While I prepare to run some more comparisons between the BIOSTAR MB-8433UUD-A & Asus PCI I-486SP3G running the AMD AM5x86-P75 / AMD X-5-133ADZ, here's another one to throw into the mix. This should wet your appetite a little more.. 😀

System 4 Specs:
Asus VL/I-486SV2GX4 Rev 2.1 running latest available BIOS (0402.001 BETA BIOS from 1999)
Cyrix 5x86 120 ("M1sc")
32MB RAM
256KB L2 Cache
STB LightSpeed/ET4000 W32P VLB
4GB CF Card (CF->IDE Adapter) on VLB Promise Controller

Motherboard Settings
Hardware Trap: [Cyrix] [JP5 -> 2&3] [JP6 -> 1&2]
CPU Selection: [Cyrix] [JP11 -> 1&2]
CPU Type Selection: [Cyrix] [JP16 -> 1&2 5&6] [JP17 -> 1&2 5&6] [JP18 -> 1&2] [JP19 -> 1&2 3&4 5&6] [JP20 -> All Open (3x CLK)] [JP21 -> All Open] [JP22 -> 1&2]
CPU External Clock: [40Mhz] [JP23 -> 2&3] [JP24 -> 2&3] [JP25 -> 1&2]
VESA Clock Delay: [Delay] [JP26 -> 1&2] Could not get system to boot with this set to [No Delay] [JP26 -> 2&3]
VL Bus Clock & Wait State: [<=33Mhz] [JP28 -> 1&2] [JP29 -> 1&2] I didn't notice a change when setting this to [>= 40Mhz] [JP28 -> 2&3] [JP29 -> 2&3]
Cryix Voltage: [3.6V] [JP32 -> 2&3]

System 4 Benchmark Results:
Note, these results were generated without enabling any of the Cryix 5x86 enhanced registers - the processor is running in its stock form.
SpeedSys Overall Score: 56.63
DOOM: 2134 gametics in 1583 realtics
Superscape: 865 fps
PC Player Bench: 18.2
Quake: 969 Frames 81.7 seconds 11.9 FPS

All cased up!

BIOS Settings

CPU Identification

SpeedSys Overall

SpeedSys Memory/Cache

CacheChk Read

CacheChk Write

The benchmarks

feipoa wrote:

Could you run Cachechk v7 on both systems and share the results? […]
Show full quote

Could you run Cachechk v7 on both systems and share the results?

Read speed:
cachechk -d -t6

Write speed:
cachechk -d -w -t6

Could you also upload the screenshot of the full Speedsys report?

What cache and memory settings are set in the BIOS for both motherboards?

System 3 Specs:
Asus PCI I-486SP3G Rev. 1.8 running latest BIOS available (0306 from 04/21/99)
AMD AM5x86-P75 / AMD X-5-133ADZ
32MB RAM
256KB L2 Cache
STB S3 Trio64V+ (2MB)
4GB CF Card (CF->IDE Adapter)

Motherboard Settings
Found via Google Groups:
CPU Type Selector: [] [JP34 -> 2&3] [JP35 -> 1&2] [JP36 -> 1&2 = Enables L1 Write-Back Mode]
CPU SL Selection: [Non-SL] [JP30 -> 2&3] [Short JP31 Pin 1 with JP32 Pin 1]
CPU External Clock: [33Mhz] [JP18 -> 1&2] [JP20 -> 1&2]
CPU External Clock Multi: [X2] [JP33 -> 2&3 = Tells the X5-133 to quadruple]
Misc: [JP38 -> Open]

BIOS Settings

SpeedSys Overall

SpeedSys Memory/Cache

CacheChk Read - What's going on here????

CacheChk Write

Here's where I'm at so far...

I was going to build a classic 486DX2/66 system, but I already have a Gateway 2000 and a couple of Packard Bells running with this setup. Plus, it's so damned fun messing around with these higher-speed processors. 😀

Build 1:
BIOSTAR MB-8433UUD-A Rev. 3.1
AMD - AMD-X5-133ADZ (Am5x86-P75) - 133Mhz - 33Mhz - 16KB L1 Cache running in Write-Back Mode) (11/6/1995)
32MB RAM
256KB L2 Cache
STB S3 Trio64V+ (2MB)
4GB CF Card (CF->IDE Adapter) using onboard IDE controller
GOTEK USB Floppy Emulator

