News Sealed AMD Athlon XP 2000+ CPU stands the test of time — 22-year-old single-core "Thoroughbred" relic has never been turned on

[Admin]

A user at X has got their hands on an untouched, box-packed, and brand-new AMD Athlon XP 2000+ dating back to 2002.

Sealed AMD Athlon XP 2000+ CPU stands the test of time — 22-year-old single-core "Thoroughbred" relic has never been turned on : Read more

 

[abufrejoval]

These devices were meant to be used!

If they were not, that's a failure!

And of course, that CPU was turned on various times during production: that's needed for QA and binning!

Bringing some kind of virginity cult to IT is just very bad anthropomorphisation and best not done.

And the idiocity of collectors assigning value to things that simply got rare by design or accident does not need reenforcement.

 

[Thunder64]

"It offered a slow base clock of 133 MHz, but even that's over 33% faster than the Pentium 4 2.2 (Northwood)."

This is poorly worded at best. There was no "base clock" back then. The FSB was 133MHz. And it was "double pumped" so 266MT/s.

 

[toffty]

These devices were meant to be used!

If they were not, that's a failure!

And of course, that CPU was turned on various times during production: that's needed for QA and binning!

Bringing some kind of virginity cult to IT is just very bad anthropomorphisation and best not done.

And the idiocity of collectors assigning value to things that simply got rare by design or accident does not need reenforcement.

I think you're taking this way to the extreme.
If your points were in regards to virtual things (bitcoins for instance) I'd agree; but anything physical can be valued and it's up to the market (buyer) to decide it's worth.

 

[baculum]

I never thought I'd do this, but I signed up to whine about how bad this article is.

Lack of SSE2 was never a problem at this time given it was just recently introduced.

There was no concept of 'base' or 'boost' -- these constructs wouldn't be introduced for another two (?) generations.

The socket was called 'Socket A' and no one referred to it by how many pins it had (462).

And as noted by the previous poster, 133 Mhz was the FSB, which is the clock of the chipset and has nothing to do with the clock of the CPU. It's funny because this still does exist, but the author just doesn't seem to understand how computers work.

This article should also be a case-study in why over-reliance on AI makes you look stupid.

And one last thought: 2.2? 2.2 what? This article should be deleted.

 

[chaz_music]

Let's hope it has not suffered from ELDRS radiation damage. Over time, the CMOS threshold voltage goes down until the enhancement CMOS devices start to behave like depletion mode parts. i.e., they are on with 0V on the gate. That 'tis a bummer, especially with power MOSFETs.

Wiki article on radiation hardening with a quick mention on ELDRS:

https://en.wikipedia.org/wiki/Radiation_hardening

 

[HideOut]

I never thought I'd do this, but I signed up to whine about how bad this article is.

Lack of SSE2 was never a problem at this time given it was just recently introduced.

There was no concept of 'base' or 'boost' -- these constructs wouldn't be introduced for another two (?) generations.

The socket was called 'Socket A' and no one referred to it by how many pins it had (462).

And as noted by the previous poster, 133 Mhz was the FSB, which is the clock of the chipset and has nothing to do with the clock of the CPU. It's funny because this still does exist, but the author just doesn't seem to understand how computers work.

This article should also be a case-study in why over-reliance on AI makes you look stupid.

And one last thought: 2.2? 2.2 what? This article should be deleted.

Well I'm glad you signed up. THG "aint what she used to be" for sure. Its mostly manufacturer data posted as a story, with a link that gives them a kickback. ChatGPT stories are seemingly common these days too for those that do not provide a kickback. Its kinda sad

 

[Pemalite]

The Author needs to stop writing articles on what they struggle to comprehend.
The Athlon XP 2000+ had a clock of 1667Mhz verses the Northwoods 2200Mhz.

Not a "base clock" of 133Mhz verses the Northwood 2.2. - That comparison is nonsensical.
The Athlon XP 2000+ had a FSB of 133Mhz that was double-pumped to 266Mhz.
The Pentium 4 Northwood often employed a 100Mhz/133Mhz/200Mhz/266Mhz quad-pumped FSB. (400/533/800/1066Mhz respectively.)

 

[Thunder64]

The Author needs to stop writing articles on what they struggle to comprehend.
The Athlon XP 2000+ had a clock of 1667Mhz verses the Northwoods 2200Mhz.

