Tualatin on GA-BX2000+ / SlotT

[Batyra]

Hi!
- I have Gigabyte GA-BX2000+ currently runnig slot 1 Pentium III 1GHz / 100FSB.
- I wanted to run P3 Tualatin 1.4 on it so I bought myself an Upgradeware SlotT adapter
http://duhvoodooman.com/powrleap/LFL/Slot-T_1.htm

- I installed the latest official bios, and later unofficial bios (BX2000p.fdc) - flashed without any problem.
- I set up jumpers on Slot-T adapter as written in manual (fsb133, voltage 1.5)
- I set up jumpers on motherboard (fsb133, multiplier x10.5)

IT DOES NOT BOOT!

I tried other setting with other sfb's and miltipliers both on adapter and motherboard,
I tried diferent power supply
I tried different memory modules (mainly I'm using OCZ fire PC150 256mb CL2)

---
I know this topic was moved a few years ago and AFAIK @Skyscraper manage to run it on this motherboard and SlotT adapter - if I understood correctly

PLEASE HELP...

[The Serpent Rider]

Does it work with Celeron Tualatin?

[Skyscraper]

If I remember correctly the Gigabyte GA-BX2000+ would not run any really fast CPUs at 133 MHz FSB like the Coppermine 1000(133) or the Tualatin P3-S 1400.

The ones with 100 MHz FSB did run flawlessly like the Coppermine Celeron 1100 and the Tualatiin Celeron 1400.


It's not rellay an issue with the motherboard not handling 133 MHz FSB as the P3 800(133) ran fine and I could even overclock to what ever max FSB the motherboard supported when I used a Coppermine 600(133) or similar. It must be some odd timing issue/bug with the combination of high FSB and a really fast CPU.

I used a MSI 6905 Master Slotket and Korea modded Tualatins but I dont think that makes anything any different as the Coppermines suffered from the same issue.



If you can find a Celeron 1400 I think that would be a safe... -ish bet.

If the Slot-T has FSB selection jumpers try forcing 100 MHz FSB. The Tualatin P3-S @1050 is almost as fast as the Tualeron 1400 (and much faster than the Coppermine) and perhaps you can get away with 112 or 115 MHz FSB with either one.



Edit.

112 MHz FSB was the best I managed with the Tualatin P3-S 1400 on the Gigabyte BX2000+ and this ruined my whole i440BX vs i815 vs VIA Apollo Pro benchmarking and CPU scaling project.



P3-S @112 MHz FSB

Celeron 1400 @112 MHz FSB

The P3-S wins out but I think the Celeron would fare somewhat better with a faster video card.

results.txt

(9.53 KiB) Downloaded 6 times

(only some of the game results here, the rest are still screenshots only)


/Edit

[appiah4]

What revision is your BX2000+? I believe not all of them handle Tualatins.

[Skyscraper]

What revision is your BX2000+? I believe not all of them handle Tualatins.

The Slot-T is a FC-PGA2 to Slot-1 adapter, the whole point is to get Tualatins running in boards they are not suppsed to run in.

Officially no i440BX board supports Tualatins. The BX2000+ is a bit more quirky than many other late BX boards though.

[Batyra]

Thanks for all the replyies guys!
I've checked few thing again and I have some very interesting conclusion... but this later, first I'll answer all of you.

@The Serpent Rider - I don't have celeron tualatin to check it ;(
@Scyscraper - thanks! that are very usefull informations.
@appah4 - I have revision 1.2 (couldn't find any info about it)

Now conclusion... My friend asked me what exactly CPU I use... and I taken off radiator of slot-T taked of CPU etc... than I realised my CPU was in socket the oposite way... OK... you'll say - you can't put cpu in s370 the wrong way... oh yes you can - if there is a corner pin missing in cpu ;/

It appears that CPU I bought has missing one corner pin...

But this is not the end of the story... After placing CPU in right position system boots, and cpu was recognised but it get allways stuck on screen before memory test... always when the FSB is set above 100MHz... if fsb is 100 system loads, if higher stuck before memory check...

I've tested all the memory i have PC100, PC133, PC150... always the same... I hope it is the matter of that missing pin... I'm waiting for a second Tualatin from a friend - I'll chceck then if it works properly - bul I'll have Scyscraper's post and info about fsb in mind...

Thanks!

[meljor]

You can download the intel pdf about the cpu and check the pinout and see if the pin has an important function or if it is just a groundpin.

[Standard Def Steve]

But this is not the end of the story... After placing CPU in right position system boots, and cpu was recognised but it get allways stuck on screen before memory test... always when the FSB is set above 100MHz... if fsb is 100 system loads, if higher stuck before memory check...

I've tested all the memory i have PC100, PC133, PC150... always the same... I hope it is the matter of that missing pin... I'm waiting for a second Tualatin from a friend - I'll chceck then if it works properly - bul I'll have Scyscraper's post and info about fsb in mind...

Thanks!

The PIII-S tends to be a little more picky about which board it will go into than the PIII-A and Tualeron. My IBM PC300PL also gets stuck before the memory count with a PIII-S @1050. However it works fine with a PIII-A 1200 @900 or Celeron-1400!

