19-Apr-2024 04:51 GMT.		 UNDER CONSTRUCTION
Anonymous, there are 8 items in your selection
[News] Voodoo3 benchmarked on Mediator and G-RexANN.lu
Posted on 18-Mar-2001 18:56 GMT by Christian Kemp8 comments
View flat
View list
Czech Amiga News posted the results of a Mediator and G-REX benchmark test originally posted on the G-Rex mailing list by Olivier Martin. The tests, done with P96Speed 1.2 on a 68040/33 A1200 with Voodoo3 card place the Mediator as the clear winner in 19 out of 21 categories, ahead of the G-Rex by differences between 16% and 1012% (!).
Voodoo3 benchmarked on Mediator and G-Rex : Comment 1 of 8ANN.lu
Posted by Christophe Decanini on 17-Mar-2001 23:00 GMT
Are this tests really fair ?
I would like to see some different program tests on different configurations. What about some test on 608060 ? Some tests on morphos would also be interesting.
Voodoo3 benchmarked on Mediator and G-Rex : Comment 2 of 8ANN.lu
Posted by Jon Åslund on 17-Mar-2001 23:00 GMT
No, the tests are not fair. The G-REX system use CyberGfx3 and The Mediator use Picasso96. I get different result with my CV64/3D if I use CyberGfx3 or Picasso96. 1012% from DrawEllipse doesn't say much. I get a couple of 100% off between some tests. The most important thing for me is the "feel" of the speed in different "useful" situations. For me Picasso96 feels faster than CyberGfx, but then again, I haven't used CyberGfx 4.
Voodoo3 benchmarked on Mediator and G-Rex : Comment 3 of 8ANN.lu
Posted by Amifan on 17-Mar-2001 23:00 GMT
In reply to Comment 2 (Jon Åslund):
I've used CyberGFX4.1 with my BVision. It's a nice and fast GFX system. Now I use Picasso96 with my voodoo3, and the system feels a lot faster in the high resolution now....
But I can't judge wich is one is faster because I use a different RTG system on a faster GFX card :)
Comparing the Picasso96 voodoo drivers to CGFX ones is nonsens, because you're almost only testing the RTG systems speed ( and state of the driver which will improve undoubtly) because I don't think that bandwidth is such an issue on slow 68k CPU's.
The Mediator show it's limitations in PPC apps, because the Mediator suffers from the "amature" bus of the BlizzardPPC. But than again...This isn't an issue anymore on Next generation PowerPC accelerators like the shark because it's on the right side of the bottleneck. A SharkPPC with 68k emulation and a warpOS driver will outperform a BlizzardPPC + G-REX anytime in both busperformance and GFX performance. You can expect the busperformance to be at least 3 times faster then the G-REX. But then again, we don't know anything about the software which is running on the shark.
Elbox??
Testing performance on MorphOS vs. Picasso 68k is as useless as comparing Picasso96 68k running on top of blackbox (Haage&Partners 68k emulation) on a 500Mhz G3 vs. the MorphOS/CybergfxPPC combo...
Voodoo3 benchmarked on Mediator and G-Rex : Comment 4 of 8ANN.lu
Posted by Alkis Tsapanidis on 17-Mar-2001 23:00 GMT
Oh come on guys this guy has no idea about what he is talking about.
Come on he is testing a CGX driver with a P96 benchmark. P96Speed
has cgx 3 support but not CGX 4. CGX 4 defaults in fast ram. Every
bitmap is in fastram except it is specified by the program while P96 puts
everything in VRAM unless it doesn't fit. Blitting from fast is much
slower than in vram. Try another benchmark like AmiGOD. P96Speed also
says that my BV is much slower than a P96 while all other bechmark
programs state overwise.
Here are results from wspeed:
WSpeed © 1994 JAMI Soft Development
Description:
Screenmode: Voodoo3: 16Bit 1024 x 768, 0 Colors
----------------------------------------
Put Pixels 843067
Draw Lines 33835
Draw Hor/Ver 144995
Draw Circles 95472
Draw Ellipse 6097
Draw Boxes 96544
Scroll X 14462
Scroll Y 14973
Print Texts 30152
CON: Output 3763
Open Windows 248
Size Windows 428
Move Windows 136
Swap Screens 500
Areafill 596
----------------------------------------
I think that neither this is right. Actually the only difference
between the busboards is the driver. The bus of the boards is equal
if you use the medi on apollo boards while it is slower on blizzard
PPC boards.Do you know ANY way to get faster results from the same
card with an equal bus speed except changing the driver? NO. so....
Voodoo3 benchmarked on Mediator and G-Rex : Comment 5 of 8ANN.lu
Posted by Christophe Decanini on 17-Mar-2001 23:00 GMT
In reply to Comment 3 (Amifan):
I never said that a MorphOS test would be good for a comparison. i just said it would be interesting in addition of some fair comparison tests.
Voodoo3 benchmarked on Mediator and G-Rex : Comment 6 of 8ANN.lu
Posted by victor # on 18-Mar-2001 23:00 GMT
In reply to Comment 1 (Christophe Decanini):
These tests simply shows the actual case, using the currently available drivers.. BTW, I haven't seen bustest results with the G-Rex yet.
Voodoo3 benchmarked on Mediator and G-Rex : Comment 7 of 8ANN.lu
Posted by Nian on 18-Mar-2001 23:00 GMT
Well all these speed tests are interesting but to see real world tests,
can someone please use SysSpeed.
Then we will have an idea of the real speed diff.
Voodoo3 benchmarked on Mediator and G-Rex : Comment 8 of 8ANN.lu
Posted by Amifan on 18-Mar-2001 23:00 GMT
In reply to Comment 6 (victor #):
I've seen some busspeed tests. 68K ofcourse, because bustest can't test the PPC <-> G-REX speed and I'm sure that I don't suprise you when I say that they are exactly the same as a BVision.....(no suprise because the G-REX uses the same PPC bridge, and the 040/33Mhz is the limitating factor).
Bustest shows 14MB/sec on a 040/33 BlizzardPPC to the G-REX, just like to the BVision.
Anonymous, there are 8 items in your selection
Back to Top