Not a "base clock" of 133Mhz verses the Northwood 2.2. - That comparison is nonsensical.
The Athlon XP 2000+ had a FSB of 133Mhz that was double-pumped to 266Mhz.
The Pentium 4 Northwood often employed a 100Mhz/133Mhz/200Mhz/266Mhz quad-pumped FSB. (400/533/800/1066Mhz respectively.)

This article is hot garbage. I wonder if the author was even alive when these came out. I had an Athlon XP 2000+ (Palomino) and it kicked ass becuse it had a strong FPU while the P4 depended on SSE2 which wasn't widely used until years later.

Also XP was for "Experience" not "Extended Performance". A quick google search could've told you that. What a crap article.

 

[sfjuocekr]

Base clock means front side bus, that is the clock before the (locked) multiplier.

Today's CPU's all run at 100MHz.

 

[Thunder64]

Base clock means front side bus, that is the clock before the (locked) multiplier.

Today's CPU's all run at 100MHz.

There is no FSB today. Hasn't been for a long time. Base clock might now mean 100MHz, but that is really just for convenience. Base clock has meant the guaranteed minimum a CPU will run at for a long time.

Back then the FSB was how fast the CPU could communicate with everything else. And since the memory controller was off die, that was important.

 

[derekullo]

My first computer that I built myself used an Athlon XP 2200+, 1 gigabyte of ram, Geforce 3 Ti500 and a 160 gigabyte hard drive!

 

[Dntknwitall]

Not one specification is correct about this athlon XP 2000+. It was manufactured on a 0.18 micron node, not 0.13. It was a 1.6 Ghz frequency not 133 mhz, although it did have a bus speed of 266 mhz so that could have been the confusion (133mhz × 2) double data rate and intel was using quad data rate at 100 mhz, so its FSB was 400mhz. I owned a few AMD CPUs from this family the 2000+, 2400+ 2800+ 3000+ and the 3200+. There might have been another but those i know for sure. That was a great time for AMD because they were putting the boots to Intel with those CPUs. I just wish the authors or writers would know what they are talking about when they write these articles, it just destroys facts and ruins the perspective of a time past. Tom's Hardware stop letting your writers use AI and make sure they have knowledge of what they are writing about, this is like the 10th article in a month about older tech that made no sense.

 

[Thunder64]

Not one specification is correct about this athlon XP 2000+. It was manufactured on a 0.18 micron node, not 0.13. It was a 1.6 Ghz frequency not 133 mhz, although it did have a bus speed of 266 mhz so that could have been the confusion (133mhz × 2) double data rate and intel was using quad data rate at 100 mhz, so its FSB was 400mhz. I owned a few AMD CPUs from this family the 2000+, 2400+ 2800+ 3000+ and the 3200+. There might have been another but those i know for sure. That was a great time for AMD because they were putting the boots to Intel with those CPUs. I just wish the authors or writers would know what they are talking about when they write these articles, it just destroys facts and ruins the perspective of a time past. Tom's Hardware stop letting your writers use AI and make sure they have knowledge of what they are writing about, this is like the 10th article in a month about older tech that made no sense.

Palomino was 0.18, Thoroughbred was 0.13. That's probably the one thing the author got right. The 2000+ was made in both. The Palomino core was square in shape, whereas Thoroughbred was rectangular. In the picture it is rectangular. Barton made it even more rectangular. Also it was 1.67GHz. 1.6GHz would be the 1900+.

Palomino

Thoroughbred and Barton

 

[HerbJerb]

Single core “back then”…. Was this written by an intern?

 

[sillysaxon]

Wonder what my BNIB still sealed AMD Duron from 2000 is gonna be worth one day?

 

[iocedmyself]

H....how do you have a job?


Athlon XP had a 133mhz FSB....which was double pumped for 266/MTs
The XP 2000 had a clockspeed of 1667mhz, while the P4 northwood chips ran at 1600, 1800, 2000 and 2200mhz. The naming scheme for the Athlon XP was a designation of comparable performance to Intel's P4s.....meaning that the XP 2000 had effective performance on par with Intel's 2000 mhz chip.

Athlon XP + came in multiple different cores, Palomino, Thoroughbred, Thorton and Barton, Spanning Desktop, mobile and Server designations, with the various flavors operating between 1.25v for server, SFF and mobile chips up to 1.75 on the palomino cores, and TDP's that ranged from 35w on the mobile parts up to 79w on the top tier barton cores.

