Question about CPUs. (Warning: A stupid one)

[keropi]

if it was true, my 15+ year old 386sx would be an XT by now :lol:

[motorollin]

keropi wrote:
if it was true, my 15+ year old 386sx would be an XT by now :lol:

And what would our Amigas have become?  ;-)

--
moto

[Fats]

Lando wrote:
No.  A CPU either works or doesn't work, there isn't a degradation in performance over time like with mechanical items.

A little bit more detail: there is wear internally in a microprocessor and one of the mechanisms is called electromigration, another is the degradation of the gate oxide that should act as an insulator but can break down (google is your friend !).
99% of the CPUs is externally clocked and you won't notice the wear out until the signals in the transistor can't reach the destination within the clock cycle. If that happens the CPU will abruptly fail or become unreliable. What can happen is that the power consumption goes up over time due to increased leakage currents in the transistors but most of the time this phenomenon degrades quite abrupt when it starts to be noticable.

greets,
Staf.

PS: I work here so I'm a little bit familiar with this material.

[CLS2086]

Actual harddrive must be replace at least each 2 years if they are used 10h/days, after that they lost 50% of the speed output.

My source : my main pecee made in 2001 (Sisoft Sandra 2003/2004, chipset HTP Raid 370, swap+temp on an another HDD)
IBM 46 Gb : after 1 year, around 40% of perforce lost, then brokedown
Maxtor 20 Gb : after 3 years, around 50 % of perforce lost
Seagate 120 gb : after 1 year, around 20 % of perforce lost
Maxtor 300 Gb : after 1 year, around 10 % of perforce lost

Capacitors can be also problematic.
I changed all of them on a 2004 WinFast Nforce2 (they explosed  one after one...), and some on 2003 MSi Nforce2. They now both work like a charm, and performances are back.

[AndyFC]

@ Van_M

I remember that Register article. I also knew exaclty which date to search for it in the archive.

I went to 2007, and then April, and then Sunday 1st:

http://www.reghardware.co.uk/2007/04/01/cpu_time_dilation/

(sad that I remember it but that isn't the saddest example of my memory. I was in Ikea a few years ago and they had some mags on a table in a living room setup. One was CU Amiga, and had a picture of the 'New' A1200 on it. I knew then (but not now - I would guess at October/November 1992) which month and year it was - this was about 8 years after it was published!)

[Zac67]

A CPU running slower after some time is complete nonsense.

But:
- I had to downclock my Athlon XP 1700+ after ~5 years from 1866 to its nominal 1533 MHz - the system kept rebooting in spite of proper cooling. Probably a voltage raise would have removed the problems, but the board didn't allow for that and more voltage speeds up electromigration thus aging.
- I noticed my P4 Prescott at work was only throughputting 2 Mkeys/s in distributed.net's RC5-72 - the cooler was pretty clogged up. After a little bit of cleaning the noise was massively reduced and the speed was back up to ~7 Mkeys/s.

@CLS2086:
Have you defragmented your drives completely or formatted them prior to measuring?
My drives run 24/7 for years (usually 4-5) and I've never observed a significant (maybe barely measurable, but surely unnoticeable) decrease of speed (seek, throughput)...
They're not RAIDed however, in certain RAID modes you may see significant speed decreases when the drives are not spindle synchronized.

[AmigaMance]

I remember that Register article. I also knew exaclty which date to search for it in the archive.

I went to 2007, and then April, and then Sunday 1st:

http://www.reghardware.co.uk/2007/04/01/cpu_time_dilation/

 Now, who am i suppose to believe? All you ignorants or Professor Asteio Artikolos?! My money on him. :lol:



@CLS2086

Actual harddrive must be replace at least each 2 years if they are used 10h/days, after that they lost 50% of the speed output.

