• Please review our updated Terms and Rules here

Looking for volunteers to help test a new benchmark

I upgraded my Pentium Pro 200 "Tyan" motherboard computers to Pentium Pro Overdrive 333MHz chips long ago. This was basically a Pentium II core with integrated L2 cache. I've used these computers for almost 20 years, are very solid.

[UID670913DCAF]
MemoryTest=10
OpcodeTest=5
VidramTest=228
MemEATest=4
3DGameTest=3
Score=210
CPU=Intel Pentium Pro OverDrive
CPUspeed=333 MHz
BIOSinfo=war,Copyright (C) 1984-98, Award Software, Inc. (05/08/98, rev. 0)
BIOSdate=19980508
BIOSCRC16=6709
VideoSystem=VGA
VideoAdapter=VGAOAK OTI-087, VESA, 256kb Video Memory, 2048kb Video Memory (V
Machine=AT clone
Description=
Submitter= BrianS

I still have the Pentium Pro 200's, 256KByte and 512KByte cache chips in a drawer. The Voltage regulator also had to be changed out with the Overdrive chip. The upgrade improved performance of real-mode code dramatically, these systems were mostly used to run custom data acquisition code that ran under 32-bit PharLap extended DOS, Microway NDP compilers. I just like them, have lots of PCI slots.
 
Last edited:
Found another couple of quirks, although interestingly, I've found a review of the machine that made the same mistake identifying the graphics controller. Unless it really is a Tseng chip with a Chips BIOS? (edit: nope, there's a teardown showing it's definitely Chips: https://www.youtube.com/watch?v=NMTkyObiYeI)

Also, the weird configuration I'm running should be functionally identical to an OmniBook 425 - as far as I'm aware, the only differences between a 425 and 430 were the silkscreening on the case, the application shortcut key combos on the case, the firmware card, and the hard drive having a preload on the 430 (instead of being loaded from the firmware card on the 425).

The device boots showing the following:

Code: 
Chips 65510 VGA BIOS
Version 1.4.0
DECOMPILATION OR DISASSEMBLY PROHIBITED
Copyright (C) 1992 Chips and Technologies, Inc.  All Rights Reserved.

The other quirk is with the CPU speed detection - it's rounding to 33 MHz, when the actual speed is 25. So, the result's being submitted as a -r speed.

[UID6953993EB]
MemoryTest=250
OpcodeTest=109
VidramTest=448
MemEATest=155
3DGameTest=91
Score=49
CPU=Cyrix 486SLC/e
CPUspeed=25.51302 MHz
BIOSinfo=Copyright (c) 1985,1986,1987 Phoenix Technologies Ltd (07/31/93, rev. 0)
BIOSdate=19930731
BIOSCRC16=6953
VideoSystem=VGA
VideoAdapter=VGA, Chips&Technologies F65510, DRAM, 256kb Video Memory (BIOS)
Machine=Hewlett-Packard OmniBook 430
Description=OmniBook 430 running OmniBook 425 1.1S ABB firmware. Video misdetected as "Tseng ET", but video BIOS clearly states that a C&T 65510 is present on boot.
Submitter=Eric Rucker <bhtooefr@bhtooefr.org>
 
Last edited:
Looking through the source, I think I see the problem for video detection - DETECTGR.PAS doesn't bail out when it gets a hit. So, most likely, what's happening is it's actually detecting a Chips F65510 as it should, but then it keeps testing... and hits the Tseng loop, where the test probes it and determines that it is a Tseng of some sort, setting "Tseng ET ". At that point, it can't find any valid model (not actually being a Tseng chipset at all), so it doesn't add the string. Basically, I'm guessing there's a certain point in the Chips test, where if it passes, it should bail out...

I could maybe see skipping the Tseng test with a break statement if anything else has passed, maybe, to avoid false positives? The Tseng test doesn't look foolproof...
 
No test is foolproof.

The graphics detection isn't actually the fix I was talking about; I think the CPU rounding 25.5 up to 33 is something that can be addressed. To address the Tseng mis-detection, I'd need access to your system to debug it.
 
I can't seem to save the results to a named file. I copied the 3 TopBench files to my HDD in a directory named TOPBENCH, then made that my current directory C:\TOPBENCH, then ran it each of two different ways:

TOPBENCH FN.FT -l (also tried -i)

TOPBENCH -l (also tried -i) FN.FT

I did get results (Score 30 on my PC/XT with and Intel InBoard and a Cyrix 486 upgrade running at an effective 33MHz), but neither generated an output file with the specified file name and file type. I am sure its user error :)

Regards,
Mike
 
I'm not sure what you're trying to do. If you just want to send me results, you can either email me the entire DATABASE.INI file after your additions are in it, or you can cut'n'paste your specific entries here and I'll see them.
 
