Hardware durability (an opinion piece)

[AnitaPeterson]

Hi folks,

I came across an interesting phenomenon, and I'd like to share this with you. I figured I might get further anecdotal confirmation from some of you - perhaps your experiences are completely different! - but it's an interesting discussion nonetheless.

In any case... some of you know that in the past I've been involved in a charitable activity to refurbish older computers and send them to Eastern Europe. Although I'm no longer involved directly in this project, I still like to repair older machines and give them away... And I've recently come to think that AMD-based machines have a shorter lifespan than their Intel counterparts.

Before you accuse me of fanboyism, consider that my top-of-the-line gaming system is an hexacore AMD; and over the years, I consistently supported the AMD/ATI underdogs against the Intel/nVidia onslaught. But my hands-on experience seems to indicate that long-term reliability is higher on Chipzilla's side.

Consider two examples:
1) Back in 2004, I built two separate machines - one around an AthlonXP 2500, the other one around a 2.4 GHz Intel P4. They were put together using some of the best parts recommended here on AT and on other forums... They saw equal amounts of work... and yet, the AMD motherboard (Abit NFS-7) died an ugly death back in 2007, forcing me to retire the whole system. Meanwhile, the P4 still works to this day.
2) Last year, I gave an Athlon XP/nForce2 Gateway machine to an older couple (who uses it for basic web browsing and light office work)... The machine died two weeks ago, after a power blackout, and refused to come back to life. I told the couple to buy a UPS - which they did! - and I replaced the dead computer with a similar one (only this time a HP). Guess what? It also died a week later - I couldn't even figure out why!

This got me thinking... If I add up the amount of times I've seen AMD-based systems kick the bucket, versus Intel, the numbers are much higher than expected. It could be that AMD was sabotaged by its partners (IMHO, the nForce motherboards, albeit popular, had an disproportionately high number of failures, on both AthlonXP and Athlon64 architectures)... But at this point, I'm kinda leaning towards thinking that - enthusiasm aside - AMD machines offset their relative price advantage with shorter lifespans.

Anyone care to comment?

 

[monkeydelmagico]

yes cheaper dies sooner but I think it's the mobo's not the chips. Haven't bricked an athlon/phenom chip yet..... knock on wood.

 

[Insert_Nickname]

Personally I have seen about equal amounts of dead AMD and Intel machines. I find it hard to give a blanket statement that one is better then the other, because its very rarely the CPUs themselves that are the problem. Usually the mainboard or PSU goes way before the CPU ever does. I have seen one truly dead CPU in the last 10 years (not counting the self-inflicted ones fried OC'ing), coincidently a socket 939 Athlon64 (and I suspect that one was self-inflicted too due to overheating, being very badly placed).

There are no guarantees with hardware, the most expensive kit can fail the day after you installed it, and the cheapest crap you can buy can last 10+ years. You never know...

By the way, you're quite right about nForce boards being very temperamental. Those have the highest failure rates along with VIA and SiS based boards in my experience. That goes for both the Intel and AMD versions. I think, without any guarantee of being right, that this is where the problem is. AMD CPUs has almost always been paired with shoddy 3rd party budget chipsets, whereas Intel's own chipsets have always been rock-solid.

 

[The Day Dreamer]

These days, HDDs are dyeing no matter how expensive they are. I am kind of annoyed as visiting WARRANTY center is pretty time consuming

 

[magomago]

Abit and Nvidia isn't AMD?

Even if it uses an AMD chipset, chances are something like a capacitor went bad on the motherboard more than anything else.

 

[AnitaPeterson]

AMD CPUs has almost always been paired with shoddy 3rd party budget chipsets, whereas Intel's own chipsets have always been rock-solid.

Yes! That's my feeling as well, and that's what I was trying to convey. Of course, the CPUs themselves have very little chance of going bad - it's the rest of the hardware that drags everything down....

 

[Torn Mind]

AMD makes the chips. They aren't responsible for everything else attached to the motherboard. A "bare" mobo has a bunch of holes, waiting for solder and sometimes a component to fill them up. When a board "dies", something in the electrical circuit is busted. Obviously, the board itself could have had manufacturing defects, but that would have been noticed immediately. But it could be a cold solder joint showing itself. Or a component such as a capacitor, transistor, MOSFET, diode, integrated circuit, etc failed.

Without a proper diagnostic, we really don't know whether it is something tied to AMD, poor component selection on behalf of the mobo manufacturers, bad soldering, or something like a power surge kill the board. Maybe the boards had capacitors that were just defective.

