I realize that they might still work even though they say for collection only, but he gets himself off the hook with Paypal by not selling his chips as functioning parts but as collectibles only. You can't make a complaint to Paypal to get your money back because you were never buying a functioning chip to start with. If it does just happen to work when you get it, that's great, but you would have no right to complain if it didn't. The people I buy from state in their listings whether the chips work or not so I have a little more leverage for getting refunds or replacements.

Something else to consider when dealing with these guys from China, Taiwan, or Hong Kong, is the wait time for the parts to arrive. The last package I got from China took about 5 weeks to arrive and that was before the terrorists started sending explosives via UPS and FedEx. After that happened, wait times for all packages went up. I ordered a part from a guy in Germany that took 7 weeks to arrive after that. I wasn't waiting more than 10 days for packages from Europe before then. Wait times on packages from China will probably be held for inspection at least as long.

I think you may be reading a little too much into this. I think we know who we are talking about from the eBay titles and the auction description. I don't think his English is that good, which may account for the apparently deceiving auction. But if you call him, his Mandarin is perfect and his manner is very polite. I am pretty sure that he isn't looking for a scam, where he is saved by his clever eBay wording. I think that is purely a manifestation of the translation.

This guy goes out of his way to find what you want and ships same day, even for items that may not yet be posted. I have never had a non-working CPU from him, or the other Chinese vendor mentioned above. If you buy a lot from him, he may even ship EMS Express for free (usually $20), which will only take 10 days max to receive the package (in Canada from Guangzhou). Considering the provence he is in, he may even speak Cantonese as well.

Alright, I recieved my Tillamook from China today.

CPU seems to be working and I get it to Post at 350MHz (4x Multi isn't working right now).
The only problem: It won't leave the bios screen unless I disable the L2 cache. I somewhat expected this, but I was hoping that the Aladdin V chipset would work.

Any suggestions?

350MHz without L2 cache is kinda pointless 🙁

I'll try some other boards/chipsets.

Oh and I can confirm the "Talimook" from the picture I posted earlier. Mine tells me the same thing 😉

What is the maximum CPU speed for which you do not need to disable the L2 cache for the system to boot?

So far not a single setting worked.

With a regular Desktop P55C the system booted fine with 100MHz FSB and L2 cache enabled at 250MHz, so I don't think it's the FSB.

The Tillamook won't leave the bios at 2x100 unless I disable the L2 cache. I'll try lowering the FSB for testing purposes.
I've read somewhere that the cache architecture was slightly modified for the Tillamook, so that might be the reason for it not working.

The CPU itself runs fine, it's a beast. 440MHz (110x4) and it's not even getting handwarm. A higher FSB won't work, something on the mainboard doesn't take it. At that speed it slowly starts to recover from the disabled L2 cache, but the performance increase isn't that much compared to the desktop P55C @250Mhz.

Edit: Even with the FSB all the way down at 66MHz (266MHz default) it won't boot with L2 enabled.

Mystery wrote:

So far not a single setting worked. […]
Show full quote

So far not a single setting worked.

With a regular Desktop P55C the system booted fine with 100MHz FSB and L2 cache enabled at 250MHz, so I don't think it's the FSB.

The Tillamook won't leave the bios at 2x100 unless I disable the L2 cache. I'll try lowering the FSB for testing purposes.
I've read somewhere that the cache architecture was slightly modified for the Tillamook, so that might be the reason for it not working.

The CPU itself runs fine, it's a beast. 440MHz (110x4) and it's not even getting handwarm. A higher FSB won't work, something on the mainboard doesn't take it. At that speed it slowly starts to recover from the disabled L2 cache, but the performance increase isn't that much compared to the desktop P55C @250Mhz.

Edit: Even with the FSB all the way down at 66MHz (266MHz default) it won't boot with L2 enabled.

Damn...

What board are you using?

But otoh good news that it at least posts, it's a good start 😀

Gigabyte GA-5AA with ALI Aladdin V Chipset, Revision G, 128MB Cacheable Area (with L2 cache *g*).
To increase system stability, I've only installed one AGP card and nothing else. Floppy controller and secondary IDE controller are disabled.