Trixter said:
I'm not sure what you're trying to do. If you just want to send me results, you can either email me the entire DATABASE.INI file after your additions are in it, or you can cut'n'paste your specific entries here and I'll see them.
Click to expand...

How can I add the results when no file containing the results is generated?

Regards,
Mike
 
You have to use the menus if you want to write the results to database.ini. The -i and -l options are just for display and don't write to files. With some software, you can add > file.txt to the command line and capture text output but I don't think topbench uses stdout, so it won't work with topbench, not that it needs to since using the menus is the proper way.
 
Yo, just a heads-up -- doing a clean boot off a DOS 6.22 floppy on a Gateway 4DX2-50V (The one that you passed on to me, funnily enough), the Database Ops -> Add this System option seems to cause a machine hang at the Detecting: CPU speed... phase.

Thus far, it's recognized a PS/2 Model 25, a HP 100lx and a AT&T 6300 correctly. Need to dig out my iTronix T2000 EFP and run it on that!

(I've referred the program to a friend of mine, by the way -- he's submitted his Aptiva 330!)
 
Odd, since there is another 4DX2-50 in the database, I'm almost certain. You can try -r for disabling the MHz rounding, and if it's truly hung, use -s to skip the CPU detection tests. TOPBENCH will work, but when you add your system to the database, you'll have to fill in the CPU details before committing the information. Also, a clean boot may help.
 
Here is one more system:

[UID77F313C2F5]
MemoryTest=1851
OpcodeTest=878
VidramTest=1346
MemEATest=967
3DGameTest=932
Score=8
CPU=Intel 8088
CPUspeed=9.54 MHz
BIOSinfo=EURO PC BIOS V2.08B
BIOSdate=19890411
BIOSCRC16=77F3
VideoSystem=MDA
VideoAdapter=Hercules/CGA on board
Machine=Schneider Euro PC (PC/XT)
Description=set to "9.54 MHz" and "Mono"
Submitter=RobertK (VCF)

The Schneider Euro PC was quite popular in Europe (and it still is among retro computerists) because its design made it look like a "home computer".
 
Cool, always wanted a EuroPC in the list :) Can you confirm something for me, though? I've never run across an 8088 running at 9.54 MHz, can you confirm it is an 8088, or is it an 8086? Also, can you re-run topbench with "-r" to double-check if it's really running at 9.54 MHz? (the -r is "rounding" and it is sometimes buggy)
 
Sure, I will check that for you the next time I have everything set up again (I would love to have the space to keep all my systems set up all the time :)).

The CPU is definitively an 8088. Here in the 10th picture you can read "SIEMENS SAB 8088-1-P" on it.
And if my Euro PC for some reason had an 8086 instead of 8088 I would have certainly noticed this when I had the case open to replace the battery (because for me the 8088 has always been the standard XT CPU, and the 8086 was something "unusual").
 
Commodore PC40-III

Commodore PC40-III

Here's mine:

With the builtin VGA active:

Code: 
[UIDEE0A147A4B]
MemoryTest=615
OpcodeTest=264
VidramTest=737
MemEATest=345
3DGameTest=260
Score=22
CPU=Intel 80286
CPUspeed=12 MHz
BIOSinfo=80286 BIOS, COPYRIGHT (C) 1988 Commodore Business Machines (08/25/88, rev. 0)
BIOSdate=19880825
BIOSCRC16=EE0A
VideoSystem=VGA
VideoAdapter=VGA, Paradise PVGA1A, 8-Bit-Video, 8-Bit-ROMfixed-sync, 256kb Vi
Machine=Commodore PC40-III @ 12 MHz VGA
Description= 
Submitter=kolrabi

[UIDEE0A148302]
MemoryTest=939
OpcodeTest=402
VidramTest=821
MemEATest=528
3DGameTest=397
Score=16
CPU=Intel 80286
CPUspeed=8 MHz
BIOSinfo=80286 BIOS, COPYRIGHT (C) 1988 Commodore Business Machines (08/25/88, rev. 0)
BIOSdate=19880825
BIOSCRC16=EE0A
VideoSystem=VGA
VideoAdapter=VGA, Paradise PVGA1A, 8-Bit-Video, 8-Bit-ROMfixed-sync, 256kb Vi
Machine=Commodore PC40-III @ 8 MHz VGA
Description= 
Submitter=kolrabi