 

[aigomorla]

depends totally on the chip and cpu.

Example... if its a kentsfield or bloomfield die, GOOD LUCK trying to kill those chips off, when running them on stock conditions within operational standard parameters.
These chips were true masterpieces of durability.

The same can be said about some opterons i have had.
Good luck trying to kill those off.


However all vendors have a black sheep...
Intel it was Yorkies/Wolfdales... fast as cheetahs... but also endurance wise simular... and also the original Gulftowns.


Also it matters how hot the chips ran... which chips had what load on them.
Intel in those generations could throttle, while AMD's couldnt.
Every 10C u lower cpu temp, you effective double the life of the processor.

If the Intel was throttling while the AMD was operating at exceeding values, the AMD was doing more work, and then you would have to calculate how much work each PC did in relationship to how and when it died.
But thats besides the moot... im pretty sure within 1000000000000% confidence the AMD XP did a lot more work then your space heater P4 did or will do even if your P4 could stay alive another 3 years....

In the generation your looking at, AMD was in its golden years... dominating and smashing intel left and right in almost every test u threw at them.... oh how i miss those days....

 

[Insert_Nickname]

If the Intel was throttling while the AMD was operating at exceeding values, the AMD was doing more work, and then you would have to calculate how much work each PC did in relationship to how and when it died.
But thats besides the moot... im pretty sure within 1000000000000% confidence the AMD X2 did a lot more work then your space heater P4 did or will do even if your P4 could stay alive another 3 years....

That reminds me of the classic tomshardware overheating video years ago. The P4 continued to operate (slowly) on the IHS, while the Athlon fried itself reaching something like 370C...

 

[sm625]

Anecdotally I would agree. I only have one remaining AMD system that is operational. There are just so many cheap off lease Core 2 Duo machines out there that it makes no sense to go with AMD because they never had the volume 5 years ago to allow for the kind of great deals you can get today on used hardware. And every single Core 2 machine that I've bought for $80-$100 has worked perfectly. I'm actually surprised none of them have died yet. (knock on wood...)

 

[Atreidin]

The only processors I've actually seen die in normal service were Cyrix processors, and those were by heat death from crappy cooling fans that kept dying and not being replaced fast enough.
If I discount the motherboards that died from the capacitor plague of whatever years those were, I've seen pretty equal reliability between Intel and AMD platforms.

 

[Cerb]

Meanwhile, I have a stack of K7S5As that just won't die; and every A64 or AII/PhIIs I've personally seen dead was near one of a few local lakes known for storm problems.

I think it's much more that the cheaper PCs used AMD more often than higher quality ones, except for a brief period of time between ~2004 and early 2006. I've yet to see an HP DC model (the various-named predecessors to the current Elite line), nor a white box Gigabyte board, with older AMDs in it, die, without some external cause, at least (PSU blowing up, bad storm taking out other components, etc.). EPoX and Iwill boards even ran rock stable with puffed out bad caps, IME (RIP EPoX).

I've had some C2D machines die, too: all Dells, or white boxes w/ RaidMax or Bestec PSUs.

 

[mindless1]

It's twofold. Cheaper systems used AMD CPUs but the failures you are talking about have a different issue.

Athlon XP era systems commonly had the HLT idle disabled in the BIOS, largely due to a legacy problem where their transition out of idle state required too rapid a current increase when they were still using 5V PSU rail for CPU power. For some reason the trend to disable HLT idle in motherboard bios continued even after they switched to 12V PSU power rail.

The result was these systems ran hot a lot of the time, causing more thermal contractions and expansions, more wear on the thermal compound, more wear on the motherboard capacitors, and more wear on the PSU with these latter two factors even more significant when budgetized motherboards and PSU were used.

Today things have changed a lot. People are buying 80+ PSU and all but the lowest of the low end motherboards have non-electrolytic capacitors where it counts, as well as AMD power management states being well supported, and CPUs now being so powerful that the majority of the time they're barely more than idle even if you're playing a video that an Athlon XP would choke on... though I have to give some credit to GPU accelerated assistance.

 

[bononos]

It's twofold. Cheaper systems used AMD CPUs but the failures you are talking about have a different issue.

Athlon XP era systems commonly had the HLT idle disabled in the BIOS, largely due to a legacy problem where their transition out of idle state required too rapid a current increase when they were still using 5V PSU rail for CPU power. For some reason the trend to disable HLT idle in motherboard bios continued even after they switched to 12V PSU power rail.
.....

