旼컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴� � Trademarked names are the property of the respective owners � 읕컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴� Copyright (c) 1989, 1990, 1991, F.M. de Monasterio Licensed Material - All rights reserved VIz version 4.16 ---------------- BENCHMARKS The data were obtained with the utilities HLTEXT.EXE and HLWINDOW.EXE of the PC TECH JOURNAL benchmarks 1.0 (c) 1988 Ziff Communications Co., BENCH24.EXE of the PC MAGAZINE benchmarks 5.0 (c) 1989 Ziff-Davis Corp., and VIZMARK.EXE version 2.22 (c) 1991 F.M. de Monasterio. These data represent the mean of 5 sequential runs of each utility, after the initial run (that was discarded), expressed as the factor in accelerating operations relative to the benchmark results obtained without a video accelerator. In other words, the higher the factor the shorter the actual benchmark times for a given test. The accelerator being tested was installed as the only resident program that intercepted the BIOS video interrupt in: 1. 25-Mhz, 486 computer with a 32-bit bus, an Award BIOS version 3.12.00d- 21, and a 16-bit nonplanar VGA adapter. 2. 25-Mhz, 386 computer with a 32-bit bus, a Phoenix BIOS version 1.10 10a, and an 8-bit nonplanar VGA adapter. 3. 16-Mhz, 386 computer with a 16-bit bus, a 1984-IBM AT BIOS, and a 16-bit nonplanar VGA adapter. 4. 10-Mhz, 8088 computer with an 8-bit bus, a 1988-EPSON BIOS, and an 8-bit nonplanar VGA adapter. A 25 row, 80 column screen was used in all machines. The 32-bit bus machines were run under IBM PC-DOS 3.30 (386) and MS-DOS 4.01 (486) while the other 2 machines were run under COMPAQ MS-DOS 3.31 (386 and 8088). All VIz versions being tested used word OUT addressing. The BIOS (video) was not remapped to faster RAM in the 386 and 486 machines; in fact, since use of a 386-specific memory manager can sometimes decrease video performance, these machines were tested without such a manager being installed. ------------------------------------------------------------------------------ 1. VIDEO ACCELERATION (relative to benchmarks without an accelerator): Bus-CPU/Mhz 32bit-486/25 32bit-386/25 16bit-386/16 8bit-8088/10 旼컴컴컴컴컴커 旼컴컴컴컴컴커 旼컴컴컴컴컴커 旼컴컴컴컴컴커 �486-VIz 4.16� �386-VIz 4.16� �286-VIz 4.16� �086-VIz 4.12� 郞袴袴袴袴袴袴拷 팠袴袴袴袴袴孤 팠袴袴袴袴袴孤 팠袴袴袴袴袴孤 팠袴袴袴袴袴孤 � Switches � � /Z- /Z+ � � /Z- /Z+ � � /Z- /Z+ � � /Z- /Z+ � 쳐컴컴컴컴컴컴캑 쳐컴컴컴컴컴캑 쳐컴컴컴컴컴캑 쳐컴컴컴컴컴캑 쳐컴컴컴컴컴캑 � HLTEXT.EXE � � � � � � � � � � BIOS scroll � � 2.95 2.95 � � 2.88 2.88 � � 2.83 2.83 � � 2.91 2.91 � � DOS scroll � � 2.57 3.75 � � 2.32 3.80 � � 2.23 5.29 � � 2.26 4.31 � � Misc scroll � � 2.97 2.97 � � 3.47 3.47 � � 3.27 3.27 � � 1.85 1.85 � � Wind scroll � � 5.21 5.21 � � 4.74 4.74 � � 3.65 3.65 � � 2.40 2.40 � � � � � � � � � � � � HLWINDOW.EXE � � 4.65 4.65 � � 4.31 4.31 � � 3.68 3.68 � � 2.21 2.21 � � � � � � � � � � � � BENCH24.EXE � � � � � � � � � � no scroll � �14.23 14.23 � � 8.26 8.26 � � 5.00 5.00 � �11.90 11.90 � � scrolling � � 4.16 4.16 � � 3.58 3.58 � � 3.42 3.42 � � 4.09 4.09 � � � � � � � � � � � 쳐컴컴컴컴컴컴캑 쳐컴컴컴컴컴캑 쳐컴컴컴컴컴캑 쳐컴컴컴컴컴캑 쳐컴컴컴컴컴캑 � VIZMARK.EXE � � � � � � � � � � DOS � � 4.16 8.88 � � 2.45 7.04 � � 1.95 11.05 � � 2.29 10.15 � � CON fast � � 6.11 6.11 � � 4.32 4.32 � � 3.17 3.17 � � 5.56 5.56 � � ANSI � � -NA- -NA- � � -NA- -NA- � � -NA- -NA- � � -NA- -NA- � � BIOS miscel � � 8.51 8.51 � � 5.08 5.08 � � 3.83 3.83 � � 3.35 