[UIDEE0A1489C9]
MemoryTest=1263
OpcodeTest=540
VidramTest=949
MemEATest=719
3DGameTest=533
Score=12
CPU=Intel 80286
CPUspeed=6 MHz
BIOSinfo=80286 BIOS, COPYRIGHT (C) 1988 Commodore Business Machines (08/25/88, rev. 0)
BIOSdate=19880825
BIOSCRC16=EE0A
VideoSystem=VGA
VideoAdapter=VGA, Paradise PVGA1A, 8-Bit-Video, 8-Bit-ROMfixed-sync, 256kb Vi
Machine=Commodore PC40-III @ 6 MHz VGA
Description= 
Submitter=kolrabi

With my Hercules card active:

Code: 
[UIDEE0A1452EF]
MemoryTest=614
OpcodeTest=263
VidramTest=792
MemEATest=346
3DGameTest=260
Score=22
CPU=Intel 80286
CPUspeed=12 MHz
BIOSinfo=80286 BIOS, COPYRIGHT (C) 1988 Commodore Business Machines (08/25/88, rev. 0)
BIOSdate=19880825
BIOSCRC16=EE0A
VideoSystem=MDA
VideoAdapter=Hercules Graphics Card
Machine=Commodore PC40-III @ 12 MHz HGC
Description=
Submitter=kolrabi

[UIDEE0A144471]
MemoryTest=939
OpcodeTest=401
VidramTest=993
MemEATest=528
3DGameTest=397
Score=16
CPU=Intel 80286
CPUspeed=8 MHz
BIOSinfo=80286 BIOS, COPYRIGHT (C) 1988 Commodore Business Machines (08/25/88, rev. 0)
BIOSdate=19880825
BIOSCRC16=EE0A
VideoSystem=MDA
VideoAdapter=Hercules InColor Card
Machine=Commodore PC40-III @ 8MHz HGC
Description= 
Submitter=kolrabi

[UIDEE0A144CDE]
MemoryTest=1263
OpcodeTest=540
VidramTest=1041
MemEATest=719
3DGameTest=534
Score=12
CPU=Intel 80286
CPUspeed=6 MHz
BIOSinfo=80286 BIOS, COPYRIGHT (C) 1988 Commodore Business Machines (08/25/88, rev. 0)
BIOSdate=19880825
BIOSCRC16=EE0A
VideoSystem=MDA
VideoAdapter=Hercules InColor Card
Machine=Commodore PC40-II @ 6 MHz HGC
Description= 
Submitter=kolrabi
 
Trixter said:
Cool, always wanted a EuroPC in the list :) Can you confirm something for me, though? I've never run across an 8088 running at 9.54 MHz, can you confirm it is an 8088, or is it an 8086? Also, can you re-run topbench with "-r" to double-check if it's really running at 9.54 MHz? (the -r is "rounding" and it is sometimes buggy)
Click to expand...

Ok, I have re-run it now, and this time I have left the result untouched (except for the machine name):

[UID77F313D609]
MemoryTest=1851
OpcodeTest=878
VidramTest=1340
MemEATest=967
3DGameTest=926
Score=8
CPU=Intel 8088
CPUspeed=10 MHz
BIOSinfo=unknown
BIOSdate=19890411
BIOSCRC16=77F3
VideoSystem=MDA
VideoAdapter=Hercules InColor Card
Machine=Schneider Euro PC (PC/XT) set to "9.54 MHz" and "Mono", without -r
Description=
Submitter=

[UID77F313C597]
MemoryTest=1851
OpcodeTest=878
VidramTest=1353
MemEATest=967
3DGameTest=926
Score=8
CPU=Intel 8088
CPUspeed=10.0406 MHz
BIOSinfo=unknown
BIOSdate=19890411
BIOSCRC16=77F3
VideoSystem=MDA
VideoAdapter=Hercules InColor Card
Machine=Schneider Euro PC (PC/XT) set to "9.54 MHz" and "Mono", with -r
Description=
Submitter=


Previously I had corrected CPUspeed to 9.54 MHz, I probably wasn't supposed to do this...

The other additions/corrections were:

BIOSinfo=EURO PC BIOS V2.08B
VideoAdapter=Hercules/CGA on board
Machine=Schneider EuroPC (PC/XT)
Description=set to "9.54 MHz" and "Mono"
Submitter=RobertK (VCF)

I don't know how your program detected that a "Hercules InColor" card would be present. The Euro PC definitvely has Hercules and CGA graphics on board (switchable in the BIOS setup). I don't know much about Hercules InColor, but as I have seen here, it should work with an EGA monitor, but there is no such option in the BIOS setup.
 
