socket 775 forever... no, no, no...

[eccerr0r]

Sigh. Now I'm getting a whole bunch of S775 boards. Luckily I got rid of one...

Wondering what people thinks about chipsets for S775 boards and their long term longevity. I had a few boards that failed.

MSI Neo3 (865) - gone but it was working (ATX)
Asus ??? (915) - crashes randomly (ATX)
Foxconn G9657MA (G965, onboard graphics) - dead (µATX)
Gigabyte EP43-UD3L (P43) - bad USB ports but otherwise still works (ATX)
Gigabyte G41M(???) (G41) - still working (µATX, only 2 DDR2's)
Intel DQ45CB (Q45, onboard graphics) - still working (µATX)
Foxconn P43AL (P43) ... - still working (ATX)
Biostar I45 (P45) - still working (ATX)

Wonder which of these one would gamble on lasting the longest, whether I should just keep on using the DQ45CB in my server. I suspect the crashing and dead boards are truly DQ'ed, but anyone have an opinion if any board would last longer, perhaps I should just stick with the µATX but having more SATA and PCI ports is tempting for a server...

It's too bad none of these boards will take x4 ddr2 chips so no matter what I'm stuck with 8GB if I don't get $ARM$LEG$ fancy 4G DDR2's

[piernov]

Re: socket 775 forever... no, no, no...

I've seen a lot of P4/Core 2 boards, very rarely seen a failed Intel ICH or MCH, like a couple of "randomly" shorted ICH (possibly caused by external events), one shorted MCH, and one GMCH acting up randomly. Also with these if you have the tools and skills they're not too difficult to replace and no trouble buying them.
Of course there have been a lot of other failures on these boards (anything can fail really), and not particularly more from one manufacturer.
Other chipset vendors, thankfully less common, are more troublesome both from a hardware and a software perspective.
I'd worry more about having an adequate CPU VRM and a proper BIOS without too many issues, as well as avoiding exotic controllers with poor drivers.

Consumer Intel chipsets aren't designed to support x4 RAM ICs so no surprise about the RAM there. If you want a lot of RAM on a Core 2 platform you should rather be looking at workstation-class hardware that supports registered RAM.

[eccerr0r]

Re: socket 775 forever... no, no, no...

Hmm... just a bit weird, of those 8 S775 boards I've had, three had problems. Pretty bad hit rate. Some of these I acquired second hand, but some I acquired new back when.

Just wondering which I can expect to not die on me if I left it running 24/7... Not sure if S775 was the worst I had in terms of long term reliability or I just had so many of them.

Next I need to redistribute the S775 chips which is yet another issue, though I have a bit more faith in the CPUs longevity... chips run the gamut from P4's to Core2 Quads...

[Topcat]

Re: socket 775 forever... no, no, no...

I have piles of LGA775 stuff....all working... Intel boards tend to be my favorite of the LGA775 flavor, as based on their chipsets, they tend to support Xeons mods with no firmware trickery.

[eccerr0r]

Re: socket 775 forever... no, no, no...

FSCK. The Foxconn P43AL just died on me after trying to install a PCIe GPU. No more POST.

[Dan81]

Re: socket 775 forever... no, no, no...

I still have a semi-dead ABIT iP35 Pro somewhere in my closet. It worked absolutely fine for a while w/ a C2D E6750, HD5450 1GB, 4x1GB sticks and an recapped Allied 8400BTX PSU, then one day it started to give errors on POST. It would hang at various stages on the port 80h LED display - usually C3, 25, 52 and very rarely 7F. It struggles to boot successfully regardless of the RAM sticks I throw at it, be it Corsair XMS sticks or just plain generic stuff.

Surprisingly, the most responsive board I've had, which is now a complete machine, is an ASUS P5E-VM DO. Same stuff that was on the IP35, just moved over and replaced the C2D with a Quad Q9400, as well as the GPU with a ASUS HD4850 512MB (though I have an untested R7 265/HD7850 2GB that I might swap in, if I have the luck to get it to work.)

[kevin!]

Re: socket 775 forever... no, no, no...

I gave my friend my trusty old Gigabyte ep45-ds3, it's 14 years old, it hasn't failed at all yet.
I've seen more socket 478, 462 boards fail than socket 775.
But 775 had many different chipsets, the ones I've seen fail the least are from Intel, the only dead intel chipset I've seen from Intel was a G31.
Although the G31 had 60,000h of use.
The most reliable boards for me are the socket 775 ones that use os-con capacitors all over the PCB, those boards never die. (For example the Gigabyte EP45-DS3 uses os-con capacitors.)

[eccerr0r]

Re: socket 775 forever... no, no, no...

Supposedly my EP43-UD3L uses all polymer capacitors but the USB ports died. Not just one or two, but all of them...implying a chipset issue whether or not it was damaged by ESD (which it should have some protection if it was a well designed board?) The funny thing is that I don't think I even used the USB ports... just that the board now fails to POST if I don't disable the USB ports.

My Foxconn G9657MA croaked due to bad capacitors I think, as well as extended high temperatures most likely. But I don't think temperatures were much unlike the DQ45CB situation...