The 4x multiplier can't be accessed by the mainboards jumpers, only goes up to 3.5. I made it available by connecting two of the CPU pins, since for some reason intel decided to change some of the pins. A multi above 4 would theoretically work (the CPU cooler doesn't even get warm at 440MHz), but anything higher than 4x is interpreted as 4x (tried 4.5 and 5). Some desktop P55Cs are locked in a similar fashion. You can lower the multi, but can't increase it above the default setting.

I think it's either the L1 cache that's supposed to be larger, or the IO voltage that's REALLY out of specs on a desktop board. (+1V or something like that, but it works and I can even run benchmarks etc.).

Edit: With L2 enabled I can enter the bios without any problems, but the system comes to a complete halt after the HDD detection is finished.

feipoa wrote:

I realize that they might still work even though they say for collection only, but he gets himself off the hook with Paypal by not selling his chips as functioning parts but as collectibles only. You can't make a complaint to Paypal to get your money back because you were never buying a functioning chip to start with. If it does just happen to work when you get it, that's great, but you would have no right to complain if it didn't. The people I buy from state in their listings whether the chips work or not so I have a little more leverage for getting refunds or replacements.

Something else to consider when dealing with these guys from China, Taiwan, or Hong Kong, is the wait time for the parts to arrive. The last package I got from China took about 5 weeks to arrive and that was before the terrorists started sending explosives via UPS and FedEx. After that happened, wait times for all packages went up. I ordered a part from a guy in Germany that took 7 weeks to arrive after that. I wasn't waiting more than 10 days for packages from Europe before then. Wait times on packages from China will probably be held for inspection at least as long.

I think you may be reading a little too much into this. I think we know who we are talking about from the eBay titles and the auction description. I don't think his English is that good, which may account for the apparently deceiving auction. But if you call him, his Mandarin is perfect and his manner is very polite. I am pretty sure that he isn't looking for a scam, where he is saved by his clever eBay wording. I think that is purely a manifestation of the translation.

This guy goes out of his way to find what you want and ships same day, even for items that may not yet be posted. I have never had a non-working CPU from him, or the other Chinese vendor mentioned above. If you buy a lot from him, he may even ship EMS Express for free (usually $20), which will only take 10 days max to receive the package (in Canada from Guangzhou). Considering the provence he is in, he may even speak Cantonese as well.

What you say may very well be true. He might be the most honest businessman in all of China. I'm just saying that by having that line about his parts being collectibles, it does put him in a stronger position in the event of a dispute. I'd just rather not be put in the position of having to test his honesty if I buy something and it doesn't work, that's all.

sliderider wrote:

What you say may very well be true. He might be the most honest businessman in all of China. I'm just saying that by having that line about his parts being collectibles, it does put him in a stronger position in the event of a dispute. I'd just rather not be put in the position of having to test his honesty if I buy something and it doesn't work, that's all.

Yes, but when it comes to buying second hand, theres always risks involved.
For me what counted was that his feedback was good, his prices were low (for the parts I bought) and he's on cpu-world so I figured he'd at least know how to pack and ship CPU's correctly. (correctly as in, not packing it in STATIC plastic or something amazingly stupid like that).

He also posts images of hard-to-find CPUs on chipdb.org which haven't yet been posted.

@Mystery
Try this CPU in a motherboard with a different chipset and see how it goes.

It's supposed to work with older (intel) Socket5 chipsets, but that kinda defeats the purpose of using this CPU, because you can't use its potential at all (due to low fsb and lockes multi).

What So7 chipsets support 100MHz FSB?
Via MVP3
Via MVP4
ALI Aladdin V

i430FX and i430HX don't support L2 cache either or disable it automatically. Aladdin 4 disables the L2 cache after posting.

I have a 266 MHz version on its way, so I'll test it out when it comes in.