RobertK said:
Previously I had corrected CPUspeed to 9.54 MHz, I probably wasn't supposed to do this...
Click to expand...

No, actually you were. I want accurate results in the database, so the correct MHz should be there regardless of what the program detects. Thanks for the clarification.

I don't know how your program detected that a "Hercules InColor" card would be present.
Click to expand...

There are certain bits that do or do not change in the vertical retrace register based on what real Hercules card is present. It is possible that the clone hardware mistakenly reproduced behavior for a Hercules InColor.
 
For what it's worth, I decided to run the stub on a clean PC DOS 3.30 disk (device=jrconfig.sys /v32 using the latest jrconfig, files=20, buffers=20) to compare things. The first and second results aren't interesting for the database, but I ran them for internal baselines. The third, though, is something that I didn't see in the database.

ambergrs.bhtooefr.org (a (for these purposes) stock 5155 with stock late CGA card):
Code: 
[UIDCB6B396]
MemoryTest=3784
OpcodeTest=1765
VidramTest=2655
MemEATest=1935
3DGameTest=1858
Score=4
CPU=Intel 8088
CPUspeed=4.77 MHz
BIOSinfo=1501512 COPR. IBM 1981 (11/08/82, rev. 0)
MachineModel=0000
BIOSdate=19821108
BIOSCRC16=CB6B
VideoSystem=CGA
VideoAdapter=CGA
Machine=PC/XT

My PCjr, running plain (no jrconfig):
Code: 
[UID44F2224]
MemoryTest=5942
OpcodeTest=3602
VidramTest=3388
MemEATest=4407
3DGameTest=3505
Score=2
CPU=Intel 8088
CPUspeed=4.77 MHz
BIOSinfo=1504037 COPR. IBM 1981,1983PS (06/01/83, rev. 0)
MachineModel=0000
BIOSdate=19830601
BIOSCRC16=44F2
VideoSystem=CGA
VideoAdapter=IBM PCjr
Machine=PCjr

My PCjr with IBM 128 kiB sidecar upgraded to 512 kiB, with jrconfig /v32 - I didn't see a memory expanded, but stock clock jr in there:
Code: 
[UID44F2349]
MemoryTest=3700
OpcodeTest=1741
VidramTest=2578
MemEATest=1899
3DGameTest=1837
Score=4
CPU=Intel 8088
CPUspeed=4.77 MHz
BIOSinfo=1504037 COPR. IBM 1981,1983PS (06/01/83, rev. 0)
MachineModel=0000
BIOSdate=19830601
BIOSCRC16=44F2
VideoSystem=CGA
VideoAdapter=IBM PCjr
Machine=PCjr

What's really interesting is that the PCjr is consistently faster with expansion memory. That makes sense, though - the PC has to wait for DRAM refresh, whereas I believe the 74S409 in the sidecar isn't making the jr wait (I think?).

Note that I could've done the first and third tests with the full TOPBENCH, but I don't actually have a working keyboard for the jr (I will tomorrow, though), and wanted everything to be equal anyway (and the second test needed the stub).

Edit: And, for completeness, although my surprise levels are zero, a result with jrconfig.sys /v32 /s0 - except for the opcode test (which might've reached slightly into sidecar RAM?), results are exactly in line with the no jrconfig test:

Code: 
[UID44F2350]
MemoryTest=5937
OpcodeTest=3437
VidramTest=3384
MemEATest=4301
3DGameTest=3373
Score=2
CPU=Intel 8088
CPUspeed=4.77 MHz
BIOSinfo=1504037 COPR. IBM 1981,1983PS (06/01/83, rev. 0)
MachineModel=0000
BIOSdate=19830601
BIOSCRC16=44F2
VideoSystem=CGA
VideoAdapter=IBM PCjr
Machine=PCjr
 
Last edited:
bhtooefr said:
What's really interesting is that the PCjr is consistently faster with expansion memory. That makes sense, though - the PC has to wait for DRAM refresh, whereas I believe the 74S409 in the sidecar isn't making the jr wait (I think?).
Click to expand...

The first 128K of PCjr system memory is all video memory which has an additional wait state. When you load a memory expansion and run the driver, it takes up the remaining space of the first 128K, which forces programs to load in the area beyond that which is all faster RAM.
 
You must log in or register to reply here.
Back
Top