Adrian_ wrote:

Thanks a lot, I'll see what I can work out with the shipping. (I'm in Europe but the shipping still costs 50% more then the item […]
Show full quote

Thanks a lot, I'll see what I can work out with the shipping. (I'm in Europe but the shipping still costs 50% more then the items).

I'm thinking maybe at purchasing 4 of these since I actually have a second AP550 mobo at home (but sadly lacking the VRM for the second CPU, which means more $ particularly for shipping).

I have already modded one Tualatin for an Asus cubx (did the mod on the CPU) Problem running Asus CUBX at 133FSB with 1Gb RAM so working on the slotkeys should be much easier.

Btw, did you tried Win7 and 2Gb RDRAM on your machine?

Well, it should still be around €20 for 2 new in box MSI slotkets. Not a bad deal, in my book.

I have not tried W7 on this machine, I am quite happy with a Win 2000 and Win ME dual-boot at the time. I think the machine is perfect for a Win 2000 or Win XP rig.

About your mod: Vtt_PWRGD should not be connected to PWRGOOD. I know its all over the internet and it "works", but according to the datasheet;

"The PWRGOOD signal input is a 1.8 V signal level and must be pulled up to VCC_CMOS1.8 . The VTT_PWRGD is not 1.8 V tolerant and must be connected to VTT ( 1.25 V) . Other CMOS inputs (A20M#, IGNNE#, INIT#, LINT0/INTR, LINT1/NMI, PREQ#, SMI, SLP#, and STPCLK#) are only 1.5 V tolerant and must be pulled up to VCC_CMOS1.5 . The CMOS, APIC, and TAP outputs are open drain and must be pulled to the appropriate level to meet the input specifications of the interfacing device."

pag 22; Intel ® Pentium ® III Processor Based on 0.13 Micron Process Up to 1.33 GHz Datasheet, Document Number: 249765-002

Unless you modify the supply going into the CPUs, your VTT will already be above spec at 1.5 V, there's no need to stress the CPU even further.

Yes, I know what you mean, but I never read any report of a Tualatin being killed/damaged by this mod. Pretty much the same thing like all those nVidia AGP cards running for years in overclocked Bx boards without a glitch.

I suppose I could move the wire to the correct pin but honestly there's a very good soldering done there (wire twisted all te way around the pin then soldered) so I'm a bit lazy about re-doing it.

A recap after all…

I recently started a new project and one of the first things to do was recap the whole motherboard. Since I needed to add some extra dollars to the order to get free shipping from mouser (recommended, btw, super-fast shipping and delivery), I figured I'd order new caps for this board. Behold,

The old Sanyo OS-CON caps. They were still in good shape so I'll save them.

oldOSCONs.JPG

The OR840, mid-surgery.

readyForSurgery.JPG

The new bling: Nichicon organic polymer capacitors with even LOWER ESR than the OS-CONs

aftermath.JPG

I'll test it out and report back.

(EDIT: this was my 100th post! A big thanks to the community for the support and ideas while building/modding this system.)

wow, not many caps...

luckybob wrote:

wow, not many caps...

Thankfully, makes for easy maintenance (the board of my other project has 54 capacitors to exchange… it made up for this one).

The system boots up properly, btw. Since I am currently recapping another board I haven't had time to test more than that, but at least I didn't outright fry it 😉

ACPI Woes

In a previous post I had mentioned some trouble when trying to install the ACPI HAL under Windows XP: a normal installation would recognize the computer as an MPS Multiprocessor PC, and trying to force the ACPI HAL during installation by pressing F5 before the F6 extra drivers prompt at the beginning would result in an unbootable system. This frustrated me so I installed Windows 2000 Pro and have been happily running it for quite a while now, but the other day I got curious as to why Windows 2000, Linux, and Windows 98/ME recognize the machine as ACPI capable, but Windows XP refuses to do so.

With Windows XP Microsoft put its foot down and decided that only ACPI 2.0 capable machines would be allowed to run the ACPI HAL, as a way to pressure board makers into the new standard. The way Windows XP decides what constitutes an "ACPI 2.0 capable machine" is not straightforward at all, though. Basically, if the BIOS in question is dated 1999-01-01 or later and it is not in the [NWACL] list in the i386/txtsetup.sif file—the "bad BIOS list" for short—then it gets to use the ACPI HAL. If the machine's BIOS is dated before 1999-01-01 but is on the [GoodACPIBios] list in the aforementioned file, it gets to use the ACPI HAL. And finally, if the BIOS came later than 1999-01-01 but is on the "bad BIOS list" it gets no ACPI… Guess in which list is the OR840?