Build 2:
Asus VL/I-486SV2GX4 Rev. 2.0 running latest available BIOS (0402.001 BETA BIOS from 1999)
Intel 486 DX4-100 (80486DX4100) (SK096) (&EW -> 16KB L1 Cache running in Write-Back Mode) (1994)
32MB RAM
256KB L2 Cache
TSENG ET4000/W32P - Diamond Stealth 32 2MB VLB (1993)
4GB CF Card (CF->IDE Adapter) on VLB Promise Controller
GOTEK USB Floppy Emulator

Build 3:
Asus PCI I-486SP3G Rev. 1.8 (Intel Saturn with Memory Interleaving) running latest available BIOS (0306 from 04/99)
AMD - AMD-X5-133ADZ (Am5x86-P75) - 133Mhz - 33Mhz - 16KB L1 Cache running in Write-Back Mode) (11/6/1995)
32MB RAM
256KB L2 Cache
STB S3 Trio64V+ (2MB)
4GB CF Card (CF->IDE Adapter) using onboard IDE controller
GOTEK USB Floppy Emulator

Build 4:
Asus VL/I-486SV2GX4 Rev 2.1 running latest available BIOS (0402.001 BETA BIOS from 1999)
Cyrix - 5x86-120GP ("M1sc") - 120Mhz - 40Mhz - 16KB L1 Cache in Write-Back Mode (10/30/1995)
32MB RAM
256KB L2 Cache
TSENG ET4000/W32P - STB Lightspeed VLB Video card (with 135MHz RAMDAC) (1993)
4GB CF Card (CF->IDE Adapter) on VLB Promise Controller
GOTEK USB Floppy Emulator

Artex, thank you for sharing the BIOS screen of that Saturn chipset board; I don't think I've seen one before. On the upside, the L1 and L2 speeds seem to be in the correct ballpark. On the downside, the main memory speeds are about half of what I would expect. I think memory interleaving is not functioning properly. Did you try filling ALL memory slots with RAM sticks? I recall Anonymous Coward mentioning this was required on 386 boards which support memory interleaving, so perhaps it is also true for the PCI I-486SP3G.

feipoa wrote:

Artex, thank you for sharing the BIOS screen of that Saturn chipset board; I don't think I've seen one before. On the upside, the L1 and L2 speeds seem to be in the correct ballpark. On the downside, the main memory speeds are about half of what I would expect. I think memory interleaving is not functioning properly. Did you try filling ALL memory slots with RAM sticks? I recall Anonymous Coward mentioning this was required on 386 boards which support memory interleaving, so perhaps it is also true for the PCI I-486SP3G.

Yeah, all slots are filled.

If you are using EDO RAM, could you try FPM RAM?

feipoa wrote:

If you are using EDO RAM, could you try FPM RAM?

Perhaps a dumb question... but is there a way physically discern the two types?

The only sure way I know of is to look up the part number on the individual memory chips and find the datasheet. You can usually find it in a few minutes online.

Artex wrote:

feipoa wrote:

If you are using EDO RAM, could you try FPM RAM?

Perhaps a dumb question... but is there a way physically discern the two types?

Another way is to plug them into a board that posts if FPM or EDO ram is being used.

I found this article: http://markmail.org/message/kckyjqo7gv3aac2m

Artex wrote:

"A good'' 486 PCI motherboard is the ASUS PCI/I-486SP3G, it has a built in NCR810 controller. Uses bank interleaved memory (you must install 72 pin simms in pairs just like a Pentium 😀) to get decent memory speeds. And in general it just works. It has 1 annoying bug that has been found, and that is you must run the external cache in write through mode."

And this quote from the Asus FAQ: http://www.menet.umn.edu/~bob/FAQ/asus-faq

Artex wrote:

Q04) What is the better cache sheme: write-through or write-back ? How can I upgrade my SP3g with a dirty Tag RAM ? Does it incr […]
Show full quote

Q04) What is the better cache sheme: write-through or write-back ?
How can I upgrade my SP3g with a dirty Tag RAM ? Does it increase
the performance ?