Hmm... actually it wasn't the Foxconn P43AL that I was working with, it was the Biostar I45 that died for a moment... for a moment, as it seemed to have woken back up from the dead. I think there are some cheap PCI/PCIe sockets on it or something, or it doesn't tolerate being outside of a case very well.

[piernov]

Re: socket 775 forever... no, no, no...

I personally think LGA775 boards with Intel chipsets are decently reliable. Some of them may be worse than others, and if they have electrolytic caps there's a high chance they'll have to be replaced.

Doesn't mean they cannot fail. I've seen a lot of different issues that can happen to any motherboard like burnt Ethernet controller (pretty common actually due to lightning damage) with shorted TVS diodes if they are present and sometimes burnt transformer, burnt audio IC (can be caused by people plugging in device with phantom power), burnt fuse or shorted TVS diode on USB (physical damage to USB connector or bad USB device), short in RAM slot with burnt trace, other physical damage with ripped off components, damaged traces or whatever.
Of course these are mostly caused by external damage. But there's also the typical corrupt BIOS, randomly shorted power MOSFET (CPU VCore, fan, other voltage regulator), dead buck controller, leaky RTC dual diode, shorted MLCC or random IC, maybe a dead SIO or two, and probably a lot of other stuff I'm forgetting about. Failed sockets can happen too. But nothing surprising compared to any other boards, and almost always repairable.

The main issue apart from these typical failures is that with older boards like this and older RAM sticks, it is *very* common to have bad contact, sometimes hard to get to work reliably, and can fail again when moved around. Also applies to PCI slot and sometimes AGP/PCIe but not as bad.

Intel chipset reliability is really an issue since 8-series (4th gen CPU).

And for CPU, I have never seen a single dead Pentium 4 or Core 2, even ones that took 12V multiple times due to shorted high-side MOSFET. I only have one that's not in working order right now because I took the bottom caps to fix another where they were ripped off, but that's not difficult to repair.
I store the LGA775 CPUs lose in random bag without any kind of protection. Of course don't do that with pins or newer CPUs, the pads and the substrate are much easier to damage, and newer CPUs also die randomly sometimes.

[Dan81]

Re: socket 775 forever... no, no, no...

Burnt Ethernet chips were a thing since a lot of time

eccerr0r - Biostars were cheap as hell since ever IMO, up there with PCChips if not even worse. Their only plus were that in my experience, they were very tolerant to what components you used.

Some of the stuff I've encountered with 775 boards:

ABIT IP35-E - quietly died despite me using a Delta PSU on it. Wouldn't power on although 5vsb leds showed up, and 5VSB on the PSU was in spec

MSI P6N SLi - F v2 - dead as a doornail. IIRC the thing that was shorted on it was MSI's own monitoring chip

ABIT IP35 Pro listed above - this one only worked with a sketchy as heck MOSFET contraption done to it and was probably its demise. It worked pretty nice when it did. Might take a look at what can be fixed on it, and if nothing else then I'll probably look for another IP35 Pro.

ASUS P5ND2 - I swear I'm at loss as to what this one has wrong with it. It legit won't power on at ALL - 5vsb LED comes up, but front panel power switch doesn't even work. Shorting GREEN and BLACK on the main ATX power literally powers it on and POSTs just fine

ASUS P5B-VM DO - this one was DOA. Wouldn't POST and was overly poor in features as opposed to the actual P5E-VM DO I run my Q9400 on.

DFI Lanparty P45-T2RS - nothing much to say about this. Would occasionally get RAM issues with it since I was running 3 Corsair XMS sticks and a puny ADATA stick alongside

Gigabyte P35-DS4 - scrapped, had a burnt CPU clockgen. Surprisingly everything else (4GB worth of GeIL RAM w/ LEDs, Q6600, Scythe cooler) survived, so dunno whatever PSU took a big phat dump on the ISL clockgen, because I suppose that thing ran on at least a FSP or a Seasonic, judging by the other parts that the seller had (GTX 560, HD7950 and HD5970)

Gigabyte 945GZM-S2 - scrapped after I flashed its BIOS chip and transplanted it to an Acer 5740 that had a dead EC BIOS (this was rather long before I got the CH341A programmer I have to this day.). Also had a zapped Realtek LAN chip that I removed from it a year before scrapping it, so nothing of value was really lost.

ASUS P5K-V - bad socket BGA.

Ones that I still have working:

MSI G41M-P26 - originally wanted to do a compact Quad build with it, but the fact that DDR3 wasn't cheap didn't leave me much choice but to set it aside for another day, maybe a low power Core 2 Duo or Pentium build w/ whatever craptacular discrete GPU I can assign to it.

MSI 865PE Neo3-V - on my to do list - it's going to be a quite good candidate for a "dual core" AGP machine, and hopefully I'll manage to ever find a good heatsink to stick a Hercules 9700 Pro 128MB on this puppy. It's small and compact (almost the size of an run-of-the-mill Socket 370 mobo!) but it can totally pack a punch if done right IMO.

ASRock 775i65G - retired. Supposedly has Kentsfield support although I don't really buy it... also the BIOS is pretty crippled in comparision with the MSI.

