i am not sure about tx, but super7 chipsets seem to have 5:4 divider so they can run at 75/60 and 83/66.

I am using an Asus TX97-E in combination with a K6-III clocked at 450MHz. The system runs with a 75MHz FSB, multiplier x2 and a Vcore of 2.5V. The K6 needs proper cooling but apart from that the system is stable so far.

My QDI Titanium IIB runs fine on a 75MHz bus, using a K6-2 and K6-3+. 83MHz though is a no go. It won't even POST.

A couple of months ago I ran a system based on the Fordlian/RedFox TX-5IB2 mainboard for a few weeks (actually it was the first and only system I've ever finished Quake2 on 😀).

I tried 75MHz with an AMD K6-2/450 CPU (multiplier set to 2x) - everything was fine under Win98SE except when I tried to activate DMA for the IDE drives, which made everything boot/open/run like a hundred times slower. So I ran it at 66MHz with a K6-2/400 (the difference in performance between 450 and 400 wasn't too obvious, and neither was the difference between DMA on and DMA off with the 2.5GB Maxtor drive).

There is no documented 83MHz FSB option in the manual - however, the board's layout is identical to that of Acorp 5TX29, for which I found a Russian web page with both detailed jumper settings and a hacked BIOS that is supposed to enable support for K6-2+/III/III+ CPUs, so I still have plenty more to experiment with...

pyrogx wrote:

I am using an Asus TX97-E in combination with a K6-III clocked at 450MHz. The system runs with a 75MHz FSB, multiplier x2 and a Vcore of 2.5V. The K6 needs proper cooling but apart from that the system is stable so far.

Same here, I've used it from 1999 until ~2002 without any problems. 83 MHz FSB basically worked, but the onboard IDE controllers had to be slowed down to PIO2, and even then stability was somewhat iffy.

I run a P200MMX at 225Mhz with a 430TX based SQ575 running at 75mhz. It's very stable. I should run some benchmarks some time and see if 233@66 is any faster than 225@75. How important is that extra memory bandwidth?

used to run a tx97e with a 233mmx at 3.5 x 75 stable100% an also 3.5 x 83 98%stable

Thanks to everyone for their replies! I was also hoping to note your memory configuration as well, weather it was EDO/FPM/SDRAM and how many sticks of how many MB's. This seems to play a role, as shown below.

I'm using a 430TX which has a Jan S. modified BIOS to support K6-III/II/II+/III+ (and Cyrix MII 4x)

At,

83 MHz, one stick SDRAM - 64 MB total
K6+ at 500 MHz, 2.2 V
Everything appears to run fine, all the benchmarks run, but I did not try installing XP or doing anything that would constitute regular everyday use. I ran a few hours of benchmarks for the Ultimate 686 Benchmark Comparison. Post-benchmark testing follows below.

83 MHz, two sticks SDRAM - 256 MB total
K6+ at 500 MHz, 2.2 V
Computer won't turn on!

83 MHz, one stick SDRAM - 256 MB total (I don't think this density is really supported)
K6+ at 500 MHz, 2.2 V
Everything appears to run fine, but I got 2 errors during the entire W2K installation, Later on, XP wouldn't boot after closing Microsoft Update (IE6-based) - 2 bad sectors on the HDD, possibly due to running the PCI bus too high? Ran chkdsk /r /f and it works again, but has 2 bad sectors now.

75 MHz, two sticks SDRAM - 256 MB total
K6+ at 450 MHz, 2.2 V
Everything appears to run fine, but the computer hangs on Microsoft Update (w/Office installed). It takes about 25 min. of checking for updates before it hangs. The service pack installed fine, all programs installed fine, but MS Update will hang when it has these 40 or so Office 2003 updates to install. Once those updates were installed at 400 MHz, 450 MHz no longer has an issue with MS Update. I should point out that Windows Update (not Microsoft Update), was able to install hundreds of updates at 450 MHz without any problems. Yesturday, however, the system froze during a disk defrag operation in W2K. I have since adjusted "Fast RAS to CAS Delay" from 2 to 3 and "Fast MA to RAS# Delay" from 1 to 2 in hope that it corrects something. The system hangs are non-repeatable it seems. You just need to wait them out.

66 MHz, two sticks SDRAM - 256 MB total
K6+ at 400 MHz, 2.2 V
MS Update finished checking for officie updates and installed numerous updates without a problem. For the sake of repitition, I didn't install the updates first, I just made sure it would finish. Then I ran the computer at 75 MHz again to ensure it would hang, it did. Then went back to 66 MHz to ensure it would finish searching and installing, which it did. So there is something that 75 MHz is too much for.

