Tetrium, what was your prognosis of the Cyrix 5x86 - 133? Does it have a 4x multiplier in it? Does it run stably at 33/133?

I have two systems which run exceptionally well with the Cyrix 5x86 - 120 in WinNT4 and Win98SE (all enhancements turn on, except for branch prediction). I would be interested in testing out one of these phantom Cyrix chips if you have one, or know of someone who has one to spare.

feipoa wrote:

Tetrium, what was your prognosis of the Cyrix 5x86 - 133? Does it have a 4x multiplier in it? Does it run stably at 33/133?

I have two systems which run exceptionally well with the Cyrix 5x86 - 120 in WinNT4 and Win98SE (all enhancements turn on, except for branch prediction). I would be interested in testing out one of these phantom Cyrix chips if you have one, or know of someone who has one to spare.

I still need to have it shipped, together with a whole bunch of other CPU's. I'm not ina rush to collect it and having CPU's shipped more often only means it'll cost me more.
Sorry.

Do you know the buyer of that eBay auction? If so, can you send him my contact information, or give me his contact information? I'd like to buy one if there was a surplus.

It was a groupbuy at cpu-world.com. I'm pretty sure the seller doesn't have any more, or I would've seen an option for another groupbuy. They are seriously good at spotting these kinds of sales 😉

Tetrium wrote:

feipoa wrote:

Tetrium, what was your prognosis of the Cyrix 5x86 - 133? Does it have a 4x multiplier in it? Does it run stably at 33/133?

I have two systems which run exceptionally well with the Cyrix 5x86 - 120 in WinNT4 and Win98SE (all enhancements turn on, except for branch prediction). I would be interested in testing out one of these phantom Cyrix chips if you have one, or know of someone who has one to spare.

I still need to have it shipped, together with a whole bunch of other CPU's. I'm not ina rush to collect it and having CPU's shipped more often only means it'll cost me more.
Sorry.

Where the heck did you manage to find a Cyrix 5x86 133? I thought the 120's were hard enough to find, the 133's are almost mythical.

Can you post the link to this past group buy? I am having trouble finding it on cpu-world.

I "think" I posted it before. Hold on, looking...

Edit:Link:
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem … em=120671144113+

Edit2: Wait, did you mean the topic on cpu-world.com itself?
For that, you'll need to ask permission from one of the admins.

Yes, for the announcement that was on cpu-world. I cannot find it. If it was a non-public announcement, can you let me know which admin you are referring to? Thanks.

feipoa wrote:

Yes, for the announcement that was on cpu-world. I cannot find it. If it was a non-public announcement, can you let me know which admin you are referring to? Thanks.

I think you need to ask CPUShack.
But just a friendly note:
That section is reserved for people who are traders, it's not a discussion subforum.
It's supposed to have limited access to keep things running fluently I think, and I can understand that. You'll want people in there who you can trust 😉

feipoa wrote:

Tetrium, what was your prognosis of the Cyrix 5x86 - 133? Does it have a 4x multiplier in it? Does it run stably at 33/133?

I have two systems which run exceptionally well with the Cyrix 5x86 - 120 in WinNT4 and Win98SE (all enhancements turn on, except for branch prediction). I would be interested in testing out one of these phantom Cyrix chips if you have one, or know of someone who has one to spare.

Have you tested if your chips have a 4x multi btw?

I have several Cyrix 5x86 chips, none of which have the 4x multiplier
2x Cyrix 5x86-120
1x Cyrix 5x86-100
2x IBM 5x86-100HF in a Gainberry upgrade marked as 120 Mhz

I wish there was an easy way to short, or add some diode or resistor to force these into 4x mode. Maybe there is. Does anyone have a detailed schematic of the chip? It was probably internal only. The 5x86-100 I have runs fine at 120 Mhz, so I am thinking the 120 Mhz would run fine at 133 Mhz?

The 5x86-100 I have runs fine at 120 Mhz, so I am thinking the 120 Mhz would run fine at 133 Mhz?

My Cx100MHz used to run at 150MHz after I had applied 5V. (:
Here you are link:
My Ultimate 486

If I were You I would try overclock your Cx120MHz to 133MHz or even 150MHz.

Regards!

How would you overclock a Cyrix 5x86-120 to 133 Mhz that has a 3x multiplier? I can see using 50 Mhz FSB and a Host-PCI clock division of 2:3 for 150 Mhz. But I doubt the Cyrix 5x86-120 would be stable in Windows with 30 Mhz of extra boost. I will try it on my Biostar MB-8433UUD v3.0 and M919 v3.4 sometime.

The only other way I can see to overclock a 120 Mhz to 133 Mhz w/3x multiplier would be to use an add-on socketed adapter and implement a 1.33x frequency multiplier circuit to the Cyrix's input clock signal. That is, 33.33 Mhz FSB x 1.33 x 3 = 133 Mhz. Has anyone here tried this? Or to externally modify the motherboards FSB's using a similiar multiplier scheme, or to entirely replace the MB's FSB circuit. I'll add this to my list of things to (eventually) investivate if I cannot get ahold of a Cyrix 5x86-133.

I noticed that your Speedsys result for your overclocked Cyrix 5x86 to 150 Mhz showed worse performance than my stock Cyrix 5x86-120 Mhz chip. What cache timings were you using? What about memory wait cycles? I use 2-1-2 on Biostar and 2-1-1-1 on M919, RAM wait states of 0. This configuration is long-term stable in WinNT 4.0 with all available 6x86 nextgen enhancements on (minus BTB). I am also using FPM 60ns (two sticks of 64 MB each).

