yon%smltalk.cbm.dec.com@decwrl.dec.com (David A. Yon) (05/23/91)
In a word...
WW WW WW OOOOOOOO WW WW WW !!
WW WW WW OO OO WW WW WW !!
WW WW WW OO OO WW WW WW !!
WW WW WW OO OO WW WW WW !!
WW WW WW OO OO WW WW WW !!
WW WWWW WW OO OO WW WWWW WW
WW WW OOOOOOOO WW WW !!
I got my new 8514/Ultra last night and threw it into the following system:
386/33 64k cache
8 meg
130 meg IDE
Windows 3.0
DOS 4.01
4DOS 3.02
"Supercom" 1024x768 interlaced .28 pitch
ATI VGA Wonder 512k/v3.0 BIOS
I've been running Windows using the ATI 800x600x16 color driver, which on
my 386/33 operated quite reasonably. At least compared to my 286/12 system
from which I upgraded only seven days ago. This did not prepare me for the
boost that the Ultra gives...
I put it in and connected the VGA card using the pass-through connector
(their Ultra card which has a built-in VGA controller does not even have a
release date yet, so you *need* a VGA card in you system in *addition* to
the Ultra). I went through the configuration, telling it about my monitor,
and installed the basic Windows driver.
Windows booted at 1024x768x256, and WOW! What a difference! The old
800x600x16 driver was crisp, but this thing makes Windows *electric*!
Iconizing a window causes the background to be *instantaneously* blitted
over the area the window occupied! Pull-down menus and dialog boxes all
come up extremely fast and are painted with lightning speed! Moving
windows around the screen is an instantaneous operation. And this is at
256 colors! Even the VGA-wonder 16-color driver doesn't even come *close*
to this kind of response.
They also have an alternate Windows driver that uses something called
"CRYSTAL fonts", which use anti-aliasing to produce a simulated 400 dpi
image on the monitor. It ships with thirteen *.PFM files (perhaps Adobe
Type-1 outlines being used here? The ATI rep didn't know), which the
display driver uses. I installed the CRYSTAL driver, and rebooted Windows.
This time my jaw *really* dropped to the floor! I loaded a document into
Ami and wonder how ATI managed to retrofit a laser printer into my monitor!
It really must be seen to be believed. Even the system fonts give the
quite convincing illusion of being fully-formed characters. Text that used
to make me lean forward and squint could now be easily read from a normal
viewing distance, even the text under the icons (which is tiny in 1024x768)
was quite easy to read. The only problem I had was that the metrics of the
CRYSTAL fonts did not match the metrics of the printer fonts I was using,
but I imagine that ATI will release better support for these fonts in the
future. Doing the aliasing also eats away at the color palette, but unless
I want to use WinGIF, it doesn't really bother me.
Earlier in the week I had run PC-Magazine Windows Benchmark on my VGA
Wonder at 800x600x16, and then ran it again last night using the Ultra.
I've combined the results into one file (courtesy Emacs), where the first
timing is the VGA Wonder, and the second timing is the 8514/Ultra. The
results are at the end of this posting.
I'm going to try it in my roommate's 286/12 tonight, run some more
benchmarks, and maybe even try AutoCAD if we have time. Maybe for laughs
I'll compare the VGA Wonder at 800x600x256 to the Ultra!
More to come...