Things I checked, which did not help the 75 MHz issue.
- Using a single stick of 128 MB
- Tried K6-III-450AFX, K6-III+/450ACZ, K6-III+/400ATZ. 450AFX is currently installed.
- Tried a K6-2 thinking the onboard L2 cache of the K6-III was interfering.
- SDRAM Speculative Read (this actually should always be disabled on the 430TX. Intel released an addendum saying they found an issue with it and it should be disabled)
- Adjust Mem. Drive Str. (Ma/RA) to 16ma/16ma, which is what it should be for 2 double-side SDRAM DIMMs (4 rows)
- DRAM Refresh at 15.6 us, which is what it should be for SDRAM.
- SDRAM CAS Lat/RAS-to-CAS, I left it at 2/2 because the RAM is PC100 CL2
- Pipeline Cache Timing, options are Faster and Fastest. You can use Fastest if you have double-bank L2 cache (according to the i430tx manual), otherwise Faster. I left it on Faster, Fastest seems to increase the motherboard's cache speed by a noticable amount and boots into Windows fine. I'm not sure how to determine if there is one bank or two of L2 cache on this board. There are 2 TQFP chips which give it 512KB L2, Everest Home Edition listed it as doubled bank 512 KB under the Motherboard/Chipset section. At anyrate, this option is one I unhid with a BIOS mode, the default was faster, which is where I'll leave it for testing.
- System and Video BIOS cacheable set at disabled
- PCI 2.1 Compliance is enabled

The BIOS items I am still checking are:
- Fast RAS to CAS Delay, 2 or 3. Was at 2, now on 3.
- Fast MA to RAS# Delay, 1 or 2 clk. Was at 1, but now on 2
- Using the lowest power CPU, the AMD K6-III+ ATZ at 2.0 V and 450 MHz.

If the above 2 BIOS options don't cure it, then:
- DRAM Page Idle Timer, 2, 4, 6, 8 clks. Left at 2 clks, but I can increase it.
- DRAM Enhanced Paging, enabled, or disabled. Left enabled, but I can disable it.
- DRAM CAS Lat/RAS-to-CAS, from 2/2 to 3/3
-Run the CPU at 4x75 instead of 6x75 to determine if the voltage regulators are having trouble delivering enough current at 6x75.

To circumvent any issues with overrunning the PCI bus in terms of frequency, I am using a GeForce 6200 because it supposedly supports 66 MHz PCI speeds. I am also using a Promise TX2 Ultra100 card, which also supports 66 MHz. The item I am not sure about is the Intel Pro 100S ethernet card. Yesturday, after all this testing was finished, I decided to add in a Yamaha PCI YMF724F sound card. I'm not sure how tolerant this card is at speeds above 33 MHz, but I was at the point in setting up my system that I needed sound. This is a quad-boot setup, with Win98SE, NT4, W2K, and XP Pro. NT4 seems the least likely to crash.

It is important for me to find someone who has run a 430TX system as an everyday computer back when it was popular. I'm interested in 75 and 83 MHz users only - What CPU and RAM were used, and what was the system used for.

JaNoZ wrote:

used to run a tx97e with a 233mmx at 3.5 x 75 stable100%

Was the system used as an everyday computer and for how long? How much RAM and how many sticks? What OS?

Hatta wrote:

I run a P200MMX at 225Mhz with a 430TX based SQ575 running at 75mhz. It's very stable.

Was the system used as an everyday computer and for how long? How much RAM and how many sticks? What OS?

5u3 wrote:

Same here, I've used it from 1999 until ~2002 without any problems. 83 MHz FSB basically worked, but the onboard IDE controllers had to be slowed down to PIO2, and even then stability was somewhat iffy.

Was the system used as an everyday computer or an occasional gamer? How much RAM and how many sticks? What OS?

iulianv wrote:

I tried 75MHz with an AMD K6-2/450 CPU (multiplier set to 2x) - everything was fine under Win98SE except when I tried to activate DMA for the IDE drives,

To what extent was the system used and tested? Many systems may seem fine initially, until you start to use them more frequently. What RAM were you using? I'd recomend using a some cheap Promise Ultra100 card to avoid your DMA IDE issues.

F2bnp wrote:

My QDI Titanium IIB runs fine on a 75MHz bus, using a K6-2 and K6-3+. 83MHz though is a no go. It won't even POST.

Was the system used as an everyday computer and for how long? How much RAM and how many sticks? What OS?

pyrogx wrote:

I am using an Asus TX97-E in combination with a K6-III clocked at 450MHz. The system runs with a 75MHz FSB, multiplier x2 and a Vcore of 2.5V. The K6 needs proper cooling but apart from that the system is stable so far.

