gravitone wrote:

The kt133(A) is for all intents and purposes, the most unholy chipset of them all.

Why? I love mine.

ok lets see:

Writeback buffer corruption when using UDMA on IDE channel 1.
SLow memory controller, VIA disabled support for memory banks 7&8 due to stability issues.
PCI bus mastering not properly implemented,(try a sblive and you'll see)
IRQ allocation issues (bios?)
Soft reboot bug which causes the screen to remain black (can be fixed in some cases by upping the Vcore voltage).

It may also be a slight personal bias, but I have yet to encounter a decently stable kt133 system. Having owned one, and having to troubleshoot a whole load of them throughout the years has given me a sour taste for VIA chipsets that will never disappear again.

The km133 series is a nightmare to work with with its horribly slow integrated savage 4 graphics core draining the already weak memory bandwidth to the point of starvation for any decent CPU.

VIA is (or at least was) terrible. My worst experience was with VIA Apollo-based (Super?) Socket 7 boards with AMD K6-2 CPUs. Notoriously unstable due to PCI compatibility issues.

AGP performance was *terrible* on both Apollo and KT133 chipsets in my experience, until you installed chipset drivers.

My opinion is that VIA became pretty good with KT266A and newer.

Part of the problem was that they were the chipset of choice for ultra cheap mobos. Those mobos had their own problems, such as horrible BIOSs, that certainly didn't help anything.

CPU testing update. I've just tried another CPU in the first working QDI board - the one featured in the O.P. The CPU is an XP-M mobile Barton. Its CPU code id is AXMH2400FQQ4C, and it can be found on cpu-world here -

http://www.cpu-world.com/CPUs/K7/AMD-Mobile%2 … H2400FQQ4C.html

The one I have is brown in colour, and has a curious layer of brown plastic "see through" layer covering the small "things" clustered around the CPU head.

When I look at the BIOS POST screen, the mobo correctly identifies this CPU as a Mobile CPU, but incorrectly reports its speed as 800mhz. I get to the Windows 98 desktop OK, and then run SiSoftware Sandra, to test out the speed of the CPU. It's slow. So, there's a problem here. I don't know why the mobo thinks this CPU is only 800mhz, when it should be 2.4Ghz. (Please note that this mobo correctly accepts a 2.4Ghz green coloured T-Bred.)

retro games 100 wrote:

CPU testing update. I've just tried another CPU in the first working QDI board - the one featured in the O.P. The CPU is an XP […]
Show full quote

CPU testing update. I've just tried another CPU in the first working QDI board - the one featured in the O.P. The CPU is an XP-M mobile Barton. Its CPU code id is AXMH2400FQQ4C, and it can be found on cpu-world here -

http://www.cpu-world.com/CPUs/K7/AMD-Mobile%2 … H2400FQQ4C.html

The one I have is brown in colour, and has a curious layer of brown plastic "see through" layer covering the small "things" clustered around the CPU head.

When I look at the BIOS POST screen, the mobo correctly identifies this CPU as a Mobile CPU, but incorrectly reports its speed as 800mhz. I get to the Windows 98 desktop OK, and then run SiSoftware Sandra, to test out the speed of the CPU. It's slow. So, there's a problem here. I don't know why the mobo thinks this CPU is only 800mhz, when it should be 2.4Ghz. (Please note that this mobo correctly accepts a 2.4Ghz green coloured T-Bred.)

Weird, maybe it has a 6x multiplier or something 😜

According to cpu-world.com, this AXMH2400FQQ4C CPU has a multiplier of 13.5. I wonder if the QDI's BIOS has no "understanding" of that specific 13.5 multiplier value. So, as it cannot set up the CPU at 13.5 multiplier, it must pick another. It seems to pick 6x, as the CPU speed is reported as being 800mhz (6x133). That's a shame. Inside the QDI BIOS, there is no area where you can set your own multiplier value setting. There is only a FSB OC'ing section.

On my motherboard, when left at default it seems to juggle it's self between 12.5x, 11x, and sometimes even 9x. If you really cared you test this behavior be hitting the reset and stuff.

So rg100.... since you seems to have a whole crew of kt133a mobos around, wanna try a couple tests for me 😉 ???????

prophase_j wrote:

So rg100.... since you seems to have a whole crew of kt133a mobos around, wanna try a couple tests for me 😉 ???????