David Yon
yon%smltalk.cbm.dec.com@decwrl.dec.com
yon@world.std.com
--------------------------------
PCLabs Windows Benchmark version 1.1
Fri May 17 23:01:13 1991
Iteration count = 5 5
Window size = 800 x 562 1024x714
# Planes = 4 1
# Bits/Pixel = 1 8
Benchmark Results
BITBLT ALIGNMENT
Source Aligned Destination Aligned:
32 x 32 0.48 ms 0.19 ms
64 x 64 1.24 ms 0.20 ms
128x128 4.20 ms 0.57 ms
256x256 16.06 ms 2.82 ms
Source Aligned Destination Not Aligned:
32 x 32 2.59 ms 0.16 ms
64 x 64 7.36 ms 0.20 ms
128x128 24.07 ms 0.60 ms
256x256 86.50 ms 2.96 ms
Source Not Aligned Destination Aligned:
32 x 32 2.63 ms 0.18 ms
64 x 64 7.51 ms 0.27 ms
128x128 24.38 ms 0.86 ms
256x256 87.26 ms 3.25 ms
Source Not Aligned Destination Not Aligned:
32 x 32 2.80 ms 0.21 ms
64 x 64 7.72 ms 0.23 ms
128x128 25.09 ms 0.88 ms
256x256 88.46 ms 3.28 ms
BITBLT OVERLAP
Horizontal Overlap: 1.26 ms 0.20 ms
Vertical Overlap: 1.25 ms 0.22 ms
BITBLT ROP'S (average time per call)
Screen to Screen:
SRCCOPY 3.56 ms 0.22 ms
SRCPAINT 8.97 ms 0.28 ms
SRCAND 8.86 ms 0.28 ms
SRCINVERT 9.08 ms 0.28 ms
SRCERASE 10.24 ms 0.28 ms
NOTSRCCOPY 7.20 ms 0.11 ms
NOTSRCERASE 10.43 ms 0.28 ms
MERGECOPY 7.69 ms 0.93 ms
MERGEPAINT 9.75 ms 0.28 ms
PATCOPY 2.07 ms 0.66 ms
PATPAINT 10.32 ms 1.79 ms
PATINVERT 7.58 ms 0.67 ms
DSTINVERT 1.36 ms 0.19 ms
ZERO 1.18 ms 0.08 ms
ONE 1.27 ms 0.30 ms
Memory to Screen:
SRCCOPY 0.85 ms 0.62 ms
SRCPAINT 6.22 ms 0.41 ms
SRCAND 6.44 ms 0.52 ms
SRCINVERT 6.33 ms 0.63 ms
SRCERASE 7.63 ms 0.41 ms
NOTSRCCOPY 4.46 ms 0.52 ms
NOTSRCERASE 7.60 ms 0.52 ms
MERGECOPY 5.07 ms 1.23 ms
MERGEPAINT 7.20 ms 0.41 ms
PATCOPY 1.96 ms 0.66 ms
PATPAINT 7.84 ms 1.79 ms
PATINVERT 7.58 ms 0.66 ms
DSTINVERT 1.25 ms 0.19 ms
ZERO 1.41 ms 0.19 ms
ONE 1.16 ms 0.30 ms
Memory to Memory:
SRCCOPY 1.16 ms 2.96 ms
SRCPAINT 1.97 ms 5.01 ms
SRCAND 1.87 ms 4.90 ms
SRCINVERT 1.98 ms 4.90 ms
SRCERASE 2.31 ms 5.16 ms
NOTSRCCOPY 1.70 ms 3.22 ms
NOTSRCERASE 1.96 ms 5.15 ms
MERGECOPY 2.45 ms 3.57 ms
MERGEPAINT 1.99 ms 5.30 ms
PATCOPY 1.61 ms 3.06 ms
PATPAINT 2.95 ms 6.07 ms
PATINVERT 2.25 ms 5.00 ms
DSTINVERT 1.63 ms 4.64 ms
ZERO 1.13 ms 4.89 ms
ONE 1.02 ms 4.99 ms
SCREEN/MEMORY BITBLT (64x64 mono)
Memory To Screen: 1.10 ms 0.49 ms
Screen To Screen: 7.56 ms 0.26 ms
PARTIAL ARCS
Major Axis X 20.37 ms 14.73 ms
Major Axis Y 22.57 ms 16.54 ms
FULL ARCS
Major Axis X 126.42 ms 75.60 ms
Major Axis Y 101.08 ms 59.89 ms
SINGLE LINES
Horizontal 0.40 ms 0.18 ms
Vertical 1.27 ms 0.17 ms
Diagonal 3.21 ms 0.17 ms
POLYLINES
Horizontal 0.48 ms 0.14 ms
Vertical 2.22 ms 0.21 ms
Diagonal 6.04 ms 0.24 ms
LINES ALL DIRECTIONS
Long 0.98 ms 0.18 ms
WIDE LINES
Width 3 17.51 ms 12.64 ms
Width 5 20.47 ms 13.09 ms
Width 10 20.19 ms 16.47 ms
Width 20 23.99 ms 14.28 ms
AREA ALTERNATE FILL 80.76 ms 36.00 ms
AREA WINDING FILL 124.99 ms 62.54 ms
RECTANGLE BOUNDRY & INTERIOR
Square Corners 3.80 ms 0.72 ms
Rounded Corners 16.80 ms 7.23 ms
RANDOM RECTANGLES 7.21 ms 2.37 ms
PATTERNS
12% GREY 54.93 ms 12.12 ms
24% GREY 54.21 ms 24.73 ms
36% GREY 54.99 ms 24.73 ms
48% GREY 54.60 ms 25.10 ms
60% GREY 54.60 ms 24.73 ms
