blog:286_shootout_vlsi_vl82c201

Unknown VLSI VL82C201

A fast VLSI 286 motherboard. With details on configuration and software tools. Testing and benchmark data.

Although not one of the latest 286 chipsets, the mid-life VLSI VL82C201 seems to be a strong performer, pulling in some excellent benchmark scores:

This board was offered to me by a member of the Vogons.org DOS gaming forum, in fact the same individual who benchmarked a similar board to the results seen in the second of those links above.

From left to right: pshipkov's VL82C201, Ekb's VL82C201, my VL82C201

Compared to the boards in the benchmark results above, mine has only 6 ISA slots (4x 16bit + 2x 8bit) versus 7 in the others (although 3x 8bit slots is not massively useful for gaming purposes).

Other than that, the layout and memory configuration looks to be the same.

They all have the same revision of the VL82C201 chipset, all specifically the VL82C201-16Q configuration. They all have 4x 30pin SIMM slots (or the equivalent SIPP sockets) plus a bank of DIP sockets. All have socketed processors. All use twin 27C256 EPROM's for BIOS.

The one thing which I could see counting against similar performance of my board is that the chipset is not quite as closely clustered together; the shorter transmission lines of the other two boards would definitely be beneficial at clock speeds which are otherwise borderline.

One thing which needs consideration with these boards is that they use a oscillator speed divisor of 1/4, rather than the typical 1/2. So if you want to play with the board/processor frequency, then you'll need oscillators of 80MHz for a 20MHz processor, 100MHz for 25MHz cpu, and so on. Your typical 32MHz oscillator on another 16MHz board will only drive the processor at 8MHz LOL

There is a higher-specced version of the chipset available in the VLSI chipset database; the VL82C201-20Q which is rated at up to 20MHz bus speed, but I've yet to see any examples of it in the wild.

The BIOS on this board is extremely rudimentary, offering only Video and System BIOS shadowing and no other performance options whatsoever. Unless I can find out the jumper configuration for the board (and therefore if there is an option for ISA clock speed, memory parity settings or wait-state configuration), I will have to assume that this is the only change I can make.

It's fairly poor and typical of very early AMI BIOS interfaces.

The Intel 287XL was fitted into the co-processor socket and no onboard jumpers changed (mainly because they are not documented). It seems to work with whatever settings the board is configured to use.

Configuration 0

  • Harris CS80C286-25
  • 64MHz oscillator (16MHz CPU speed)
  • Standard ISA bus clock
  • 4x 1MB 60ns SIMM
  • VGA 1 Trident TVGA9000 512KB VGA
  • VGA 2 Tseng Labs ET4000AX 1MB VGA

The BIOS of this motherboard does not feature any performance options, so what would have been a 'least aggressive' tuning config on other boards is just what you get on this.

VGA 1

VGA 2


Configuration 1

  • Harris CS80C286-25
  • 80MHz oscillator (20MHz CPU speed)
  • System + Video BIOS shadowed
  • 4x 1MB 60ns SIMM
  • VGA 1 Trident TVGA9000 512KB VGA
  • VGA 2 Tseng Labs ET4000AX 1MB VGA
VGA 1

VGA 2


Configuration 2

  • Harris CS80C286-25
  • 100MHz oscillator (25MHz CPU speed)
  • System + Video BIOS shadowed
  • 2x 1MB 60ns SIMM
  • VGA 1 Trident TVGA9000 512KB VGA
  • VGA 2 Tseng Labs ET4000AX 1MB VGA

I could only get this configuration working with 2 of my 4 1MB 60ns SIMM modules. I suspect that it's going to need some higher quality modules, or higher rated (50ns?) chips.

As it is, this is unreliable - not quite as unreliable as the Headland HT12 at 24 or 25MHz, but you certainly wouldn't want to use the system in this state. Again it seems more willing to lock up with the Tseng Labs card than the Trident, just as the other 286 system was.

VGA 1

VGA 2

Test Config 0 Config 1 Config 2
Processor Speed 16MHz 20MHz 25MHz
Wait states ? ? ?
Norton Sysinfo CPU 12.3 15.4 19.4
Norton Sysinfo HDD 1114KB/sec 1319KB/sec 1471KB/sec (2)
Norton Sysinfo Overall 11.3 13.7 16.7
CheckIt Dhrystones 4041 5139 6374 (3)
CheckIt Whetstones 534.5K 633.8K 831.4K
CheckIt Video chars/sec VGA 1 3197 3965 5112
CheckIt Video chars/sec VGA 2 3450 4248 5425
CheckIt HDD Transfer 1091KB/sec 1292KB/sec 1437KB/sec
Landmark CPU 24.76MHz 31.28MHz 39.17MHz
Landmark FPU 17.87MHz 22.50MHz 28.15MHz
Landmark Video VGA 1 2943.2 3052.9 3110.89
Landmark Video VGA 2 6922.8 8623.1 10685.2
3DBench VGA 1 6.3 7.9 9.7
3DBench VGA 2 6.8 8.6 11.6 (1)
ATPerf RAM Read 0 0 0
ATPerf RAM Write 0 0 0
ATPerf ROM Read 2 2 2
ATPerf Video Write VGA 1 8 11 14
ATPerf Video Write VGA 2 2 2 2
CompTest RAM thruput 7552KB/sec 9473KB/sec 11915KB/sec
CompTest Effective WS 0.1 0.1 0.1
CompTest MFLOPS 0.46 0.058 0.073
CompTest Video BIOS VGA 1 7881 9829 12144
CompTest Video BIOS VGA 2 7743 9511 12065
CompTest Video Direct VGA 1 4651 5814 7299
CompTest Video Direct VGA 2 5556 6993 8772
CompTest Dhrystones 4253 5322 6676

Games tests are run with the fastest VGA option (normally VGA #2).

Games Test Config 0 Config 1 Config 2
Wolfenstein 3D 12.7 15.8 19.7 (4)
F1GP Bench 1 156-172 127-135 105-110

Software Versions

  • Norton Sysinfo 8.0
  • CheckIt 2.1
  • Landmark 6.00
  • 3DBench 1.0
  • ATPerf 2.00
  • Comptest 260

Notes

  • (1) - 3DBench did not finish correctly. The 3D visuals disappeared and the counter continued until it reached 0 then displayed the calculated 11.6fps total.
  • (2) - Norton Sysinfo crashed the first time it was ran.
  • (3) - CheckIT 2 crashed several times until I was able to get a full run from it.
  • (4) - Once the system is up to temperature it is difficult to get this to run on the ET4000 at 25MHz. Starting from cold it runs at an ultra fast (for a 286) 19.7FPS.

The Last Byte supports the VL82C802 memory controller used in the VLSI chipset on this board. However, I had problems enabling it.

Whichever way you set shadow RAM in the BIOS, it always ended up with an error message about 'Shadow RAM relocated':

Unfortunately there's no labelled jumpers on this motherboard m(

Through trial and error, and reading the chipset.doc documentation included with The Last Byte, it seems that shadow RAM control is by pin #69 of the chip. Checking for continuity from that led me to a set of jumpers just above it - which were shorting to ground. Removing the jumper from J16, below:

Note to self: it was in position 3-4, furthest away from the board edge, this lead the board to start up with 3712KB of RAM, versus 4096KB. A good sign. On booting Dos and watching The Last Byte load, I was greeted with this more successful looking banner:

However.

Once The Last Byte has loaded, the system is incredibly unstable - with the keyboard or screen locking up at frequent intervals.

(Go back to the 286 PC main page)

  • blog/286_shootout_vlsi_vl82c201.txt
  • Last modified: 2021/05/27 18:47
  • by john