I believe that all Transmeta CPUs were BGA. Certainly never come across one that wasn't.

feipoa wrote:

Oh wow, I totally forgot about Transmeta. Didn't their CPUs rely on some kind of software decoding compared to the conventional CPU? I sorta remember something like that. I don't own any Transmeta chips. Not sure if they had a PGA package. There's something about a PGA which makes me want to collect the chip.

Yep. Transmeta's cores ran Code Morphing Software, which dynamically translated code streams to an internal VLIW instruction set. If you're interested in the technical details, there's some great articles on RealWorldTech about them.

In reality, it didn't work all that well.

Horizons wrote:

[...]

In reality, it didn't work all that well.

Actually it did. Even with all the delays and unexpected ramping up of speeds by Intel &co, when it hit the market the Crusoe did what it was supposed to, which was deliver the best x86 performance per Watt. The only problem was that at the time no one was interested - and that Transmeta couldn't ramp up performance/speed (due to being stuck on outdated process technology compared to AMD and Intel) to be competitive in the high end. But that had nothing to do with the concepts or even the engineering implementation, which were sound.

Intel clearly took notice, as they copied the design philosophy of maximizing performance per Watt for the Pentium M, even if they chose a very different route to get there. That Pentium M was specifically designed as a Transmeta-killer (and its release was the final nail in their coffin) and is the granddaddy of all subsequent Intel Core CPUs, so you could even argue that the success of the Transmeta design (even if it was a commercial flop) was responsible for pushing Intel towards the most succesful designs in recent history 😉

dionb wrote:

Horizons wrote:

[...]

In reality, it didn't work all that well.

Actually it did. Even with all the delays and unexpected ramping up of speeds by Intel &co, when it hit the market the Crusoe did what it was supposed to, which was deliver the best x86 performance per Watt. The only problem was that at the time no one was interested - and that Transmeta couldn't ramp up performance/speed (due to being stuck on outdated process technology compared to AMD and Intel) to be competitive in the high end. But that had nothing to do with the concepts or even the engineering implementation, which were sound.

Intel clearly took notice, as they copied the design philosophy of maximizing performance per Watt for the Pentium M, even if they chose a very different route to get there. That Pentium M was specifically designed as a Transmeta-killer (and its release was the final nail in their coffin) and is the granddaddy of all subsequent Intel Core CPUs, so you could even argue that the success of the Transmeta design (even if it was a commercial flop) was responsible for pushing Intel towards the most succesful designs in recent history 😉

Dynamic translation-oriented CPUs have consistently produced results that are, at best, inconsistent. Branchy code streams on both Crusoe and Efficeon were not especially fast, although they could do pretty well on streams that were "run this innermost loop a billion times." Sure, they were low power, but that's what happens when you have a core microarchitecture that's basically a DSP.

olivil wrote:

I'm having the hardest of times finding a P2 400MHz in MMC-1, most of those are MMC-2!

https://www.impactcomputers.com/pmf40002001aa.html

There you go. I just searched the part number. PMF40002001AA it's forsale $59.95

😀

what about the nCube?

http://www.cpushack.com/2013/11/01/ncube-and- … the-hypercubes/

Not sure if it's been mentioned before, but the 80386-12 is pretty rare. I can't remember the exact story about these, but apparently the 386 was planned for 16MHz and many of them fell short and were sold as 12MHz parts. I think most of them might have ended up in the original Compaq Deskpro 386, but I think some may have been recalled due to the 32-bit bugs (non sigma CPUs).