Yup, bad BIOS list, along with other pretty nice boards like ABIT's BP6 and ASUS' P2B-D. Here's the full "bad BIOS list" from the official XP Pro SP3 CD downloaded from Microsoft's site;

1[NWACL]
2ABitBP6=ABitBP6
3AMIINT01/18/00=AMIINT01/18/00
4ASUSP2B-D=ASUSP2B-D
5ASUSP2B-F=ASUSP2B-F
6ASUSP2B-DS02/03/99=ASUSP2B-DS02/03/99
7ASUSP2B-DS10/21/99=ASUSP2B-DS10/21/99
8ASUSP2L97-DS02/02/99=ASUSP2L97-DS02/02/99
9ASUSP5A-B=ASUSP5A-B
10ASUSP5A03/12/99=ASUSP5A03/12/99
11CompaqArmada3500=CompaqArmada3500
12CompaqArmada6500=CompaqArmada6500
13CompaqM500=CompaqM500
14CompaqPanther=CompaqPanther
15CompaqPresario1692=CompaqPresario1692
16CompaqPresario1925=CompaqPresario1925
17COMPAQTHUNDER05/18/99=COMPAQTHUNDER05/18/99
18COMPAQRACEBAIT12/06/99=COMPAQRACEBAIT12/06/99
19DellCP???01/27/99=DellCP???01/27/99
20DELLCPt=DELLCPt
21DellCPtA02=DellCPtA02
22DellLatitudeLT=DellLatitudeLT
23DellInspiron3500=DellInspiron3500
24DellPE1300=DellPE1300
25DellPersonalWorkation1300=DellPersonalWorkation1300
26FICPA2013=FICPA2013
27FICVB601=FICVB601
28FujitsuCAPRICOR=FujitsuCAPRICOR
29FujitsuM63E=FujitsuM63E
30FujitsuM65=FujitsuM65
31FujitsuSPRINT=FujitsuSPRINT
32FujitsuTANDEM=FujitsuTANDEM
33FujitsuElm=FujitsuElm
34FujitsuMaple=FujitsuMaple
35FujitsuSycamore=FujitsuSycamore
36FujitsuV66NAS=FujitsuV66NAS
37GatewayE1200=GatewayE1200
38GatewayE1400=GatewayE1400
39GatewayProfile=GatewayProfile
40GatewaySolo=GatewaySolo
41Gigabyte=Gigabyte
42HitachiFlora220CX=HitachiFlora220CX
43HitachiFlora220MP=HitachiFlora220MP
44HitachiFlora270EX=HitachiFlora270EX
45HitachiFlora270VX-NH7=HitachiFlora270VX-NH7
46HPAshaki02/24/99=HPAshaki02/24/99
47HPHPBDD=HPHPBDD
48HPHPCCD=HPHPCCD
49HPKayakXUMP=HPKayakXUMP
50HPLH4=HPLH4
51HPPuma02/12/99=HPPuma02/12/99
52HPScimitar=HPScimitar
53IBMAPTIVA145=IBMAPTIVA145
54IBMTP240SandShark=IBMTP240SandShark
55IntelKauai=IntelKauai
56IntelL440GX=IntelL440GX
57IntelOR84009/14/99=IntelOR84009/14/99
58IntelSea2=IntelSea2
59IwillDBS100=IwillDBS100
60MicronAtlanta=MicronAtlanta

