debs3759 wrote on 2022-06-05, 20:33:

Sphere478 wrote on 2022-06-05, 15:44:

If that’s a 433gp core like I think it is you can possibly get to 350mhz I wouldn’t try any higher than 2.4v though

All MII with the same CPUID / register values are the same. There is no PR433 core, just cores that are capable of that speed, some of which were marketed as such. You could just as well call them PR300 cores.

(people know it easily by saying that, like you could say k6-3+ core and people would know you mean the 18nm mobile core)

You don’t happen to know the process size or the code name do you? Because I can find only conflicting reports about the cyrix line.

Fun fact btw. As these cores got faster they actually lost performance per clock.

If you ran a 2.2v 433gp core and a 2.9v 300gp both at 300gp speeds, the 2.9v one is faster per clock. Fun fact. But the higher clocks the 2.2v core is capable of more than makes up for it. Making it much faster at max clock vs max clock.

Anonymous Coward wrote on 2022-06-06, 00:59:

The celeron 1400 is multiplier locked, correct? Was it overclocked by increasing the fsb?

Rikintosh wrote on 2022-06-06, 01:18:

As for the multiplier I can't say, ...

Tualatin multi is always locked except perhaps for some ESs

Tetrium wrote on 2022-06-06, 09:53:

Anonymous Coward wrote on 2022-06-06, 00:59:

The celeron 1400 is multiplier locked, correct? Was it overclocked by increasing the fsb?

Rikintosh wrote on 2022-06-06, 01:18:

As for the multiplier I can't say, ...

Tualatin multi is always locked except perhaps for some ESs

If I remember correctly, Cpushack wrote an article on Intel's multipliers.

Nexxen wrote on 2022-06-06, 10:00:

Tetrium wrote on 2022-06-06, 09:53:

Anonymous Coward wrote on 2022-06-06, 00:59:

The celeron 1400 is multiplier locked, correct? Was it overclocked by increasing the fsb?

Rikintosh wrote on 2022-06-06, 01:18:

As for the multiplier I can't say, ...

Tualatin multi is always locked except perhaps for some ESs

If I remember correctly, Cpushack wrote an article on Intel's multipliers.

Trying to google it, nothing. curious if it has any info on the pentium 1/ pentium mmx / tillamook.
got a link?

Sphere478 wrote on 2022-06-05, 15:44:

If that’s a 433gp core like I think it is you can possibly get to 350mhz I wouldn’t try any higher than 2.4v though

Yes, it boots with 350 MHz at stock 2.2V, the bios says MII 500, a nearly 100 MHz overclock! It also boots with 2.0V, but unstable, also with 2.1V.

Excellent!!! Unfortunately with those cores that’s usually about all you’ll get. But do report your max if you feel like pushing the envelope :p

Speaking of rare CPUs, have you met Transmeta? I think the most popular product with this CPU was the HP TC1000 tablet pc.

Here is the wiki about this CPU: https://en.wikipedia.org/wiki/Transmeta_Crusoe

The idea was pretty cool, it had its own instructions, and it wasn't a real x86, but it had a "hardware-level emulator" that translated the x86 instructions to the cpu. Unfortunately it didn't live up to the performance expectations, but it could have been great, a "multiplatform" processor.

Rikintosh wrote on 2022-06-08, 20:30:

Speaking of rare CPUs, have you met Transmeta? I think the most popular product with this CPU was the HP TC1000 tablet pc.

Here is the wiki about this CPU: https://en.wikipedia.org/wiki/Transmeta_Crusoe

The idea was pretty cool, it had its own instructions, and it wasn't a real x86, but it had a "hardware-level emulator" that translated the x86 instructions to the cpu. Unfortunately it didn't live up to the performance expectations, but it could have been great, a "multiplatform" processor.

