Looking at the results of your ABIT AB-PB4 compared to that of a M919 (Um8881), I see the following.

ABIT - X5-160
L1 = 92.8 MB/s
L2 = 55.3 MB/s
RAM = 38.96 MB/s

M919 - X5-160
L1 = 141 MB/s
L2 = 59 MB/s
RAM = 45 MB/s

ABIT - X5-133
L1 = 77.0 MB/s
L2 = 45.8 MB/s
RAM = 32.3 MB/s

M919 - X5-133
L1 = 117 MB/s
L2 = 49 MB/s
RAM = 37 MB/s

The most striking is the L1 result. I wonder what is going on there?

The L1, L2, and RAM results are 34%, 6.3%, and 13.6% lower, respectively.

Interesting. I'm pretty sure I have used a Kingston Upgrade chip which is a X5-133 soldered to an interpaoser. This was because it would run on any 5V 486 motherboard (and those with the VRM damaged, which for some reason happens to often on my 486 boards). It seems I should try testing it with a real X5-133.

I wonder if the L1 cache is being put into write-through mode?

DRIVERS :

PB4

Filename: PB4_IDE1.EXE
Date: 10/16/95
NOTE :

PB4 PCI IDE Driver Rev: 1.01.

Source: http://leoromo.net/mirrors/www.elhvb.com/mobo … /pb4/index.html

2nd option: http://ftp.isu.edu.tw/pub/Hardware/ABIT/drive … b4/pb4_ide1.exe

hello guys, after reading whole thread, ive decided to reply for help with my like new pb4 v 1.5, after some tinkering with various cpus from my collection ive placed x86-X5-133-ADZ with no results, so i got "dumb" idea to flash bios chip with F2 bios file. Of course without any further experience all went wrong way and now board is dead piece of silicon. except it starts boot block and was able to boot my floppy disk with awdflash on it and bios file, but every try to reflash ends with error msg.
Last attempt starts flashing that chip, but ends on the start of second line, and now even boot block has gone.
Can someone give me advice how to solve this, i have lots of other boards but none of them has Ali chipset M1487 like this one. Iam frustrated with all the day attempts, but still know board is fully functional before so trash can is not an option yet. Thanks for any advice to LAMA.

You can use an Ethernet card with ROM socket on it to reflash the chip or use the 'hot swap' method on another board:

https://flashrom.org/Flashrom

Haven't tried this approach personally, but I am in the same situation as you with my board (boot block is all I get) except my board is a 1.3 revision. I have noticed the EEPROM chip on this board has more pins than BIOSes on my other motherboards.

If I had the time (I am moving house at the moment) I have spare flash chips - I could flash the right BIOS and mail it to you. Sorry! Maybe some other kind-hearted soul here can help.

Sorry guys for dragging this thread from the dead but I see no sense in making another thread if my issue is about this very motherboard (plus it's not a terrifically long time ago anyway).

Anyway, after alot of delays and moving house, I finally got round to sorting some of my machines, (I got plenty to show off so il certainly be making a gloat thread soon, but I digress!).

Just a quick question; I'm a little lost on the manual about the jumper settings for the CPU (BTW, massive thanks for uploading!). It had originally a ST 486 DX2 66 and I wish to drop in an Am5x86 133.

What jumpers to I need to set exactly? I'm certain they are the set of 14 pins just south of the CPU socket. If someone could perhaps decipher it in a (probably moron) way, that would be great, as I honestly think the way I think it's saying is probably not right!

Let me know if pictures are required. Many thanks in advance!

What I would really like to see out of one of these AB-PB4 motherboards is someone using that brown PISA slot for a CPU card. This is the only PCI-, ISA- or VLB-based 486 board I know of with a PISA slot. And how would you delegate the tasks between the PISA CPU and the motherboard's CPU? Could you use the system like a quasi SMP system? Or is it more like you have two desktops shoved into one case whereby the two desktops can only communicate with each other, say, between an ethernet card in sort of beowulf-like cluster?

Forgive my idiocy, the only thing I know about PISA cards is most (if not all) seem to be entire systems on a card. Are up you basically asking how it is implemented? Would they be two separate entities or would they work in tandem? If so I can only answer with; "I don't know" and probably may never know, as I have no PISA hardware what so ever and nor have I ever seen any in the flesh.

Since I'm constantly learning about this stuff, like you I'm eager to learn more about it, but right now all I want is a working system! 🤣

Yes, that is what I'm asking. I find the concept fascinating on a PCI-based 486.

