Re: Pentium 1 366mhz The Infamous Golden Tiger

I’m gonna give some advice here about certain types of tillamook packages.

The golden tigers and the interposer film chip types should probably be avoided. Except purely as a unused collection piece.

The ppga ones are the ones you guys want. (Look like plastic pentium mmx)

Meanwhile in the world of working combinations... 😀
Scored SL2Z4 Tillamook for a whopping 5 euros on local auctrion site and finally got a chance to try the mods listed in the thread, 15 minutes with soldering iron and put the CPU into my PC Partner MVP3BS7 that I got from the same auction site a year or so ago for also whopping 5 euros and voilaa! It works!
It needed 2.2 volts to be 100% stable at 4x100 MHz but I am glad that everything works, cache shows 512kb and it is quite fast.
I threw in Asus Geforce2 MX400 64MB and just for fun installed NFS Hot Pursuit 2 😁 It "ran" or to be more precise walked at best. All settings at minimum @640x480 it made ~10-15fps, sometimes dropping below 10 at heavy areas. But it worked and that was very good test for stability, gave proper workout 😀 I obviously uninstalled the game afterwards, at least now can say that this CPU is quite fast!

"23 to county, he's jerking up the road like a glitchy teleporter"

Heh, first hot pursuit should be nice on that though. I put HP2 on my PII366/Rage Mobility compaq and it was a mess of broken graphics and struggle.

Paadam wrote on 2023-01-22, 20:55:

Meanwhile in the world of working combinations... :) Scored SL2Z4 Tillamook for a whopping 5 euros on local auctrion site and fi […]
Show full quote

Meanwhile in the world of working combinations... 😀
Scored SL2Z4 Tillamook for a whopping 5 euros on local auctrion site and finally got a chance to try the mods listed in the thread, 15 minutes with soldering iron and put the CPU into my PC Partner MVP3BS7 that I got from the same auction site a year or so ago for also whopping 5 euros and voilaa! It works!
It needed 2.2 volts to be 100% stable at 4x100 MHz but I am glad that everything works, cache shows 512kb and it is quite fast.
I threw in Asus Geforce2 MX400 64MB and just for fun installed NFS Hot Pursuit 2 😁 It "ran" or to be more precise walked at best. All settings at minimum @640x480 it made ~10-15fps, sometimes dropping below 10 at heavy areas. But it worked and that was very good test for stability, gave proper workout 😀 I obviously uninstalled the game afterwards, at least now can say that this CPU is quite fast!

Try a radeon 9200/9250/7500 they usually work pretty good on these builds and are pretty zippy.

The gf2 isn’t a horrible card by any means though. I just kinda wonder if a radeon might play that game better. It might not. Curious.

This card plays it very well, it is totally CPU bound. With PII 400 it played smoothly at 1024x768/medium settings.

Paadam wrote on 2023-01-23, 04:06:

This card plays it very well, it is totally CPU bound. With PII 400 it played smoothly at 1024x768/medium settings.

Very well 😀

Are you going to try higher fsb?

Not likely, I am happy with it. Going any higher will not make any miracles happen and I am satisfied to have Pentium being on par with K6-2 CPU's.

Nemo1985 wrote on 2019-01-23, 12:09:

Just for reference, the tillamook doesn't work on Epox EP-MVP3G5 (and probably EP-MVP3G-M or EP-MVP3G2), with the latest bios (vp3c0c21), but it works on Epox EP-51MVP3E-M with latest bios (vp300411).

I have three EP-51MVP3E-M, two are defective. One of the two, when it worked, used to detect and label an unmodified Tillamook as "Tillamook" (I never went further).
On the one that is still working now, nothing happens if I try to use a Tillamook. I tried to swap bios chips with the one that used to recognize it but still no display (fans are spinning). The board works with a K6-2 no matter which bios chip. Lattest bios in both cases (one of them is the modified version from Jan Steunebrink).
The only differences are :
the one that recognized the CPU is a REV0.3, the currently working one is REV1.0.
Cache chips are not the same : UMC branded ones for the REV0.3, EliteMT for the REV1.0. But I disabled it in the Bios anyways.
Expansion cards and RAM but I don't see why it would matter.
I will test my Tillamook on another board some day, but it's unlikely to be the cause.

Which exactly is your MVP3E ?
Have you tested only with a modded CPU ?