I have a Transmeta based system somewhere, just haven't seen it in forever. Probably in my outdoor storage space. Gotta get round to finding and testing it someday 😀

Sphere478 wrote on 2022-06-06, 04:06:

(people know it easily by saying that, like you could say k6-3+ core and people would know you mean the 18nm mobile core) […]
Show full quote

debs3759 wrote on 2022-06-05, 20:33:

Sphere478 wrote on 2022-06-05, 15:44:

If that’s a 433gp core like I think it is you can possibly get to 350mhz I wouldn’t try any higher than 2.4v though

All MII with the same CPUID / register values are the same. There is no PR433 core, just cores that are capable of that speed, some of which were marketed as such. You could just as well call them PR300 cores.

(people know it easily by saying that, like you could say k6-3+ core and people would know you mean the 18nm mobile core)

You don’t happen to know the process size or the code name do you? Because I can find only conflicting reports about the cyrix line.

Fun fact btw. As these cores got faster they actually lost performance per clock.

If you ran a 2.2v 433gp core and a 2.9v 300gp both at 300gp speeds, the 2.9v one is faster per clock. Fun fact. But the higher clocks the 2.2v core is capable of more than makes up for it. Making it much faster at max clock vs max clock.

Here is a generic list, there was another site that listed serials VRS layers and process nodes.
Cyrix made the same chips with many layouts, which is not represented by the list

rmay635703 wrote on 2022-06-08, 23:04:

Here is a generic list, there was another site that listed serials VRS layers and process nodes. Cyrix made the same chips with […]
Show full quote

Sphere478 wrote on 2022-06-06, 04:06:

(people know it easily by saying that, like you could say k6-3+ core and people would know you mean the 18nm mobile core) […]
Show full quote

debs3759 wrote on 2022-06-05, 20:33:

All MII with the same CPUID / register values are the same. There is no PR433 core, just cores that are capable of that speed, some of which were marketed as such. You could just as well call them PR300 cores.

(people know it easily by saying that, like you could say k6-3+ core and people would know you mean the 18nm mobile core)

You don’t happen to know the process size or the code name do you? Because I can find only conflicting reports about the cyrix line.

Fun fact btw. As these cores got faster they actually lost performance per clock.

If you ran a 2.2v 433gp core and a 2.9v 300gp both at 300gp speeds, the 2.9v one is faster per clock. Fun fact. But the higher clocks the 2.2v core is capable of more than makes up for it. Making it much faster at max clock vs max clock.

Here is a generic list, there was another site that listed serials VRS layers and process nodes.
Cyrix made the same chips with many layouts, which is not represented by the list

B54EA607-0984-4852-8853-EC2BF6F62F43.png

Yeah see, that’s what I’m saying, those lists are always suspicious. There is no way that all those cores have the same transistor count. Adding mmx alone would change the count.

It may be right about other stuff but I take it with a grain of salt.

And also, I happen to know for a fact that the 266 came in at least two different process nodes. Including what appears to be the 433gp core (i have one)

Yeah, the more I look at that list the more I find flaws with it.

For instance, I also have a 300gp that appears to share the 433gp core and that list calls it a 18nm core but doesn’t list that on the 300gp

I’m honestly not convinced that the 433gp core is 18nm I think it may actually be 25nm

Like I said, none of the info on these passes the sniff test.

Perhaps we should make a thread trying to get to the bottom of this.

Anyone willing to help research and pool data on this?

Scrounging up datasheets may help. And maybe old archived cyrix website notes?

Ultimate Cyrix spec data repository.

I remember there being quite a bit of discussion on the Cyrix MII fab process in Cyrix appreciation thread
This website claims it was 0.18 um, https://www.pcstats.com/articles/1187/index.html

Have you gone through cyrix/via's site on wayback to see what they claim it to be? I looked through the Cyrix MII databook from 1999 but could not find a reference to the fab process. I also looked through the Cyrix MII-433GP addendum, but no reference to fab process.

feipoa wrote on 2022-06-09, 01:29:

I remember there being quite a bit of discussion on the Cyrix MII fab process in Cyrix appreciation thread
This website claims it was 0.18 um, https://www.pcstats.com/articles/1187/index.html

Have you gone through cyrix/via's site on wayback to see what they claim it to be? I looked through the Cyrix MII databook from 1999 but could not find a reference to the fab process. I also looked through the Cyrix MII-433GP addendum, but no reference to fab process.

