sliderider wrote:

i740 cards in AGP are actually more common than you think.There's usually between two and ten of them for sale on ebay USA on any given day. It's only the PCI cards that are anywhere close to rare because they weren't released until right before Intel decided to give up on marketing i740 as discrete graphics and integrated it into their chipsets instead.

There's a bunch of them here right now and I'm sure I could find more by broadening my search.

I know that the Starfighter PCI is considered rare.
OK, so i740 cards are somewhat common in the US. When I had a look on German ebay I had to wait quite a while until one appeared.

So it seems that the same stuff is not common everywhere. For example I rarely see a Virge GX/2 here, Virge 325/DX/GX are common, Trio3D (/2X) are common. In the US GX/2 doesn't seem to be too rare. And I had the impression that Voodoo4/5 are more common in the US, although it doesn't seem so at the moment on ebay.

I'm testing some parts now, starting with the P3B boards because they seem most interesting. Both POST and appear to work okay, and even still store the current time although one is nearly a week behind. That last one seems a bit flaky too, because if I pick it up from the table while powered on, the cpu fan sometimes slows down to about half speed and the system will hang. Might just be improper grounding or something... Or maybe I shouldn't touch it while running in the first place 🤣

One other thing I noticed, with both boards, after powering up it takes about ten seconds before the POST screen appears. Is that normal? On my P2B I'm nearly on the DOS prompt at the 10 second mark.

All graphics cards are still good, but the i740 flashes (changes resolution?) four or five times before the POST screen. Seems weird but then this might be normal for an i740.

I tried a 256MB module from the stack and to my amazement the P3B detected its entire 256MB. So I tried the 256MB ECC modules I normally use and like the P2B it still only detects half of it. Anyone knows why?

edit: if anyone's interested, I'm editing the first post as I test new parts. So look there if you want to stay updated on this 😀

Touching motherboards when they are running is not something I would recomennd 😀

I was tierd a few days ago and must have hit the power supply switch twice.
I was very surprised when the motherboard powered on when I inserted a memory module.
No harm done though.

Its not the ECC that your board has problems with.
The BX chipset only accepts double sided 256mb modules.
Single sided 256mb modules or double sided 512mb modules will be detected as half their size.

Skyscraper wrote:

Touching motherboards when they are running is not something I would recomennd :) […]
Show full quote

Touching motherboards when they are running is not something I would recomennd 😀

I was tierd a few days ago and must have hit the power supply switch twice.
I was very surprised when the motherboard powered on when I inserted a memory module.
No harm done though.

Happened to me once too, when I installed a CPU with the PSU still connected. Didn't know that could happen with RAM too!

Skyscraper wrote:

Its not the ECC that your board has problems with.
The BX chipset only accepts double sided 256mb modules.
Single sided 256mb modules or double sided 512mb modules will be detected as half their size.

The ECC modules are double sided, actually. There are nine chips on each side (not counting the ECC chips) while the non-ECC module has eight.

jwt27 wrote:

The ECC modules are double sided, actually. There are nine chips on each side (not counting the ECC chips) while the non-ECC module has eight.

Then I think the modules use chips with 32Mx4 design and not the normal 16Mx8.
Those are not fully supported either.
The extra chip might be a sign that the modules are registered, BX useally accepts those but with the limitations above.

Skyscraper wrote:

Then I think the modules use chips with 32Mx4 design and not the normal 16Mx8. Those not not supported either. The extra chip mi […]
Show full quote

jwt27 wrote:

The ECC modules are double sided, actually. There are nine chips on each side (not counting the ECC chips) while the non-ECC module has eight.

Then I think the modules use chips with 32Mx4 design and not the normal 16Mx8.
Those not not supported either.
The extra chip might be a sign that the module is registered, BX usally accepts those but not the chipkill function.

That must be it then, I have some single-sided 128MB ECC modules too which are correctly detected. These explicitly state "8/16Mx72" on the PCB. And yes, both are ECC/Registered modules.