Paadam wrote on 2023-01-22, 20:55:

Meanwhile in the world of working combinations... :) Scored SL2Z4 Tillamook for a whopping 5 euros on local auctrion site and fi […]
Show full quote

Meanwhile in the world of working combinations... 😀
Scored SL2Z4 Tillamook for a whopping 5 euros on local auctrion site and finally got a chance to try the mods listed in the thread, 15 minutes with soldering iron and put the CPU into my PC Partner MVP3BS7 that I got from the same auction site a year or so ago for also whopping 5 euros and voilaa! It works!
It needed 2.2 volts to be 100% stable at 4x100 MHz but I am glad that everything works, cache shows 512kb and it is quite fast.
I threw in Asus Geforce2 MX400 64MB and just for fun installed NFS Hot Pursuit 2 😁 It "ran" or to be more precise walked at best. All settings at minimum @640x480 it made ~10-15fps, sometimes dropping below 10 at heavy areas. But it worked and that was very good test for stability, gave proper workout 😀 I obviously uninstalled the game afterwards, at least now can say that this CPU is quite fast!

What is this fan used to cool your processor?

Cheers,

Hi, after fighting with some issues I managed to reuse old de-soldered mobile Tillamook in TCP320 package (I have only this one).
described in separate thread: Interposer for mobile Pentium MMX Tillamook

Now I'm trying it to run in some S7 MBs I have handy:
* PC-Chips M590 - this MB needs VCC2DET bridge otherwise onboard VRM keeps shutdown and MB will not even POST. This is because CPU voltage autodetection - this MB can set Vcore only via SETUP menu, no jumpers. L2 cache works on this MB without bridges. Problem is that this MB use very high VIO - I measured 3,86V instead 3,3V that is too high for Tillamook. I think they did it to be able to overclock the crappy chipset to boast supporting 100MHz FSB (without letting user know they overvolted his CPU and RAM heavily to do this). I'm planning to replace onboard VIO regulator (some strange circuit using power N-FET and small BJT?) with normal 3,3V LDO. Problem is that FET and LDO, both in TO-252 cases, have swapped in and out pins so I cannot reuse the copper plane for cooling, I don't know what current the VIO consumes for proper dimensioning the LDO...

* Gigabyte GA-5AA rev 2.2 - this MB needs cache bridges mod otherwise I got blank screen after POST/before boot until I disable L2 in SETUP. After the mod L2 is working but I have a problem with L1 - SETUP list this option as "Reserved" and while I can change to "enabled" or "disabled" it doesn't change after save settings and L1 don't work. This degrades performance to some 486 level. Is there some DOS tool that can enable L1? I remember people used some utils to disable L1 to decrease system power for some old games...
BTW this MB use VIO set to 3,5V and maybe there's normal regulator with resistor feedback network that I could tune, didn't investigated yet until L1 problem will be solved. Any newer unofficial BIOS this MB? I read here someone succeeded with GA-5AX...

I found and flashed latest Gigabyte beta BIOS F7. Now my CPU is detected as "Talimook" (fortunately not Taliban 😀 and both L1 and L2 caches now works. I found another weird behavior that CPU acts like it has locked multiplier to 4x. Thats not make sense, it's 233MHz part and so it should be 3,5 x 66MHz. I tried all 8 jumper combinations for multiplier but it always run at 4 x 66MHz = 266MHz. I managed to overclock it stable at 4 x 83 = 333MHz but cannot run it stable at 95 or 100MHz even when I pushed Vcore to 2,8V and I don't want push more. It simply didn't even finish the POST or restarts during POST. At 83MHz FSB I can go stable with 2,2V. Strange that for not so big freq. step is not enough such voltage boost. And the MB itself should support 140MHz FSB so 100MHz should be piece of cake (with PC133 SDRAM).
BTW the used VIO regulator is 7,5A LDO by Linfinity LX8383A so VIO can be easily tuned via resistor feedback network.

RayeR wrote on 2023-11-21, 06:27:

I found and flashed latest Gigabyte beta BIOS F7. Now my CPU is detected as "Talimook" (fortunately not Taliban 😀 and both L1 and L2 caches now works. I found another weird behavior that CPU acts like it has locked multiplier to 4x. Thats not make sense, it's 233MHz part and so it should be 3,5 x 66MHz. I tried all 8 jumper combinations for multiplier but it always run at 4 x 66MHz = 266MHz. I managed to overclock it stable at 4 x 83 = 333MHz but cannot run it stable at 95 or 100MHz even when I pushed Vcore to 2,8V and I don't want push more. It simply didn't even finish the POST or restarts during POST. At 83MHz FSB I can go stable with 2,2V. Strange that for not so big freq. step is not enough such voltage boost. And the MB itself should support 140MHz FSB so 100MHz should be piece of cake (with PC133 SDRAM).
BTW the used VIO regulator is 7,5A LDO by Linfinity LX8383A so VIO can be easily tuned via resistor feedback network.

Your problem is with BF2 (intel, not bf2 amd, different pin) it is probably being pulled high.

I have BF2 wired to a solder jumper on my interposer and currently I configured it at HIGH (VCC3). This is correct for higher multipliers 4x and above, not? In the datasheet the highest described mul is 4x as BF2:0 = 1 0 0. I belived that further combinations like 1 0 1, 1 1 0, 1 1 1 gives me higher multiplier but for some reason it's capped at 4x. So it means that CPU treats this invalid combinations as 4x. If I set BF2 to 0, then I would probably get lower multis but this is not that I want. I don't remember how older Pentiums behaved but I though they didn't have capped multiplier but also they didn't overclock much so it didn't make sense anyway.
Well, so if a real 300MHz version exists and use mul 4,5x then it would probably use combination 1 0 1 as valid and maybe even higher - anybody has this CPU and tried? Intel could individually use LASER trimming to enable/disable max valid mul after chips binning.

RayeR wrote on 2023-11-23, 05:46:

I have BF2 wired to a solder jumper on my interposer and currently I configured it at HIGH (VCC3). This is correct for higher multipliers 4x and above, not?

Multipliers switching mechanism for Mobile Pentium MMX BF0/1/2: 243571-002.pdf page 17 and 18. BF1 uses a different wiring compared to BF0/2.

RayeR wrote on 2023-11-23, 05:46:

I belived that further combinations like 1 0 1, 1 1 0, 1 1 1 gives me higher multiplier but for some reason it's capped at 4x. So it means that CPU treats this invalid combinations as 4x. If I set BF2 to 0, then I would probably get lower multis but this is not that I want. I don't remember how older Pentiums behaved but I though they didn't have capped multiplier but also they didn't overclock much so it didn't make sense anyway.

From my brief testing some weeks ago: 4x was the maximum on the Tillamook 266 SL2Z4.
For normal desktop Pentium MMX multipliers, see tests at the bottom of this thread, page 1:
Pentium MMX Multipliers
But there I did not consider or toggle the later intel BF2 pin, which is not the same as the AMD/Cyrix BF2 pin/jumper.

RayeR wrote on 2023-11-23, 05:46:

I have BF2 wired to a solder jumper on my interposer and currently I configured it at HIGH (VCC3). This is correct for higher multipliers 4x and above, not? In the datasheet the highest described mul is 4x as BF2:0 = 1 0 0. I belived that further combinations like 1 0 1, 1 1 0, 1 1 1 gives me higher multiplier but for some reason it's capped at 4x. So it means that CPU treats this invalid combinations as 4x. If I set BF2 to 0, then I would probably get lower multis but this is not that I want. I don't remember how older Pentiums behaved but I though they didn't have capped multiplier but also they didn't overclock much so it didn't make sense anyway.
Well, so if a real 300MHz version exists and use mul 4,5x then it would probably use combination 1 0 1 as valid and maybe even higher - anybody has this CPU and tried? Intel could individually use LASER trimming to enable/disable max valid mul after chips binning.

If I recall correctly, if you pull it high it will enable 4x and may even lock it to 4x

Floating or pull down will enable 2x, 2.5x, 3x, 3.5x

If I recall from my experiments

———————————

I have found examples of p55 (and maybe p54?) who’s multis were locked and even scrambled.

P54 the max is 3x (can’t recall about locking but there were two cores, the prior didn’t go all the way to 3x)
P55 the max is 3.5x locks and scrambles have been observed
Tillamook the max seems to be 4x as the existence of 4.5 I have not as of yet been able to observe.

This is all from memory hopefully I remembered correctly feel free to correct.

There is no doubt that a 300mhz tillamook exists from intel, but I haven’t been able to observe a 4.5x in action or get anyone to post info from one, let alone find any verified info about its fsb/multi I kinda wonder if it is a 75mhz fsb chip?

Okay, we seem to actually have gotten confirmation of 4.5x today

Re: Interposer for mobile Pentium MMX Tillamook

I didn't know that the 300mhz version was rare enough that people didn't know whether it was a 4.5x66 or a 4x75. I own a 4.5x66 Vaio.

Yeah, I’ve been trying to get my hands on a functioning interposed specimen.

So you can independently verify that indeed it does have a 4.5x multi?

Can you by chance probe the BF pins configuration? Or post a high res illuminated shot of it so that we can see traces behind the film. Perhaps light filters will help transparency

It's in my parents' place, I'll be able to check on that in late december.

It doesn't have any OS currently because I lost its external CD-ROM reader power supply, but I'll try to install DOS and check with System Speed Test (Speedsys).

I'm new to DOS and I never installed it so I'll do my best.

I’m wondering if maybe a separate pin than we are currently aware of was used. This would explain how we were unable to set it on 266,233

Has anyone noted a tillamook datasheet that specifically notes a 300mhz part, and how it’s configured?

A continuity check of every cpu pin to gnd and vcc3 should reveal the configuration if something interesting is to be found. No power is required. Removal of all power is advised.

I must warn though, these pins are impossibly small.

If too much of a task, knowing config of all three bf would be nice at least