ASUS P5E-VM DO - currently my secondary PC, running Win10 Enterprise LTSC on a 500GB Seagate (not my wisest choice in matter of HDD, but considering I didn't pay much for it and it's still got life left in it, why not. It's a 7200.10 by the way.), along with a HD4850 512MB, Quad Q9400 and 4GB worth of DDR2. Surprisingly NFS Hot Pursuit runs fluently on this - I've seen Sandy and Ivy Bridge machines (not necessarily powerful, though not iGPU level either.) stutter in this game and this Q9400 just runs it like it was made for it.

ECS G41T-M5 - this thing is just tiny. Shame it doesn't run RAM higher than 800MHz, even when using a 1066FSB CPU.

ASRock G41M-VS3 - similar to the ECS but actually has an competent BIOS. Can clock rather well but had issues with it where it would corrupt the screen all of a sudden. Not sure if it was because I was stupidly trying to run a HD7870 + Q6600 on a "500W" Allied (which would probably top out at 350W and nothing more) or if my DDR3 sticks were rather iffy. (Kingmax and GoodRAM aren't something I should be proud of - I'm thankful these two are running properly in a GB H55M-S2V + i3 540 combo. Will probably drop in a HD7850 just to see how it fares today and will either sell it or replace the Q9400 machine with it and sell the Core 2 Quad)


Ones that I've sold:

-Gigabyte EP45T-DS3R (complete w/box)
-ASUS P5QL Pro (in a build)
-ASUS P5QD Turbo
-ASUS P5GC-MX/1333 (office build w/ E5700)
-ECS G31T-M7 (traded for a retro Slot A EPOX EP-7KXA build that I still have to this day)
-DFI Lanparty P45-T2RS (traded for a MSI GTX750Ti 2GB that I later sold)
-MSI P965 Platinum (traded for a modded 500GB OG Xbox locally - I just wanted to get rid of it because it was a horrid nightmare to replace all the KZGs it had on it.)
-ASUS P5K-SE/EPU (traded for a silver PS2 when I was still in highschool)


And that's pretty much it. The list could go MUUUCH longer if we count AMD boards as well, but that's for another day.

[eccerr0r]

Re: socket 775 forever... no, no, no...

still trying to milk computational bandwidth out of these machines, so I need to get 2GB RAM per core on my machines. Which is unfortunate as that 2-DDR2 slot board has a quad in it. Really should stick a duo in that, and put the quads in boards that have 4 DDR2 slots.

That board is kind of far away so I can't mod it...

[stj]

Re: socket 775 forever... no, no, no...

the problem with intel chipsets is the 945, 965 and maybe others get far too hot.
i always fit a fan on them.

[eccerr0r]

Re: socket 775 forever... no, no, no...

Do only the IGP chipsets get hot or all of them including the ones that don't have a GPU onboard?

wondering if the, say, G41 should get hotter than a P43... shouldn't the G41 have many more transistors on it...

Then there's the Q-series ones with GPUs but they are not clocked the fastest at all...

[stj]

Re: socket 775 forever... no, no, no...

g41 gets hot as fuck.

i dont think it's about transistor count, but the size of them.
the "nm" die size

[eccerr0r]

Re: socket 775 forever... no, no, no...

Well the G41 and P43 should be the same process, I would think.

[piernov]

Re: socket 775 forever... no, no, no...

I haven't noticed these Intel GMCH getting unexpectedly hot, unlike the ATi stuff of that era that's also much less reliable, and NVidia too.

[ribcage]

Re: socket 775 forever... no, no, no...

I have Asus P5L 1394 without onboard graphics and the Intel 945 Northbridge gets too hot to touch even on idle, stock Intel CPU cooler pushes some air through the northbridge's heatsink, but if you have a cooler that does not, then the northbridge is totally overheating.

[piernov]

Re: socket 775 forever... no, no, no...

You can feel a burning sensation at 45-50°C, although of course the die will be slightly hotter than the heatsink. Even then, I personally haven't noticed the heatsink getting burning hot right away on any board I worked on. 82945G is rated up to 99°C with a TDP of 22.2W.
Of course it requires a heatsink that's not too small, and would benefit from a little bit of airflow either from CPU fan or case fan, but shouldn't be anything of concern.

[eccerr0r]

Re: socket 775 forever... no, no, no...

Hopefully TDP > idle power and thus should not be full heat.

So far each of the GMCH's I've dealt with they do indeed get quite toasty, though I haven't had a working non-GMCH for a while to compare to.

The interesting thing about the Biostar board is that it has a heatpipe on the MCH. I don't think any of the other boards have fans or heatpipes including the GMCHs...

[Dan81]

Re: socket 775 forever... no, no, no...

I don't think any of the other boards have fans or heatpipes including the GMCHs...

From what I remember, my ABIT's IP35 Pro and its Pro XE variant also have a heatpipe going all through the board, from the VRMs, to the GMCH and to the southbridge as well.

Same goes for the Lanparty P45-T2RS I've had. VRMs and GMCH had a heatpipe going in a sort of a loop.