72% GREY 54.61 ms 24.73 ms
84% GREY 54.97 ms 24.36 ms
96% GREY 54.61 ms 25.10 ms
Hatch Patterns 54.23 ms 12.59 ms
TEXT - IMAGE FONTS
System Font 1.83 ms 2.18 ms
TmsRmn 21pt Font 3.62 ms 2.13 ms
TmsRmn 27pt Font 4.01 ms 2.62 ms
Helvetica 20pt Font 3.30 ms 1.92 ms
Helvetica 24pt Font 3.99 ms 2.29 ms
TEXT - BACKGROUND MIXED
LeaveAlone 5.13 ms 2.56 ms
OverPaint 1.83 ms 2.14 ms
GREYED TEXT (27 Characters) 4.09 ms 3.71 ms
MIXED FONTS 3.45 ms 2.96 ms
SET/QUERY CURRENT POSITION
Set 0.01 ms 0.01 ms
Query 0.01 ms 0.01 ms
STRETCH BLT
32 x 32 5.01 ms 5.79 ms
43 x 48 9.84 ms 9.27 ms
80 x 80 14.49 ms 13.45 ms
96 x 96 15.82 ms 14.40 ms
192x192 23.77 ms 20.51 ms
32 x 48 7.51 ms 7.13 ms
48 x 96 12.25 ms 11.89 ms
48 x 32 7.06 ms 5.92 ms
108x 96 17.43 ms 15.65 ms
21 x 21 3.12 ms 2.85 ms
105x105 17.21 ms 15.68 ms
IMAGE DATA 0.44 ms 0.41 ms
DIALOG BOXES (Show & Destroy) 351.49 ms 230.63 ms
RUBBER BANDING 1.71 ms 0.51 ms
SCROLL WINDOW
Byte Aligned 14.17 ms 5.42 ms
Not Aligned 32.17 ms 5.32 ms
ERASE WINDOW 58.82 ms 12.40 ms
RECTANGULAR CLIPPING
BitBlt 0.32 ms 0.25 ms
Lines 0.45 ms 0.25 ms
Text 0.58 ms 1.94 msplim@hpsgwp.sgp.hp.com (Peter Lim) (05/24/91)
/ roell@informatik.tu-muenchen.de (Thomas Roell) / 9:43 pm May 23, 1991 / writes: $ >Has anybody had any experience with the new ATI 8514/Ultra card? The $ >magazine ad I read about it claimed it was up to 5 times faster than $ >even a TI 34010 equipped card. $ $ I didn't try out the ATI 8514/Ultra, but many different other 8514 clones. $ And the result was that a normal board outperformed all these highend $ boards with a 32MHz 34020. Note that this test didn't include CAD or other $ This is a little confusing. Are you saying the "32MHz 34020 boards" are the "highend boards" ? or that ATI 8514/Ultra is one of the "highend boards" ? In my knowledge, the ATI Ultra/8514 cost around US$500 a pop (or is it cheaper by street price ?). .... I am just waiting for the dust to settle before I try one. Regards, ___o``\________________________________________________ ___ __ _ _ Peter Lim. V````\ @ @ . .. ... .- -> 76 MIPS at under US$20K !! --- -- - - /.------------------------------------------------ === == = = >--_// . .. ... .- -> 57 MIPS at under US$12K !! `' . If you guessed SUN, IBM or DEC, you are wrong ! E-mail: plim@hpsgwg.HP.COM Snail-mail: Hewlett Packard Singapore, Tel: (065)-279-2289 (ICDS, ICS) Telnet: 520-2289 1150 Depot Road, Singapore 0410. #include <standard_disclaimer.hpp>
roell@informatik.tu-muenchen.de (Thomas Roell) (05/27/91)
>This is a little confusing. Are you saying the "32MHz 34020 boards" are the >"highend boards" ? or that ATI 8514/Ultra is one of the "highend boards" ? Sorry, I wanted just to say that a ATI 8514/Ultra is for GUI applications faster than the '32MHz 34020' boards. - Thomas -- _______________________________________________________________________________ E-Mail (domain): roell@lan.informatik.tu-muenchen.de UUCP (if above fails): roell@tumult.{uucp | informatik.tu-muenchen.de} famous last words: "diskspace - the final frontier..."