Unfortunately, the PIII-S, even at 1050MHz, is the fastest CPU for 440BX. When I tested PIII-S @ 1050 vs Celeron 1400 in my Dell XPS T550, the PIII-S won every time. But I'm sure overclocking the Celeron to 1568MHz (112FSB) would take care of the slight speed deficit.

[Batyra]

I tried two other PIII Tualatin SL5XL and SL6BY...
Situation is the same @fsb 112 (so 1.17GHZ) it runs... and above 112 it get stuck before the memory test
It seems like fsb jumpers on Upgradeware adapter are not working at all... so motherboard fsb has higher priority... similar with multiplier if I want to set to 14x100 it is seemed as 10.5x100 so 1.05... nothing more.

What else I can do to run it on 1.4?

Help!

[appiah4]

I tried two other PIII Tualatin SL5XL and SL6BY...
Situation is the same @fsb 112 (so 1.17GHZ) it runs... and above 112 it get stuck before the memory test
It seems like fsb jumpers on Upgradeware adapter are not working at all... so motherboard fsb has higher priority... similar with multiplier if I want to set to 14x100 it is seemed as 10.5x100 so 1.05... nothing more.

What else I can do to run it on 1.4?

Help!

BX board do not have a 1/2 AGP divider, only 2/3 or 1/1 and this is also the case for your board, so to run 133MHz Tualatins you need a graphics card that will tolerate the 2/3 divider option, therefore 83MHz AGP Bus. This is why it is rarely a good idea to be stubborn about running Tualatins P3s on BX chipsets, you should probably just stick to Tualatin Celeron 1300 and 1400MHz CPUs.

Regardless, have you tried a different graphics card on that motherboard? One that is known to be more tolerant of high AGP Bus speeds?

[Batyra]

@Apphia4
I tried with Canopus X20 (GF3) and GF4 Ti... PCI didn't want to boot.
This board also have two options for fsb133 one wigh PCI @33 and second one of CPI @43.3 (or something like that)

Do you have any idea what card I may use?

[appiah4]

@Apphia4
I tried with Canopus X20 (GF3) and GF4 Ti... PCI didn't want to boot.
This board also have two options for fsb133 one wigh PCI @33 and second one of CPI @43.3 (or something like that)

Do you have any idea what card I may use?

440BX has 1/4 divider for PCI, which is the PCI 33Mhz option there, so make sure you choose that, but it's not the PCI bus that's causing the issue but the AGP bus most likely.

I'm assuming the GF3 and GF4 cards were AGP?

Try booting the system with a PCI card, or something like a Voodoo 3 AGP and see if that helps. If the PCI card works you can use something like a Radeon 9250 PCI with the P3-S CPU and call it a day. If you can't find a fast PCI card, I would just stick to a Celeron 1300 or 1400 and use a GF3/4 with that. You won't miss out on much.

This is the Tualatin dilemma, you can either:

440BX: AGP issues at 133MHz
694T: Poor southbridge performance
i810E2: No AGP
i815EP: No ISA

Of the lot, 694T or i815EP are my usual pains of choice, but to each their own.

[matze79]

The Tualeron has slower Cache, thats why it falls behind.

[Skyscraper]

With a slow CPU (P2 or slow P3) my BX2000+ will happily run 153 MHz FSB (AGP @102MHz, PCI @38MHz) with Nvidia video cards although perhaps not always 3D stable depending on the video card.

The faster the CPU the lower the max FSB gets. This motherboard (at least my board) has some odd timng bug/flaw/issue/"feature".

[Batyra]

Tried different cards i.e. Voodoo 3 both PCI and AGP... it boots every time but stuck before memory count...

Once it started!! On all setting "as they should be"... but only once! when I turned it of it didn't want to bood again... So it can boot?!?

Do you think it may be caused by a capacitor? One seems a little bulky, take a look.

[Batyra]

There is "Patch ID: none" so that may be the problem.
I have seen in the internet that there were two FDc bios'es... form 2002 and from 2004...
Here I found info about the 2004 one http://drivers.mydrivers.com/drivers/99_39828.htm
https://biosagentplus.com/bios/2A69KG0L?ref=822

Maybe I'm useing wrong one?

@Scyscraper - did you also had "Patch ID: none"?

Can anyone of you help me find the bios... (I think I might used wrong one... so I'll try to flash it again, but it is hard to find it..)

[Skyscraper]

There is "Patch ID: none" so that may be the problem.
I have seen in the internet that there were two FDc bios'es... form 2002 and from 2004...
Here I found info about the 2004 one http://drivers.mydrivers.com/drivers/99_39828.htm

Maybe I'm useing wrong one?

@Scyscraper - did you also had "Patch ID: none"?

Can anyone of you help me find the bios... (I think I might used wrong one... so I'll try to flash it again, but it is hard to find it..)

As the CPU works at 1050 MHz (and in my case also 1176 MHz) I kind of doubt it's it has something to do with Patch ID but I guess a different BIOS could make a difference? My motherboard is in storage so I can not do any tests right now.

