VOGONS


First post, by ShovelKnight

User metadata
Rank Oldbie
Rank
Oldbie

Hi,

I have two Super Socket 7 motherboards that are both highly regarded: Shuttle HOT-597 and Gigabyte GA-5AX (rev 4.1).

They both work and appear to be stable. The HOT-597 was recently recapped with high quality polymer capacitors. But the GA-5AX has jumpers instead of DIP switches to set the FSB frequency, which makes it very easy to change the frequency between 66, 75, 83 and 100 MHz using a couple of SPDT switches installed onto the front panel of the case. This should give some additional flexibility when slowing down the CPU.

I heard that VIA MVP3 is considered more stable than the ALI chipset, but will I be able to tell the difference in daily use?

The plan is to use a K6-3+ and Nvidia GeForce 2 MX400 to create a PC compatible with most DOS and early Windows games.

Last edited by ShovelKnight on 2019-10-26, 12:31. Edited 1 time in total.

Reply 1 of 6, by kolderman

User metadata
Rank l33t
Rank
l33t

Use one with the k6 and build the other into a pentium mmx system. Between them and setmul you will have lots of slow down options. Also gives you more options to try different sound/graphics cards aas well.

Reply 3 of 6, by RoberMC

User metadata
Rank Newbie
Rank
Newbie

For me VIA MVP3 has not been more stable than ALI V, in fact my experiences have been much better with ALI + K6-2+ or K-6-III+ compatibility, even on boards that did not officially support them with some bios tweaking. Stability has been the same, they are pretty stable systems when the mainboard is well made. Ali performs better in some tasks/games, and VIA in others, and for me, ALI + Windows 98 is easier and more reliable, as there are only a set of final drivers and they work straight away, with VIA, there are tons of 4 in 1 versions, ones containing VXD drivers, others WDM, or a mix, and i suffered some VIA inconsistencies with AGP graphics cards that i had to solve trying different versions of the drivers. For VIA, newer drivers does not mean better drivers.

The problem with ALI is that it seems the ALADDIN V chipset might have been cheaper in its age, so there are a crap load of low quality chinese cheap motherboards out there using them. They were bad, with zero support, bad quality control, and non-existant bios updates, and some are dieing now because of bad capacitors. This is bringing bad reputation to ALADDIN V Motherboards, but it is not ALI's chipset own fault. There are very high quality ALI motherboards out there, like yours.

Right now, the Aladdin V motherboards I still keep are the BCM VP1541, also known as the Diamond Multimedia Micronics C200 (ATX), and a Jamicon 566-A-AT (Baby AT). They are both well made boards, hightly overlooked, and still can be obtained for a decent price.

The BCM VP1541 does EVERYTHING the Aladdin V platform can offer; Official K6-2+ and K6-III+ support, Big HDD support, a lot of settings for everything, etc. The Jamicon is a bit more limited, it works with the K6-2+ / K6-III+ but External cache must be disabled in BIOS (no big impact on this processors which have internal L2 cache), and it does not support BIG HDDs, but i built around it the system i am most proud of, using an old 2003 Asus barebone case modified to acomodate the Baby AT board where it fits very tightly, a Voodoo 3 (a Voodoo 2 does not fit), a Promise Sata controller, a Samsung SSD, a Sound Blaster AWE64 (A good SBPro or 16 does not fit), Floppy and CD-ROM. The system is very, very compact, stable, fast, and compatible with almost everything DOS/Win98. In fact sometimes i wonder why i keep thinkering with Pentiums, MMX, II or III when i already have the perfect system for me for that era of games. Curiosity i suppose.

I would suggest you to have the most important parts ready at your desk, and try both motherboards with the components you want to use, and see for yourself which one behaves better or you like more. Also, you can solder wires at the back of the dip-switches and install a front panel just like they were standard jumpers, if you do that just let the switches off at the motherboard and use the front panel switches as you like.