...and more interestingly, there was also a 12MHz version of the 386SX. I think these all ended up being soldered into Japanese laptops. They may have been low power versions. Not sure how rare they are, but I've never seen one in the flesh (because I don't crack open ancient laptops). I'm sure most of what existed has already been recycled.

i understand that "rare" definition also applies to some markets

for example: Here in Argentina i have never seen an athlon XP 1500+ and only one time a 1900+ Palomino. Same goes for Overdrives, despite working on IT for the last 21 years, i found only ONE dx4 100 Overdrive that was sold in september 1995.

in retrospective it makes sense, high end / expensive parts are rare in 3rd country's. Its a shame because the chances that i can pull something high end o rare pre-2005 from a computer are getting slim. The last "rare" thing was a Diamond Riva 128 4MB PCI from a sempron 754 clone (weird place to be) last year

Anonymous Coward wrote on 2020-12-26, 04:46:

Not sure if it's been mentioned before, but the 80386-12 is pretty rare. I can't remember the exact story about these, but apparently the 386 was planned for 16MHz and many of them fell short and were sold as 12MHz parts. I think most of them might have ended up in the original Compaq Deskpro 386

As far as I know, the original Compaq Deskpro clocked the 386 processor at 16MHz, and I also failed to find references on google that support the claim that the Deskpro 386 used 12MHz 80386 chips. Do you have a source for the claim that 12MHz parts ended up in the original Compaq Deskpro 386?

On my list is the RapidCAD set (at least for a price that's affordable) found one, though.
And Intel 486DX2-40 in PGA seems pretty unobtainable ..

mkarcher wrote on 2020-12-26, 08:46:

Anonymous Coward wrote on 2020-12-26, 04:46:

Not sure if it's been mentioned before, but the 80386-12 is pretty rare. I can't remember the exact story about these, but apparently the 386 was planned for 16MHz and many of them fell short and were sold as 12MHz parts. I think most of them might have ended up in the original Compaq Deskpro 386

As far as I know, the original Compaq Deskpro clocked the 386 processor at 16MHz, and I also failed to find references on google that support the claim that the Deskpro 386 used 12MHz 80386 chips. Do you have a source for the claim that 12MHz parts ended up in the original Compaq Deskpro 386?

I think I did find references to this in very old issues of Byte or PC Magazine, but CPU-World would be the best place to ask, because they know where at least some of their 80386-12s came from.

*edit* Maybe my memory isn't what it used to be. I might be getting things confused. There was definitely a 12MHz 386, but I will have to dig back through some magazines to find out where they went...

I think there're nowdays even more shades of "rare" definitions considering even most common ones are "rare" too. For example cpu like the Athlon 1000 on Slot is not that easy to find or the K62+@570 and the K63+@550 actually I've never seen these and I doubt I'll ever see them. Only found a K63-400@2,4v and a K62+@550 that already are not that common in the middle of most K6-2@400 or 450Mhz.
I found a 486 Overdrive DX4 100Mhz with write-through cache in a computer fair years ago. It's still nice to have with the heatsink and the voltage regulator already on the top of the cpu even if using it without a fan on it is almost scary.. heatsink temperatures goes up incredibly high.

Well, there's a seller on a popular auction site that has over 14k pieces of K6-2+/570 so I wouldn't think it qualifies as 'rare' anymore 😀

how the f...a seller ends with 14k processors from 20+ years?

quicknick wrote on 2020-12-26, 13:45:

Well, there's a seller on a popular auction site that has over 14k pieces of K6-2+/570 so I wouldn't think it qualifies as 'rare' anymore 😀

Impressive, how these things could happens who knows.. might be a lucky find? Some archived factory oem cpu in some forgotten company room? 😁

Anonymous Coward wrote on 2020-12-26, 13:06:

mkarcher wrote on 2020-12-26, 08:46:

Anonymous Coward wrote on 2020-12-26, 04:46:

Not sure if it's been mentioned before, but the 80386-12 is pretty rare. I can't remember the exact story about these, but apparently the 386 was planned for 16MHz and many of them fell short and were sold as 12MHz parts. I think most of them might have ended up in the original Compaq Deskpro 386

As far as I know, the original Compaq Deskpro clocked the 386 processor at 16MHz, and I also failed to find references on google that support the claim that the Deskpro 386 used 12MHz 80386 chips. Do you have a source for the claim that 12MHz parts ended up in the original Compaq Deskpro 386?

