Curious: lifespan for i7-3770S and i7-4771 when taken frequently into the 200s?

vbuggy

Golden Member
Nov 13, 2005
1,610
0
71
As I said, curious. There's already stuttering and occasional instability but I was wondering how long these could be expected to survive realistically speaking.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
10-25 years or longer.

I dont think there is a stock CPU that died of age yet. Only due to the loss of love from its owner.
 

Stuka87

Diamond Member
Dec 10, 2010
6,240
2,559
136
The chip will throttle if the temp gets too high. You can't really damage it unless it's getting heated from an external source. So watch your case temps.
 

zir_blazer

Golden Member
Jun 6, 2013
1,160
400
136
The problem is that the Heatsink that Intel supplies with their current high end Processors is soo ridiculous undersized that you WILL get into throttling scenarios. Heck, we're not even talking about AVX, which is supposed to get the most heat out of Haswell. And you also have two Xeons model above mine (Top bin is 200 MHz faster) which should consume a bit more power and carry the same Heatsink, so it should be worse for those.

As far that I know, the problem is not only heat itself, but thermal cycles. With all the power saving features plus the fact that it rapidly heats when in use, you can go from around 30-40°C in Idle to 80°C+ during a few minutes in Full Load. These thermal cycles with an heavy delta from cold to hot places stress on materials due contraction and expansion, which was supposed to be what triggered premature deaths during the famous nVidia bumpgate (Remember those Notebooks with integrated GeForce 6150 around 5 years ago?). And is what now Desktop Processors should also endure in retail conditions.

Actually, I'm starting to think than that is a not a mere design choice based on reducing Bill of Materials by using cheap Heatsinks, but on planned obsolescence. When you add in that the paste that they use between the Heatspreader and the die possibily degrades over time, making the deltas bigger, so chances are that it has an overally much reduced life expectancy than other Processors with soldered Heatspreader and less temperature delta due to better cooling.
Remember that these days computers usually last much longer than before, because technology is not superceding them fast enough and even a 6 years old computer may be good enough for browsing, which during early 2000 was impossible. So one of the ways to sell more, is to sell replacement to parts that died. Making more possible scenario choices for them to die could lead to more sales. So as a conspiracy theory, I find it realistic.


I sweared that once I get motivated enough I will make a Thread about an Athlon 64 Venice E6 which had severe overheating problems starting from around 4 years in use, and worsened over time until overheating protection shutted down the computer almost always during Windows XP boot. I expect that Ivy Bridges and Haswell will get similar symptoms some years into the future.
 

Sheep221

Golden Member
Oct 28, 2012
1,843
27
81
Dude, the manufacturing technologies and design got drastically improved since days of Athlon XP, Athlon 64, Pentium 4 and Core 2. The Core i architectures are so effective in these terms compared to older technologies that it's not correct at all what you suggest, and even then it was not valid, because those were design problems rather intentional ones.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
The problem is that the Heatsink that Intel supplies with their current high end Processors is soo ridiculous undersized that you WILL get into throttling scenarios. Heck, we're not even talking about AVX, which is supposed to get the most heat out of Haswell. And you also have two Xeons model above mine (Top bin is 200 MHz faster) which should consume a bit more power and carry the same Heatsink, so it should be worse for those.

I think your issue is something else. Under non 256bit AVX or AVX2 load Haswell simply runs cooler than IB. Either your fanspeed is not what it should, or your heatsink needs reseating.
 

zir_blazer

Golden Member
Jun 6, 2013
1,160
400
136
Dude, the manufacturing technologies and design got drastically improved since days of Athlon XP, Athlon 64, Pentium 4 and Core 2. The Core i architectures are so effective in these terms compared to older technologies that it's not correct at all what you suggest, and even then it was not valid, because those were design problems rather intentional ones.
Effective in what regard? You have the same issue: Thermal cycles. Yes, nVidia parts failed prematurely due to design issues, but what I point out, is that in Ivy Bridge, and particularily Haswell, with retail dissipation you have very big deltas that put material stress that you did not had before. If they are more prone to fail or not, we will know in a few years.

In my Athlon 64 case, the only thing I could get by googling around is a few other guys that had overheating issues that couldn't be tracked down to cooling, but a few managed to fix it by changing the grease between the Heatspreader and the die. So basically, for mine, it started to degrade after 4 years 24/7. If Intel last longer, I don't know.


I think your issue is something else. Under non 256bit AVX or AVX2 load Haswell simply runs cooler than IB. Either your fanspeed is not what it should, or your heatsink needs reseating.
I'm not comparing Haswell with Ivy Bridge, I'm comparing it to itself. Fanspeed is set from BIOS to max - I through I could have seated the Heatsink improperly, but temperatures are around what should be expected according to all the "Is my Haswell too hot?" Threads. And besides I don't have thermal grease to replace the Thermal Pad at the moment for when I have the will and spare time.
 