Anyway as you want to use a K6-2+/K6-III+ processor, that front panel switches will be really unneccesary, as this processors multiplier can be controlled using command line tools or windows apps. You can make them run at 386SX speeds or at full 600Mhz with a single command line, i have never seen a K6-2+ or K6-III+ that is not stable at 100x6 with the right voltage, they run pretty cool, and passive cooling with custom made heatsink is feasible.

Reply 4 of 6, by ShovelKnight

User metadata
Rank Oldbie
Rank
Oldbie
RoberMC wrote:

Anyway as you want to use a K6-2+/K6-III+ processor, that front panel switches will be really unneccesary, as this processors multiplier can be controlled using command line tools or windows apps. You can make them run at 386SX speeds or at full 600Mhz with a single command line, i have never seen a K6-2+ or K6-III+ that is not stable at 100x6 with the right voltage, they run pretty cool, and passive cooling with custom made heatsink is feasible.

You can only control the CPU multiplier on the fly with SetMul, but at the low end of the spectrum, 66 MHz FSB vs 100 MHz FSB does make a difference, especially with the caches disabled. 😀

Thanks for your suggestions, I guess I will go with the Gigabyte (ALI-based) for this build.

Reply 5 of 6, by dionb

User metadata
Rank l33t++
Rank
l33t++

I'll second that. Most of the problems reported with the Aladdin V are related to AGP compatibility, but a lot of those have been fixed in later BIOS and driver releases. In any event IDE and PCI are rock solid. With MVP3, there were multiple southbridges that could be used, each with their own weaknesses and ideal drivers. It's entirely possible to get it all working (in the day my main system was a K6-2 on a DFI MVP3 board running Win2k flawlessly), but it can be a real pain in the rear end.

Reply 6 of 6, by RoberMC

User metadata
Rank Newbie
Rank
Newbie
ShovelKnight wrote:
RoberMC wrote:

Anyway as you want to use a K6-2+/K6-III+ processor, that front panel switches will be really unneccesary, as this processors multiplier can be controlled using command line tools or windows apps. You can make them run at 386SX speeds or at full 600Mhz with a single command line, i have never seen a K6-2+ or K6-III+ that is not stable at 100x6 with the right voltage, they run pretty cool, and passive cooling with custom made heatsink is feasible.

You can only control the CPU multiplier on the fly with SetMul, but at the low end of the spectrum, 66 MHz FSB vs 100 MHz FSB does make a difference, especially with the caches disabled. 😀

Thanks for your suggestions, I guess I will go with the Gigabyte (ALI-based) for this build.

In practice i do not think you need such a bus speed resolution:

-You have the kind of games that will thank all the power you can give to them (Quake, Duke3D, etc.). You play them at 100x6

-You have games that run a little too fast at maximum power, games that needs a slower CPU to boot but then they work fine, or games that just have some weird speedups in some parts or control problems, like scrolling or transitions (Warcraft, AITD1-3 etc). You play them at 100x2 or even at 100x6 with L1 and/or L2 cache disabled, whatever fits best for that game in question.

- You have problematic games that needs a 386 age cpu to not be unplayable fast (Wing Commander, etc), You play them at 100x2 with L1, L2, and L3 caches disabled, or even at 100x4 with all caches disabled if you want a little better performance. When all caches are disabled the processor's performance is seriously crippled, in that situation the difference in runing at 66 vs 100 Mhz bus is so negligible that all the work involved in modding the motherboard/case is wasted in my opinion, as for me, all 286/386 era games run fine at 100x2 with some or all caches disabled.

Even if there were a game running a little too fast at 100x2 with all caches disabled, it would be by a very marginal small amount, the use of tools like "throttle" to slow the CPU down just a little would fix the problem. I do not like "trhrottle" tool to do a big CPU slow down, as it is inconsistent, but it works great for fine tuning.

But anyway, try the games you want to run in the test bench, and then decide if you need such a speed resolution at the motherboard bus before modding.