Just giving my feedback... I have had plenty of ASUS S478 boards on my hands in the last 3-4 years, I can't quantify how many, but I'd say rather plenty. And I must say I didn't encounter the problems described. I've tried plenty with various chipsets (Intel, SiS, VIA) and so far they have been rock solid. Can't remember a broken one honestly (except a few that were mishandled - various scratches on the back). I quite like the ASUS ones, usually they come with decent caps (Rubycon/Sanyo) and performance is ok. Here in Italy they were quite popular so they're easy to find and for nice prices.

SSTV2 wrote on 2023-06-11, 10:06:

A remnant of the solder ball in the left corner indicates that it's a BGA socket. You can instantly tell which type of mounting the socket is using by turning M/B over.

This is P4P800 it's indeed BGA. Last photo from ASUS P4SP (SIS chipset) with different socket soldering.

Roman555 wrote on 2023-06-12, 19:41:

First symbols of serial numbers of Intel ICH5 faulty chips:
32,33,34,40,41,42,43,44,45,46,47,48,49,4А,4В,4С

Strange. I was sure that serial number decodes like this:
First letter - facility code (L or F)
First digit - MF year
Next 2 digits - MF week

I had pritty good ratings with Asus sk 478 .... I need to recheck my boards. except for a few cap's here and there... maybe bios corruption? don't remember disasters.

The most problem from asus I had was with the P5N32-E-SLI Nforce680i chipsets , I broke 3 of them during waranty.

ediflorianUS wrote on 2023-06-16, 14:57:

The most problem from asus I had was with the P5N32-E-SLI Nforce680i chipsets , I broke 3 of them during waranty.

That's Nvidia Bumpgate, most of these boards are doomed anyway.

Roman555 wrote on 2023-06-12, 19:41:

First symbols of serial numbers of Intel ICH5 faulty chips:
32,33,34,40,41,42,43,44,45,46,47,48,49,4А,4В,4С

Is it known whether there ever were non-faulty ICH5 chips?
I just examined a P4P800-E Deluxe (one of the later 865PE/ICH5R boards) and it sports an F42 serial number on the southbridge.

When it is said that it was "supposedly" fixed in new revisions - does it mean that the chips were improved or that the board vendors applied workarounds?

does it mean that the chips were improved or that the board vendors applied workarounds?

Yes.

The Serpent Rider wrote on 2023-06-16, 19:39:

does it mean that the chips were improved or that the board vendors applied workarounds?

Yes.

This is either or question and was answered with one yes.

Which one is and please clarify yours for both questions?

Cheers,

The Serpent Rider wrote on 2023-06-10, 15:25:

ICH5 south bridge and USB ESD protection or the lack of thereof, to be precise. Well-known issue. ASUS and several other manufacturers supposedly fixed that in new revisions.

I/O Controller Hub wiki page says:

In particular, when connecting USB devices via front panels, the chips died by discharges of static electricity. Intel reacted to the problem by shipping ICH5 with increased ESD tolerance. Effective ESD preventive measures on USB ports are difficult and costly, since they can impair the quality of the USB-2.0 high-speed signals. Many motherboard manufacturers had omitted the necessary high-quality safety devices for front panel connectors for cost reasons.

Is it known does this issue affect mainly or only front panel USB? I've been looking for LGA 775 board with AGP but haven't yet pulled the trigger thanks to their price, but if the 80€+ board might croak just by plugging USB stick in I'm out.

I've been running a p4p800se and had no problems with it. Then again I don't use front USB headers with it.

Roman555 wrote on 2023-06-12, 19:41:

First symbols of serial numbers of Intel ICH5 faulty chips:
32,33,34,40,41,42,43,44,45,46,47,48,49,4А,4В,4С

Guys, that literally does not make any sense. I inspected huge pile of MoBos starting socket 7 and ending 775. Serial number always match XYWWddddd pattern. Where X - mf facility code (usually 8, F, L). Y - year, WW - week, d - some digits. So strictly speaking first symbol of S/N is "F" not "3" of "4". I can see you guys just drop "F" to make it more sense. But it doesn't. "3" and "4" look pretty decent for mf year. But for first digit of a week valid values are 0,1,2,3,4,5. Not 6,7, 8,9,A,B,C.
This sequence from ixbt forum originates to rom.by forum which is long gone.

BreakPoint wrote on 2023-06-17, 07:07:

Roman555 wrote on 2023-06-12, 19:41:

First symbols of serial numbers of Intel ICH5 faulty chips:
32,33,34,40,41,42,43,44,45,46,47,48,49,4А,4В,4С

Guys, that literally does not make any sense. I inspected huge pile of MoBos starting socket 7 and ending 775. Serial number always match XYWWddddd pattern. Where X - mf facility code (usually 8, F, L). Y - year, WW - week, d - some digits. So strictly speaking first symbol of S/N is "F" not "3" of "4". I can see you guys just drop "F" to make it more sense. But it doesn't. "3" and "4" look pretty decent for mf year. But for first digit of a week valid values are 0,1,2,3,4,5. Not 6,7, 8,9,A,B,C.
This sequence from ixbt forum originates to rom.by forum which is long gone.

Agree. The number appears to be called "FPO number". I've seen a letter in 6th position. So maybe those numbers are about 5th-6th positions of FPO numbers
I've found out this:

1st letter or digit = plant code: 0 = San Jose, Costa Rica 1 = Cavite, Philippines 3 = Costa Rica 6 = Chandler, Arizona 7 = Phil […]
Show full quote

1st letter or digit = plant code:
0 = San Jose, Costa Rica
1 = Cavite, Philippines
3 = Costa Rica
6 = Chandler, Arizona
7 = Philippines
8 = Leixlip, Ireland
9 = Penang, Malaysia
L = Malaysia
Q = Malaysia
R = Manila, Philippines
Y = Leixlip, Ireland

2nd digit = Year of production

3rd & 4th digits = week

5th - 8th digits= lot number (batch number )

10th - 13th digits = serialization code

Still doesn't make much sense. I mean, usually such changes would be represented in changed stepping/revision number.

Ok. I found explanation for this weird sequence. Original text form ixbit forum says "Serial numbers of faulty chips". But later on this topic they ran into same trouble as we did 😀 So other guy corrected this statemen. It's not "serial number of chip" it's "serial number of ASUS motherboard". Guess it's one from a sticker on LPT port.

Ok, so only one board in my possession is affected, although both are Revision 2.0.

BreakPoint wrote on 2023-06-17, 17:43:

Ok. I found explanation for this weird sequence. Original text form ixbit forum says "Serial numbers of faulty chips". But later on this topic they ran into same trouble as we did 😀 So other guy corrected this statemen. It's not "serial number of chip" it's "serial number of ASUS motherboard". Guess it's one from a sticker on LPT port.

It's really interesting. Mine has a s/n started from 4A . It's in working state so far

Roman555 wrote on 2023-06-17, 18:44:

BreakPoint wrote on 2023-06-17, 17:43:

Ok. I found explanation for this weird sequence. Original text form ixbit forum says "Serial numbers of faulty chips". But later on this topic they ran into same trouble as we did 😀 So other guy corrected this statemen. It's not "serial number of chip" it's "serial number of ASUS motherboard". Guess it's one from a sticker on LPT port.

It's really interesting. Mine has a s/n started from 4A . It's in working state so far

I may actually consider using a PCI USB card and avoid using the onboard ones altogether, if it turns out one of my s478 boards is one of the affected ones and still in working condition. If it can help to prevent sucha board from dying, then why not?
Not sure in how much the USB ports behind the i/o shields are affected btw.

Very interesting. I was installing sound blaster drivers one day on a P4P800 not long ago and it just froze, and never came back on.

Tetrium wrote on 2023-06-17, 21:50:

Not sure in how much the USB ports behind the i/o shields are affected btw.

AFAIK the main culprit is front IO port(s). ATX IO is shielded better.

I have two boards that are both p4p800 le and all caps are still good and both boards have been well used over the years. Both still work perfectly although on both the front usb were never hooked up. The board flex is true as both when removed and inspected had the taco warp. I did not effect performance though. I personally think these are great quality boards for the time.

Tommaso