Kinda weird though, both use 7.5ns chips but the 256MB modules are rated PC133 CL3 and the others PC100 CL3. The latter refuse to run at 150MHz CL2 while the first work okay at that speed.

jwt27 wrote:

That last one [P3B] seems a bit flaky too, because if I pick it up from the table while powered on, the cpu fan sometimes slows down to about half speed and the system will hang. Might just be improper grounding or something... Or maybe I shouldn't touch it while running in the first place 🤣

One other thing I noticed, with both boards, after powering up it takes about ten seconds before the POST screen appears. Is that normal? On my P2B I'm nearly on the DOS prompt at the 10 second mark.

Some months ago I was messing with an Asus P2B. It never booted, until one time I put some pressure on the board. Turned out that the back of a Mosfet had come loose, while the legs still kept it in place, as if it was OK. IIRC it took long to boot when the Mosfet was pushed against the board, but booted fine again when properly soldered to the board.
May be an idea to check the Mosfets on your asus boards.

The SuperMicro Socket T on the right looks like a BTX board. Finding a case that'll fit it properly is kind of a chore (especially if you don't want an old Gateway or Dell tower case). As far as making a DX9 machine from them - there are PCI Express graphics cards that don't require x16 slots, and I also remember reading about folks modifying the slot to accept longer cards (although I've never tried this one personally; a board you got for free sounds like a good candidate for such an experiment though...if you're so inclined).

Quick search for x1 graphics cards yields these two:
http://www.amazon.com/Zotac-64-Bit-Video-Grap … /dp/B00CIDFP70/
http://www.amazon.com/HIS-Silence-DisplayPort … /dp/B004GL72K0/

Would probably be fine for earlier DX9 games that the Pentium 4 would be a good match for at least. Probably worth looking the two boards up on SuperMicro's site as well - I get the feeling the BTX board is the better spec'd of the two based on its power connector arrangement. But that could be entirely mistaken; it's just a guess. One or both may support dual-core processors as well.

The HP server board looks like it takes Socket 603/604 Xeons, but I'd probably just return it for scrap - it also looks like it's survived a fire event of some sort and been stripped for parts or physically damaged (its missing some retaining latches, the battery, potentially a few caps, etc), and I'm guessing it's probably incomplete as well (it probably has a riser card that provides expansion slots and all those black plugs probably connect to a backplane or something).

The PCI IDE controller looks like a Rosewill - you'll have to look closely to determine if its a RAID capable model (there's a few versions of that card, some do RAID, some don't). The non-RAID version that I've used in the past was pretty good for what it did, but does add to the system start-up time. I have no experience with the RAID enabled version, but I don't think it's a complete hardware solution.

The Asus Socket 478 board has a Radeon 9100 IGP for its chipset (the ATI southbridge gave it away). In my experience ATI chipsets (and their associated drivers) were never that great; but I have not used the 9100 IGP specifically (I've used/seen a few Xpress series chipsets, and generally the machines were on the temperamental side of the spectrum). Here's a review if you're curious:
http://www.pcstats.com/articleview.cfm?articleID=1601 The "SurroundView" feature looks pretty cool - assuming you've got a Radeon 9xxx sitting around to hook it up. Depending on how the IGP interfaces with the rest of the system, you might be able to get SoftTH to run with reasonable performance in some games on there. 😀

obobskivich wrote:

The SuperMicro Socket T on the right looks like a BTX board. Finding a case that'll fit it properly is kind of a chore (especially if you don't want an old Gateway or Dell tower case). As far as making a DX9 machine from them - there are PCI Express graphics cards that don't require x16 slots, and I also remember reading about folks modifying the slot to accept longer cards (although I've never tried this one personally; a board you got for free sounds like a good candidate for such an experiment though...if you're so inclined).

I just had a look at both boards again and now I don't know why I called the left board a SuperMicro.. I can't even find any brand or model name on it.
The board on the right is a SuperMicro X7SBE rev 1.1 and it's in normal ATX form factor. A quick search reveals that it's still in production and costs about €250 new 😳 Can't image why anyone would scrap this stuff instead of selling it.
Their website confuses me a bit. On the board is printed "UNB DDR2 ECC ONLY" while the website states non-ECC works too. I assume the Xeon I got came from this board since it supports that.
On second thought, I might just test and sell these two boards, I don't have all that much use for them really.

obobskivich wrote:

The HP server board looks like it takes Socket 603/604 Xeons, but I'd probably just return it for scrap - it also looks like it's survived a fire event of some sort and been stripped for parts or physically damaged (its missing some retaining latches, the battery, potentially a few caps, etc), and I'm guessing it's probably incomplete as well (it probably has a riser card that provides expansion slots and all those black plugs probably connect to a backplane or something).

