• Please review our updated Terms and Rules here

Looking for volunteers to help test a new benchmark

That's really quite bizarre...? Try topbench -p -i -s -v -k and see if it will at least run with that.

200lx with 2mb ram: runtime error 200@114e:17de.
200lx with 1mb ram: reboots.

Both units have new backup and main batteries and pass full self-test diagnostics.
 
Even though this may not make the official database, I'd like to demonstrate the sort of speed you will get if you set a Tandy 1000 TX, TL or TL/2 to the slow mode :

[UIDB44476E6D]
MemoryTest=2393
OpcodeTest=1066
VidramTest=1656
MemEATest=1420
3DGameTest=1062
Score=7
CPU=Intel 80286
CPUspeed=4 MHz
BIOSinfo=unknown
BIOSdate=19880615
BIOSCRC16=B444
VideoSystem=CGA
VideoAdapter=Tandy 1000 SL/TL
Machine=Tandy 1000 TL (4MHz)
Description=
Submitter=Great Hierophant

I have a Make-it-486 Accelerator I will be installing in the machine soon and hope to report those benchmarks as well.
 
I've never had one in my possession, so I'm not surprised. Much of the CPU detection routines were from a 3rd-party library, and the rest (20%) were written by me based on the hardware I have/had available.

If you're willing to ship me a 486dx-120 system I would be happy to debug the code :) but for now, just use the -s option to skip the CPU detection and then write in the proper values when you edit the entry. (And I'd be curious to see that entry, since the only 120MHz entry in the database is an AMD x586 CPU -- is that what you have? That reminds me, I never benchmarked my own Pentium 120... I'll have to pull that out and add it to the database. Uses a 60Mhz bus instead of 66MHz...)
 
I've never had one in my possession, so I'm not surprised. Much of the CPU detection routines were from a 3rd-party library, and the rest (20%) were written by me based on the hardware I have/had available.

If you're willing to ship me a 486dx-120 system I would be happy to debug the code :) but for now, just use the -s option to skip the CPU detection and then write in the proper values when you edit the entry. (And I'd be curious to see that entry, since the only 120MHz entry in the database is an AMD x586 CPU -- is that what you have? That reminds me, I never benchmarked my own Pentium 120... I'll have to pull that out and add it to the database. Uses a 60Mhz bus instead of 66MHz...)

nah that wouldn't be practical.
Code:
[UID9D5AC8890]
MemoryTest=55
OpcodeTest=22
VidramTest=180
MemEATest=37
3DGameTest=17
Score=186
CPU=AM486DX4
CPUspeed=120 MHz
BIOSinfo=(C)1985-1994,American Megatrends Inc.,All Rights Reserved,6145F Northbelt Parkway,GA-30071,USA.(404)-263-8181. (10/10/94, rev. 0)
BIOSdate=19941010
BIOSCRC16=9D5A
VideoSystem=VGA
VideoAdapter=S3 ViRGE
Machine=Custom Built / Acer AP43 motherboard
Description=Homebrew 486 machine with Acer AP43 PCI motherboard
Submitter=smeezekitty
It isn't a 5x86. It says Am486 DX4-120
A80486DX4-120NV8T
on it
 
I was trying to run it with my IBM AT with an Intel Inboard 386 card just to see what happens, but when I launch it, I get a message that I'm not running in Real-Mode.
I'm running DOS 6.22. I understand you want a clean boot, but I booted normaly since there are device drivers for the Inboard Card. I ran the switch to run the program anyway, and I got results similar to the 8MHZ IBM AT that modem7 must have submitted in the database.
 
I was trying to run it with my IBM AT with an Intel Inboard 386 card just to see what happens, but when I launch it, I get a message that I'm not running in Real-Mode.

That's because of QEMM, EMM386, or similar drivers emulating EMS using protected mode. If you disable just that driver but keep the other drivers, the "not running in real mode" message should go away and your system will still be usable.
 
I removed the memory managers and did not get the message.
Odd when I have the 386 card set to the highest speed 4/4 I score 13 and when I have it at the lowest I score an 8.
I would have thought it would have been rated higher than a normal at. At the lowest it's compared to a commodore PC-10. Funny I used to have one of those.
 
