feipoa wrote on 2025-02-13, 07:41:

bertrammatrix wrote on 2025-02-12, 04:38:

... unless of course my hardware is glitchy and some boards may be able to run faster settings with some bios, which can certainly be the case

May be. This is something I need to test for again. It may be an edge case situation. My board can do 180 Mhz with 1024K L2, and 128 MB EDO with 2-1-2 SRAM and DRAM at 1 ws read, 0 ws write. It may have been that I needed to set DRAM to 2 ws with the 5/6/1996 BIOS, but I don't recall.

Ugh 2-1-2, I wish. I still have to use the slowest setting with 60mhz FSB. I was honestly hoping this would improve seeing the 1024 module has some nice fast chips, comparatively to the original 256k one (which would also only work on the slowest setting on a 60mhz fsb)

Mmm maybe I will try to slow the dram down a tad and see if that helps.

I had setup Cyrix 5x86 120MHz system running at 133MHz before. (33MHz*4, Cyrix 5x86 enhancement registers enabled)
As for my experience (with the help of feipoa), Cyrix 5x86 CPU is very sensitive to heat and also proper setting of voltage is required to get stable.

At first, it looked stable at 3.52V in open case condition but with the case closed it failed soon when run GL Quake. Finally I had to set voltage at around 3.7V and had to use several fans to remove heat to outside of case through open ISA slot. In this condition this system did not get failure during the test. (Running GL Quake continuously in Windows95 kept around 30hrs until the test ended. There was no sign of failure and I just decided to stop the test)

I set up several OS including Windows NT, Windows98 and even Windows ME besides DOS in this system and it runs without problem as to my purpose (although this CPU seems not qualified stable at 133MHz by Cyrix. This CPU has only 120GP mark on it)

I’m using PCCHIPS M919 v3.4F (with 128MB FPM, 1024kB L2 cache, fastest ram timing, 1996/5/6 Bios), Matrox G200 PCI, Voodoo2, Orpheus II, SAAYM, Tandy Nano, Adaptec 2940U2 with ZuluSCSI V6.4 and so on.

There seems some Cyrix 5x86 CPUs with 4x capability exist though they are very rare. And it surely is very fun to play with, but it should be noted that voltage and heat removal is very important to get them run stable.

Assuming you don't have air conditioning in your house/room, try runing the test again on a hot summer day.

feipoa wrote on 2026-04-20, 20:47:

Assuming you don't have air conditioning in your house/room, try runing the test again on a hot summer day.

Actually testing was done at the room temperature in my room (around 23degree C). If in hot summer without air conditioner, it could be the system can get failure in shorter time. I could feel hot air came out from inside even in room temperature, obviously it will be more harsh condition if ambient temperature is higher.

I wonder how 133MHz/4X CPU is in terms of heat stability. For example, upto which ambient temperature it can run without failure? Certainly it will run at higher ambient temperature than mine considering it is official 133MHz rating and mine is only 120MHz.

I find playing the games to be more telling of failures as opposed to letting the games run in auto loop. Even better is if you have kids, let them play a few games for several hours.

Did you measure the ambient temperature inside your case? Should be near 33-38 C.

feipoa wrote on 2026-04-21, 23:40:

I find playing the games to be more telling of failures as opposed to letting the games run in auto loop. Even better is if you have kids, let them play a few games for several hours.

Did you measure the ambient temperature inside your case? Should be near 33-38 C.

I didn’t measure and I agree it should be hot and I expect it should be more than 30 degree C. (Here in summer ambient temperature is more than 30 degree C and goes up to 35 degree C or higher. I felt the air temperature came from inside the case was similar to that of wind in hot summer)

I should agree that I didn’t play not that many hours than testing GL Quake auto loop. It could be it is not that stable than auto loop test in real world use.

In other point of view, I may just be satisfied with results upto this point than getting more perfect stability. Running Cyrix 5x86 at 133MHz is very exciting experience but I put more attention on how easy to change performance.

This system is very unique to me in that I connected turbo button to FSB to select 25MHz or 33MHz. Therefore it can run at 100MHz when turbo off and 133MHz when turbo on. And as you know, multiplier can be changed using setmul or cpuspd. Combining these, it can cover 25MHz to 133Mhz easily. Furthermore toggle on/off of Cyrix enhancement features, performance can be adjustable further.

In this point of view, Cyrix 5x86 4X capable CPU is better than 3X CPU. Multiplier can be selected among 1, 2 and 4, Resultantly this system can run at 25MHz, 33MHz, 50Mhz, 66Mhz, 100Mhz and 133Mhz covering nearly full range at that era.