The ASUS SP3G can drive write-through or write-back cache. You can use both
modes with and without dirty tag RAM present. In write-through mode, dirty
tag RAM has no function. The BIOS default is "write-back", resulting in
some 5% performance loss, because if there is no dirty tag RAM, then there
are lots of unnecessary memory writes, since each cache line refill will
need to be preceeded by a write back of four DWORDs, even if none of them
had been modified.
In write-back mode, dirty tag RAM increases real-world performance by
approx. 1% over write-through. The small positive effect of write-back on
PC compatible hardware is caused by the lack of write-allocate. In case
of the Saturn chipset, the number of clocks for some transactions is
higher for write-back and the Saturn has already very efficient write
posting buffers which compensate much of the inefficiency of not having a
dirty tag RAM.
Starting with rev. 1.8, the SP3G got a socket for a dirty tag RAM. With
this RAM installed. Suitable chips for 256kB cache size are 64K x 1 RAMs
with 15 ns access time, e.g. Cypress Cy7C187, AT&T ATT7C187-15,
Matra MHS HME 65787-15, Motorola MCM6287C-15 and Toshiba TC5562-15 or any
other equivalent.

I have a rev 1.8 board, but I low and behold I don't have this dirty TAG RAM up there..

...so I need to run the L2/external cache in Write-Through mode. The BIOS Option for "Cache Update Policy" refers to the L2 cache. L1 cache WB mode is configured by the jumpers. I changed the BIOS option to Write Through (L1 is still in WB mode) and here are the updated results:

CacheChk Read

CacheChk Write

[/quote]

What is the part number on your memory chips?

That seemed to help. Perhaps this board's memory interleaving doesn't work with L2 set to write-back mode. The read/write are still a little slower than the Biostar's and other PCI 486 boards based on the UMC and SiS chipsets. I wonder what is holding it back? Perhaps some extra conservative settings imposed by the Intel chipset?

A properly tuned Biostar with UMC 8881/8886 chipset and an AMD X5-133 should have an effective RAM access times of 106/61 ns (read/write). Your Saturn-II-based board is at 114/92 ns.

Perhaps these conservative settings make use with a 40 MHz bus more stable when using a Cyrix 5x86-120. While even UMC 8881/8886-based boards should be fine at 40 MHz with 256 KB cache and the fastest cache settings, it is when the cache exceeds 256 KB that the fastest L2 timings can become unstable. My tests showed that the X5-160 is not quite as suseptable to this phenomenia, but still affected. Matching up L2 speeds to FSB speeds on various motherboards can be an art on its own. In short, I suspect this Saturn-II board with a Cyrix 5x86-120 may be a good mate if you use 1024 KB of cache and 128 MB of memory. Since you need to have the L2 set to write-through mode, even 512 KB of cache and 128 MB of memory would make a good fit.

Are you able to test this board with 512KB or 1024KB of cache and 128 MB of memory using HIMEM, Windows 9x, and Memtest 4.0?

feipoa wrote:

What is the part number on your memory chips? […]
Show full quote

What is the part number on your memory chips?

That seemed to help. Perhaps this board's memory interleaving doesn't work with L2 set to write-back mode. The read/write are still a little slower than the Biostar's and other PCI 486 boards based on the UMC and SiS chipsets. I wonder what is holding it back? Perhaps some extra conservative settings imposed by the Intel chipset?

A properly tuned Biostar with UMC 8881/8886 chipset and an AMD X5-133 should have an effective RAM access times of 106/61 ns (read/write). Your Saturn-II-based board is at 114/92 ns.

Perhaps these conservative settings make use with a 40 MHz bus more stable when using a Cyrix 5x86-120. While even UMC 8881/8886-based boards should be fine at 40 MHz with 256 KB cache and the fastest cache settings, it is when the cache exceeds 256 KB that the fastest L2 timings can become unstable. My tests showed that the X5-160 is not quite as suseptable to this phenomenia, but still affected. Matching up L2 speeds to FSB speeds on various motherboards can be an art on its own. In short, I suspect this Saturn-II board with a Cyrix 5x86-120 may be a good mate if you use 1024 KB of cache and 128 MB of memory. Since you need to have the L2 set to write-through mode, even 512 KB of cache and 128 MB of memory would make a good fit.

Are you able to test this board with 512KB or 1024KB of cache and 128 MB of memory using HIMEM, Windows 9x, and Memtest 4.0?

Not sure how far I wanna go with the testing on this one.. The board cost me an arm and a leg to begin with, and I don't know how much I'd gain real-world performance-wise by throwing down some additional $$ for these chips and extra memory. I kinda just want to enjoy the board now. 😀

What are your thoughts with the Cyrix 5x86 120 on the Asus VL/i-486SV2GX4 after seeing the numbers?