3.35 � � no scroll � �15.46 15.46 � �10.36 10.36 � � 5.37 5.37 � � 8.73 8.73 � � scrolling � � 4.56 4.56 � � 3.75 3.75 � � 3.56 3.56 � � 4.49 4.49 � � � � � � � � � � � 읕컴컴컴컴컴컴켸 읕컴컴컴컴컴켸 읕컴컴컴컴컴켸 읕컴컴컴컴컴켸 읕컴컴컴컴컴켸 -------- 2. COMPARISONS WITH OTHER RESIDENTS (1): Computer 旼컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴커 16bit-386/16 � 286-VIz QUICKCRT PCKSCRN VSCREEN THEFAST WARP � � 4.16 (2) (3) (4) (5) (6) � 郞袴袴袴袴袴袴袴袴拷 팠袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴孤 � Resident (bytes) � � 912 1456 4592 1040 912 1296 � 쳐컴컴컴컴컴컴컴컴캑 쳐컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴캑 � Load switches � � /Z+ /F /A+/Q+ none none /S � 쳐컴컴컴컴컴컴컴컴캑 쳐컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴캑 � HLTEXT.EXE � � � � BIOS scrolling � � 2.83 2.65 2.41 1.52 0.96 1.00 � � DOS scrolling � � 5.29 2.03 1.92 1.38 0.97 1.78 � � Misc scrolling � � 3.27 1.95 2.83 1.91 2.01 1.00 � � Wind scrolling � � 3.65 2.66 2.43 2.52 2.81 1.00 � � � � � � HLWINDOW.EXE � � 3.68 2.49 2.29 2.38 2.58 1.00 � � � � � � BENCH24.EXE � � � � no scroll � � 5.00 4.17 3.33 3.49 0.94 0.99 � � scrolling � � 3.42 3.16 2.74 1.77 0.95 1.01 � � � � � 쳐컴컴컴컴컴컴컴컴캑 쳐컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴캑 � VIZMARK.EXE � � � � DOS � � 11.05 1.84 1.73 1.57 0.95 1.69 � � CON fast � � 3.17 2.87 2.53 2.07 0.94 1.00 � � ANSI � � -NA- -NA- -NA- -NA- -NA- -NA- � � BIOS miscellan � � 3.83 3.23 2.60 1.70 2.48 1.00 � � no scroll � � 5.37 4.21 3.24 3.65 0.91 1.01 � � scrolling � � 3.56 3.19 2.79 1.87 0.94 1.00 � � � � � 읕컴컴컴컴컴컴컴컴켸 읕컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴켸 郞袴袴袴袴袴袴袴袴拷 郞袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴拷 � Output factor (7)� � 4.51 2.87 2.57 1.98 1.45 1.12 � � Factor/kbyte (8)� � 5.06 2.02 0.57 1.95 1.63 0.93 � 突袴袴袴袴袴袴袴袴槁 突袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴槁 NOTES: (1) ZENO.COM (version 2.20) could not included in the above comparisons because its 'teletype' service garbled the DOS prompt/command line. (2) Copyright 1986 T. Overfield (version 1.00). (3) Copyright 1987-89 MultiSoft Corporation (version 1.05). (4) Copyright 1985-87 Golden Bow Systems (version 1.06). (5) Copyright 1986-87 Soft Programming Services (version 1.04). (6) Written by E. Tauck (version 1.00). Could not use the faster version WARPF.SYS or WARPVF.SYS because testing int-29h services via VIZMARK crashed the machine (these two CON device drivers themselves use int 29h for 'fast' output, but rely on the internal CON driver to handle this interrupt). WARP provides rudimentary ANSI support. (7) Sum of the acceleration factors divided by the 12 tests. (8) Mean video acceleration factor divided by the resident size (rounded to the next larger paragraph [16 bytes] boundary) in kbytes (1 kbyte = 1,024 bytes). This is an approximate index of the effectiveness of the resident code in implementing video acceleration. -------- 3. VIDEO ACCELERATION (relative to benchmarks with ANSI.SYS): Bus-CPU/Mhz 32bit-486/25 32bit-386/25 16bit-386/16 旼컴컴컴컴커 旼컴컴컴컴커 旼컴컴컴컴커 � 486-VIz � � 386-VIz � � 286-VIz � � 4.16 (1) � � 4.16 (2) � � 4.16 (3) � 郞袴袴袴袴袴袴袴袴拷 팠袴袴袴袴孤 팠袴袴袴袴孤 팠袴袴袴袴孤 � Resident (bytes) � � 1856 � � 1728 � � 1696 � 쳐컴컴컴컴컴컴컴컴캑 쳐컴컴컴컴캑 쳐컴컴컴컴캑 쳐컴컴컴컴캑 � Load switches � � /Z-/A+ � � /Z-/A+ � � /Z-/A+ � 쳐컴컴컴컴컴컴컴컴캑 쳐컴컴컴컴캑 쳐컴컴컴컴캑 쳐컴컴컴컴캑 � HLTEXT.EXE � � � � � � � � BIOS scrolling � � 2.95 � � 2.76 � � 2.83 � � DOS scrolling � � 2.83 � � 3.10 � � 2.70 � � Misc scrolling � � 2.97 � � 3.47 � � 3.27 � � Wind scrolling � � 5.21 � � 4.74 � � 3.62 � � � � � � � � � 쳐컴컴컴컴컴컴컴컴캑 쳐컴컴컴컴캑 쳐컴컴컴컴캑 쳐컴컴컴컴캑 � VIZMARK.EXE � � � � � � � � DOS � � 4.65 � � 8.14 � � 3.10 � � CON fast � � 6.60 � � 6.39 � � 6.41 � � ANSI � � 7.70 � � 6.62 � � 4.21 � � � � � � � � � � BIOS miscellan � � 8.51 � � 5.08 � � 3.83 � � no scroll � � 15.46 � � 8.92 � � 5.37 � � scrolling � � 4.56 � � 3.76 � � 3.56 � � � � � � � � � 읕컴컴컴컴컴컴컴컴켸 읕컴컴컴컴켸 읕컴컴컴컴켸 읕컴컴컴컴켸 NOTES: (1) Relative to ANSI.SYS distributed with MS-DOS (version 4.01). (2) Relative to ANSI.SYS distributed with IBM PC-DOS (version 3.30). (3) Relative to ANSI.SYS distributed with COMPAQ MS-DOS (version 3.31). -------- 4. ANSI COMPARISONS WITH OTHER RESIDENTS (1) Computer 旼컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴� 16bit-386/16 � 286-VIz ANSI.COM NANSI.SYS ZANSI.SYS ZAVT.SYS � � (2) (3) (4) (5) (6) � 郞袴袴袴袴袴袴袴袴拷 팠袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴� � Resident (bytes) � � 1696 2336 3168 2720 3536 � 쳐컴컴컴컴컴컴컴컴캑 쳐컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴� � Load switches � � /Z-/A+ FAST/B0 none none none � 쳐컴컴컴컴컴컴컴컴캑 쳐컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴� � VIZMARK.EXE � � � � DOS � � 3.10 1.80 2.42 2.84 2.53 � � CON fast � � 6.41 2.29 3.78 6.82 4.13 � � ANSI � � 4.87 1.88 3.13 3.43 3.60 � � � � � � BIOS miscellan � � 3.83 1.00 1.00 1.00 1.00 � � no scroll � � 5.37 1.00 1.00 1.00 1.00 � � scrolling � � 3.56 1.00 1.00 1.00 1.00 � � � � � 읕컴컴컴컴컴컴컴컴켸 읕컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴� 郞袴袴袴袴袴袴袴袴拷 郞袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴� � Output factor (7)� � 4.79 1.99 3.11 4.36 3.42 � � Factor/kbyte (8)� � 2.89 0.87 1.01 1.64 0.99 � 突袴袴袴袴袴袴袴袴槁 突袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴袴� NOTES: (1) Relative to ANSI.SYS distributed with COMPAQ MS-DOS (version 3.31). (2) Version 4.16; the resident accelerator uses 912 bytes of memory; ANSI plus Cursor service resident size is less than 800 bytes. Notice that DOS acceleration switch is off (/Z-). (3) Copyright 1989 Ziff-Davis Corp., Author: M.J. Mefford (version 1.20). Switch /B0 yields the smallest resident size; part of such a resident handles keyboard remapping services (not included in the benchmarks). Switch 'SLOW' generates an output slower than that of ANSI.SYS alone. Resident can be turned off and removed from memory. (4) Copyright 1986, D. Kegel (version 2.2). Includes a keyboard remapping service. Driver cannot be turned off or removed from memory. (5) Copyright 1986, 1987, T. Hanlin III (version 1.2). Does not include keyboard remapping services. Driver cannot be turned off or removed from memory. (6) Copyright 1989, 1990, Luns Tee (version 1.1). Also includes an Avatar level 0 handler. Used ZAVTNK.SYS, which is smaller than ZAVT.SYS and has the keyboard remapping service disabled. Driver cannot be turned off or removed from memory. (7) Sum of the acceleration factors for DOS, CON, and ANSI divided by 3. (8) Mean DOS/CON/ANSI output factor divided by the resident size (rounded to the next larger paragraph [16 bytes] boundary) in kbytes (1 kbyte = 1,024 bytes). This is an approximate index of the effectiveness of the resident code in implementing this video output subset. ------------------------------------------------------------------------------ [END]