AMD chips were overclocking beasts, that was the era i got into overclocking, and i had several dozen Athlon XP chips across all 4 core designations that would hit 2.6-3.15ghz on air, while maintaining temps below 55c, more than double their rated clock speed. Comparatively, the P4 laptop that my father had at the time, had the barrel jack for the power connector MELT OFF THE MOTHERBOARD multiple times because it was such a hot running power hungry PoS.

Barton core Athlon XP's were the precursor to Athlon 64, where Pentium 4 was so terrible and power hungry that intel abandoned it completely and went badk to Pentium 3 to use as the basis for their Core 2 chips.

If you're going to use AI, at least fact check what it spits out, or ya know dig through the archives of the site that employs you so you don't look like an ignorant tool.

 

[Dr3ams]

Would love to have kept some of my CPUs originally packed, but I'm a working man and they were too expensive to just collect.

Some used AMD CPUs I've got in my hording drawer:

• AMD K8 Athlon 64 3200+ 2.0 GHz - release date 2005
• AMD K8 Athlon 64 2800+ 1.8 GHz - release date 2003
• AMD Sempron 2200+ 1.5 GHz - release date 2004
• AMD K7 Athlon XP 2400+ 2.0 GHz - release date 2002
• AMD K7 Athlon 850 MHz - release date 2000
• AMD K6-2 300 MHz - relese date 1998

The oldest Intel CPU I still have is an Intel Pentium 133. It was in my system before I bought the K6-2. 👀

 

[mitch074]

H....how do you have a job?


Athlon XP had a 133mhz FSB....which was double pumped for 266/MTs
The XP 2000 had a clockspeed of 1667mhz, while the P4 northwood chips ran at 1600, 1800, 2000 and 2200mhz. The naming scheme for the Athlon XP was a designation of comparable performance to Intel's P4s.....meaning that the XP 2000 had effective performance on par with Intel's 2000 mhz chip.

Athlon XP + came in multiple different cores, Palomino, Thoroughbred, Thorton and Barton, Spanning Desktop, mobile and Server designations, with the various flavors operating between 1.25v for server, SFF and mobile chips up to 1.75 on the palomino cores, and TDP's that ranged from 35w on the mobile parts up to 79w on the top tier barton cores.

AMD chips were overclocking beasts, that was the era i got into overclocking, and i had several dozen Athlon XP chips across all 4 core designations that would hit 2.6-3.15ghz on air, while maintaining temps below 55c, more than double their rated clock speed. Comparatively, the P4 laptop that my father had at the time, had the barrel jack for the power connector MELT OFF THE MOTHERBOARD multiple times because it was such a hot running power hungry PoS.

Barton core Athlon XP's were the precursor to Athlon 64, where Pentium 4 was so terrible and power hungry that intel abandoned it completely and went badk to Pentium 3 to use as the basis for their Core 2 chips.

If you're going to use AI, at least fact check what it spits out, or ya know dig through the archives of the site that employs you so you don't look like an ignorant tool.

All true. I'll add that these chips needed some tinkering to reach their top performance, because they were extremely sensitive to RAM latency - sync'ing the FSB and RAM bus meant an extra 20%, tightening up the timings could bring in another 5%, making sure the PCI bus was properly dispatching data could yield some more percents... Heck, I had a Duron 950 from that time period, it was actually much faster when downclocked because it meant I could raise the FSB from 100 to 133 and lower the multiplier from 9.5 to 7.

 

[YSCCC]

the second gen of CPU I had built on, it was fun back then they just use different jumper pins to seal the multiplier of a chip which is perfectly fine for higher clocks to lower tier one, back in the days using some sort of soldering a jumper wire or some silver paint could gain some significant performance jump and not only choose some settings in the BIOS is part of the fun.

 

[Corporate_goon]

In addition to the many other issues folks have already pointed out, I'll add to this that the assertion that the Athlon XP ran hotter than the Pentium 4 is ridiculous. The Pentium 4 is notoriously one of the hottest CPUs in history - one of the big advantages of the Athlons were that they drew less power and ran cooler than their P4 equivalents.

This entire paragraph is incoherent. "Fine wine"? What are you talking about? Lack of SSE2 crippling compatibility? This never happened!

