VOGONS

Common searches


Reply 921 of 1005, by midicollector

User metadata
Rank Member
Rank
Member

Apologies for the noob question here, but why are these AMD K5 CPUs going for so much? For a second, I th0ught the processor market had gone crazy, but I did some searching on Ebay and 200mhz, 100mhz and 486 AMD CPUs are all going for around $20 and under.

Reply 922 of 1005, by ThinkpadIL

User metadata
Rank Oldbie
Rank
Oldbie
midicollector wrote on 2023-09-07, 01:33:

Apologies for the noob question here, but why are these AMD K5 CPUs going for so much? ...

That's simple - it's because buyers are ready to pay so much. 🙂

Reply 924 of 1005, by bestemor

User metadata
Rank Oldbie
Rank
Oldbie
midicollector wrote on 2023-09-07, 06:18:

I mean, is there anything special or unique about that model that makes people want it? Like, is there something that makes that model desirable or high priced?

Re: Market value for popular or rare classic computer hardware

Reply 925 of 1005, by acl

User metadata
Rank Oldbie
Rank
Oldbie
midicollector wrote on 2023-09-07, 06:18:

I mean, is there anything special or unique about that model that makes people want it? Like, is there something that makes that model desirable or high priced?

They were more advanced than Pentium internally. Closer to Pentium Pro.

  • Speculative execution : make statistics (really basic ones) on code branching. For example : "the last 3 times i evaluated this condition, it was "true" so i will assume this will be true again". This allow the CPU to fill the pipeline with the code of the "true" branch. Without speculative execution, the CPU would have to wait until the condition is executed to know what to execute after. This helps to fill the pipeline efficiently. And if the prediction is wrong, then you just drop what's inside the pipeline and load the correct branch.
  • Register renaming : in previous CPUs, registers like eip / eax / bx etc... are fixed. They are stored in a predined physical small memory. With register renaming, registers are allocated dynamically on chunk of registers memory in the CPU. That means that you can allocate two or more instance of the same register. This allow out of order execution. And to evaluate simultaneously two instructions that uses the same register for different purpose. The two instructions will have their own instance of the same eax register for instance.
  • Out of order execution : evaluate code in a different order from the regular program instruction flow. Again, as an optimisation to fill the pipeline more efficiently.

(for the experts : sorry if these descriptions contains aproximative statements, English does not help 😁)

So it's a technically interesting CPU.
The K5 PR200 is also genuinely rare because it has been produced in a small number.

"Hello, my friend. Stay awhile and listen..."
My collection (not up to date)

Reply 926 of 1005, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++

Living through the times, it seemed to me that the K5 was quite a way down the list in socket 5/7 CPU commonality, with Intel and Cyrix/IBM being more available. Even during it's shorter than competing CPUs production runs, it still felt about 3 times harder to find. More common than WinChips still though, but they were a year later or so.

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 927 of 1005, by pentiumspeed

User metadata
Rank l33t
Rank
l33t

Keep in mind early K5 had several features disabled. Later one K5's had them redesigned which allowed AMD to r enable them.

Cheers,

Great Northern aka Canada.

Reply 928 of 1005, by Unknown_K

User metadata
Rank Oldbie
Rank
Oldbie

K5's were the budget Pentium class chips and I purchased one for a spare motherboard I had to build a system so my friend could play C&C with me in the spare room.

Most of the oddball K5, Winchip, 486sx/2, 3rd party 486 chips I have were found at a recycler that parted out very low budget retail systems and no name laptops. When the major brands would not touch the left over stock they ended up in Europe and Asia. I think my K5 was the PR133 model with the nice gold top that unfortunately got scratched up badly.

Collector of old computers, hardware, and software

Reply 929 of 1005, by mkarcher

User metadata
Rank l33t
Rank
l33t
acl wrote on 2023-09-07, 07:53:
  • Speculative execution : make statistics (really basic ones) on code branching. For example : "the last 3 times i evaluated this condition, it was "true" so i will assume this will be true again". This allow the CPU to fill the pipeline with the code of the "true" branch. Without speculative execution, the CPU would have to wait until the condition is executed to know what to execute after. This helps to fill the pipeline efficiently. And if the prediction is wrong, then you just drop what's inside the pipeline and load the correct branch.

(for the experts : sorry if these descriptions contains aproximative statements, English does not help 😁)

Your explanation is quite good, but it doesn't describe "speculative execution", it describes "branch prediction".