I'm not sure about this board, but weren't PISA slots often used for PCI/ISA riser cards?

Hey guys! I also have this mobo but someone has removed all jumpers and olso RJ1 RJ2 RJ3 jumpers... Can someone please tell me what those RJ1 RJ3 RJ3 are? Are they just 0hm resistor packs or what?... Thanks in advance..

40mhz front side bus should yield in excess of 40mb memory transfer speed in speedsys(Memory Throughput).
At 50mhz I get 56MB/s on a LS486E and at 40mhz I think it was something like 48MB/s.

UMC boards(888x) are slightly slower but were faster regarding second level cache speed(up to 77MB/s @ 50mhz vs 70MB/s on the SiS49x).

The Kingston chip I tested is slightly slower than a normal 5x86 on the LS486E as it's L1 cache is running in WT mode. I have no idea why this is. Nonetheless a great little processor.

Hi
I have Abit AB-PB4 v1.3 with Am5x86@160MHz, EDO ram, 15 ns cache.
SRAM read timing 3-1-1-1, write timing 0 wait and DRAM "Faster".
Mamory bandwith looks lower than Yours, are You using 12 ns cache?

indomitus wrote on 2015-11-23, 18:58:

Hello...
this is my AB-PB 4 mobo; I do not remember if it works. I'll check tonight's.

indomitus, I am most intrigued - although your motherboard is the same revision as mine (1.3) you have JP12 and JP31 present, which are missing on my board. This seems to make it impossible for me to set 50MHz bus speed. I have reinstated a jumper block at the postition of JP12 (it was hardwired to position 1-2 before) but JP31 doesn't exist at all. I would love to see a close-up of that part of your board, if possible.

lukas12p wrote on 2020-06-21, 12:08:

Hi I have Abit AB-PB4 v1.3 with Am5x86@160MHz, EDO ram, 15 ns cache. SRAM read timing 3-1-1-1, write timing 0 wait and DRAM "Fas […]
Show full quote

Hi
I have Abit AB-PB4 v1.3 with Am5x86@160MHz, EDO ram, 15 ns cache.
SRAM read timing 3-1-1-1, write timing 0 wait and DRAM "Faster".
Mamory bandwith looks lower than Yours, are You using 12 ns cache?

I have also got around to testing this board with the 5x86 at 160MHz. The only difference between my setup and yours is that I used 'fastest' DRAM timings in the BIOS, and I have compact flash for storage. It's a shame that the board cannot do 45MHz, as I believe this board would perform very well at 180MHz with a capable CPU (I have 3 candidates).

brassicGamer wrote on 2021-04-04, 01:08:

I am most intrigued - although your motherboard is the same revision as mine (1.3) you have JP12 and JP31 present, which are missing on my board. This seems to make it impossible for me to set 50MHz bus speed. I have reinstated a jumper block at the postition of JP12 (it was hardwired to position 1-2 before) but JP31 doesn't exist at all. I would love to see a close-up of that part of your board, if possible.

The same findings here. AB-PB4 rev 1.3 has JP12 hardly shortened btw pin 1 and 2. It can be restored to full capable jumper. But cant find JP31. Cant find him on the net, pictures etc. All board pictures, on the internet, have no JP31. I dont know what exactly does JP12 and JP31, but 50MHz FSB can be set by existing JP4 and JP11. I am trying Amd486DX4 100MHz SV8B here. When opened JP4 and JP11 for 50MHz FSB, AB-PB4 gives only blank screen. With shorted jumper JP7, which is normaly opened for whole CPU, expect for Amd 5x86 133/160MHz in the list, it boots to BIOS post as 486 DX2 100MHz and NSSI says 50MHz FSB with 2x multiplier. When we suppose 5x86 133/160, it equals to 4x33 and 4x40MHz. It means multiplier4x with different FSB. It seems, this multiplier is not supported by enhanced AMD 486DX4 SV8B, which alows 2x or 3x. I am able to start Win98 with SV8B and 2x50. It definitely seems 50MHz FSB is working in default config with shortened pins 1 and 2 on JP12 and no JP31. The question is what realy does JP12 and nonexisting JP31. Because it is specific behaviour only for 5x86 150MHz, it means 3x50MHz. This multiplier also supports enhanced AMD486DX100 SV8B, but maybe 50MHz FSB is too high for him and thats why he is causing blank screen in 3x50 mode. But CPU works fine with 3x40 like DX4 120. The conclusion is, if you want 50MHz FSB, no JP12 and JP31 is needed.