becap@cs.mcgill.ca (Brian CAPSON) (03/05/91)
Hi fellow netters;
It seems that I've got a bit of bad news for those interested in the
new ATI VGA Wonder+ Card. When I read the ATI ads that stated the
Wonder+ was "faster than VRAM and up to 350% faster than other VGA
systems" I was intrigued. As an avid Windows user, when I heard that
I could upgrade my current board to the Wonder+, I was even more
interested. Luckily I was able to convince a local vendor to let me
take home the Wonder+ and try it out. The results I got were not
encouraging. I did some benchmark test on both card under identical
circumstances both in Windows and in DOS, and some of the results are
listed below. The results are in milliseconds (so lower is better),
and suggest that not only is the Wonder+ not as fast as they claim,
but is in fact SLOWER than the old VGA Wonder! One thing must be
mentioned; the Wonder+ card had 256k on it, while th Wonder had 512k
on board. This limited the modes available to the Wonder+ for testing,
but I don't think effected the speed of the card (I may be wrong,
_please_ correct me if I am). The windows tests were done in 386
enhanced mode, displaying 800x600x16 colours. They were done with
QEMM's ROM shadowing on, (using ATI's rambios.sys driver had negligible
effect on performance over the QEMM shadow) and used the latest drivers
availible from ATI (March 1, 1990). Being an avid user of ATI products
for quite a while, I am dissapointed with these results, and would ask
anyone with an explanation for them to come forward. Otherwise, ATI
has alot to answer for in terms of their marketing.
Machine: Gateway 2000 '386 20 MHz
DOS Mode Performance (PC Mag. Benchmark ver 5.6)
Wonder Wonder+
VIDEO PERFORMANCE TESTS:
Direct Screen Access 2.74 2.47
Teletype Without Scrolling 0.83 0.82
Teletype With Scrolling 1.93 2.47
---- ----
5.5 5.76
Windows (386 enhanced mode) Performance (PC Mag Windows Bench. ver 1.1)
Not all test are listed here, but they represent a fair sample.
DIALOG BOXES (Show & Destroy) 633.65 780.40
PATTERNS
12% GREY 57.59 76.69
24% GREY 57.66 75.28
36% GREY 57.69 77.00
48% GREY 57.59 76.93
60% GREY 57.69 77.03
72% GREY 57.69 76.97
84% GREY 59.59 77.00
96% GREY 57.69 77.00
Hatch Patterns 55.74 74.49
BITBLT ALIGNMENT
Source Aligned Destination Aligned:
32 x 32 0.84 0.97
64 x 64 1.56 2.58
128x128 5.86 9.76
256x256 18.12 32.38
Source Aligned Destination Not Aligned:
32 x 32 3.53 4.01
64 x 64 10.11 11.13
128x128 32.24 34.72
256x256 113.90 128.22
Source Not Aligned Destination Aligned:
32 x 32 3.76 4.28
64 x 64 10.53 11.51
128x128 34.12 37.58
256x256 115.06 128.38
Source Not Aligned Destination Not Aligned:
32 x 32 3.81 4.08
64 x 64 10.71 11.75
128x128 32.48 37.10
256x256 116.14 128.40
Total for all tests done: 3666.85 4210.48
- Brian Capson becap@cs.mcgill.ca
McGill University School of Computer Science
Montreal, Quebec) (03/06/91)
I'm sure (but not positive) that 256K would be slower than a 512K card. +------------------------------------------------------------------------------+ | Santanu Sircar BITNET: ssircar@umaecs.bitnet | | University of Massachusetts/Amherst INTERNET: ssircar@ecs.umass.edu | +------------------------------------------------------------------------------+
jlr1801@aim1.tamu.edu (Jeff Rife) (03/07/91)
In article <1991Mar5.035327.6809@cs.mcgill.ca> becap@cs.mcgill.ca (Brian CAPSON) writes: >Hi fellow netters; > >It seems that I've got a bit of bad news for those interested in the >new ATI VGA Wonder+ Card. When I read the ATI ads that stated the >Wonder+ was "faster than VRAM and up to 350% faster than other VGA >systems" I was intrigued. As an avid Windows user, when I heard that >I could upgrade my current board to the Wonder+, I was even more >interested. Luckily I was able to convince a local vendor to let me >take home the Wonder+ and try it out. The results I got were not >encouraging. I did some benchmark test on both card under identical >circumstances both in Windows and in DOS, and some of the results are > >Machine: Gateway 2000 '386 20 MHz > >DOS Mode Performance (PC Mag. Benchmark ver 5.6) > > Wonder Wonder+ >VIDEO PERFORMANCE TESTS: > Direct Screen Access 2.74 2.47 > Teletype Without Scrolling 0.83 0.82 > Teletype With Scrolling 1.93 2.47 > ---- ---- > 5.5 5.76 [windows benchmarks omitted] > >- Brian Capson becap@cs.mcgill.ca > McGill University School of Computer Science > Montreal, Quebec O.K., my $.02 says the Wonder+ is better, here's why: 1) 1024x768 non-interlaced support (some later versions of the Wonder also have this feature. 