61MicronTrek2U375=MicronTrek2U375
62NCRS20GX=NCRS20GX
63NECPowerMate8100=NECPowerMate8100
64NECVersaNote=NECVersaNote
65NECVersaNoteC2=NECVersaNoteC2
66NECVersaLXAGP+=NECVersaLXAGP+
67NECZ1=NECZ1
68QDIGRP010599=QDIGRP010599
69SiemensMobile750=SiemensMobile750
70SonyF290=SonyF290
71SonyN505=SonyN505
72Sonypcg777=Sonypcg777
73Sonypcg888=Sonypcg888
74Sonypcgf16=Sonypcgf16
75SonyPCVL300=SonyPCVL300
76Sonypcvs520=Sonypcvs520
77Sonypcvs720=Sonypcvs720
78Sonyz505dx=Sonyz505dx
79SuperMicro = SuperMicro
80ToshibaPortege3300=ToshibaPortege3300
81ToshibaSatellite4030XCDT=ToshibaSatellite4030XCDT
82Tyan=Tyan
83ViaVP3A=ViaVP3A

As far as I understand, attempting a Windows XP install in any of these will result in the machine using any of the non-ACPI HALs. If you try to force the HAL with F5 during installation, I can only speak as to what happens with the OR840: it will not boot. There is another way to force ACPI though, modify the option ACPIEnable = 2 to ACPIEnable = 1 under the [ACPIOptions] header of the aforementioned file. This will allow the installer to probe for ACPI compatible BIOS and install the proper HAL independent of what the "bad BIOS list" says. So, the good news is this behavior can be changed, but would you want to? Motherboards get put in the "bad BIOS list" due to instabilities, as tested by Microsoft. So what to do?

Well, I risked it based on a hunch: the OR840 is listed as IntelOR84009/14/99, and according to some google-fu and with the help of the internet archive, I realized that the BIOS versions at that date were the very first available for the board: 0127.P01, 0128.P01, or 0129.P01. The last BIOS version available for the machine is 0255.P04, released in 2001. During the various BIOS revisions, S3 states were added and removed, and my hope is that any ACPI malfunction was ironed out enough to run the system stably.

I'll report on this after some time has passed.

EDIT 2020-05-31: Here are the original sources
https://www.itprotoday.com/compute-engines/ho … ng-installation
https://www.itprotoday.com/windows-78/jsi-tip … n-though-it-not

i learned something new today. thank you.

luckybob wrote:

i learned something new today. thank you.

Nah, thank you for showing interest every time I post about this machine—glad someone's reading! 🤣

Well, these are all things I'm going to do with my system. eventually. I swear it is on the first page! I only just got my special eprom programmer adapter the other day. Had to ship it in from the UK. I love my GQ-4X but christ-on-a-stick they want so much for adapters it's literally the definition of asinine.

A couple of follow-ups

Concerning RDRAM troubles
I have commented a few times about my trouble with big RDRAM modules and how the machine becomes unstable with them. I've hypothesized basically everything under the sun: 40ns modules make the machine unstable, ECC makes the machine unstable, ICH controller is at fault, old caps, etc. After a full recap and thorough cleaning of the RAM slots I had plugged in 1.5 GB of RAM and forgotten about the issues, but recently these problems have resurfaced with a vengeance.

At the moment the facts are: Using 2x 256 MB modules in Channel 1 and CRIMMs in Channel 2 works perfectly. Plugging in 2x 512 MB modules in Ch1 while leaving CRIMMs in Ch2 will not boot into Windows. Using 4x 256 MB modules to fill up both channels will not boot into Windows. Using CRIMMs in Ch1 and 256 MB modules in Ch2 will not boot into Windows… I tried these configurations with Tualatins and Coppermines, both dual and single setups. I cleaned the memory slots again early in the morning, but no dice: at the moment 512 MB of RDRAM sit in the machine while I write this and wonder wtf is wrong. I think something is up with Channel 2, and whatever it is it meddles with the ability to address 512 MB modules in Ch1. I am thinking as I write, but from what I remember reading about how RDRAM works, its sounds plausible and not too crazy.

All this said, I guess I can refute some theories: it wasn't the caps, because problems persisted after a full recap. The 40ns module theory is nonsense, as is the ECC module theory: the machine currently has 40ns, ECC modules and is working fine. The ICH misbehaving might be a thing, but at the moment I'm inclined to say its not the ICH that's at fault.

I'll take a look at some schematics and datasheets this weekend and see if I can figure out what's wrong with the RAM. Hopefully I can fix it and be rid of these issues once and for all.