I removed the memory managers and did not get the message.

Cool.

Odd when I have the 386 card set to the highest speed 4/4 I score 13 and when I have it at the lowest I score an 8.
I would have thought it would have been rated higher than a normal at.

You're still limited to the XT's 8-bit bus to access all the other peripherals (including the video card). So there are limitations to what any inboard-type card could do.
 
You sure can see a difference when it loads the database though.

Here's the slowest setting if you want to add it.

[UID35CBAF3D3]
MemoryTest=1519
OpcodeTest=760
VidramTest=2005
MemEATest=1019
3DGameTest=742
Score=8
CPU=Intel 80386DX
CPUspeed=8 MHz
BIOSinfo=61X9266 COPR. IBM 1981, 1985 PARITY CHECK 1 (11/15/85, rev. 0)
BIOSdate=19851115
BIOSCRC16=35CB
VideoSystem=EGA
VideoAdapter=EGA
Machine=PC-AT 319 or 339, 8MHz Speed Setting 1
Description=IBM 5170 with Intel 386/AT card installed slow speed
Submitter=offensive_jerk

somehow the fastest setting didn't save. I'll have to redo it.

What's that parity check 1 all about in the BIOS string? I don't have any error messages like that on boot.
 
What's that parity check 1 all about in the BIOS string? I don't have any error messages like that on boot.

The BIOS "copyright" string is just a linear search of the BIOS in a few areas looking for a copyright message. The actual code looks for:

Code:
CopyNotices:array[1..maxCopyStrings] of string[9] =
    ('(C)', 'COPR.', 'COPYRIGHT', 'ORACLE', 'TOSHIBA'); {Must be upper-case}

... and once it finds one of those strings, it works backwards until it finds a non-printable character, then works forwards until it finds a non-printable character, and takes whatever is inbetween them. Your BIOS just happens to have a PARITY error message right up against the copyright string. Nothing to worry about.
 
Quick question:

Your program detected 8 mhz on setting 1 and 4 on my 386 card. Is your program able to read the actual mhz of the CPU even if it's on a card like mine? Or, does the Intel speed setting probably turn some cache off or something like that, and keeps the speed of the CPU the same?
 
The speed measurements are determined by first getting the CPU model correct, then noting the current timer tick 16-bit value out of the PIT, then running a known series of instructions that behave consistently on that specific CPU model, then reading the timer tick value again to see how much time has gone by. There are things that can trip this process up, such as guessing the wrong CPU model, or the CPU "advertising" itself as the wrong model (more applicable to Cyrix and Via CPUs), or the entire system having bottlenecks not related to the CPU. Add-in card accelerators are vulnerable to the latter, which is affecting the speed measurement.

Put another way: Your 16 MHz 386 (what I'm assuming is on your inboard card) is indeed running at 16Mhz but, because of the other bottlenecks created using an accelerator board, it has the effective throughput of only running at 8MHz.
 
Trixter: the work you did here is fantastic, i've joined recently and only last week read and test about this program,
after reading the sources, wow! very good job sir!

i think this program will be more and more important as time passes.
 
Thanks :) It's already answered some of my long-standing questions, such as "just how much was an 8086 system faster than an 8088 system", "just how much speedup, and where, did an NEC V20/V30 bring to a system", "how did a 386dx-40 compare to a 486sx-25", etc. So I'm glad I did it, and I'm glad others on the forum helped.
 
Looks like there have been calls to run this on a MediaGX system. I'll try it out on my MediaGXi 180 MHz system and see what it reports.
 
Wacky numbers from a wacky system

Code:
MemoryTest=54918
OpcodeTest=5
VidramTest=38
MemEATest=54916
3DGameTest=54902
Score=402
CPU=Cyrix MediaGXi
CPUspeed=180 MHz
BIOSinfo=Compaq
BIOSdate=19970609
BIOSCRC16=0518
VideoSystem=VGA
VideoAdapter=Cyrix XpressGRAPHICS (MediaGX) 2MB VRAM (Shared)
Machine=Compaq Presario 2200
Description=Rare Cyrix MediaGX-based Compaq Presario 2200
Submitter= Rob Ivy
 
Last edited:
Back
Top