Re: Did such Mobo configurations ever exist?
Simply beacuse it's a mixed brand solution: AMD 761 northbridge and Via 686B southbridge.
Mixed brand chpsets means two things:
- the manifacturer spared a few cents by using a lower priced, pin compatible southbridge. This usually happens on cheaper micro-atx boards, and most of time the oem cuts on other parts too (capacitors' brand and capacity, vrm section, lower quality pcb and solderings, minor brand ICs ...);
- you can get easily troubles in installing chipset drivers: usually all-in-one chipset drivers for Windows install both northbridge and southbridge drivers. If you are lucky drivers don't fight with themselves but you should expect lower performances than a single brand solution, because low level parameters such as Pci timings are optimized for the last one. Linux drivers, on the other hand, relies heavily on chipset identification: non standard solutions can be easily misidentified, leading from low performances to data corruption or kernel panic at boot.
This applies on single brand, mixed old and new solutions as well: I remeber a few unreliable socket 7 boards with a Via MPV3-like northbirdge [seems an older revision] and a MPV2 southbridge. Both linux and windows drivers corrupt any data if Dma transfer is used, because they misidentify for a real MVP3 board and MVP3' way of dealing with Dma is different from the MPV2' one. Forcing the old Pio mode transfer avoids this issue, but it's so sloooooow.
Zandrax
Originally posted by gdement
Mixed brand chpsets means two things:
- the manifacturer spared a few cents by using a lower priced, pin compatible southbridge. This usually happens on cheaper micro-atx boards, and most of time the oem cuts on other parts too (capacitors' brand and capacity, vrm section, lower quality pcb and solderings, minor brand ICs ...);
- you can get easily troubles in installing chipset drivers: usually all-in-one chipset drivers for Windows install both northbridge and southbridge drivers. If you are lucky drivers don't fight with themselves but you should expect lower performances than a single brand solution, because low level parameters such as Pci timings are optimized for the last one. Linux drivers, on the other hand, relies heavily on chipset identification: non standard solutions can be easily misidentified, leading from low performances to data corruption or kernel panic at boot.
This applies on single brand, mixed old and new solutions as well: I remeber a few unreliable socket 7 boards with a Via MPV3-like northbirdge [seems an older revision] and a MPV2 southbridge. Both linux and windows drivers corrupt any data if Dma transfer is used, because they misidentify for a real MVP3 board and MVP3' way of dealing with Dma is different from the MPV2' one. Forcing the old Pio mode transfer avoids this issue, but it's so sloooooow.
Zandrax
<----Computer says I need more beer.
[damn *ç°"!&%£$"!)=!£)(!£"*é*DU{§§"E£"Y/ 
] : it's a dual socket 370, not a dual slot 2 one, you can't use your Xeons. 
what did we say about hoarding too much stuff and depriving others of them?! *cough* socialism *cough* lol! 
Comment