YES! 😁

Sweet. Basically I want to know about 2 different functions that the chipset is capable of, but non-functional as observed with mine. I have seen reports on the web that they're possible though.

1. Software Cooling.

I can get this to work on mine, but it isn't stable. The actual name of the feature is called "disconnect while idle". What I have read is that when this is enabled the amount of power needed by the processor shifts too often/too fast for the voltage regulator to cope with with, and eventually an under-voltage with arise and lockup. It's not clear weather this is inherent to the design of the particular board or a function of crappy caps.

There are two easy ways to do this: H.oda's coolon, and also by maunally hitting a register with wpcredit.

2. Software Multipler adjustment. This trick sounds really promising, but every time I have tried it resulted in instant lockup. Again, my research leads me to believe that it is a function of the MB's design. For this to work you need a mobile or otherwise unlocked chip.

This one is pretty easy too, I tried to use , and it seemed like it should work. Depending on how the chipset is programmed, you may have to change a register with wpcredit first.

Here are some files to help you out:

Here is the link to get CPUMSR, which is what I tried to use to change the speed

http://www.cpuheat.wz.cz/html/Download.htm

One last addendum:

The register for multiplier adjustments is 55 bit 2, it needs to be set to 1. The register for cooling is 55 bit 1 = 1, and if you look at register 52 it will be 6B when it is off, and EB when it is on.

OK! Here goes -

I am using this barton 2400 CPU: AXMH2400FQQ4C. Inside the BIOS, I have set the core voltage to its lowest possible setting of -0.1V. Running Speedfan, the core voltage value is 1.52V (cpu-world tells me this CPU is 1.45V).

So, I first run the Coolon program. I tick the Cooling checkbox on the PCR tab. (I notice in the coolon program that the register changes from 6B to EB. I also notice that the "value in box number 52" inside the wpcredit program is set to EB.) I then look at Speedfan's idle temperature monitor value. Unfortunately, this idle temp (35) does not decrease. I minimise Coolon so that it continues to run in "Cooling mode"...

I then run the setup.exe program for the lladrv low level driver. This is from reading the set up instructions on the CPUHeat & CPUMSR projects homepage - from your web link provided above. This program installs then asks for the machine to be rebooted. This reboot occurs. However, I don't see any new "hardware icon" inside Control Panel -> System area. Should there be anything new to see in this System area?

Anyway, I then rerun Coolon.exe again, and tick the Cooling checkbox again. I minimise this cooling app, then run the main cpumsr.exe program. This program shows lots of error messages, starting with this pop up: "Driver initialisation failed". And then the next error message pop up box says: "Driver not running". About 3 more errors appear, until I am finally at the cpumsr program itself. I look at the multiplier change options section, and I see that I am only allowed to choose up to 11.5x. I think it's 11.5. Maybe it was only 11. Sorry, I forget. Anyway, I need 14 Edit: Sorry, I meant I need 13.5 for this AXMH2400FQQ4C CPU, not 14., so I guess it's time to check out wpcredit...

I run wpcredit, and this is where I am now unsure as to what to do. I see the "value in box number 52" as EB, and then I move my cursor a few places to the right, to where box number 55 is. I see the value 09 in it. I am not sure what to change this value to. Please advise.

BTW, I think the cpumsr.exe is really upset about the "Driver not running" thing. Do you think this will mess up any more tests?

Sorry for not being more clear about this. Where your seeing things like "EB" and "09", that is a hexadecimal representation of the bit field. Look where you see the "01000000" or what ever, the string of 0's and 1's. You'll count the rightmost as bit 1, the next as 2 and so forth. That's where you can adjust it that way. Then click set and it should take effect.

Defiantly going to need the LLA driver working. Try to uninstall, reinstall, reboot. Other wise I think there was another tool out there i'll just have to look for ya.

Progress made - well, kind of. I got the LLA driver successfully installed: I used the Add new hardware wizard. Now, I run the cpumsr.exe, then I get this pop up error message: "Processor could not be changed! Unknown error occured". I click OK, then I'm taken to the cpumsr program itself. If I try and set the multiplier to 13.5 @ 1.45V and then click the SET button, my computer freezes. I restart my machine, and try this again, but pick a "safe" value of 12.5 - which is the current multiplier setting. No freeze, that works. So, I try 13x multiplier. Freeze, reboot time.

