386DX40 wrote:

[ very interesting stuff ... ]

I will try CTCM.exe and CTCM7.exe. Thanks! (Are they easily found on the net?)

For Speedsys (as seen in the O.P., and others), please see this thread (half way down for the D/L link)

Finally!! A stable Socket 3 System!!

386DX40 wrote:

Was this VIA chipset just solid and reliable but not very fast perhaps?

Hmmm! I wonder?!

Check this site: ftp://ftp.heise.de/pub/ct/ctsi/
All the other freaking system tools are worth a closer view too.

elianda wrote:

Check this site: ftp://ftp.heise.de/pub/ct/ctsi/

Thanks a lot! 😀

I have set up the FIC mobo again, for some cache testing. I went to the "ftp.heise" link above, and DL'd CTCM7, but I could not find CTCM. (Where can this be DL'd please?)

I run CTCM7, and see this information on the screen -

Processor Timing: cx586, cx86gx
Processor CPUID: unknown
Processor Name: Cx586 (M1SC)3:1 Sample, ID=2Dh, Rev=13h
Actual clock rate: 99.912 HMz
Primary Cache (L1) : 16 KByte, 4way associative
Secondary Cache (L2) : 256 KByte,direct mapped
Code Cache (L1) : unified with Data
Main memory : 32 MByte, no Memory holes found
Cacheable Area L1 : 32 MByte, no noncacheable Areas found
Cacheable Area L2 : 32 MByte, no noncacheable Areas found

MMOVI

(The program stops at this point - I have to press CTL-ATL-DEL)

Amigaz wrote:

... and stuff an AMD 5x86 and run it at 40 or 50mhz FSB since you have a PCI divider jumper as I do on my PIO-2

Can you recommend which AMD 5x86 CPU to buy please? There seems to be various CPUs available, with different codes written on them. Here's one, at random:

http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewIt … em=230383851715

Any good? Thanks.

retro games 100 wrote:

Can you recommend which AMD 5x86 CPU to buy please? There seems to be various CPUs available, with different codes written on t […]
Show full quote

Amigaz wrote:

... and stuff an AMD 5x86 and run it at 40 or 50mhz FSB since you have a PCI divider jumper as I do on my PIO-2

Can you recommend which AMD 5x86 CPU to buy please? There seems to be various CPUs available, with different codes written on them. Here's one, at random:

http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewIt … em=230383851715

Any good? Thanks.

Never seen that one before...looks odd

Try and get this one
http://www.cpu-world.com/CPUs/80486/AMD-AMD-X5-133ADZ.html

It's the coolest running 5x86.

Or try and get your hands on the Evergreen or Kingston 5x86 cpu's
They are 486 "upgrade" cpu's

I put on my environmental harzard suite, and ventured forth in to the attic of doom. There, lurking in the shadows of horror (tm), I found a mobo that I bought from some guy in Sweden. Hehe! It's a UMC chipset based board, model number UM8810PIAO, PCB revision 2.1. It's got a Voltage Regulator Module (VRM) attached to it, with an AMD P75 CPU. This CPU is labelled: AMD-X5-133ADW

I removed the two 16MB Fast Page Memory sticks (32MB total) from the FIC mobo, and put them in to this UMC board. I grabbed the nearest graphics card to hand (an S3 Virge 325 2MB PCI), and ran some benchies. As seen in the Speedsys screenshot, the memory bandwidth is better on this mobo, compared to the FIC mobo.

I ran "cache check 7", and this is what it said:

This machine seems to have both L1 and L2 cache. [reading]
L1 cache is 16KB-- 135.4 MB/s 7.7 ns/byte (548%) (315%) 3.9 clks
L2 cache is 256KB-- 42.9 MB/s 24.4 ns/byte (173%) (100%) 12.4 clks
Main memory speed -- 24.7 MB/s 42.4 ns/byte (100%) [reading] 21.6 clks
Effective RAM access time (read) is 169ns (a RAM bank is 4 bytes wide).
Effective RAM access time (write) is 124ns (a RAM bank is 4 bytes wide).
"AuthenticAMD" 486(5x86) Clocked at 133.4 Mhz. Cache ENABLED.