I have a Socket7 MB that works well with K6-III+/500 and Cyrix-433GP CPUs and has an Intel TX chipset, so this is what I'll test it on.

If CLKMUL is locked at 4x on these Tillimock's, then the fastest I'll be able to get it is 333 MHz. The board also has the full range of voltage regulation, 2.1, 2.5, 2.8, 2.9, 3.2, 3.3, 3.5 Volts.

Mystery wrote:

What So7 chipsets support 100MHz FSB? Via MVP3 Via MVP4 ALI Aladdin V […]
Show full quote

What So7 chipsets support 100MHz FSB?
Via MVP3
Via MVP4
ALI Aladdin V

Theres also SiS530, which can be clocked at 133Mhz FSB (but often only with it's L2 cache disabled and it has no AGP slot) 😉
Edit:Or I should say, it has a 1:4 PCI divider, so it's possible to run the board at 133Mhz FSB without having to overclock any of it's busses

Just found this in the Tillamook specs from intel:

Signals Removed from the Low-Power Embedded Pentium® Processor with MMX™ Technology Signal Function ADSC# Additional Address Sta […]
Show full quote

Signals Removed from the Low-Power Embedded Pentium® Processor with MMX™ Technology
Signal Function
ADSC#
Additional Address Status. This signal is mainly used for large or standalone L2
cache memory subsystem support required for high-performance desktop or
server models.
BRDYC#
Additional Burst Ready. This signal is mainly used for large or standalone L2
cache memory subsystem support required for high-performance desktop or
server models.

Might that be related to our problem?
If it is, why does it only affect newer chipsets, but not the older ones on many Socket5 boards?

Mystery wrote:

Just found this in the Tillamook specs from intel: […]
Show full quote

Just found this in the Tillamook specs from intel:

Signals Removed from the Low-Power Embedded Pentium® Processor with MMX™ Technology Signal Function ADSC# Additional Address Sta […]
Show full quote

Signals Removed from the Low-Power Embedded Pentium® Processor with MMX™ Technology
Signal Function
ADSC#
Additional Address Status. This signal is mainly used for large or standalone L2
cache memory subsystem support required for high-performance desktop or
server models.
BRDYC#
Additional Burst Ready. This signal is mainly used for large or standalone L2
cache memory subsystem support required for high-performance desktop or
server models.

Might that be related to our problem?
If it is, why does it only affect newer chipsets, but not the older ones on many Socket5 boards?

Interesting. I think you may have hit something there.
This could indicate Tillamook "could" function on a HX board with 512MB of it's memory cached.
All you'd need is a voltage regulator module for the Tillamook to sit in.
Bad news would be, it could be very hard to make it work with high end desktop chipsets

Edit:Could you post the link where you found this please?
Edit2:Nvm, found it. It's here: http://www.intel.com/design/intarch/datashts/27318403.pdf

Edit3:After having read that document (well, the relevant bits) and the one for the desktop MMX chip (just skip to "Table 2. Quick Pin Reference") it appears that the 2 blocked pins were doubles, which themselves have not been blocked?

This is interesting and could mean several things. One of those is that the Tillamook (-->and I'm merely speculating here<--) won't boot when there is more then 64MB of memory cached or something?

If this is the case and it's non-circumventable for us, it would mean that either we use it with only 64megs of memory, or we use more memory but accept that not all of it is cached.

Made a new post instead of another edit.

I found this in the Pentium MMX (desktop) datasheet:

ADS# O The address strobe indicates that a new valid bus cycle is currently being driven
by the Pentium processor with MMX technology.
ADSC# O The address strobe (copy) is functionally identical to ADS#.

and

BRDY# I The burst ready input indicates that the external system has presented valid data
on the data pins in response to a read or that the external system has accepted
the Pentium processor with MMX technology data in response to a write request.
This signal is sampled in the T2, T12 and T2P bus states.
BRDYC# I The burst ready (copy) is functionally identical to BRDY#.

Think it's possible that the way Super 7 is implemented changed something with these pins, causing Tillamook to not function correctly anymore?