The idea of speculative execution is to go one step further, and not only fill the pipeline, but also already execute those instructions, but do it in a way that the effects of the instructions can be undone/discarded if the prediction turned out false. Register Renaming (another feature you listed) helps a lot for speculative execution, as this just means that you need to keep the last confirmed value of a logical register in a different hardware register than the current value with speculative execution. If the prediction turned out to be right, your redeclare the "speculative current register" as "confirmed register". If the prediction turned out to be wrong, you just reset the register view to the registers that contain the last non-speculative values.

Reply 931 of 1005, by Mandrew

User metadata
Rank Member
Rank
Member

You guys forgot the simplest reason: building a collection.
These people don't give af about K5 features and performance, they just want every single production model behind a display case.
They never sell either because processors need very little space. So what's left? Stuff found by recyclers later.
Now onto the insane price issue: would you buy a K5-200 for your collection for top $$$ if you had every single production model already but not that one?
Let's say your monthly budget for retro is $100 but you see a K5-200 for $500. Would you skip it because $500 is way too much for something like that or would you buy it to complete your collection and not buy anything else in the next 5 months? You could risk waiting for something cheaper but it's like waiting for a miracle when it comes to rare models, highly unlikely.
Starting a collection is easy but completing it is hard, especially in countries where supply is limited to Ebay offerings from other countries. Also $500 is a lot for one guy but not that much for another so sometimes people are just out of their league.
And don't forget that most of this stuff won't get cheaper anytime soon so you can easily sell later if you get bored of it. Storage costs are zero, they fit in a sock drawer and don't require any special attention.
I know some guys are mighty pissed here because collectors drive the market but it is what it is.
It's a competition.

Reply 932 of 1005, by acl

User metadata
Rank Oldbie
Rank
Oldbie
mkarcher wrote on 2023-09-08, 00:08:
acl wrote on 2023-09-07, 07:53:
  • Speculative execution : make statistics (really basic ones) on code branching. For example : "the last 3 times i evaluated this condition, it was "true" so i will assume this will be true again". This allow the CPU to fill the pipeline with the code of the "true" branch. Without speculative execution, the CPU would have to wait until the condition is executed to know what to execute after. This helps to fill the pipeline efficiently. And if the prediction is wrong, then you just drop what's inside the pipeline and load the correct branch.

(for the experts : sorry if these descriptions contains aproximative statements, English does not help 😁)

Your explanation is quite good, but it doesn't describe "speculative execution", it describes "branch prediction".

The idea of speculative execution is to go one step further, and not only fill the pipeline, but also already execute those instructions, but do it in a way that the effects of the instructions can be undone/discarded if the prediction turned out false. Register Renaming (another feature you listed) helps a lot for speculative execution, as this just means that you need to keep the last confirmed value of a logical register in a different hardware register than the current value with speculative execution. If the prediction turned out to be right, your redeclare the "speculative current register" as "confirmed register". If the prediction turned out to be wrong, you just reset the register view to the registers that contain the last non-speculative values.

Thank you very much.
That makes a lot of sense !

"Hello, my friend. Stay awhile and listen..."
My collection (not up to date)

Reply 933 of 1005, by feipoa

User metadata
Rank l33t++
Rank
l33t++
Mandrew wrote on 2023-09-08, 07:14:

These people don't give af about K5 features and performance, they just want every single production model behind a display case.

I think for most people, they are in so deep that it is not possible or practical to create a display case. Neatly stuffed into drawers or boxes is the most.

Mandrew wrote on 2023-09-08, 07:14:

I know some guys are mighty pissed here because collectors drive the market but it is what it is. It's a competition.

Confessions from a former VOGONer/collector who left the forum and hobby, "i was a dribbling loon back in the day. "

I too am waiting for this grand epiphany.

Plan your life wisely, you'll be dead before you know it.

Reply 934 of 1005, by ThinkpadIL

User metadata
Rank Oldbie
Rank
Oldbie
feipoa wrote on 2023-09-08, 08:08:

Confessions from a former VOGONer/collector who left the forum and hobby, "i was a dribbling loon back in the day. "

I too am waiting for this grand epiphany.

Don't know about others, but I personally didn't take this hobby seriously from the start. That's why I spend on it as much time and as much money as I really want to. Missed some rare item on eBay - I don't care. Started some project and never finished it - I don't care. That's why I still enjoy this hobby very much. For me the process of searching, bidding, buying, tinkering, talking on forums with interesting people is much more important than a goal. I actually have no goals, I just enjoy the process. 😎

Reply 935 of 1005, by Mandrew

User metadata
Rank Member
Rank
Member
feipoa wrote on 2023-09-08, 08:08:

I think for most people, they are in so deep that it is not possible or practical to create a display case. Neatly stuffed into drawers or boxes is the most.

True, I went with the library cabinet style with easy access, much safer and more practical than plastic display cases.

Reply 936 of 1005, by Mandrew

User metadata
Rank Member
Rank
Member
ThinkpadIL wrote on 2023-09-08, 09:07:

I just enjoy the process. 😎

Building a collection is a ton of fun but it can get extremely stressful over time. It's a fine balance between going forward and going broke with doubts and regrets along the way.

Reply 937 of 1005, by ThinkpadIL

User metadata
Rank Oldbie
Rank
Oldbie
Mandrew wrote on 2023-09-08, 09:41:
ThinkpadIL wrote on 2023-09-08, 09:07:

I just enjoy the process. 😎

Building a collection is a ton of fun but it can get extremely stressful over time. It's a fine balance between going forward and going broke with doubts and regrets along the way.

Well, I think the main problem in collecting as well as in gaming is that people are trying to achieve or practically unachievable or hardly and expensively achievable goals. They spend a lot of money, effort and time and suffer all the way, comforting themselves with the hope of finding happiness at the moment of achieving this cherished goal. But when (and if) the goal is achieved, their reward is only a couple of minutes of euphoria, and then comes devastation and a feeling of wasted time, effort and money.

That's why I don't care about achieving goals and instead enjoy the process. If at some stage of the process I'll achieve some goal - great! But if not, I really won't care much. 🙂

Reply 938 of 1005, by Mandrew

User metadata
Rank Member
Rank
Member
ThinkpadIL wrote on 2023-09-08, 10:16:

Well, I think the main problem in collecting as well as in gaming is that people are trying to achieve or practically unachievable or hardly and expensively achievable goals. They spend a lot of money, effort and time and suffer all the way, comforting themselves with the hope of finding happiness at the moment of achieving this cherished goal. But when (and if) the goal is achieved, their reward is only a couple of minutes of euphoria, and then comes devastation and a feeling of wasted time, effort and money.

That's a very accurate description about the hardships of collecting stuff, including the part about the struggle to reach 100% and the terrible feeling when something is out of reach at the moment.
But it's also the reason why collecting stuff is so much fun, there is always something to look forward to. The challenge is always there and if collecting stuff would be easy and cheap then I wouldn't see the appeal at all. Yes, some hardware are expensive and some are very hard to find no matter how much money you save, but watching the collection grow is more fun than people'd think. Sometimes I spend hours looking at my stuff, organizing them, fiddling with the database and photos I built and comparing them to items other people shared online. Each piece has a story attached to it.

Reply 939 of 1005, by ThinkpadIL

User metadata
Rank Oldbie
Rank
Oldbie
Mandrew wrote on 2023-09-08, 12:00:
ThinkpadIL wrote on 2023-09-08, 10:16:

Well, I think the main problem in collecting as well as in gaming is that people are trying to achieve or practically unachievable or hardly and expensively achievable goals. They spend a lot of money, effort and time and suffer all the way, comforting themselves with the hope of finding happiness at the moment of achieving this cherished goal. But when (and if) the goal is achieved, their reward is only a couple of minutes of euphoria, and then comes devastation and a feeling of wasted time, effort and money.

That's a very accurate description about the hardships of collecting stuff, including the part about the struggle to reach 100% and the terrible feeling when something is out of reach at the moment.
But it's also the reason why collecting stuff is so much fun, there is always something to look forward to. The challenge is always there and if collecting stuff would be easy and cheap then I wouldn't see the appeal at all. Yes, some hardware are expensive and some are very hard to find no matter how much money you save, but watching the collection grow is more fun than people'd think. Sometimes I spend hours looking at my stuff, organizing them, fiddling with the database and photos I built and comparing them to items other people shared online. Each piece has a story attached to it.

Well, I'm not a masochist so I don't need any hardships or challenges in order to have fun. I collect things because they are interesting to me and not because they are rare, expensive or hard to find. And if something I'm looking for is out of reach, I don't really care. I'll keep looking for it till I'll finally find it or will loose interest, but till then I won't really care that I don't have it and will keep being a happy person.

And yes, I agree, to watch the collection grow, to look at it, to organize your stuff is also part of a fun.