For speedsys 5x86-120, I got:
L1: 173 MB/s, L2: 68 MB/s, RAM: 48 MB/s

For speedsys 5x86-150, you reported:
L1: 145 MB/s, L2: 63 MB/s, RAM: 44 MB/s

In the other post, you mentioned that you needed to up your CPU voltage to 5.0 V. As opposed to 3.45 V? How did you come to this determination? Did you try 4 V? Did you try booting into Windows and if so how was stability?

udam_u, could you also post what other BIOS settings were required for operation at a 50 Mhz FSB on your Cyrix 5x86-150? I didn't see it in the other post. I briefly attempted this at 5V and 50 Mhz FSB but didn't get past memory check on my M919 w/128MB EDO RAM. It went to about 109,000 KB and froze.

@feipoa

How would you overclock a Cyrix 5x86-120 to 133 Mhz that has a 3x multiplier? I can see using 50 Mhz FSB and a Host-PCI clock division of 2:3 for 150 Mhz.

Hi,
there is very simple way. You have to discover hidden 66MHz fsb setting and set multiplier to 2x. Rg100 found Biostar hidden settings. You can read about this in second post here:
UMC chipset PCI 486 mobo. AMD P90 CPU options?
I'm sure that M919 also supports unofficial FSB settings.

I noticed that your Speedsys result for your overclocked Cyrix 5x86 to 150 Mhz showed worse performance than my stock Cyrix 5x86 […]
Show full quote

I noticed that your Speedsys result for your overclocked Cyrix 5x86 to 150 Mhz showed worse performance than my stock Cyrix 5x86-120 Mhz chip. What cache timings were you using? What about memory wait cycles? I use 2-1-2 on Biostar and 2-1-1-1 on M919, RAM wait states of 0. This configuration is long-term stable in WinNT 4.0 with all available 6x86 nextgen enhancements on (minus BTB). I am also using FPM 60ns (two sticks of 64 MB each).

For speedsys 5x86-120, I got:
L1: 173 MB/s, L2: 68 MB/s, RAM: 48 MB/s

For speedsys 5x86-150, you reported:
L1: 145 MB/s, L2: 63 MB/s, RAM: 44 MB/s

As I remember I used to use 1:1/2 divider, slowest cache timings and dram waitstates set to 1. I haven't remembered other settings (also I didn't use cx enhancements in this test).

Currently my 486 is disassembled because of damaged HDD drive (appeared bad sectors). I'm going to buy CF card for testing purpose - I don't want to annihilate another vintage hard drive.

In the other post, you mentioned that you needed to up your CPU voltage to 5.0 V. As opposed to 3.45 V? How did you come to this determination? Did you try 4 V? Did you try booting into Windows and if so how was stability?

I always start OC experiments by setting bios settings to slowest and voltage to highest. I have efficient copper cooling system, so I don't worry about CPU temperature.

When system behaves stable I start to tune up bios settings (always change only one option at one try). Then I check again how system behaves - if everything is ok, I try to reduce voltage and retest system. I repeat this until system works stable.

I made cx@150MHz test with voltage set to 5V. It didn't want to boot up with lower voltage.
As for stability this system couldn't boot to Windows because HDD was too damaged. Only speedsys, cpuchk and 3dbench were readable from HDD. Those two benchmarks worked correctly (but I know they are not good stress tests) I will retest this system in the future.

I briefly attempted this at 5V and 50 Mhz FSB but didn't get past memory check on my M919 w/128MB EDO RAM. It went to about 109,000 KB and froze.

How do you cool your CPU? You should use solid cooler and reasonable amount of thermal grease when voltage is set to 5V.

I think system froze because your CPU became too hot.
Also for overclocking experiments I recommend to use only one memory module (I used to use one 60ns FPM module).

Regards! (:

I think that UMC PCI 486 mobos have things in common, and these are their ability to use EDO RAM, and to unofficially allow 60 and 66 MHz bus speed. The M919 is no exception. Here are the highest bus speed jumper settings:

Key
oo = open jumper
-- = closed jumper

60 MHz
oo
--
oo

66 MHz
--
oo
--

I'm not sure what 486 chipsets support EDO ram. I know some of the newer UMC ones support EDO, and perhaps some revisions of the SiS one too.

Btw, good posts udam_u and RG100 😉
The only thing I could possibly add is to check if your cache modules are at least 15ns before attempting 66Mhz FSB.

Thanks for the tip. I'll test the 66 Mhz, 2x, 1:1/2 trick out on my MB-8433UUD and M919 using some less-rare cpu first. L2 is all 15 ns.

From what I remember reading some time ago, if the UM8881F chipset has an E, as shown below in the given location of -EYT, then it supports EDO RAM.

UM8881F
9633-EYT
R62706

feipoa wrote:

From what I remember reading some time ago, if the UM8881F chipset has an E, as shown below in the given location of -EYT, then […]
Show full quote

From what I remember reading some time ago, if the UM8881F chipset has an E, as shown below in the given location of -EYT, then it supports EDO RAM.

UM8881F
9633-EYT
R62706

Thanks for the info. Do you know if there are any UMC 486 VLB mobos that support EDO, or was this feature limited to the UMC 486 PCI mobos?

PC Chips M919 has a single VLB slot, 3 PCI slots, and supports EDO. v3.4 has a fairly recent BIOS -- May or June 1996. I use this MB on my testbed.