Concerning ACPI Windows XP
Now for good news: the machine is happily "purring" (slightly stuck Radeon 8500 fan, gotta clean it 😜) with Windows XP running. While the amount of RAM is not ideal, I haven't suffered any slow downs, crashes or instabilities after I switched to 512 MB RDRAM only. So far, it seems any instability problem was indeed ironed out in later revisions of the BIOS and ACPI is viable with XP.

interesting.

I'm actually looking forward to getting my board out and playing with it. Once I get some other things off the list...

It would be nice for sure to have other people testing these things.

I contacted a user from the voodooalert.de forum to share the Tualatin mod, these german-speaking guys seemed to be very fond of the board, and while I saw that he shared the info on the mod I never heard back from anyone about it. I'd like someone to try it and confirm it works or bring out problems that need to be ironed out.

I'm really excited to hear you are looking forward to tinker with the machine!

Follow-up on the previous follow-ups

RDRAM troubles: the end?

So, I don't want to jinx it, but it seems the motherboard was just dirty.

After my last update I took apart the machine and proceeded to clean the motherboard very thoroughly. And I mean VERY thoroughly: IPA, toothbrush, toothpicks, cotton swabs, compressed air,… the complete package. The idea was to leave it clean so that during the weekend—today—I could look at it to try to figure out what's wrong. Half-way through the process I found myself in the densely populated area near the three power connectors and the top RDRAM slots. There's an IC and an electrolytic capacitor there. To my surprise there was some caked stuff near the IC (which is a clockgen for the RDRAM, btw EDIT 2018-02-18: the IC is marked LVC32A, a quadruple, 2 input positive OR gate. The clockgens for the RDRAM are located under the bottom Slot1 connector, on either side of the i840 chipset.): my guess is leaked substance from the electrolytic cap there, which I failed to notice and clean when I replaced that capacitor…

In any case, I removed the stuff, cleaned the rest of the board pretty well and out of curiosity set up the components on my test bench, i.e. an empty motherboard box. After a solid POST and quick Windows boot the machine has been running with 2 GB of RAM for about 35 straight hours without complain. No instabilities while running some AIDA torture tests, which would bring the machine down previously. In any case, I'll leave it overnight doing some memtest and report back.

ACPI Windows XP: the end?

Yep. I ran into some trouble—finally—when attempting to replace the noisy Radeon 8500 with a GeForce4: after installing the nvidia drivers Windows would refuse to boot up just in the same way it refuses to do it when you force ACPI during install (POSTs, starts the boot up and shows you the WinXP loading screen, screen goes blank and resets). I removed the GF4, put the Radeon back in, uninstalled the nVidia drivers and installed Catalysts: the machine works again. If I try any other nvidia card, Windows won't boot after installing the drivers.

I guess I have to backtrack and recommend against doing this. Windows XP is still usable if you do a normal install from the CD: it will use the MPS Multiprocessor HAL and everything will be peachy.

Current plans

I have just received a new sata controller, a Promise SATA150 TX4, which I got it because it has drivers for Windows 98. Tomorrow I will assemble the machine in an old Antec case I have laying around, install Windows 2000, Windows 98 SE, and Linux and test what I expect will be the final hardware configuration of this machine: 2x Tualatins at 1.4 GHz (sticking with them over the Coppermines), 2 GB of RDRAM (will have to mess around with the Win98 install), a GeForce4 Ti4600, Voodoo2 SLI, an Audigy2 Zs, a USB2 card, and the aforementioned controller. I will buy a new case for the machine, too. I have a special one in mind, but I'll leave that as a "surprise" for later.

The potential root cause for all your RD-RAM issues may have been finally identified? Great news, indeed!!! I hope, no new issues will show up during testing and that the system remains absolutely stable... Good luck and all the best 😀

slivercr wrote:

I have just received a new sata controller, a Promise SATA150 TX4, which I got it because it has drivers for Windows 98. Tomorrow I will assemble the machine in an old Antec case I have laying around, install Windows 2000, Windows 98 SE, and Linux and test what I expect will be the final hardware configuration of this machine: 2x Tualatins at 1.4 GHz (sticking with them over the Coppermines), 2 GB of RDRAM (will have to mess around with the Win98 install), a GeForce4 Ti4600, Voodoo2 SLI, an Audigy2 Zs, a USB2 card, and the aforementioned controller. I will buy a new case for the machine, too. I have a special one in mind, but I'll leave that as a "surprise" for later.