Last edited:

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
The oldest CPU I gave away to someone I remember is a 10 year old 2Ghz Pentium M. Still runs fine today. Same 100C Tjmax.
 

zir_blazer

Golden Member
Jun 6, 2013
1,160
400
136
The oldest CPU I gave away to someone I remember is a 10 year old 2Ghz Pentium M. Still runs fine today. Same 100C Tjmax.
That is my point. Most old Processors are still working, is very rare to see a Processor fail. I gave away a machine with a K6-II 500 MHz that was still working after 13 years. This means that from a sales perspective, you're not selling a replacement for the dead part if it is going to last forever. However, your TJMax comment is not reelevant at all, unless the Processor spended those 10 years with heavy thermal cycle duty (Open a game or something, get it to Full Load to heat it up close to TJMax, close it, let it enter a deep sleep C-State to cool down, repeat).

We were used to big advancements in the performance front that made old parts obsolete because they weren't fast enough to cope with new Software, but considering that mainstream machines from 2008 or so are still good enough for most daily usage but games (Unless you LOVE bloatware), you will only replace them either if they're not fast enough any longer, or if a part fails and cost too much to replace, in what case you just go and purchase a new computer. I consider that allowing more scenarios where a Processor would get damage is positive from a sales perspective, reason why it makes sense that they have a more realistic lifetime.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
"Good enough" is a poor mans term or someone that tries to fuel the ARM hype.

That pentium M is anything but fast today. Its slow and utter rubbish. And only still in use for the novelty factor.

Even the oldest actually used system in our household today with an i5 661 is not exactly a speedracer.

"Good enough" is a false term to describe the lack of upgrades during the crisis. But if you do some research you would notice its not just CPUs. Its TVs, clothes, appliances and everything else.

Check the average household savings around countries. In Denmark alone we "lack" to spend over 40 billion $ on consumer products since the crisis. Simply added into saving accounts instead. The smartphone market only lives on its trend factor. But thats already starting to erode.
 

zir_blazer

Golden Member
Jun 6, 2013
1,160
400
136
I live in Argentina, and here is common to see every now and then someone that doesn't have the money to replace his dinosaur computer. And believe it, if they have plenty RAM they're still "good enough" as long as they're not filled with crapware. Not everyone wants to play Crysis. Office-type applications doesn't demand too much performance, the only thing that makes these machines feel slow is Flash content, and for as long as they don't upgrade to the latest, greatest and bloatest versions of some applications, they will just keep going.

Heck, I just remembered that some years ago we had the Netbook craze with those slow as hell Atoms and hyperbloated Vistas. Yet the Joe Averages that purchased them became used to their slowness...
 
Last edited:

vbuggy

Golden Member
Nov 13, 2005
1,610
0
71
The chip will throttle if the temp gets too high. You can't really damage it unless it's getting heated from an external source. So watch your case temps.

That's kind of the crux of the question.

e.g. I used to have heat death issues with Prescotts and Northwoods, which were also able to throttle.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
That's kind of the crux of the question.

e.g. I used to have heat death issues with Prescotts and Northwoods, which were also able to throttle.

The shutdown temp is 130C.

P4 suffered death due to overvoltage. Also known as the sudden death syndrome. Not due to heat.
 

zir_blazer

Golden Member
Jun 6, 2013
1,160
400
136
The Northwood Sudden Death Syndrome affected Northwoods A and B if I recall correctly. I think they had aluminum interconnects. I suppose that Northwood C (800 MHz Bus and Hyper Threading) used copper interconnects by then because I don't recall that these were susceptible to NSDS. And yes, it was related to overvolting and electromigration damage, not temperature.

"Damage" is relative. Its not like if you're going to do something that kills it instantly, but the wear-and-tear of daily usage.
 

vbuggy

Golden Member
Nov 13, 2005
1,610
0
71
The shutdown temp is 130C.

P4 suffered death due to overvoltage. Also known as the sudden death syndrome. Not due to heat.

For OCers, relevant but I'm also talking about normal high-load use of lower-end business desktops. It was certainly heat as systems with the same CPU's but better cooling survived much longer.

The Northwood Sudden Death Syndrome affected Northwoods A and B if I recall correctly. I think they had aluminum interconnects. I suppose that Northwood C (800 MHz Bus and Hyper Threading) used copper interconnects by then because I don't recall that these were susceptible to NSDS. And yes, it was related to overvolting and electromigration damage, not temperature.

"Damage" is relative. Its not like if you're going to do something that kills it instantly, but the wear-and-tear of daily usage.

See above.
 
Last edited: