Current Barton XP life

[LRguy]

I'm deciding on buying a new system right now. Ya'll know when and how long this wonderchip is going to last without any changes? What speed can it achieve before it finally poops out? If I buy an Nforce Ultra board with 400fsb now and when I want to upgrade I want the fastest compatible cpu.

Please don't talk about overclocking, I don't want to hurt my computer that way

 

[Jeff7181]

Overclocking doesn't "hurt your computer" if you do it correctly.

The fastest processor you'll be able to buy is an XP3200 which runs at 2.2 Ghz. It will run indefinately, duh, lol... neither AMD nor Intel sell processors that "finally poop out" in your life time.

 

[gsethi]

Originally posted by: Jeff7181
Overclocking doesn't "hurt your computer" if you do it correctly.

The fastest processor you'll be able to buy is an XP3200 which runs at 2.2 Ghz. It will run indefinately, duh, lol... neither AMD nor Intel sell processors that "finally poop out" in your life time.

NOT CORRECT. each computer component has a maximum life span associated with it. They wont run forever even if you run them at default settings. Normal life span for CPU's is 10 years (i dont remember if it is 10 years @ 24/7 or not). The heat generated by the component will melt the raw material (silicon, boron, phosphrus, aluminium etc...) that is required to make a transistor. This melting rate is very very slow if you run your processor @ default, but if you overclock it, thats where the melting rate increases but it is melting very very very........ very slowly each day you use it.

Now, i am pretty sure that most of us wont be using the components for 10 years (heck, i replace mine in about 6 months - 1 year), so we dont notice anything. But if you want to test it out, but a brand new system today, and just keep it on for ~10 years or so and you will notice that it will die (but ofcourse who has the time to wait 10 years, right )

EDIT: Why do you think that your burn a chip if you dont use a heatsing/fan ? The heat generated by the chip melts down the transistors (since you are not dissipating the heat quickly enough) and your chip malfuctions. When you use a heatsink/fan, the melting rate is much slower but still the heat does melts your transistors.)

 

[sc0tty8]

10 years is the limit huh...

I have had MANY components well past that mark that still worked, I have had stuff 15 yrs old that ran at the very least 12 hours a day that worked. I still have some 'ye olde stuff' that still works.

My dad=the king of old computers, the fastest comptuer on his work netowrk is a celeron 400 with 32 megs of ram:Q

he gets a computer form this guy that he knows, they are all old 486 era computers, and it is all my dad needs, I think his 2nd fastest computer at work is a p90, and I am sure it will be going for years to come.....

 

[cow123]

i think when jeff said "doing it right" to not lower the lifespan of the chip, i think he means NOT increasing the vcore - thus a negligible temperature increase even if the chip was clocked substantially higher. imo, if you arent running a processor full whack on default vcore, then its underclocked.

 

[LRguy]

So the 3200+(2200mhz) speed is the limit for the 400fsb Barton? No more? They just came out with it. Damn, I can't afford an A64.

 

[sandorski]

Originally posted by: LRguy
So the 3200+(2200mhz) speed is the limit for the 400fsb Barton? No more? They just came out with it. Damn, I can't afford an A64.

There is always the possibility that AMD might release a faster Barton, but that would depend on how well and how long Socket A cpus continue to sell. If AMD has moved to .09micron manufacturing and SocketA cpus are still selling well, they possibly might make some at .09micron SocketA cpus, which would likely be faster. They did something similar back when the Athlon was first introduced, the K6s' were still selling well and they changed over from .25 to .18, so they released some newer K6-2/3 cpus. I wouldn't count on it, but it is a possibility.

 

[LRguy]

If they moved the bartons to the .09 process, will those be compatible with the 400fsb boards now?

 

[Soulkeeper]

nobody really knows if they will make any more socket A cpu's
but they are basically gonna start the long "phase out" process
it is possible they might release .09u bartons at higher clocks but it would steal sells away from the athlon64 and not really be a profitable business move

 

[Mingon]

but it would steal sells away from the athlon64 and not really be a profitable business move

of course it would make profit, the bartons are cheap to produce and are good performers.

 

[KF]

>The heat generated by the component will melt the raw material (silicon, boron, phosphrus, aluminium etc...)
>that is required to make a transistor. This melting rate is very very slow...

Maybe it doesn't matter what it's called; there is an ongoing process, they say, that deteriorates the transistor characteristics. But the materials don't come near to the melting point, and the deterioration process they describe doesn't sound like melting to me. Any chemical or quantum process is accelerated by temperature.

As for the lifetime, I don't think they are saying the chip will quit after some period of time. It is some sort of statistcal thing where some percentage of chips are expected to deteriorate to the point they won't meet spec (but will still function), and ten years would be a guarantee.

Be that as it may, my Athlon TBird 800 works better than it did when it was new a few years ago, if how far it OCs is an indication. Besides being old, it has had a rough OCed life, and it has a chip off the slug, and was overheated unintentionally to the point the white heatsink goo solidified and turned blackish brown. Then again, heatsinks and chipsets have come a long way in that time, and maybe that's what's increasing the OC.

