Gigabyte GA-5AX Rev3.0 to Rev5.1 - The road to K6 victory
 

Hey guys,

It has been a while since I took my K6/2 500 for a 720MHz spin, but I'm back with another (read: better) motherboard. Yes, I needed a new one, as my P5A completely maxed out at 120x6 ... maxed out as in this is the maximum I can set the board to work at theoretically.

I needed to find another board, one that can do way more than 120MHz ... and I found one!



The Gigabyte GA-5AX Rev5.2 goes all the way up to 140MHz FSB, which suits me perfectly. Having looked on Ebay Belgium, but not finding anything, I switched to Ebay Germany as those pages have an incredible amount of hardware ready to be purchased. Veni, Vidi, Vendo ... yes, I found a GA-5AX and after a week it got here.

BUT ... (I guess you all saw this one coming, no?)



When I opened the box to have a first glance at my board of glory, I noticed this wasn't a Rev5.2 board ... it was a Rev3.0! Optimistic as I am, I quickly tried to find out if the revision might not be as bad as I thought it would be.

http://tw.giga-byte.com/Support/Moth...ProductID=1578

Perfect! A motherboard that has different manuals for different PCB revision, it doesn't look good.



Nope, it doesn't look good, only 100FSB at most ... this board wasn't worth the euros I spent on it and the worst part: it costed me quite some golden coins to bring the board to Belgium. Already brought down, I decided to have quick look at the Rev5.2, just to see what the big difference is.



Yeap, that's what I was looking for: 140MHz FSB. But wait, JP9 ... I came across that multiple times already. Yes, in the first manual as well and ...





You see that, lads and gents? I can turn this baby into a Rev5.1 with one simple modification. Turning this board into a Rev5.2 requires me to find the missing JP10, which I haven't found on this board.



Now, all is well, you'd think, but this is MADSHRIMPS, my friends, we are always pushing things over the edge. How can I eventually make this board run higher than 135FSB? In theory it's possible, the PLL supports 140FSB, the only thing I need to figure out is how I can find that last jumper: JP10. That'll be something for Geoffrey, I'm afraid :ws:.

I have to work in Kortrijk next monday, maybe I can stop at your place, don't know the exact hour though. In between 9u and ... :p

Or maybe next thursday, then I'll be in Waregem.

Next thursday I'm in Gent to do my first exam, tomorrow is not a problem, though. Back in the lab in the attic? :p

No time tomorrow, neither sunday.

Woops, I meant monday :)

Where did you find the datasheet if the ICS 9148BF-36 PLL?

Are you 100% sure that rev3 and rev5.2 use the same PLL?

I'm pretty sure the ICS is the exact same, there exist windows overclocking tools for the GA-5AX and there's no difference between the revisions.

Datasheet was from another pll: the ICS9148BF-04. I reckon they work the same?

I know the ICS 9148BF-04, recap of its datasheet:



3 Frequency Select pins, maximum front side bus clock = 83.3MHz.

The Pcchips Socket 7 board I have over here has the ICS 9148BF-17 PLL, which has the option to set up to 100MHz front side bus clock via BIOS, no datasheet available either. The main pin functionality remains the same probable.

This datasheet should be it: http://www.idt.com/?partID=9148BF-36

Up to 100Mhz, so says the datasheet. So ICS 9148BF can have different frequency select with different Pattern Numbers. -04 up to 83MHz, -36 up to 100MHz. What precise pattern number do you find on the rev5 board?

Haven't found a large picture of a rev5.2 board yet, but, as the JP9 is present and connected (two pins are soldered together), it should be possible to cut that connection and change it with jumpers to hit 135MHz, no?

It should be possible to replace it with some kind of jumper, no problem, but it is not being said that it will then be able hit more then 100MHz FSB. Can you use rev5.2 BIOS?

The bios has nothing to do with settings FSB mhz, the fsb is set by those jumpers and cannot be set in the bios.

FSB cannot be set via BIOS? Even the pcchips board could do that :D

P5A can't, 5AX can't and those two boards are noticably better than the PCCHips :p

What does this say about the modifications?

If you could use BIOS of the rev5 board, you would not need the extra jumper, and looking at the ICS 9148BF-36 datasheet it seems that the PLL can only have 3 jumpers as these are the only the frequency select bits. I guess you need to continue your hunt for rev5 boards.

What's with you and the bios? Are you having an affair behind my back??? :D


The only difference between Rev3 and Rev5.2 is that JP10. Nothing else, the bios is exactly the same!

Rev3: JP7 and JP8 (and JP9)
Rev5: JP7, JP8, JP9 and JP10.

Look at the picture of JP9. Can't I just cut that connection and add a jumper on the three soldering points? If yes, I already have 135MHz fsb.

JP10 just isn't visible on the Rev3 board.