Why did you choose 2.5 V? There were two voltage ratings of the K6-III, 2.2V (AFX) and 2.4V. Was a lower voltage found to be less stable? Was the system used as an everyday computer and for how long? How much RAM and how many sticks? What OS?

No, it wasn't a primary use computer. It was for late dos/early 9x gaming with 6.22 and Win 95 OSR2. I had 2x64mb of SDRAM in it at the time. It did get used for many >4hour gaming sessions in both operating systems, along with running memtest86 and prime95 overnight to ensure stability.

I've since switched to 64mb(4*16mb) of EDO, I found that it was faster. But I haven't used the computer that much since then so I don't know if it affected stability.

Hatta wrote:

I've since switched to 64mb(4*16mb) of EDO, I found that it was faster. But I haven't used the computer that much since then so I don't know if it affected stability.

EDO RAM faster than SDRAM? How did you determine it was faster and what BIOS settings were you using? I suspect EDO would not be faster if you adjust the BIOS timings such that the system is stable.

This is my TX97 chipset configuration:

I have a 2.4V K6-III/400 but I found that it needs a VCore voltage of 2.5V in order to run stable at 450MHz. A K6-2 400 clocked at 450MHz was fine with its 2.2V default setting. Atm there is a single PC133 128MB SDRAM stick in one of the DIMM slots.
I am still using the system almost every day (in fact it is running right now and has been since several hours without any problems), mostly for playing older games and toying around with old and obscure hardware under varoius OSes (Windows 98, DOS, Linux, OS/2). None of my sound cards or graphics cards has refused to work with the PCI clock above 33MHz so far, even the most brain-damaged PCI soundcards work fine.

pyrogx wrote:

This is my TX97 chipset configuration:

Ok, thanks for the additional information. The usage of the K6-III-400 explains the 2.5V requirement.

Can you run your system with 2 sticks of 128 MB RAM (16Mx64 stick configuration)? Our BIOSes are pretty similar. I've attached mine.

feipoa wrote:

5u3 wrote:

Same here, I've used it from 1999 until ~2002 without any problems. 83 MHz FSB basically worked, but the onboard IDE controllers had to be slowed down to PIO2, and even then stability was somewhat iffy.

Was the system used as an everyday computer or an occasional gamer? How much RAM and how many sticks? What OS?

It was my everyday machine, which started out as P5-133 with 64 MB of EDO RAM, but got upgraded later with a K6-233 and finally a K6-III-400 with 256 MB SDRAM (2x 128 MB PC-100 CL2).
The OS were Win98SE and Linux. The Linux kernel's verbose error logging helped me to find out about the IDE problems at 83 Mhz FSB speeds.

Since I had the crappy 2.4V AHX model, good cooling was important, I had a Socket A HSF installed, as well as big heatsinks for the voltage regulators (these are a weak point in TX boards, most of them weren't designed to provide the 30W needed for 450 MHz). It also didn't work with my old AT PSU, so I upgraded to ATX (the board had connectors for both).

I was surprised too, and it was a pretty small advantage to the EDO. Unfortunately I'm drawing a blank as to what tool I used, and I don't immediately see it on the hard disk. I also installed another 128K of cache, so that might change my results. I'll reinvestigate this weekend and make another thread.

P5TX-Apro works pretty good at 83mhz fsb with everything else stock and the pci works very well at 41.5mhz. The only thing that sucks about the board is the jumpers and they can be picky if the quality isn't perfect.

5u3 wrote:

It was my everyday machine...

You've made an important observation about the voltage regulator for the CPU voltage, especially if it is a linear regulator. Do you have a photo of this mod? I'll try to measure the heat output on my regulator.

I have the K6-III/450AFX 2.2V chip installed now, but maybe it would be better if I ran the K6-III+/400ATZ 1.6V chip at 450/2.0v. I also have the AMD K6-III+/450ACZ 2.0V chip. Any idea if one of these two + chips runs cooler at 450/2.0v?

I tried soldering on the ATX header to my motherboard, but the board wouldn't turn on, so I need to use the ATX-to-AT adapter cable.

When you upgraded to 256 MB RAM, was the computer still your everyday use machine?

feipoa wrote:

You've made an important observation about the voltage regulator for the CPU voltage, especially if it is a linear regulator. Do you have a photo of this mod?