It's at this point I realise I've forgotten to set the "55 bit 2" to 1 value inside wpcredit. Sorry. I do that, and now the 55 value is 00001011. (It was 00001001) I decided to reboot, and run wpcredit again just to make sure my new bit setting had been saved correctly. It hadn't. My change was missing. Bit 2 was back to 0 again. I set it to 1 again, saved then exited. I reran wpcredit without rebooting, and my bit changed had been saved OK. So, I guess you can't reboot if you want that bit change to "stick".

Then, I rerun cpumsr.exe, select 13.5 multiplier @ 1.45V, and click the SET button but the PC instantly freezes. I notice that 13.5 is the highest value I'm allowed to pick. This board's BIOS allows you to pick 14. If I run a t-bred 2400, the board somehow manages to "pull out of the hat" a 15x multiplier to run it correctly! Please note: I'm currently doing these tests on the Epox revision 1.1 board. Perhaps it's now time to switch over to that QDI board, and rerun all of these tests? (Actually, I'll do that...)

Your getting it. At least now you can say that you have experience with programming the registers of a chipset. Seems like your getting the same result that I get when I try it, maybe the QDI will be different. I surprised that you didn't see any change with the software cooling, it looks like everything was in place for it.

OK, I've got the QDI board operational, with the aforementioned 2400 barton CPU in it. The QDI board doesn't seem to cope with this barton chip as well as the epox 1.1 board does, because the speed is reported as only 800mhz, which is 600x133fsb, with a multiplier of just 6x. It should be 13.5x, according to cpu-world.

I run wpcredit, and set both right bits to equal the value of 1. So, before I saw 00011000 (interesting, this differs from the epox's 00001000 value), and now after setting both right bits I see: 00011011. I save my changes and exit wpcredit.

I run speedfan, and note the idle temperature. It is 30C. I then run coolon, and before I tick the "cooling checkbox", I see that the "off" value for this field is 6C, which differs from the epox's "off" value of 6B. I tick the cooling checkbox, and I now see the value EC (not EB, from the epox board). I look at Speedfan's idle temperature for about 1 minute, and it doesn't alter. I run wpcredit, and see that the "52 value" is indeed "EC". I decide to minimise coolon, and let it carry on trying to cool the CPU - but I fear that something is amiss and it cannot achieve this.

I run cpumsr, and again I see the error message pop up box saying: "Processor could not be changed! Unknown error occured". I click OK, then I'm taken to the cpumsr program itself. If I try and set the multiplier to 13.5 @ 1.45V and then click the SET button, my computer freezes - just like the epox board.

Edit: I have tried all of the above on a 3rd board, an Abit KT7A revision 1.1 non-raid mobo. I get exactly the same results.

I'm currently testing probably my last KT133A chipset based mobo. This one is made by Jetway, and it's called 663AS Ultra, revision 3.2 board, with a "BOM" (whatever that means) of 4. The following jetway support page says that with this board revision and BOM number, and with the latest A08 BIOS, you can use a Palomino 2100+ CPU @ 133 FSB. I successfully flashed the BIOS from A07 to A08, and the BIOS POST datestamp changed to "I can't remember/2001" to 07/30/2002.

http://www.jetway.com.tw/jetway/CPU-support/663ASU32.htm (shows CPU support list for this board)

I should have read that page more carefully, because I wasted half an hour wondering why no t-bred CPU would POST in the mobo. I'm using this jetway board alongside my "new" Enermax PSU, which has a meaty 5V rail of 40A, and which is powering an ATI Radeon 9800 Pro graphics card. Out of all the KT133A boards I've tested, this is the one that won't work with any t-bred CPU. Nevertheless, I've got a Palomino 2000+ handy, and so in it goes...I then run 3dMark2001 first edition at maximum settings (including 6x AA), and I get 6544, which seems OK.

I notice that the RAM sticks look a bit different. They are taller than my other SDRAM sticks, and they are single sided, with a sticker saying DIMM128MBSD133. Current CPU idle temp is 44C. That's quite high, but not too bad. Of couse, it would be lower if I was able to use a t-bred or barton, etc, but I can't use them in this board. Several of the caps look a bit domed. I gathered from the seller that it's been in use up to last week, so I guess the board has seen a lot of service! Inside the BIOS, I cannot see any specific overclocking options, other than some basic voltage adjustment - plus and minus values.

RG100, Glorious Knight of Valor(dation) for the Archaic