I think it is a very good choice keeping the Tualatins instead of the Coppermines! Like I wrote before, they are a physical testament of your great achievement (in addition to the virtual one here at vogons) 😀 I am already excited to read about the new case soon 😀 And some time later, see some games heating up the system 😉

Got a new fridge during the holidays! Detailed pics on the weekend.

Corsair Carbide Quiet 600Q? Nice! I like it. A worthy home for this great system 😀 Looking forward to the additional pictures! I hope the hardware migration completes without any issues.

voodoo5_6k wrote:

Corsair Carbide Quiet 600Q? Nice! I like it. A worthy home for this great system 😀 Looking forward to the additional pictures! I hope the hardware migration completes without any issues.

Sorry to keep you—and anyone following this thread—waiting for pics, here are a few quick and not so good ones of the system in the new case. The coppermines are in place while I modify one of the socket 370 heatsinks I use in the slotkets (or find a new one), since I just noticed that after changing the motherboard caps the bottom part of the heatsink sits flush against them, which makes me very uneasy.

Also, you were close but its actually a Carbide 600C, the "Clear" version with the window side panel. I normally steer clear of windows, RGB, etc., but I couldn't resist this one. I dont think I'll go for lights or RGB just yet, but looking inside the case is pretty awesome so far, I like how the dual Slot 1 CPUs look.

A few things to do still:

• Make a shorter SLI cable,
• Get a round floppy cable,
• Mod a PCI card cover to accommodate the USB cables going out,
• Plug the CD-In cable, which I just noticed I forgot,
• Make a new Front Panel Audio cable, so that the front panel connectors are all usable,
• Find a way to silence the GeForce4,
• Get some localized airflow on the Voodoos and the RDRAM,
• Figure out the cooling solution for the CPUs: airduct again, or fans on each CPU for display purposes?

The build is very quiet, even with the case fans set to maximum in the included fan controller. The only audible thing is actually the GeForce's fan, which has a very characteristic high pitch sound to it. Passive cooled as they are the CopperMines max out at 65 ºC or so under load, around 6 or 7 ºC cooler than the recommended max temperature, so they're good. I'll have to see about the Tualatins because past experience tells me they won't allow passive cooling.

Overall I am very happy about this case choice, its cool and quiet. Besides taking care of the things mentioned before—and maybe getting a black floppy drive—the system is mostly in its final configuration.

slivercr wrote:

Also, you were close but its actually a Carbide 600C, the "Clear" version with the window side panel. I normally steer clear of windows, RGB, etc., but I couldn't resist this one. I dont think I'll go for lights or RGB just yet, but looking inside the case is pretty awesome so far, I like how the dual Slot 1 CPUs look.

Well, close enough 😉 I too don't like these windowed cases (or window case mods back then), or this modern all-tempered-glass-nonsense. But to each his own... I do however see the point of having the ability to lean back and enjoy the looks of cool hardware installed. Unfortunately, often people then tend to make the case look like the showcase of the lighting department in your building supplies store. I will never get this... Anyhow, I can understand making an exception for this system. Will you place the tower on your desk so you can have a good look?

slivercr wrote:

Find a way to silence the GeForce4,
Get some localized airflow on the Voodoos and the RDRAM,

If you could eliminate one of the expansion cards (most likely the USB one), you could move the Sound Blaster and the two Voodoo2 cards each one slot down (well, "up" in this case 😉). This way, you have two slots for the GeForce4 and could mount a Zalman VF700-Cu or VF700-AlCu (like I have on my GeForce4 Ti 4600 and FX 5900 Ultra, Retro 3 which is linked in my signature). For the Voodoo2 cards, you could use something like this here to mount one or two smaller fans directly above the cards (can be found on ebay or amazon for OK prices):

• The attachment pci fan mount.JPG is no longer available

slivercr wrote:

The coppermines are in place while I modify one of the socket 370 heatsinks I use in the slotkets (or find a new one), since I just noticed that after changing the motherboard caps the bottom part of the heatsink sits flush against them, which makes me very uneasy.

Good catch! I had a similar issues on the CUBX-E (although not related to exchanging caps). There are a few tiny caps very close to the socket and the cooler I intended to use would slightly push on those caps. As I don't want additional heat transfer to the cap or an unknown amount of weight/force on the cap, I searched for a comparable heat sink with smaller base plate (which I luckily did find pretty quickly).