I remember running rain/waterfall utilities and enabling hlt in bios because win95-2000 didn't do the job well enough.

 

[Joeydubbs]

I have built a few machines, all intel based, and the only PC part to ever fail on me was an AMD video card...so my minimal anecdotal evidence supports your theory lol

 

[mfenn]

(RIP EPoX).

9NDA3J for life. :rose:

 

[mindless1]

I remember running rain/waterfall utilities and enabling hlt in bios because win95-2000 didn't do the job well enough.

It's not that Win2k (not sure about 9x) didn't do it well enough, it's that it and later OS too, depended on the chipset bit being set correctly in the bios for HLT idle to work while most mobo manufacturers deliberately disabled it. The same would happen with XP, Vista, Win7 and 8, or any non-windows OS as well.

You could actually edit a bios to have the bit set right for it then flash that, with no need to run the utilities on every boot to do it.

Regardless, systems of that era had really bad performance to power consumption ratios. Best thing to do is replace them before they fail. Even for a job that didn't require much in the way of performance, if the platform could handle the other hardware and memory required then I'd sooner step backwards to a PIII Coppermine era system than forward to an Athlon XP because HLT idle worked on those. On the other hand integrated video wasn't very good in that era so add the power and purchase cost of a video card unless it's a headless system.

 

[Berliner]

The only system in the last 15 years that has ever "died" on me was due to a mainboard with faulty capacitors, and that is not even completely dead yet.

I don't think it has anything to do with AMD or Intel.

 

[piasabird]

I prefer Intel over AMD. However, a hard drive can give up at any time. I have seen a few hard drives bite the dust. I just have a feeling that Intel is higher quality. I have built both Intel and AMD systems over the years.

I do think that Intel is overpricing their processors because they think they can get away with it. I also think that operating systems are overpriced. After 2 years no operating system should be allowed to be sold at full price. Especially how Mocrosoft has a virtual monopoly in the desktop sector and the business sector.

 

[VirtualLarry]

I think that most OEM boxes with Intel were generally, though not always, built better.

AMD-based OEM boxes were built to a price-point that meant that the OEM was cutting corners. Remember how long NV 6150SE chipset based boxes were on the market?

Only to be replaced by E-300 boxes on the low end. I seriously question if that is progress.

 

[schmuckley]

I've killed: 1 1075T,2 2500Ks,1 e8400,1 Xeon x3450,1 i3 540
not going to list motherboards :$
IME Intel chips die easier.
..and boards

 

[aigomorla]

I've killed: 1 1075T,2 2500Ks,1 e8400,1 Xeon x3450,1 i3 540
not going to list motherboards :$
IME Intel chips die easier.
..and boards

and how many of those were overclocked beyond stock settings?
And how many % increase over stock settings did u get on the intels...

u need to factor in that aspect too...

 

[CoolRunnings]

I've built several hundred computers since 2002. Starting in 2006 I started seeing more and more problems with the AMD systems and by 2008 I stopped using their CPU's in any computer I built. Of those computers, 3x of the Intel boxes have died (motherboard failure, PSU, etc. but the chips were all still goodP) and at least 25 of the AMD systems died - 10 of which were bad CPU's. I know it's anecdotal but it's enough to cause me to stick with the Intel stuff. Additionally, the AMD systems I still have out there gave me far more problems with drivers, faulty motherboards (Gigabyte, I hate you!), and random instability.

 

[mindless1]

In my experience there is almost never a truly "bad CPU" failure that doesn't manifest itself right away during a system build.

Months to years later, maybe the thermal compound dried up, or the fan died, or the PSU failed or an external AC surge occurred and made a spike all the way to the CPU, but otherwise CPUs ran within reasonable limits (even overclocked, just not extremely overclocked to the point of running over 70C temp.) have a much longer lifespan than 11 years.

How can an AMD system give more trouble with drivers when it's the same situation for either?? You won't get Intel chipset drivers for an AMD system but otherwise the AMD, Sis, Via, nVidia were the same situation for either CPU, while ethernet, sound, etc were often the same driver.

Personally I did find ATI drivers to be a bloated mess about 10 years ago but now they're a bit better and everyone else's are also a bloated mess, lol. Remember when driver meant 3MB or less of zipped up files? Quite often they used to come with the product on a floppy.

 

[CoolRunnings]

I agree, reliability should be looked upon as a platform-as-a-whole situation, not just the CPU. Hence why I don't build AMD systems anymore.