Yeah it's incomplete, and is probably both useless and worthless without all its proprietary parts. I don't have any way to test it and I don't think anyone would want to buy it in this state either.

I don't see any fire damage though, aside from the brown stuff at the power connector maybe. The PCB is just black in colour and the blue spot in the middle is a reflection from the window above me 😉

obobskivich wrote:

The Asus Socket 478 board has a Radeon 9100 IGP for its chipset (the ATI southbridge gave it away). In my experience ATI chipsets (and their associated drivers) were never that great; but I have not used the 9100 IGP specifically (I've used/seen a few Xpress series chipsets, and generally the machines were on the temperamental side of the spectrum). Here's a review if you're curious:
http://www.pcstats.com/articleview.cfm?articleID=1601 The "SurroundView" feature looks pretty cool - assuming you've got a Radeon 9xxx sitting around to hook it up. Depending on how the IGP interfaces with the rest of the system, you might be able to get SoftTH to run with reasonable performance in some games on there. 😀

I'll give it a try and see if it's any good. That Surroundview looks really cool but I don't even have enough space for two CRTs here, let alone three! 🤣

gerwin wrote:

Some months ago I was messing with an Asus P2B. It never booted, until one time I put some pressure on the board. Turned out that the back of a Mosfet had come loose, while the legs still kept it in place, as if it was OK. IIRC it took long to boot when the Mosfet was pushed against the board, but booted fine again when properly soldered to the board.
May be an idea to check the Mosfets on your asus boards.

Had a quick look, all of them appear to be well connected to the boards. I'll try pushing them down a bit while running, see if I can replicate the issue that way.
Have you seen my updates on the first post? I can't get this P2B to power up either...

jwt27 wrote:

One other thing I noticed, with both boards, after powering up it takes about ten seconds before the POST screen appears. Is that normal? On my P2B I'm nearly on the DOS prompt at the 10 second mark.

I believe this is normal if the BIOS settings have been lost and the board have to reconfigure to default settings on boot. Happened to me when the system had been powered down for some time and was running on the original coin battery (which was getting bad). Can also happen if you change alot of hardware and boot up.

The long bootup time (for POST) should only happen once though! If you power down and on again it should show POST screen immediately.

jwt27 wrote:

I just had a look at both boards again and now I don't know why I called the left board a SuperMicro.. I can't even find any bra […]
Show full quote

I just had a look at both boards again and now I don't know why I called the left board a SuperMicro.. I can't even find any brand or model name on it.
The board on the right is a SuperMicro X7SBE rev 1.1 and it's in normal ATX form factor. A quick search reveals that it's still in production and costs about €250 new 😳 Can't image why anyone would scrap this stuff instead of selling it.
Their website confuses me a bit. On the board is printed "UNB DDR2 ECC ONLY" while the website states non-ECC works too. I assume the Xeon I got came from this board since it supports that.
On second thought, I might just test and sell these two boards, I don't have all that much use for them really.

I recently bought a used X7SBE on eBay for $30 shipped. It's a pretty decent server board. Non-ECC RAM does work in it.

Quick update:

I gave the Athlon64 X2 5000 to my sister and swapped it with the Athlon64 3600 in her "old" PC. Works fine and it's quite a bit faster now 😀

Tested the Intel socket 7 board too which I've identified as Intel Advanced/EV Endeavour. Even with the missing cap it still works, but I intend to recap it anyway before I decide do anything with it.