Overall, I like the new home of your system too! However, I can't wait to see the Tualatins back in action to really complete it 😀

voodoo5_6k wrote:

Well, close enough 😉 I too don't like these windowed cases (or window case mods back then), or this modern all-tempered-glass-nonsense. But to each his own... I do however see the point of having the ability to lean back and enjoy the looks of cool hardware installed. Unfortunately, often people then tend to make the case look like the showcase of the lighting department in your building supplies store. I will never get this... Anyhow, I can understand making an exception for this system. Will you place the tower on your desk so you can have a good look?

Hahaha, true about the lights. When I was a teenager I remember the beginning of the "modding craze": people adding side fans, then windows, making fan controllers, etc. I remember back then I really wanted a case with a window, but never had the tools or money to attempt it. I guess I'm fulfilling a teenage dream? I will still not go for lights though 😜 It will not sit directly on the desk, but on top of a mini fridge I have next to the desk: I'll be able to look inside

voodoo5_6k wrote:

If you could eliminate one of the expansion cards (most likely the USB one), you could move the Sound Blaster and the two Voodoo2 cards each one slot down (well, "up" in this case 😉). This way, you have two slots for the GeForce4 and could mount a Zalman VF700-Cu or VF700-AlCu (like I have on my GeForce4 Ti 4600 and FX 5900 Ultra, Retro 3 which is linked in my signature). For the Voodoo2 cards, you could use something like this here to mount one or two smaller fans directly above the cards (can be found on ebay or amazon for OK prices)

The problem is I really need the USB2 card since the motherboard only has a couple of USB1 ports. While I don't need the faster transfer speed, it is nice to have if I need to transfer small files. Also modern game controllers do not play well with USB1, and I need a controller to play the occasional THPS2.

I've thought about a bigger cooler though, in particular I thought about getting an AGP riser card or extension and being able to use the slot on top of (or under, in this case) the AGP card. This would essentially free space for a big cooler, but unfortunately the RDRAM gets in the way.

My plan now is either fitting a small flower cooler I took from a non-working Leadtek geforce fx, but I think it will touch the voodoo's memory chips; or looking for a bigger, passive cooler like what you would see on a geforce 6200, and then set up a bracket like the one you mentioned to blow over both voodoos and the geforce. I'm not sure how well it would work though, and I don't have a thermocouple to test 😢

voodoo5_6k wrote:

Good catch! I had a similar issues on the CUBX-E (although not related to exchanging caps). There are a few tiny caps very close to the socket and the cooler I intended to use would slightly push on those caps. As I don't want additional heat transfer to the cap or an unknown amount of weight/force on the cap, I searched for a comparable heat sink with smaller base plate (which I luckily did find pretty quickly).

Overall, I like the new home of your system too! However, I can't wait to see the Tualatins back in action to really complete it 😀

Yeah, I'm glad I caught it! I ordered a couple of smaller heatsinks which should fit nicely. They are slightly shorter, so I may be able to fit a fan on each without troubles to cool down the Tualatins.

I will update the pictures once the Tualatins are in place!

slivercr wrote:

It will not sit directly on the desk, but on top of a mini fridge I have next to the desk: I'll be able to look inside

Cool 😀 Maybe you can show the entire setup once everything runs and the system is sitting on its final spot.

slivercr wrote:

Also modern game controllers do not play well with USB1, and I need a controller to play the occasional THPS2.

I see. But what about using a gameport controller via the additional bracket of the Sound Blaster? This would free up a slot and still let you use a decent controller! I'm using Microsoft SideWinder game pads (gameport version) for games like Need for Speed or sports games. I haven't seen a better controller yet... Let alone more robust. This thing is virtually indestructible and built to last forever.

slivercr wrote:

I ordered a couple of smaller heatsinks which should fit nicely. They are slightly shorter, so I may be able to fit a fan on each without troubles to cool down the Tualatins.

Sounds good! Those fans don't need to push a lot of air as you have a decent case ventilation thanks to the modern quality case! So they can be relatively slow and quiet although they are small (I'd guess they are 60mm fans or smaller).