Im using the 01/15/02 BIOS as can be seen in the pictures I posted. It runs the Tualatin Celeron 1400 just fine at 1568(112) MHz or the P3-S 1400 at 1176(112) MHz.

[shamino]

Do you think it may be caused by a capacitor? One seems a little bulky, take a look.

I don't see any obvious bulging in the picture, but if you notice it in real life then consider it bad. They don't start to bulge until after they have already failed badly, and they don't always bulge at all.
If it's hard to tell, then it can help to feel several of the other caps and see if you notice a difference between them.

The green caps in the picture look like Sanyos. Those are unlikely to fail, but if they're bulged then replace them anyway. The 3 towards the bottom of the picture are some that I'd be suspicious of. They look like Hermei but whatever they are, they aren't any of the reputable brands. Since they're not one of the good brands, I wouldn't trust them regardless of what they look like.

I would replace those cheaper caps (and all like them on the motherboard), but that's because I'm used to doing it. If you aren't used to recapping then you might be more hesitant. I don't know if replacing them will actually fix your issue, but even if it doesn't, at least that potential failure will have been prevented.

With a slow CPU (P2 or slow P3) my BX2000+ will happily run 153 MHz FSB (AGP @102MHz, PCI @38MHz) with Nvidia video cards although perhaps not always 3D stable depending on the video card.

The faster the CPU the lower the max FSB gets. This motherboard (at least my board) has some odd timng bug/flaw/issue/"feature".

That sounds like it might be a limitation of the onboard power supply or voltage regulation. Higher clocked CPUs might be more sensitive, and if they draw more power then the board's VRM might not be as stable with them.

[infiniteclouds]

I have the same issue on my GA-6BXC Slot 1 board with MS6905 Adapter when using one of the modified III-s 1.4 Tualatins. Certain Tualatin compatible BIOS might POST but will immediately lock up at 133FSB but with most of the Tualatin-BIOS it won't even POST. 100FSB is no problem... and the board does 133Mhz FSB just fine with VIA C3 CPUs.

I contacted the seller of the modified CPUs and he offered to modify my board for free but I'll have still have to pay postage to South Korea and back. There is some language barrier but here is what he said of it:

And an overclocked BX M/B can not be guaranteed a working of P-IIIs 1.4G.
Major reason is signal quality do not meet timing specific.

GA6BXC uses a ICS9148-26 for Frequency Generator.
But, this IC is designed for 100MHz FSB.
=> https://www.idt.com/document/dst/9148-26-datasheet
( on page 3, => Note1. Performance not guaranteed )
This means actual FSB can be reached up to 131%(174MHz) for one clock timing level.
By this reason, CPU can be overclocked for short time.
If lucky, ICS9148-26 does not go to 174MHz or CPU works at 174MHz for short time.
But, any case it is out of specific.

Next can help you to understand what is need to work CPU.
=> http://download.intel.com/design/Pentiu ... 526408.pdf
( page 22 to 55 )

Fixing will be simple and it is analog level works.
I do not have intent of changing any M/B function.
But, before measurement, I can not tell you which part or value will be changed.
And if big change is needed, I will ask you with details.

And M/B, MS6905 and CPU have signal quality control parts.
I will adjust only these parts.

All basic function will not be changed.
Modification key point is signal quality adjustment.
It is not adding or removing IC part.
After measurement of signal quality, adjusting value of resistor or capacitor.
Maybe, you will not know what is changed.

He also says he can find a way to get it running on my SE440BX-2 board if I wanted to use that.

[Skyscraper]

With a slow CPU (P2 or slow P3) my BX2000+ will happily run 153 MHz FSB (AGP @102MHz, PCI @38MHz) with Nvidia video cards although perhaps not always 3D stable depending on the video card.

The faster the CPU the lower the max FSB gets. This motherboard (at least my board) has some odd timng bug/flaw/issue/"feature".

That sounds like it might be a limitation of the onboard power supply or voltage regulation. Higher clocked CPUs might be more sensitive, and if they draw more power then the board's VRM might not be as stable with them.

Yea but the strange thing is that the Tualatin Celeron runs fine at 1568 Mhz with 112 MHz FSB and at that speed it uses more power than the Pentium III-S would at 1400. Both CPUs use the same voltage so it would be kind if strange if it would be a current problem.

I don't think the issue has anything to do with the "Korea mod" at all, I experienced the same issue with the Coppermine PIII 1000(133). The Coppermine Celeron 1100(100) runs fine even overclocked again using the same voltage and even more current than the Coppermine 1000 PIII does running at stock.

I'm pretty sure it's a timing issue/bug/flaw/"feature". When the going gets fast... the slow can't keep up... It could very well be that my board is a bad apply but this thread kind of indicates the issue could be affecting more BX2000+ motherboards.

I have not experienced this with any other motherboard with the proper 1/4 PCI divider, only 3D stability issues with some video cards because of the out of spec AGP bus. Even my first revision Asus P2B from the spring of 1998 runs the modded Tualatin PIII-S 1400 flawlessly but seriously overvolted with the MSI slotket asking for 1.8V and the motherboard providing about 1.68V actual after drop and droop.