I think I did find references to this in very old issues of Byte or PC Magazine, but CPU-World would be the best place to ask, because they know where at least some of their 80386-12s came from.

I had a 12mhz 386 Turbosport laptop years after it was obsolete. So yes they exist.

One of my friends had a very old “budget” 386-12mhz tower built from shady computer shopper sources, I don’t remember if he was able to run Windows when that came around making it a very expensive albeit fast XT/MSDOS machine.

I would argue these Chips are rare
8086-6
8088-6
8087-6

My 64k 5150 came with an 8088-6 making it an easy overclocker but why did Intel make these when no one made a 6mhz XT?

They were all made before 1982 but no idea what they were for

quicknick wrote on 2020-12-26, 13:45:

Well, there's a seller on a popular auction site that has over 14k pieces of K6-2+/570 so I wouldn't think it qualifies as 'rare' anymore 😀

That's just amazing. It looks like many people are buying 10+ at a time.

I think this goes down as the largest influx of a specific rare old PC component I've seen since I started getting into this stuff. If they really have 14,000 of them, there's simply no way they'll sell out of them before they've met demand and saturated the market. If they were cheaper there would be no reason not to buy one for every compatible ss7 system. But I don't believe there are 14,000 ss7 boards out there anymore.

Unless some crazy person spends a few hundred thousand dollars and buys them all, I doubt these will ever be out of stock under normal circumstances. Until the seller retires.😀

That said: everyone should send low offers if interested in them, to keep the perceived value low. If hundreds sell at the asking price, they'll jack it up and stop taking offers.

386SX wrote on 2020-12-26, 13:32:

I think there're nowdays even more shades of "rare" definitions considering even most common ones are "rare" too. For example cpu like the Athlon 1000 on Slot is not that easy to find or the K62+@570 and the K63+@550 actually I've never seen these and I doubt I'll ever see them. Only found a K63-400@2,4v and a K62+@550 that already are not that common in the middle of most K6-2@400 or 450Mhz.
I found a 486 Overdrive DX4 100Mhz with write-through cache in a computer fair years ago. It's still nice to have with the heatsink and the voltage regulator already on the top of the cpu even if using it without a fan on it is almost scary.. heatsink temperatures goes up incredibly high.

I just saw a K6-III+ 550 on Ebay sell for $460.00! I wonder if the buyer will actually put it in a motherboard and fire it up?

Ozzuneoj wrote on 2020-12-26, 16:22:

quicknick wrote on 2020-12-26, 13:45:

Well, there's a seller on a popular auction site that has over 14k pieces of K6-2+/570 so I wouldn't think it qualifies as 'rare' anymore 😀

That said: everyone should send low offers if interested in them, to keep the perceived value low. If hundreds sell at the asking price, they'll jack it up and stop taking offers.

Wow, the seller upped the price x2! Some months ago, they were selling lots of two at that price. Now they are selling single parts.

Repo Man11 wrote on 2020-12-26, 16:31:

386SX wrote on 2020-12-26, 13:32:

I think there're nowdays even more shades of "rare" definitions considering even most common ones are "rare" too. For example cpu like the Athlon 1000 on Slot is not that easy to find or the K62+@570 and the K63+@550 actually I've never seen these and I doubt I'll ever see them. Only found a K63-400@2,4v and a K62+@550 that already are not that common in the middle of most K6-2@400 or 450Mhz.
I found a 486 Overdrive DX4 100Mhz with write-through cache in a computer fair years ago. It's still nice to have with the heatsink and the voltage regulator already on the top of the cpu even if using it without a fan on it is almost scary.. heatsink temperatures goes up incredibly high.

I just saw a K6-III+ 550 on Ebay sell for $460.00! I wonder if the buyer will actually put it in a motherboard and fire it up?

We tried a group buy for 50 of them on CPU-World. A member saw it and outbid us, then set up a new eBay account to sell them. I advise against buying off that seller until lack of interest pushes the price down.