Tried one of the P3B-F boards again too (the one that seemed unstable, and where the clock was a week behind). I restarted it a few times without changing any parts and each time I had to wait about 10 seconds before it would POST. So that seems wrong, to start with.
Next I tried pushing on some mosfets, starting with the ones in the top-right corner. On the first try I got it to crash, on a second try too but then the PSU started making funny noises. And now I can't get it to power up again. I resoldered the mosfets, swapped cpu, ram, power supply and graphics card, but it doesn't do anything anymore. When I switch it on, the keyboard lights blink once and the on-board LED lights up, but that's it 🙁
Any ideas what to do now?

Wow, so it seems to react to pushing and soldering the mosfets, but not in a positive way at all. sorry to hear that.
I found my old notes on the Asus P2B (knowing you are dealing with a P3B-F):
"Asus P2B Before the Mosfet fix: No post with just blinking power and HDD LED, no Fans spinning up, no Beeps, and the 'Fault' pin of the VRM chip being 'high'."

Also, I found the the P2B can typically 'play dead'. After trying a VIA CPU on a cheap slotket for example (I blame the slotket). It then seems to be dead afterwards, but when trying again the next morning; it works again. I don't like these jokes.

Xeon x3220 SLACT is a rebranded Core Quad Q6600, one of the best CPUs for early 1066 Mhz LGA 775 boards. I own an x3230 and it outperforms older core i3s and runs a little slower than my i5-450M. Definitely keep it!

Am I the only one who looks at that Intel 486 board, sees all the horrible damage to it and thinks "what if it WORKS?" 0.o

And you should probably keep the riser board from 486 mobo!

jwt27 wrote:

Tried one of the P3B-F boards again too (the one that seemed unstable, and where the clock was a week behind). I restarted it a few times without changing any parts and each time I had to wait about 10 seconds before it would POST. So that seems wrong, to start with.

I've experienced that with a few boards over time, what usually fixes it is putting in a new clock battery.

Trying the s478 board right now, had hoped it would make a nice
replacement for my microATX board... But I can't get it to do anything
more than BEEEEP indefinitely. I'm posting this from Arachne now which
sucks, will reply to you guys once I get my P4 running again.

jwt27 wrote:

Trying the s478 board right now, had hoped it would make a nice replacement for my microATX board... But I can't get it to do an […]
Show full quote

Trying the s478 board right now, had hoped it would make a nice
replacement for my microATX board... But I can't get it to do anything
more than BEEEEP indefinitely. I'm posting this from Arachne now which
sucks, will reply to you guys once I get my P4 running again.

What CPU did you test it with? It could be that the CPU isn't supported somehow. I checked the ASUS website and apparently for Prescotts it will need a BIOS update. Perhaps that's the cause of the beeping??
If that isn't it, try to clean up the board a bit before you test any further and reseat CPU, HSF and perhaps your graphics card.

I always use some slow very compatible CPU for testing boards because they just work most of the time, are well documented, eat less power then the high end parts and are expendable 😁

Tetrium wrote:

What CPU did you test it with? It could be that the CPU isn't supported somehow. I checked the ASUS website and apparently for P […]
Show full quote

jwt27 wrote:

Trying the s478 board right now, had hoped it would make a nice replacement for my microATX board... But I can't get it to do an […]
Show full quote

Trying the s478 board right now, had hoped it would make a nice
replacement for my microATX board... But I can't get it to do anything
more than BEEEEP indefinitely. I'm posting this from Arachne now which
sucks, will reply to you guys once I get my P4 running again.

What CPU did you test it with? It could be that the CPU isn't supported somehow. I checked the ASUS website and apparently for Prescotts it will need a BIOS update. Perhaps that's the cause of the beeping??
If that isn't it, try to clean up the board a bit before you test any further and reseat CPU, HSF and perhaps your graphics card.

I always use some slow very compatible CPU for testing boards because they just work most of the time, are well documented, eat less power then the high end parts and are expendable 😁

Used the same CPU as in my other board, Northwood HT 3.2GHz. I think the
long beeps mean that the RAM is not detected or faulty.

Just reassembled the system with my old MSI board, managed to boot once,
and now I get the same beeping as with the Asus board.
Sorry for swearing, but... GODDAMNIT!

I suddenly feel very compelled to buy a new i7 machine. Let's hope Arachne
has enough Javascript support to let me order parts in a webshop.