I'll get the Biostar results for ya tomorrow btw!

Here's the two type of memory in there - one looks like 60ns and one looks like 70ns.

2x

2x

Artex wrote:

Not sure how far I wanna go with the testing on this one.. The board cost me an arm and a leg to begin with, and I don't know how much I'd gain real-world performance-wise by throwing down some additional $$ for these chips and extra memory. I kinda just want to enjoy the board now. 😀

What are your thoughts with the Cyrix 5x86 120 on the Asus VL/i-486SV2GX4 after seeing the numbers?

I'll get the Biostar results for ya tomorrow btw!

I didn't realise that you did not already own 64kx8 or 128kx8 cache chips. Many people have 8 pieces of 64kx8 cache chips laying around.

As for the Cyrix 5x86-120 on the Asus VL/i-486SV2GX4, the memory read numbers are horrifically slow. Something is up. They should be at least 2x faster. Perhaps you should put the Cyrix 5x86-120 into the Saturn-II board.

feipoa wrote:

Artex wrote:

Not sure how far I wanna go with the testing on this one.. The board cost me an arm and a leg to begin with, and I don't know how much I'd gain real-world performance-wise by throwing down some additional $$ for these chips and extra memory. I kinda just want to enjoy the board now. 😀

What are your thoughts with the Cyrix 5x86 120 on the Asus VL/i-486SV2GX4 after seeing the numbers?

I'll get the Biostar results for ya tomorrow btw!

I didn't realise that you did not already own 64kx8 or 128kx8 cache chips. Many people have 8 pieces of 64kx8 cache chips laying around.

As for the Cyrix 5x86-120 on the Asus VL/i-486SV2GX4, the memory read numbers are horrifically slow. Something is up. They should be at least 2x faster. Perhaps you should put the Cyrix 5x86-120 into the Saturn-II board.

Yeah, the world of external L2 cache is new to me, so I take what I'm given when I buy these boards. 😀 As for the Cyrix - any idea what could be the culprit? I can certainly try the chip in the Saturn board but I know that the chip is supported on that particular board/revision/BIOS.

nn514400A is FPM
http://www.datasheetarchive.com/dl/Scans-091/ … HI000108774.pdf

HYB514400B is FPM
http://www.datasheetarchive.com/dlmain/Datash … /DSA-265303.pdf

Where is the EDO?

The culprit is something to do with the motherboard or BIOS settings. I do not beleive it is specific to the Cyrix 5x86. Your X5-133 settings are probably really slow too.

feipoa wrote:

nn514400A is FPM http://www.datasheetarchive.com/dl/Scans-091/ … HI000108774.pdf […]
Show full quote

nn514400A is FPM
http://www.datasheetarchive.com/dl/Scans-091/ … HI000108774.pdf

HYB514400B is FPM
http://www.datasheetarchive.com/dlmain/Datash … /DSA-265303.pdf

Where is the EDO?

The culprit is something to do with the motherboard or BIOS settings. I do not beleive it is specific to the Cyrix 5x86. Your X5-133 settings are probably really slow too.

Darn.. I knew I got off too easy with these setups. I'll grab (what I believe is) the EDO tomorrow from the Biostar board. For now my brain is shot. 😉 I really appreciate your knowledge on these setups as this is a new world to me!

I have never noticed any performance benefit from using EDO memory over FPM on these 486 board. I actually find that FPM is a little more tolerable of faster timings with a 40 MHz FSB.

feipoa wrote:

The culprit is something to do with the motherboard or BIOS settings. I do not beleive it is specific to the Cyrix 5x86. Your X5-133 settings are probably really slow too.

Anything I should look for specifically?

This is an interesting thread. So the Cyrix 5x86 works in SP3G? Where are the speedsys results?

Anonymous Coward wrote:

This is an interesting thread. So the Cyrix 5x86 works in SP3G? Where are the speedsys results?

Certainly turning out to be interesting - that's for sure. It's definitely a learning experience as well - you really gotta WORK for it with these systems but I guess that's all part of the experience right!? 😎

I have not tried the Cyrix 5x86 in the Sp3G - only the Asus VL/I-486SV2GX4 board at this point.

Again, I really appreciate everyone's input and guidance here. Definitely relying on your expertise to see these builds through to completion, and hopefully running in their most optimal configuration.