I managed to get the MSI slotket working with P3 Coppermine CPUs.

This slotket is much better than the no name FC-PGA slotket I have been using with the Aopen AX6BC so far.
The no name slotket made the motherboard think that all Coppermine CPUs use 1.5V and with no voltage setting in the bios setup anything over ~900 mhz was unstable.
The Aopen AX6BC also had stability issues with most video cards @133 mhz FSB even if a P3 800(133) Coppermine CPU was used (The board was configured to use the 2/3 AGP setting with jumpers).

With the MSI slotket configured to 1.75V all problems seem to be gone, at least with the P3 1000(133) I am using at the moment.
This made me order a "MASTER" version of the MSI slotket with the voltage clamp chip so I dont need to worry when the modified Tualatin arrive.
Now when the AX6BC is stable at the 133 mhz FSB setting I wish I had ordered the P3 1400S and not the Celeron 1400 though.
The Celeron will be good for the Abit BH6 😀 so it is not wasted.

These days where can you order an PC-PGA compatible slotket for normal price?

Order was probably the wrong word.
I have the leading bid and I am going to win, nobody else usually bids on these items on "Swedish Ebay".
http://www.tradera.com/delar-till-aldre-penti … 02619_199283646
Only ships to Sweden so you can not steal them 😜

There is another source, a German site but the price is ~30 euro with shipping.

That v2.3 is a great catch 😉 .

Skyscraper wrote:

... Now when the AX6BC is stable at the 133 mhz FSB setting I wish I had ordered the P3 1400S and not the Celeron 1400 though ...

Which version is your Aopen AX6BC and/or what is written on this chip?:

TELVM wrote:

That v2.3 is a great catch 😉 .

Skyscraper wrote:

... Now when the AX6BC is stable at the 133 mhz FSB setting I wish I had ordered the P3 1400S and not the Celeron 1400 though ...

Which version is your Aopen AX6BC and/or what is written on this chip?:

My board is a vanilla AX6BC.
I cant find a revison number.
The board is mounted in a case so its hard to get a good look.

I did manage to take a picture of the chip though.

Fairchild RC5057M
High Performance Programmable Synchronous Dc-dc Controller For Low Voltage Microprocessors

1.3V - 3.5V 😀

I downloaded the PDF
Error information overflow.

Excellent, it's VRM 8.4 compliant and probably the board can run a Tually 1400-S OK 😉 .

Fairchild RC5057

... The RC5057 single-output device is designed for Pentium IIs and IIIs (Coppermine, Camino, Katmai and Mendocino motherboards) ...

http://www.eetimes.com/document.asp?doc_id=1288492

But those mosfets will need heatsinks, and the case decent ventilation.

Yea some sort of heatsinks on the mosfets sounds like a good idea.

I am back to beeing glad I ordered the Tualatin Celeron 1400 because the Quadro 980 XGL I use in this system does seem to have issues with 89 mhz AGP bus speed.
3D benchmarks work fine, no corruption and no instability but Star Wars Rogue Squadron freezes now and then.
I switched the CPU to the P3 850 @ 995 with 78 mhz AGP and Star Wars Rogue Squadron does not freeze any more.
It is a little bit strange because 3dmark 2001 runs without freezing even at 95 mhz AGP bus speed.

I have another 980 XGL that I will try with 89 mhz AGP bus speed and I also have a Geforce4 ti 4600 in the Abit BH6 system that I could try.
I really want a Geforce4 ti 4x00 in this system.
The P3 1400-S is not worth it if I have to step down to a Geforce 2 or Geforce 2/4 MX.

GeorgeMan wrote:

These days where can you order an PC-PGA compatible slotket for normal price?

I suppose if you ask nicely(send him a message), this fellow in bottom link below might be willing to ship it as 'Grossbrief International' (max 500g) + 'Einschreiben'/signed-for ?
(= 3,45+2,05 euro, barring any handling cost...)

http://www.deutschepost.de/de/b/briefe-ins-au … ernational.html

http://www.ebay.de/itm/370981716065
http://www.ebay.de/itm/370982665866

At least I had some luck 2 years ago.
But who knows now... 😎

That is the German company I meant.

http://www.cbo-do.de/shop/index.php is their homepage.
Better to contact them directly than go through Ebay I think.

The 89 mhz AGP bus speed adventure continues.

So far I have tested these cards.

2x Quadro 980 XGL
2x Geforce3 ti 200
1x GF4 ti 4600
1x GF4 ti 4800SE
1x Quadro 2 Pro
1x GF FX 5600 Ultra
1x GF FX 5900 Ultra
1x GF4 MX440 SE 8X 64MB (Albatron 480E with 64 bit memory bus 🙁)

All these cards works well in 3D benchmarks @ 89 mhz AGP except the Quadro 2 Pro and the ti 4600.
Only the Albatron GF4 MX card works with Rogue Squadron @ 89 mhz (and 92 mhz) AGP flawlessly.
The problem is that the 980 XGL gets 8137 points in 3dmark 2001 while the GF4 MX only gets 3317 points and do not support DirectX 8.

I have found information indicating that on some revisions of Aopen AX6BC the AGP divider jumper does not work.
With all CPUs detected as 66 mhz FSB CPUs the motherboard chooses the 1/1 AGP ratio even if the board is jumpered for 2/3 ratio.
I do not think this is the case with my board because I do not think my Albatron GF4 MX can handle 133+ mhz AGP bus speed 😁.
The P3 1000 I use gets detected as a 66 mhz FSB CPU how ever I jumper the slotket.

