13 Sep 2017 07:50
| Hi Gunnar Thanks a lot for the information. Could you comment the results shown in the image? What screenshot shows what exactly? Thank you very much.
| |
| | Gunnar von Boehn (Apollo Team Member) Posts 6263 13 Sep 2017 07:56
| Obetto Sannala wrote:
| Could you comment the results shown in the image? What screenshot shows what exactly?
|
The picture combines 3 GFX cards. LEFT = Cybervision64 in an AMIGA-4000 with 68060 CPU Middle = Picasso IV in an AMIGA-4000 with 68060 CPU RIGHT = Amiga 600 with SAGA-VAMPIRE (68080 CPU) running GOLD 2.7 prerelease You see in each picture the scores of the test. For example WRITE CHUNK PIXEL Cybervision64 = 892 PICASSO IV = 637 SAGA VAMPIRE-2 = 6580For comparison AMIGA 4000T with 604e 200mhz/060/128mb mit CybervisionPPC scores: WriteChunkyPixels = 1097
Higher result is better. Behind the score is also printed a factor. For the PICASSO IV is in white in this row a "10.33" This means SAGA VAMPIRE is 10.33 times faster than PICASSO IV in this test.
| |
| | Steve Ferrell
Posts 424 13 Sep 2017 08:12
| Wow!
| |
| | Obetto Sannala
Posts 61 13 Sep 2017 08:13
| Thank you very much This is really, really impressive! Will the VAMPIRE GFX Driver be available for free?
| |
| | Edgar Fink
Posts 47 13 Sep 2017 08:23
| Very nice! Why do the versions of Picasso96 and Workbench differ between the benchmarks?
| |
| | Samuel Devulder
Posts 248 13 Sep 2017 08:30
| Strangely enough the ScreenToFront() score of the Vampire looks suspiciously low. We have 101 for PicassoIV, but only 60 for the Vampire. Is it related to screen sync ? (60 for 60fps display)
| |
| | Gunnar von Boehn (Apollo Team Member) Posts 6263 13 Sep 2017 08:39
| Samuel Devulder wrote:
| Strangely enough the ScreenToFront() score of the Vampire looks suspiciously low. We have 101 for PicassoIV, but only 60 for the Vampire. Is it related to screen sync ? (60 for 60fps display)
|
Yes this is synced to VBL. Thereby limited to 60 FPS.
| |
| | Samuel Devulder
Posts 248 13 Sep 2017 08:43
| Ok :) It also seem that "Draw() HL/VL" needs a bit of improvement since it's off the PicassoIV's score by ~19%. But the overall result is impressive. I suspect the x10 boost in WritePixelArray() is caused by the better memory management in Vampire (no need to go though a Z-III bus to move data between FastRam and VideoRam).
| |
| | Samuel Crow
Posts 424 13 Sep 2017 08:44
| Edgar Fink wrote:
| Very nice! Why do the versions of Picasso96 and Workbench differ between the benchmarks?
|
Apollo core uses the unregistered version of P96 since it's based on the open source UAEGFX driver. The others are commercial drivers so they have registered versions.Likewise the Vampire has a custom Kickstart in flash memory.
| |
| | Vojin Vidanovic
Posts 770 13 Sep 2017 09:08
| Gunnar von Boehn wrote:
| Lets thank Henryk and Philippe for this great work!
|
Its good to see improvements on SAGA internals and overall gfx front. Hope AROS 68k gfx driver will benefit it too. Mind it, my friendly advice is to keep the hardwork on, and to promote less what is coming (except in feature list on paper as it grows). Big "advertisment" do piss non Vamp users, but for much of our crowd, its good to be excited, but its not good once you not know when it will come to your home. While SAGA result is great mind those gfx cards were Cirrus Logic GD5446 and S3 86C764 Trio64
| |
| | Keith Matthews
Posts 39 13 Sep 2017 09:18
| Nice one Gunnar
| |
| | Mr Niding
Posts 459 13 Sep 2017 09:24
| Thanks Henryk and Philippe for this great work! :)
| |
|