AMD did continue to make faster K6s as the K7s/Athlons took over, but it wasn't long before the peformance jumps of K7s relegated K6s to the bargain bin. I think you can still buy new K6s, but the cheapest Athlons are not much more and the peformance gap is tremendous.

The uncertainty about how fast the cost of producing SOI Athlon 64s can be reduced is too large to make much of a prediction about whether old-style Athlon are going to be shrunk and get very much faster. But I would guess that AMD is going to have to. AMD isn't just competing with itself, so their chips can't all be priced high. If the price of some slower Athlon 64s is below $200 within a couple of months after its introduction, I think that would indicate that AMD will not be putting much into improving old-style Athlons.

 

[mechBgon]

Originally posted by: LRguy
So the 3200+(2200mhz) speed is the limit for the 400fsb Barton? No more? They just came out with it. Damn, I can't afford an A64.

Sure about that? It looks like the A64 3200+ is about the same price as the AthlonXP 3200+. Boards start at ~$125 for the Athlon 64 and will probably be dropping after a little while. They both take the same RAM.

So if you can afford a 3200+, you can afford an Athlon 64

 

[0roo0roo]

i think the limit is 15 years. but even with a lousy overclock, halving of even quartering a cpu's life span doesn't matter. by the time it dies, it will be worth less then a starbucks coffee

 

[clicknext]

If you don't do anything wrong, you can safely assume that it will live longer than its useful lifespan.

 

[Jeff7181]

Electron migration is how a CPU breaks down. The rate of migration increases when you increase the voltage. Contrary to what some people think... heat doesn't play as big a role in processor life as voltage does. Everybody's so concerned about heat... "what heatsink should I get? what fan should I get? is the retail heatsink good enough? is 60 degrees C too high? how can I make my CPU run cooler?"
If you're overclocking, of course lower temps are better... but when AMD sells you an XP2500 that they say will run up to 85 degrees C, it will. If it doesn't, you get a new one for free. Why would they give you a heatsink/fan that won't keep the CPU alive? That would be stupid.
If you don't overclock, chances are you'll have bought 2 new CPU's or computers before this one would have "pooped out" running at the speed, voltage, and temperature that AMD intended.

 

[zephyrprime]

The heat generated by the component will melt the raw material (silicon, boron, phosphrus, aluminium etc...) that is required to make a transistor

It doesn't "melt" the processor but it will cause a slow diffusion of atoms that will eventually render the device inoperable. But i'm not sure if this is athe actual cause of failure. If I recall correctly, there may be more caustic effects at work.

 

[Soulkeeper]

Originally posted by: Mingon

but it would steal sells away from the athlon64 and not really be a profitable business move

of course it would make profit, the bartons are cheap to produce and are good performers.

yes but all i'm saying is they still make more money selling a 3200+ athlon64 for like $400 or whatever than a 1700+ for $50 or a 2500+ for $90

 

[Jeff7181]

Originally posted by: Soulkeeper

Originally posted by: Mingon

but it would steal sells away from the athlon64 and not really be a profitable business move

of course it would make profit, the bartons are cheap to produce and are good performers.

yes but all i'm saying is they still make more money selling a 3200+ athlon64 for like $400 or whatever than a 1700+ for $50 or a 2500+ for $90

Most definately... if they can afford to sell an XP1700 for $50... their cost to manufacture has to be lower than that... now lets be generous and say the A64 costs twice as much to manufacture... so at MOST it costs AMD $100 to manufacture an A64, which is HIGHLY unlikely it's that high. And it's going to sell for $450 or so? I'll take that profit any day of the week and twice on Sunday.
(BTW... there's millions, possibly billions of $$ in R&D that nobody seems to ever take into consideration... which is why they have to charge as much as they are)

 

[gsethi]

Originally posted by: zephyrprime

The heat generated by the component will melt the raw material (silicon, boron, phosphrus, aluminium etc...) that is required to make a transistor

It doesn't "melt" the processor but it will cause a slow diffusion of atoms that will eventually render the device inoperable. But i'm not sure if this is athe actual cause of failure. If I recall correctly, there may be more caustic effects at work.

hehe, i didnt want to get into that technical details (and use the exact terminology). I wanted a common person to be able to read my post and understand it. I know the technical details and stuff on how the processor deteriorates but i was trying to avoid all the technical terms in that post. If you guys want all the technical terms and everything, i can look them up in my notes (about 1.5 years old).

yes, voltage plays the most important role in this process but normally, more voltage = more temperature and again, i didnt want to get into all that technical info at the time.

If you guys want to do a search, a similar post was created about 1 year ago and i did post some detailed technical information in that post. Anyone care to look it up ?

EDIT: (again in simple language) : The chip will fail to work when the different layers of materials used in the process of manufacturing the chip fade away (due to voltage and heat). All it takes is a single transistor to spoil a perfectly good chip. and all it needs to mess up a single transister is to "fade" or "melt" or "whatever you want to call it " a small layer with a thickness which can be as small as 0.01 nanometer