Edit:If linking the 2 pins together, would this make Tillamook work again?

Edit2:I checked the location of the pins.
ADS# AJ05
ADSC# AM02

BRDY# X04
BRDYC# Y03

BRDY# and BRDYC# are right next to eachother while ADS# and ADSC# are in the same diagonal row of pins with 2 pins in between them.

Now I wonder something else: It appears that on the top of the PPGA package are little solder dots at the exact location where on the bottom are pins. Would this mean that connecting the solder dots on the top will do the same thing as when shorting the 2 pins on the bottom together?

Yes you can connect the pins on top of the CPU or at the bottom.
Some people put single strands of copper wire into the cpu socket holes and you could also solder a wire to the bottom of the mainboard, where the CPU socket is soldered to. Doesn't make a difference.

The 4x multiplier for the Tillamook is unlocked exactly the same way.

If those 2 duo's of pins are connected, think it will make Tillamook work correctly again?
I'm tempted to try out myself but am reluctant to go through all the mess in my attic right now to set up a test bench 🙁

And Mystery, great find!!

Edit:I'm thinking of putting 2 pieces of papertape (and hope it's not electrically conductive enough to mess things up) and shorting the 2 pairs of pins.

For tests that are only temporally, I usually use the copper wire method. You need a good eye and a steady hand, but it works.

There are also conductive pens that should make this pretty easy.

I have no idea if it'll work, but it's worth a try. I might give it a shot tomorrow. Not sure if I find the time, but I'll try to squeeze it in.
This would be really great, but I'm not getting my hopes up yet.

And it's only a good solution if it works with "modern" chipsets that have a high FSB and for optimal results support SDRAM and AGP, like MVP3 or AladdinV, to get the best performance out of the Tillamook 😀

Mystery wrote:

For tests that are only temporally, I usually use the copper wire method. You need a good eye and a steady hand, but it works. […]
Show full quote

For tests that are only temporally, I usually use the copper wire method. You need a good eye and a steady hand, but it works.

There are also conductive pens that should make this pretty easy.

I have no idea if it'll work, but it's worth a try. I might give it a shot tomorrow. Not sure if I find the time, but I'll try to squeeze it in.
This would be really great, but I'm not getting my hopes up yet.

And it's only a good solution if it works with "modern" chipsets that have a high FSB and for optimal results support SDRAM and AGP, like MVP3 or AladdinV, to get the best performance out of the Tillamook 😀

My idea with the paper tape was to first put the paper tape onto all of it except where it has 2 little holes where the solder dots are.
Then you can use conductive paint (where did I put my little bottle btw?) and draw a line over the paper tape.
Want the short gone? Just pull the paper tape and all that remains is the conductive paint on the solder dots 😉

At least...that's the basic idea

Tetrium wrote:

My idea with the paper tape was to first put the paper tape onto all of it except where it has 2 little holes where the solder d […]
Show full quote

Mystery wrote:

For tests that are only temporally, I usually use the copper wire method. You need a good eye and a steady hand, but it works. […]
Show full quote

For tests that are only temporally, I usually use the copper wire method. You need a good eye and a steady hand, but it works.

There are also conductive pens that should make this pretty easy.

I have no idea if it'll work, but it's worth a try. I might give it a shot tomorrow. Not sure if I find the time, but I'll try to squeeze it in.
This would be really great, but I'm not getting my hopes up yet.

And it's only a good solution if it works with "modern" chipsets that have a high FSB and for optimal results support SDRAM and AGP, like MVP3 or AladdinV, to get the best performance out of the Tillamook 😀

My idea with the paper tape was to first put the paper tape onto all of it except where it has 2 little holes where the solder dots are.
Then you can use conductive paint (where did I put my little bottle btw?) and draw a line over the paper tape.
Want the short gone? Just pull the paper tape and all that remains is the conductive paint on the solder dots 😉

At least...that's the basic idea

How about just using foil tape then? It's not cheap though, and I suppose the adhesive might hamper it's electrical conductivity, but I guarantee it's still pretty dang conductive.