While the Pentium was almost 2x larger in terms of die size, the Athlon's overly bulky package size shadowed the Pentium since AMD's at-the-time incumbent Socket 462 had a much larger footprint than Intel's Socket 478. Moreover, AMD fine wine wasn't as robust as it is now because the XP 2000+ lacked SSE2 support, which would later cripple compatibility.

I'd like to give the writer of this piece the benefit of the doubt and assume these issues are the result of bad translation, but there are so many factual errors in addition to the weird grammar that I can't help but think this is an AI-written article. Toms should be embarrassed to run something like this.

 

[8086]

The socket was called 'Socket A' and no one referred to it by how many pins it had (462).

I was one of many that would often refer to it as both Socket A and Socket 462 interchangeably. A lot of publications I read at the time would do the same, like Maximum PC.

 

[babbott92]

I never thought I'd do this, but I signed up to whine about how bad this article is.

Lack of SSE2 was never a problem at this time given it was just recently introduced.

There was no concept of 'base' or 'boost' -- these constructs wouldn't be introduced for another two (?) generations.

The socket was called 'Socket A' and no one referred to it by how many pins it had (462).

And as noted by the previous poster, 133 Mhz was the FSB, which is the clock of the chipset and has nothing to do with the clock of the CPU. It's funny because this still does exist, but the author just doesn't seem to understand how computers work.

This article should also be a case-study in why over-reliance on AI makes you look stupid.

And one last thought: 2.2? 2.2 what? This article should be deleted.

I got one of these new in 2005 when I was 13, it was the 2.2Ghz model, and the lack of SSE2 pissed me off 😅 since then I have not purchased an AMD chip, I'm still holding my grudge haha

 

[Misgar]

Some used AMD CPUs I've got in my hording drawer:

I just sorted through some old processors and the closest I can find is one A1200AMS3B and an A1200AMS3C (Socket A, 1200MHz).

Loads of Pentium 4s, Celerons and two 80486SX. I can't find my first 80286 or the 8086 I replaced with an NEC V30. Probably in another hoarding drawer or box. I doubt they'll ever be powered on again.

 

[iocedmyself]

All true. I'll add that these chips needed some tinkering to reach their top performance, because they were extremely sensitive to RAM latency - sync'ing the FSB and RAM bus meant an extra 20%, tightening up the timings could bring in another 5%, making sure the PCI bus was properly dispatching data could yield some more percents... Heck, I had a Duron 950 from that time period, it was actually much faster when downclocked because it meant I could raise the FSB from 100 to 133 and lower the multiplier from 9.5 to 7.

Yup, overclocking in those days was a balancing act between FSB and multipliers. I actually had (and probably still have tucked away somewhere) the OCZ Gamer Extreme 533mhz gold memory kit that ran at CL 1.5 and 1T, that was rated to safely operate at up to 3.5v and required a voltage booster that occupied one DIMM slot and could provide 3.9v to the memory. I managed to get a stable 600mhz on that ram (with the help of those old school "ram coolers" that clipped in over the dimm slots)

As a fun little aside, my uncle was a VP at compaq back in the 90's and early 2000's, had worked at DEC-alpha before, and knew many of the engineers at AMD. He ended up being gifted one of the first socket A 1.2ghz Thunderbird chips (it was number 7 off the line) and he ended up putting it in a PC build for me a couple years later. I knew nothing about computers at the time, and after reinstalling windows i was digging around in the BIOS for something, and changed the FSB from 100...to 166.

Now it booted, and it worked fine until i booted up a game, and then it just shut down after a couple of minutes, but then would boot right back up. I went through that series of events a couple times before giving up on gaming and going back to work....but after a couple hours it shut down again, and refused to turn on, so i had to call my uncle to pick it up and take a look.

He pulled the heatsink off (those terrible aluminum box coolers of the day) and found the chip had gotten so hot, that the metal of the heatsink had changed color significantly. He managed to get it booted into the bios and discovered the chip was operating at 2075mhz. He brought it back to me with a much beefier heatsink, dialed the settings back a little bit and everything continued to work perfectly. Which is what kicked off my interest in overclocking and hardware and facilitated several dozen ebay purchases of various socket A CPU's in pursuit of finding the golden sample that would do 3ghz with air cooling.