I think you did not understand me too well. rev3 uses the ICS 9148BF-36 PLL which has only 3 FSB frequency select pins, and so that's the reason why there probable ain't JP10 viewable on your board, because the PLL can't use it! The rev5.2 board has 4 frequency select pins so I'm wondering if they both use the same PLL. You say they do, but I'm just not so sure.

Anyway, you should start with cutting the JP9 jumper and wee what happens. You can easily make a jumper at that spot, if you don't have a spare one then just take 3 wires which you can bind together, electrically its the same :)

Hmm, I'll try to figure out whether it uses the same PLL or not.

But what does the Bios of rev5 have to do with all this?

I can find multiple 5AX boards in ebay, would it help to buy several if you ever intend to try to make a turboPLL?

IF you could overclock through BIOS then you did not have to search for missing jumpers, just flash the rev5.2 bios and then you probable get the new jumpers unlocked at system boot. Unfortunately, it does not support OC via BIOS (which is kinda stupid since the PLL is software programmable).

Looking at the rev5 manual you'll notice how at 66MHz and 100MHz the PCI and AGP speeds are standard clocked, the PCI and AGP changes together with the FSB clock. There is unfortunately no third FSB setting (f.a. 133MHz) which puts PCI and AGP back to standard specs and so every clock settings above 100MHz is pretty much like TurboPLL, except that it does only affect FSB/PCI/AGP and not the whole frequency outputs of the PLL.

Downside of using TurboPLL is that software will probable not readout that well, so the oc via traditional jumpers is most likely the best way to score hwbot points, getting the rev5.2 board is then best option you'll have. Finding a board which have 133MHz support and software overclockability (setfsb) would be even greater, but since no s7 CPU runs on 133MHz FSB I don't think such board exists.

To answer your question properly: Getting such board would not help designing TurboPLL, for starters it questions why I would build TurboPLL if the board can overclock up to 140MHz FSB out-of-the-box. Only if you would need higher then that frequency, there is a reason to build TurboPLL, but since there is a strap change at only 66MHz and 100MHz I quess the board is not so different as many other boards are and with TurboPLL activated they would both hit roughly same FSB overclocks.

Had any luck so far with the JP9 jumper? Just make the cut and see if the FSB changes.

Time, time, time ... :(

Massman, I've a GA-5AX myself, bought for 20 euro. It's a 4.1, that also supports 140 fsb, so up to 840MHz!!!!

If your interested, I can make some high detailed picture's so you can compares both motherboards ;)

Edit: in CPUFSB, you can change the FSB of this motherboard. It only has one PLL so I think our motherboard have the same PLL

I hoop ook dat we de strijdbijl kunnen begraven ;)

First thread pictures shows the PLL. How high can you set the FSB in reality? What works? What is stable? Shoot pictures, always cool :)

Just get it today ;)

4.5x120 done, first need to install windows ;)

I will make picture's in about a hour or somting

Nice, at that speed it will be very hard to find a better performing board.

Please post pictures of the area around the PLL, that should make things more clear! :)

120MHz is quite good

Chipset max's out @ 125MHz

If I do 130Mhz my cpu's name becomes @^D 4^0 &Hz :P

Some bad news, the ICS is deffirent:



And, is it a Athlon, is it a Duron? No its a K6-2+



Laptop's are the way to go btw:

Laptop:
http://www.cpu-world.com/CPUs/K6-2/A...2B-500ACZ.html

PC:
http://www.cpu-world.com/CPUs/K6-2/A...-2-500AFX.html

Laptop one has on chip L2 ;)

Btw, can't you make a switch on the JP9?

A different ICS ... got to admit: that doesn't look so good. Changes are big, though, that the different ICS was needed for that JP10 jumper setting, so I'll be giving the JP9 modification a shot, no doubt about it.

Disable L2 cache in the bios. Booting will take VERY long, but you should be able to push a bit harder.

Yes K6+ is cool for sure, smaller production process, less Vcore, better overclockable and more tech on board.

If anybody might be interested, I still have here a:
K6-2/266am2
1.9V Core 3,3V I/O
00320pmw 7

For free of course

Sure, always fun to have an extra chip to play with :-).

Ok, will send it to you after te weekend.

****, Massman was first :(

Nice vCore on that chip. I will disable de L2 today, see if that makes any diffidences

BTW:

http://timkonings.ti.funpic.org/Book1.xls

All the socket 7, slot 1 and socket 370 processor\'s ever made by Intel and AMD

Edit:

L2 doesn\'t make a difference :( . My P1 133MHz boot\'s @ 233MHz :D :D

Will open the junkjard this evening, as I have still quit some older cpu\'s here.

I think Massman and I both really like some chips, so i\'m looking forward to your sale

Oh, if you find enough cpu\'s don\'t hesitate to contact more of your team.be overclocking mates ;p

Ahum!