With the clocks on the GF4 MX bumped to 350 mhz core and 600 mhz memory the performance still sucks but it will have to do for now.

In the topic I pointed out before, I was also trying a bunch of Video cards on an Aopen AX6BC rev1.4 mainboard. Which was unintentionally running with 133MHz AGP instead of 89MHz. Surprisingly one Geforce 2 MX seemed to accept that, the other cards locked up within 10 minutes. Link: 50 to 133MHz FSB on a BX Mainboard
Only a little later did I discover that the 'AGP Frequency Select' jumper needed to be configured manually, to get 89MHz. 😜

The i440BX AGP clock rate is actually an obscure thing to set and read. Nothing reads it, could not find a diagnostic card or diagnostic software for it. My best knowledge is that a Voodoo 3 has never failed me on 89MHz so far, but does lock up on 133MHz AGP clock.

The AGP Clock rate is just one pin on the intel northbridge: "MAB12#": Host Frequency Select". On an early Asus P2B v1.04 mainboard, I added the 'AGP Frequency Select' jumper myself. But since the Intel nortbridge is Ball-grid-array, there was no easy way to identify the trace going there.

As I have written my board is jumpered for the 2/3 AGP ratio. I do know how to read a manual 😁 But I just rechecked just in case!
There was a thread on another forum discussing if that jumper is working on all revisons of the board or not.
But as I wrote the jumper seems to work fine on my board but the cards are not not stable enough @89 mhz AGP.
It seems the vanilla AX6BC just isnt the perfect board for high AGP clocks.

Do someone happen to know who made this 440BX board?
The board in the picture is not mine but I have one just like it and cant identify it.
The person with the ad I stole the picture from does not seem to know what he is selling 😀
Its marked PWB REV:B01 on the backside (PCB revison I guess) and "ver K03" in the lower right corner.
The other numbers on the board seems to be serial numbers and do not give any intresting hits.

Looks like a crossbreed between a micronics C300 and C400...
Those serial+barcode stickers, of which there are three on this board, usually indicate Intel or large OEMs such as Siemens, Dell etc.

Thanks!
I think it is the C400.

This board looks more or less identical to mine.

http://www.hardware.fr/articles/74-1/diamond- … onics-c400.html

Good, C400 then. 😀

Skyscraper wrote:

It seems the vanilla AX6BC just isnt the perfect board for high AGP clocks.

All any i440BX mainboard can do is set the MAB12# pin on the intel northbridge to have either a 2/3 or 1/1 ratio of AGP clock to FSB clock.
Normally they connect it only to the CPU slot-1 (like my asus P2B rev 1.04) to make it autodetect, and then there are boards that allow a manual override of that signal by jumpers or the BIOS (like the Aopen AX6BC).
There is no i440BX mainboard around offering any more value in that regard. It is the main drawback of the i440BX chipset itself.

AGP at 66MHz with FSB at 133MHz will require a later intel 8XX series chipset without any ISA slots AFAIK, or a VIA Apollo Pro 133 chipset.

gerwin wrote:

Good, C400 then. 😀

AGP at 66MHz with FSB at 133MHz will require a later intel 8XX series chipset without any ISA slots AFAIK, or a VIA Apollo Pro chipset.

I know 😀.
But I do think some boards are more stable at high AGP speed than others just as different boards reach different max overclocks with the same CPU.

I am testing a cheap VIA Apollo Pro board right now just to see how bad they are.
http://www.tomshardware.com/reviews/12-socket … ards,196-6.html <---- that one.
And the performance numbers in Toms test are just awful.
http://www.tomshardware.com/reviews/12-socket … rds,196-21.html

I am going to check if the performance ever got fixed with a bios update or if Biostar was satisfied with achieving 75% of the performance of an Asus board with the same chipset.

Yeah I have one VIA too, a chaintech 6AJA4.
What I forgot; if you really want to know if the BX motherboard is the bottleneck, you can run it with PCI graphics.

gerwin wrote:

Yeah I have one VIA too, a chaintech 6AJA4.
What I forgot; if you really want to know if the BX motherboard is the bottleneck, you can run it with PCI graphics.

I have even tested the board with an ISA video card 😉

The performance of the Biostar M6VCG still sucks with a bios from late 2000!
Perhaps I will flash the board with the latest BIOS from Biostars site but the last bios release often suck when it comes to performance.
(Hyperion AGP driver installed, both systems use Forceware 43.45)

3dmark 2001 scores.
Aopen AX6BC + Coppermine 1000 + Quadro 980 XGL = 8137p
Biostar M6VCG + Coppermine 933 + Quadro 980 XGL = 6124p

SuperPi 1m
Aopen AX6BC + Coppermine 1000 = 2m 33s
Biostar M6VCG + Coppermine 933 = 3m 22s

The Biostar M6VCG must be one of the worst motherboards ever made when it comes to performance.
I have now updated the board with the latest bios.

I did some tests with a Celeron 850.

SuperPi 1m scores.
Abit BH6 + Celeron 850 @ 953 mhz, memory @ 112 mhz: 3m 01s
Biostar M6VCG + Celeron 850 @ 974 mhz, memory @ 115 mhz: 4m 07s
Biostar M6VCG + Celeron 850 @ 974 mhz, memory @ 148 mhz: 3m 37s

So higher clock speed and faster memory but still much much worse performance compared to the i440BX platform,