I then ran CTCM7, and it said:

Primary Cache (L1) : 16 KByte, 4way associative
Secondary Cache (L2) : 512 KByte, direct mapped
Code Cache (L1) : unified with Data
Main memory : 32 MByte, no Memory holes found
Cacheable Area L1 : 32 MByte, no noncacheabled Areas found
Cacheable Area L2 : 32 MByte, no noncacheabled Areas found

Pcpbench (Default mode 100, LFB) = 6.7

The next thing to do is to remove the AMD P75 CPU from the UMC board, and put it in to the FIC board. I will also adjust the FIC board's FSB from 33 to 40. (Can I go as high as 50?)

I removed the Cyrix 5x86 100mhz CPU from the FIC board, and replaced it with an AMD 5x86 133mhz CPU. Pcpbench scores 6.1. Previously, using the UM8810PIAO mobo with this AMD CPU, I got 6.7. I then run cachechk7, and it says:

L1 cache is 16KB -- 137.5 MB/s 7.6 ns/byte (1155%) (222%) 3.9 clks
L2 cache is 256KB -- 61.8 MB/s 17.0 ns/byte (519%) (100%) 8.6 clks
Main memory speed -- 11.9 MB/s 88.0 ns/byte (100%) [reading] 44.8 clks
Effective RAM access time (read) is 352ns (a RAM bank is 4 bytes wide).
Effective RAM access time (write) is 92ns (a RAM bank is 4 bytes wide).

Also, please see Speedsys screenshot for this FIC mobo + AMD CPU combo below.

Things to do: is it possible to increase the FIC mobo's FSB to either 40 or 50? Also, I can't find CTCM.exe on the net. If someone can provide a DL link, I'll run it on the FIC board before I dismantle this testing rig. Thanks.

Sorry I haven't been on in so long, been too busy to mess with my old PC's! CTCM.exe can be found inside the CTCM_INT zip file found here.

http://motherboards.mbarron.net/download/

I don't understand why your memory throughput is so low on that board. I have a Micronics board with an Intel chipset, and no memory adjustments at all in the BIOS running a 66MHz DX2 that gets around 29MB/s! Are your memory timings for the L2 cache and main memory set at the fastest timings available that will allow the board to boot?

I guess the poor main memory throughput is due to the VIA chipset perhaps? I dunno. VIA never really impressed me even in the AMD Athlon XP era....had a KT133 board and a KT266 (both non A revisions) and those were both problematic and slow too!

CTCM7 will typically lock up right after showing the cacheable areas on a 486 board, as the next tests it does only seem to work on Pentium and above CPUs. CTCM.exe will work on a 486 board and give you some detailed memory throughput results.

Thanks a lot for the CTCM link. Unfortunately, I don't have that FIC mobo immediately available, but I have just tested CTCM on another 486 mobo. It's an MS4144, PCB revision 1.4. (4 ISA, 3 PCI, SIS 496/7 chipset, 256K L2 cache.) It's got an AMD DX4-100 (WB cache) CPU in it, with 2 sticks of 16MB FPM (32MB total), and a 2MB Trio64v+ graphics card. I ran CTCM, and it says:

Processor CPUID: am486DX4-WB/ Typ:00 Fam:04 Mod:09 Stp:04
Clck: 99.5 MHz
internal Bus: 32 Bit between CPU and primary Cache or Memory
FPU: i487-Typ
L1 Cache : 8 KByte, 4-way associative
L2 Cache : 256 KByte, direct mmapped
Write Strategy L1 : Write Back
Write Strategy L2 : Write Back
Dirty Tag L2 : ok

Through Put & Bus Performance : Main memory from 00120000

Best Time for 8K MOVSD (Cache /Page Hits) : 65 mcs - 125.1 MByte/s
average t. for 8K MOVSD (Miss + Hit) : 234 mcs - 35.0 MBytes/s
average t. for 8K MOVSD (if clean) : 310 mcs - 26.4 MByte/s
average t. for 8K MOVSD (if dirty) : 526 mcs - 15.6 MByte/s
Main Memory 8K MOVSD (Cache misses) : 613 mcs - 13.4 MByte/s

average with 256 KB L2-Cache /DOS (640K) : 265 mcs - 30.9 MByte/s
average with 256 KB L2-Cache /WIN (4M ) : 334 mcs - 24.5 MByte/s