I wonder if maybe there are some documents released from national semiconductor around that time that tout their capabilities around those times

What I can see seems to say that the 433gp came out around 1998.

The first amd .18 chip was what? 2-3 years later? The k6-3+ and intel didn’t do it until the freaking coppermine?

Yeah, .18 on a 433gp seems off…?

But the process size seems to have its origins around the 1998-2000 time frame..

Could cyrix really have been that far ahead?

And if so, why did that process only yeild 300mhz chips? I guess pipeline length and whatnot can affect that though.

Cyrix had 6 layer cores all the way back in late 1996

https://www.cpushack.com/CyrixID.html

Intel didn’t do that until years later

Cyrix was doing the best with poor quality fabs

Sphere478 wrote on 2022-06-09, 01:45:

What I can see seems to say that the 433gp came out around 1998.

The first amd .18 chip was what? 2-3 years later? The k6-3+ and intel didn’t do it until the freaking coppermine?

Yeah, .18 on a 433gp seems off…?

Anything above pr366 was “paper released” in 1998

Having been a person trying to order one in 1998 there were several important notes.

1. The PR433 had a higher retail cost than an AMD 400 (in the real world)

2. The chinesium place I ordered all my beige custom PCs from listed up through PR500 with a “call” next to anything above PR366

3. In mail order, IBMs price structure capped at PR333 the same time Cyrix capped at a fictional PR433, and if you wanted to order a cheap ST chip I could never find one past pr266. It’s worth noting IBM designed the faster “Cyrix” core forcing National ahead kicking and screaming, so it doesn’t make sense that Cyrix would be more advanced than IBM silver tops at that point

4. If you “forced” the issue to get a faster Cyrix in 1998 through one of the many catalog places your “please wait 3-4 weeks” turned into 3 months .

4a) if a significant number of folks could have bought pr433’s in 1998 Cyrix likely could have likely struggled on a few more years.

Maybe I am wrong but if I had to bet
getting a pr433 In 1998 was like getting an Intel pentium 200 in 1996.

In theory a small number was trickling out but due to the very poor bin ratio they probably all got consumed outside retail channels.

As an example My pr433 was date coded in the year 2000, that is a fair way off from 1998

Tetrium wrote on 2022-06-06, 09:53:

Anonymous Coward wrote on 2022-06-06, 00:59:

The celeron 1400 is multiplier locked, correct? Was it overclocked by increasing the fsb?

Rikintosh wrote on 2022-06-06, 01:18:

As for the multiplier I can't say, ...

Tualatin multi is always locked except perhaps for some ESs

I was just reading this topic and it reminded me that I have a 1.4 Celeron Tualatin Sample (QID2QS)
Is there a chance to be unlocked?
(Picture : https://retro.user-unfriendly.net/Parts ... ample).jpg)
This CPU was given by a friend and I don't have any Tualatin compatible 370 motherboard to test it...

Pentium 1 366mhz The Infamous Golden Tiger/Warrior

I am still waiting for them to arrive, but I made an amazing score of super rare golden pentiums.

Including a 366 model…😳

Not sure if this fits here but I got an ultra low power K6 2+E , runs at 1.5v @ 350Mhz. Total 7watt TDP, not sure how difficult it’ll be to get it running but I do have a few DFI Super Seven boards that can handle the low voltage.

Seen a few of these at higher speeds and voltages but none at this low a speed or voltage.

That's cool, can it work without heatsink in your opinion?
My k6-3+ could run at 200 mhz and v1.3, the whole system consumption was under 30 watt.

Nemo1985 wrote on 2022-11-12, 13:23:

That's cool, can it work without heatsink in your opinion?
My k6-3+ could run at 200 mhz and v1.3, the whole system consumption was under 30 watt.

It’s designed for fanless embedded systems, at a lower speed and voltage I doubt it would need much if any cooling.

It’s worth investigating just how little cooling it requires at a range of low voltages and speeds.

Never had an embedded K6 variant before so I don’t know if AMD changed anything internally regarding power management.

I asked because I tried to run a Tillamook without fan but the heatsink almost too hot after some use, the voltage was surely higher (v1.9), maybe it could work with the k6-3+