Anonymous Coward wrote on 2022-01-14, 15:20:

---

Did ST ever get access to the MII? I only seem to remember them doing up to the 6x86L. I think they got dropped after Cyrix was bought by NatSemi.

You could still buy brand new trays of ST PR166 chips in 2005, I was told they were used in random appliances.

Personally have never noticed anything
ST beyond 6x86 PR200, if the exist in MX form they must be rare or regional

Okay, I hooked it back up and did some investigation.

It seems that setting bf2 pin results in no post on this chip?!

I have these dip settings on this freeway design motherboard
5, 6, 7
Which I assume correspond to bf0, bf1, bf2 respectively.

So with that ASSUMPTION in hand,
0 = off 1= on
bf0 bf1 bf2
0 0 0 500gp 347mhz
0 0 1 no post
0 1 0 433gp 298mhz
0 1 1 no post
1 0 0 233gp 199mhz
1 0 1 no post
1 1 0 366gp 249mhz
1 1 1 no post

Tests were done at 100fsb. To reach other gp pr ratings it is assumed adjusting the bus clock is required.

So in summary, it seems that the chip supports 3.5x, 3x, 2.5x, and 2x.

One final remark, the processor is really fast in windows, boots very fast. Very responsive. In many things it acts like it has performance comparable to a equally clocked k63+ Which is impressive. But alas, this chip isn’t stable at 350mhz and at 300 it is a far cry from the performance of a k63+ running at 600mhz.

I can actually play halo on the k63+, but the cyrix, forget it.

I would have investigated BF2 settings at lower bus speeds, 66 or lower if possible, I believe the BF2 settings should be 4x, 4.5x and 5x at least.

BitWrangler wrote on 2022-01-14, 19:01:

I would have investigated BF2 settings at lower bus speeds, 66 or lower if possible, I believe the BF2 settings should be 4x, 4.5x and 5x at least.

That’s a good idea, good catch one sec, lemme test,

66 mhz bus test

0= off 1= on
bf0 bf1 bf2
0 0 0 500gp 347mhz 100fsb
0 0 1 100fsb no post 66fsb no post
0 1 0 433gp 298mhz 100fsb
0 1 1 100 fsb no post 66fsb no post
1 0 0 233gp 199mhz 100fsb
1 0 1 100fsb no post 66fsb 333gp 266mhz
1 1 0 366gp 249mhz 100fsb
1 1 1 100fsb no post 66fsb 333gp 266mhz

So there we go, there is the 4x setting!

Just had a go at looking some of this up, which seems to match with what you've found.

Mii datasheet I looked at had pins y33, x34 and w35 being ckmul0, ckmul1 & reserved. In a later addendum it listed w35 as ckmul2. So same pins as Intel bf0,bf1,bf2. It also gave some model numbers with speed (with the MII400 from the addendum):

1model     mult   bus   CPU
2MII-300GP x3    75MHz 225MHz
3MII-300GP x3.5  66MHz 233MHz
4MII-333GP x2.5 100MHz 250MHz
5MII-350GP x3   100MHz 300MHz
6
7MII-400GP x3    95MHz 285MHz

But no actual table showing ckmul settings and multiplier (found one for other chips in a thread about earlier versions: Subtle differences between Cyrix and IBM 6x86s ). Flicking through some pictures of the MII I can see chips labelled as 2.5x,3x,3.5x and 4x. So at least some versions went higher than 3.5x.

[edit: Ah, found a table here: https://www.pchardwarelinks.com/cpuspeed.htm
that says the highest for the MII was x4 on some versions]

snufkin wrote on 2022-01-14, 19:11:

Just had a go at looking some of this up, which seems to match with what you've found. […]
Show full quote

Just had a go at looking some of this up, which seems to match with what you've found.

Mii datasheet I looked at had pins y33, x34 and w35 being ckmul0, ckmul1 & reserved. In a later addendum it listed w35 as ckmul2. So same pins as Intel bf0,bf1,bf2. It also gave some model numbers with speed (with the MII400 from the addendum):

Copy code to clipboard

1model     mult   bus   CPU
2MII-300GP x3    75MHz 225MHz
3MII-300GP x3.5  66MHz 233MHz
4MII-333GP x2.5 100MHz 250MHz
5MII-350GP x3   100MHz 300MHz
6
7MII-400GP x3    95MHz 285MHz

But no actual table showing ckmul settings and multiplier (found one for other chips in a thread about earlier versions: Subtle differences between Cyrix and IBM 6x86s ). Flicking through some pictures of the MII I can see chips labelled as 2.5x,3x,3.5x and 4x. So at least some versions went higher than 3.5x.

[edit: Ah, found a table here: https://www.pchardwarelinks.com/cpuspeed.htm
that says the highest for the MII was x4 on some versions]

Sweet!

Thx!

Okay: so I suppose I can simplify this further with a condensed table

11= on 0 = off
2Bf0, bf1, bf2
31 0 0 = 2x
41 1 0 = 2.5x
50 1 0 = 3x
60 0 0 = 3.5x
71 0 1/ 1 1 1 = 4x

Gonna see if I can decode the multiplier options for a 2.9v 366 now.

1=on 0=off

(66fsb test actual mhz, not PR)

Bf0, bf1, bf2
0, 0, 0 233mhz 3.5x
1, 0, 0 133mhz 2.0x
0, 1, 0 200mhz 3.0x
1, 1, 0 166mhz 2.5x
0, 0, 1 66.7mhz 1.0x (otherwise known as 8ghz)
1, 0, 1 266mhz 4.0x
0, 1, 1 66.7mhz 1.0x (otherwise known as 8ghz)
1, 1, 1 266mhz 4.0x

Hahaaaaa, post screenies to HWBot 🤣

BitWrangler wrote on 2022-03-12, 04:45:

Hahaaaaa, post screenies to HWBot 🤣

Hehehe. I found that quite amusing 🤣

What would be the other cores that I need to test to find their multis? Is the next one the 3.3v cores or was there a revision between the 366 and the 166? That may have a different multi set?

But then again, those cpus probably use the same settings as the 366 but just drop 4x

Get that 66Mhz to WCPUCLK 😜