My source : my main pecee made in 2001 (Sisoft Sandra 2003/2004, chipset HTP Raid 370, swap+temp on an another HDD)
IBM 46 Gb : after 1 year, around 40% of perforce lost, then brokedown
Maxtor 20 Gb : after 3 years, around 50 % of perforce lost
Seagate 120 gb : after 1 year, around 20 % of perforce lost
Maxtor 300 Gb : after 1 year, around 10 % of perforce lost

 That's a very interesting post. Thanks, i did know about this. I have a 8GB IDE SeaGate in active use since... I don't even remember since when. I should have done some benchmarks when i first bought it so i could compare them with current benchmarks.
 I can't say that it feels slower but this might be because i'm using the standard IDE port of my Amiga 1200, which is already slow.

[jj]

Stop........De-fraging a hard-drive is pointless.  It will take hours and hours, and you will not notice the real world difference (apart from the old its supposed to be faster so it is effect).  you will never gain back the time lost in de-fraging in a performance gain.  And guess what as soon as you have de-fragged the drive it gets fragmented again.  

[motorollin]

JJ wrote:
Stop........De-fraging a hard-drive is pointless.  It will take hours and hours, and you will not notice the real world difference

The hard drive in my company laptop got to 70% fragmentation. I defragged it and it was significantly faster.

JJ wrote:
you will never gain back the time lost in de-fraging in a performance gain.

So do it overnight.

--
moto

[jj]

Ok do it once every few years.  Never bother doing as  some suggest regurarly.  It is a waste of time.

[Piru]

Proper tools will do the defrag operation at scheduled times (such as during the lunchbreak).

Defragmenting is well worth it. It is not waste of time if you're not waiting for it to finish.

Fragmentation is especially bad with the paging file and NTFS Master File Table. If paging file (swap) is fragmented it will cause massive slowdowns to swapping operations. Similarily MFT fragmentation will reduce the overall performance of the filesystem, regardless if the files are fragmented or not.

Paging file can be defragmented with PageDefrag

MFT defrag is slightly more demanding task, the one tool I've found that does the job well is "Diskeeper".

[Piru]

@CLS2086

I have to say I'm a bit sceptical about hard drives losing performance, either.

The only case I can think of is bad block remapping, which could in theory explain the slowdown. However, in practice I've never seen any performance loss.

[Kraftwerk]

I think this has much more to do with the fact that over time we ask our CPUs to do an ever increasing load without changing the CPU's speed. Thus it seems to run slower when in fact we are just using a smaller slice of the processing speed pie.

[Tomas]

AmigaMance wrote:
 Can a CPU (68k, PPC603) become slower by the passing of time, if it suffers from mechanical (bumps) or thermal stress?
 I'm not talking about the clock frequency, which will be always the same (i guess) but about the overall performance of that CPU.
 Well, i warned you that this is a stupid question, so be gentle. :-D

Nope, not possible.. It can however start calculating wrong if something becomes defect, which will most likely result in a unstable computer that crashes alot. I believe the cpu is the component in a pc that has the longest lifetime if used properly. Pretty much the only thing that is likely to kill your cpu, is heat, wrong voltage and overclocking it way up.

Actually i have noticed the opposite of your question. I found that a cpu clock lower when it is new, while after a few years i seem to be able to overclock it higher than when it was new.
One example was my old p1 133.. I first could only overclock it to 166mhz without stability issues. At 180mhz it would either crash at boot or soon after. Then later after running it at 166mhz for some time i could bump it up to 180mhz and after staying there for a while i could bump it up to 200mhz using the same exact components. It ran perfectly stable at 200mhz for years, even running seti@home 24/7 and would even nearly boot at 220mhz.

[Tomas]

6 years later I was called to format the hard drive and reinstall windows 98 because (unsurprisingly again) the old setup got seriously messed up. Again win98 , same driver set, 3dmark 2000.... guess what.... 762 marks!

Could be something simple as a missing or different version of the chipset drivers. There is just no way a cpu can get slower like that. The only way it could get slower, is if the speed was throttled down by bios due to overheating.

[jj]

@Piru

Thanks for the link to the pagefile defragger cause we all know that a fragmented pagefile does cause a real world performance hit and windows defragger does not defrag the pagefile as it is in use when the defragger is running.

Schedule to run in lunchbreak?

You must have a small hardrive or a really long lunchbreak :lol:  Wonder how long it would take to defrag my more or less full 250gig sata HDD.  Im guessing awhile