Also, a speedsys shot is included below.
Edit: I ran pcpbench, and I get a score of 7.0. I ran Superscape 3DBench, and I get a VGA score of 62.5 frames per second.

retro games 100 wrote:

I run CTCM7, and see this information on the screen -

MMOVI

(The program stops at this point - I have to press CTL-ATL-DEL)

I have the same problem when running ctcm7 on my 386 - the program stops with the message "main memory!!" and MMOVI on the bottom of the screen.

how did you solve it?

Accessing memory banks from date 2010-1-25........failure! 😳 Error message: "can't remember". To do: upgrade brain to brain 2.0

I probably never got it working. I expect a 386 can't complete this operation.

Edit: Looking back at my testing with this curiously under performing mobo, I wonder if a jumper was in the wrong position, and if true, that may explain its poor performance. The next time I test this mobo, I'll double-check every jumper. If they appear correct, and the board still performs badly, I might make some educated guesses, and change one jumper at a time, in an attempt to improve its performance.

schlang wrote:

retro games 100 wrote:

I run CTCM7, and see this information on the screen -

MMOVI

(The program stops at this point - I have to press CTL-ATL-DEL)

I have the same problem when running ctcm7 on my 386 - the program stops with the message "main memory!!" and MMOVI on the bottom of the screen.

how did you solve it?

Use CTCM 1.6n.

retro games 100 wrote:

Accessing memory banks from date 2010-1-25........failure! 😳 Error message: "can't remember". To do: upgrade brain to brain 2.0

I probably never got it working. I expect a 386 can't complete this operation.

Edit: Looking back at my testing with this curiously under performing mobo, I wonder if a jumper was in the wrong position, and if true, that may explain its poor performance. The next time I test this mobo, I'll double-check every jumper. If they appear correct, and the board still performs badly, I might make some educated guesses, and change one jumper at a time, in an attempt to improve its performance.

Main thing is, you're improving your knowledge and getting more skilled ;D

My experience with a lot of these earlier PCI-486 motherboards is that they just run slow. Full stop. Sometimes no amount of troubleshooting would speed'em up. It is probably chipset and/or BIOS related.

As most of us have by now arrived at, UMC seems to be the champ, but I have some curious hopes for SiS.

elianda wrote:

Use CTCM 1.6n.

you are absolutely right 😀

feipoa wrote:

My experience with a lot of these earlier PCI-486 motherboards is that they just run slow. Full stop. Sometimes no amount of troubleshooting would speed'em up. It is probably chipset and/or BIOS related.

As most of us have by now arrived at, UMC seems to be the champ, but I have some curious hopes for SiS.

I'm interesting in hearing about the SiS chipset. Even though they lack the PCI divider, that doesn't mean theres no use for them 😀
And besides, it's not like theres tons of 486 boards going on Ebay in stacks of 10 for €1+shipping 🤣

Only one seller on eBay has this motherboard and they want $181.91 USD for it and on Amazon it's going for $147.51. Seriously, why do 486 motherboards still go for so high from eBay, Amazon, etc? 😵

This board got decent reviews back in the day from users. Though it is more stable than it is speedy. The recommended CPU is the 5x86-133.

BTW, FIC still has BIOS downloads available for nearly all their 486 motherboard located: here. The last revision of BIOS for the FIC PIO-3 is dated 1-04-96 and requires either the optional SST Flash EPROM chip, or EPROM programmer in order to flash.

These "high price sellers" only sell a fraction of their inventory at these unrealistic prices. If you compare the amount of sales that they do, against the amount of stock that they list on ebay, you will see that about 99% of it remains unsold during each 30 day "listing period".

I've heard from these scrap sellers that it costs them about $4-7 per scrap motherboard, so you would think they would be happy with selling them for 2-3x what they paid for them. Instead, they seem diluted with greed and ask for 45x times the cost.