Unfortunately no, I don't have my TX97-E board any more. But I'm sure it had switching regulators, it's even mentioned in the manual. My mod was pretty simple though, I cut up an old 486 heatsink and epoxied the pieces on top of the regulators plastic cases. A better solution would be to raise the regulators from the board and screwing the HS to the back side, but AFAIR mine were SMD mounted and I was afraid to ruin the board.

feipoa wrote:

I have the K6-III/450AFX 2.2V chip installed now, but maybe it would be better if I ran the K6-III+/400ATZ 1.6V chip at 450/2.0v. I also have the AMD K6-III+/450ACZ 2.0V chip. Any idea if one of these two + chips runs cooler at 450/2.0v?

The + chips should run much cooler. My current retro K6 is a K6-3+ 400 ATZ running at 2.0V, and I didn't even bother to mount a fan on the CPU heatsink. A light breeze from two slow 12cm case/PSU fans is enough, even at 500 MHz.

feipoa wrote:

When you upgraded to 256 MB RAM, was the computer still your everyday use machine?

Yes. I upgraded at the same time as the K6-III CPU.

when i bought the board i paired it up with 2x 32mb sdram ti 10ns, was really something to have 64mb total and also quite expensive back then.
timing at spd do not really know as i wasnt realy into ocing. as i might recall cas2 was not running ok so i was at cas3.
i ran win98 SpecialEdition, 83mhz would give me a spontaneous reboot sometime at a hot day in the summer.
and i was using the system for about 4-6hrs a day in the evening.

about your bad sectors feipoa they are no hard errors on the disc. just corrupted miswritten sectors no big deal. have them fixed with spinrite 6 program, and they will be rewritten. but the mft or fat will keep them at bad of course.

5u3 wrote:

Unfortunately no, I don't have my TX97-E board any more. But I'm sure it had switching regulators, it's even mentioned in the manual. My mod was pretty simple though, I cut up an old 486 heatsink and epoxied the pieces on top of the regulators plastic cases. A better solution would be to raise the regulators from the board and screwing the HS to the back side, but AFAIR mine were SMD mounted and I was afraid to ruin the board.

Did the motherboard die?

Were the core and I/O switching regulators getting hot enough to need heatsinks? I see from images online that your board didn't use any heatsinks on the regulators.

My board, however, is a totally different story (attached photo). It uses linear regulators with heatsinks.

D45H2A - PNP Power Amplifier
3.3 V (collector to motherboard GND)
42 degrees C
[150 C max, 10A max, 60W max at 25 C]
This is the item with the largest heatsink in the photo (left-most). It may be responsible for the CPU I/O voltage, SDRAM voltage, and PCI voltage.

BYV118 - Rectifier Diode, Schottkey Barrier
2.07 V (cathode to motherboard GND)
48 degrees C
[150 C max, 10A max]
This is the middle regulator in the photo. I think it is responsible for the CPU core voltage.

NDP603A - N-Channel MOSFET
5 V (drain to motherboard GND)
52 degrees C
[175 C max, 25A max, 50W max at 25 C]
This is the right-most regulator in the photo. It may be respnosible for EDO/FPM SIMM voltage and PCI 5 V.

The hottest surface temperature measused was 52 C on the right-most linear regulator while running Quake1 continuously. The spec sheets mention the max operating temperature for all three regulators as being at 150 C and 175 C, so I don't think they are even getting close to overheating with the supplied heatsinks. However, the warmer they run, the less power they can deliver to the load.

5u3 wrote:

The + chips should run much cooler. My current retro K6 is a K6-3+ 400 ATZ running at 2.0V, and I didn't even bother to mount a fan on the CPU heatsink. A light breeze from two slow 12cm case/PSU fans is enough, even at 500 MHz.

Ok, it seems the + chips were fabbed with 0.18 micron, whereas the non-plus chips were 0.25 micron. Here is a nicely organised website which goes through all the power requirements of the AMD K6,
http://www.amd-k6.com/cpu-specs/

The K6-III+ 450ACZ at 2.0 V is mentioned has having a 12/16 W power requirement. The K6-III+ 400ATZ at 1.6 V is mentioned as having a 7.1/9.5 W requirement. I figure the 400ATZ run at 2.0 V and 450 MHz will have the same power requirement as the 450ACZ, so 12/16W. The K6-III 450AFX at 2.2 V is marked as 12.1/20.2 W, which is much higher on the upper end. The manual for my motherboard mentions it supports the K6 at 300 MHz (4.5x66). This chip has a power requirement of 9.2/15.4 W, which seems similar to the 450ACZ. From this comparison, I am going to assume that I don't need additional cooling on my motherboard's voltage regulators. I am now running the I am now using the 400ATZ at 2.07 V and 450 MHz.