2) 70 Hz refresh in all video modes for no flicker. 3) 16-bit video memory and BIOS operation, even with a monochrome card installed. (I could *never* get my Wonder to do this) 4) Much smaller card for less power consumption, heat, etc. Lower chip count has actually lowered the large-quantity price, and the extra 256K is just two chips, and < $30 most places. 5) It is (up to) 350% faster than OTHER cards, not their own. After all, they are trying to make money. Now, here's my benchmark puzzle, with the first two tests conducted outside of Windows, and the second two in DOS boxes. Why is the direct to screen better in the Window'ed DOS attempt? But notice that QEMM pulls BIOS-type performance down, and Windows pulls it down farther. Test Machine: QEMM installed Computed CPU: 80386SX 16.0 Mhz Direct Screen Access 4.01 Teletype Without Scrolling 0.99 Teletype With Scrolling 3.13 ------------------------------------------------------------------ Test Machine: Without QEMM Computed CPU: 80386SX 16.0 Mhz Direct Screen Access 4.01 Teletype Without Scrolling 0.55 Teletype With Scrolling 2.63 ------------------------------------------------------------------ Test Machine: Exclusive DOS Computed CPU: 80386SX 16.0 Mhz Direct Screen Access 4.23 Teletype Without Scrolling 1.92 Teletype With Scrolling 4.01 ------------------------------------------------------------------ Test Machine: Window DOS Computed CPU: 80386SX 16.0 Mhz Direct Screen Access 2.20 Teletype Without Scrolling 12.46 Teletype With Scrolling 10.16 Everything here, even the benchmarks, is IMHO. -- Jeff Rife P.O. Box 3836 | "Because he was human; because he had goodness; College Station, TX 77844 | because he was moral they called him insane. (409) 823-2710 | Delusions of grandeur; visons of splendor; jlr1801@aim1.tamu.edu | A manic-depressive, he walks in the rain."
Hubert Lai <LAIH@QUCDN.QueensU.CA> (03/11/91)
In article <13136@helios.TAMU.EDU>, jlr1801@aim1.tamu.edu (Jeff Rife) says: >2) 70 Hz refresh in all video modes for no flicker. Can the new Wonder+ do 70 Hz refresh on my trusty old NEC Multisync II? That's not the same as the Multisync 2A or 3D. If it does, will it do it in both 800x600 and 1024x768 modes? <=- Hubert
george@hobbes.catt.ncsu.edu (03/12/91)
DOS performance aside, how's the Windows performance? Does it seem any faster when dragging windows, scrolling in Word, etc? Also, how much is the upgrade price from a VGA Wonder to a VGA Wonder +? I've got the V6 VGA Wonder board. Lastly, will the 70Hz refresh stop the subtle flickering I see on my NEC 3D in 800x600 (non-interlaced) mode? george -- ------------------------------------------------------------------------ | o | George Browning george@catt.ncsu.edu | o | | o | NC State University Raleigh, NC | o | ------------------------------------------------------------------------
jlr1801@aim1.tamu.edu (Jeff Rife) (03/12/91)
In article <91069.220813LAIH@QUCDN.QueensU.CA> LAIH@QUCDN.QueensU.CA (Hubert Lai) writes: >In article <13136@helios.TAMU.EDU>, jlr1801@aim1.tamu.edu (Jeff Rife) says: > >>2) 70 Hz refresh in all video modes for no flicker. > >Can the new Wonder+ do 70 Hz refresh on my trusty old NEC Multisync II? >That's not the same as the Multisync 2A or 3D. If it does, will it do >it in both 800x600 and 1024x768 modes? > ><=- Hubert I don't know. I've used it on various monitors, but none as old as the II. My assumption is yes, as it runs 70 Hz on both 1024x768 interlaced or non- interlaced monitors, as well as 800x600 modes. It does kill flicker quite effectively. One thing to try is to lie to your card. Try *all* of the monitors listed in VSETUP. See what works best. The most recent version of VSETUP has more choices as well. If people are interested, and there is agreement that it would not violate ATI's rights, I will post it to cica. -- Jeff Rife P.O. Box 3836 | "Because he was human; because he had goodness; College Station, TX 77844 | because he was moral they called him insane. (409) 823-2710 | Delusions of grandeur; visons of splendor; jlr1801@aim1.tamu.edu | A manic-depressive, he walks in the rain."