is this safe

[Donut417]

8 hours ago, ImRyan said:

is it safe to leave my pc on for 18 hours at 100% cpu load

 

i have i7 3770k at stock and a corsair h100

I left mine on at 100% CPU for 24 hours. Thats how you do a CPU stress test, generally done when you build a new PC and want to make sure its stable. Some people even suggest longer. As long as your temps are good, it should not be an issue. 

[AshleyAshes]

5 hours ago, TheCherryKing said:

The Intel Xeon CPU's exist for a reason. They don't get very hot and most of them are low power consumption. The Intel Core CPU's run hotter and are not designed to run 24/7. Yes, it is possible to run an Intel Core processor 24/7 but it may not last as long as Intel intended. 

 

4 hours ago, Vantage9 said:

The best analogy i can think of is a car engine.  

 

An engine left running at idle uses some fuel and puts off some heat.  The wear and tear on an engine running at idle is virtually nothing, and a given engine could idle for years on end given unlimited fuel.

 

An engine running near redline RPM uses SIGNIFICANTLY more fuel and puts off way more heat.  The wear and tear on a redlining engine is exponentially more than at idle RMP.  With unlimited fuel, the same engine might only run for a few weeks at redline RPM.**

 

**This is just a hypothetical as an example, please don't start quoting engine life figures to me.

 

As long as the engine is under its redine, we're technically "within spec", but to think you'll get the same engine life out of idling vs near redline is absurd.  Electricity and voltage going through a processor has the same effect - more of it equals more wear and tear.

 

2 hours ago, TheCherryKing said:

I am well aware that Xeons are designed for more than 24/7 operation. I stated earlier that a consumer chip can also run 24/7 although it is not designed to. 

None of this makes any sense in reality however.

 

Firstly, CPUs are solid state,they are devoid of moving parts.

 

Secondly, consumer CPUs and Xeon's have few differences.  There are Xeon's produced for consumer sockets even and enthusiast i7's are running on server sockets.  All of which are produced from the same or nearly the same die.  I assure you, there is no massive re-engineering effort made between an i7 7700k and a Xeon E3-1285 v6, there are only minor differences to allow specific features.  There is for sure no 'Secret Sauce' in a Xeon that makes it capable of 24/7 operation.

 

I've seen plenty of consumer chips run 24/7 under load in servers and render farms at small companies.  Things like the 2500k still chugging out frames since they were bought new.

 

I understand the emotional gut idea that a Xeon is 'designed' for certian tasks, but outside of specific features such as ECC support, virtualization and some other specialized features, it's all a fantasy and not backed up by any facts.

[TheCherryKing]

21 minutes ago, AshleyAshes said:

 

 

None of this makes any sense in reality however.

 

Firstly, CPUs are solid state,they are devoid of moving parts.

 

Secondly, consumer CPUs and Xeon's have few differences.  There are Xeon's produced for consumer sockets even and enthusiast i7's are running on server sockets.  All of which are produced from the same or nearly the same die.  I assure you, there is no massive re-engineering effort made between an i7 7700k and a Xeon E3-1285 v6, there are only minor differences to allow specific features.  There is for sure no 'Secret Sauce' in a Xeon that makes it capable of 24/7 operation.

 

I've seen plenty of consumer chips run 24/7 under load in servers and render farms at small companies.  Things like the 2500k still chugging out frames since they were bought new.

 

I understand the emotional gut idea that a Xeon is 'designed' for certian tasks, but outside of specific features such as ECC support, virtualization and some other specialized features, it's all a fantasy and not backed up by any facts.

The secret sauce it is that Xeons are higher binned Intel Core processors. Intel disables cores and ECC memory support for their lower binned processors which are the Intel Core lineup. 

[AshleyAshes]

1 minute ago, TheCherryKing said:

The secret sauce it is that Xeons are higher binned Intel Core processors. Intel disables cores and ECC memory support for their lower binned processors which are the Intel Core lineup. 

See this is where I think you don't know how all this works and you're making it up.

 

Like, disabled cores?  You know they sold dual core Xeon's as recently as broadwell, right?  Those also had disabled cores inside a standard LGA 1150 die.

 

For sure there's binning going on but to argue that that inherently renders a consumer CPU incapable of 24.7 operation is just, well, silly.

[TheCherryKing]

Just now, AshleyAshes said:

See this is where I think you don't know how all this works and you're making it up.

 

Like, disabled cores?  You know they sold dual core Xeon's as recently as broadwell, right?  Those also had disabled cores inside a standard LGA 1150 die.

 

For sure there's binning going on but to argue that that inherently renders a consumer CPU incapable of 24.7 operation is just, well, silly.

Stop putting words into my mouth! I have never said that consumer CPU's are incapable of 24/7 operation. 

[Donut417]

1 minute ago, valdyrgramr said:

My fury was once hotter than the surface of the sun...

LOL. Thats pretty fucking hot. Not going to lie. 

 

[Donut417]

26 minutes ago, valdyrgramr said:

Good thing it was made by Sapphire.  That thing is still intact.

Yeah. Its just like Windows used to report my CPU at over 5 Ghz when it was running at stock clock. 

[AncientNerd]

16 hours ago, Vantage9 said:

 

 If you're running at 100% CPU use, then you are putting way more voltage and heat through that CPU than someone who leaves their computer running 24/7 but is only occasionally under full load. 

Sorry this is incorrect, circuits just don't work this way. you will vary current draw but the voltage is going to stay at the set value, yet the heat will be somewhat higher but as long as it is below the specified value it is fine no matter what the % CPU usage. Electronics don't have ware and tare the way mechanical parts do, because of friction. They have issues due to heating and current draw mostly damage due to excessive heat - there is a level of heat that is non-damaging. Damage due to current draw is what you see with high voltage damage you are forcing more electrons through a part of the chip than it can handle - i.e., there is enough potential difference that more electrons can flow than the material can handle (high current) which will damage the components faster than heat but just as much.

 

So yes high voltage and high heat will damage the CPU, no just running at 100% CPU for long periods of time will not cause this. With the assumptions that 1) you have not over clocked and 2) you have the correct cooler for the TDP.

[Vantage9]

2 hours ago, AncientNerd said:

Sorry this is incorrect, circuits just don't work this way. you will vary current draw but the voltage is going to stay at the set value, yet the heat will be somewhat higher but as long as it is below the specified value it is fine no matter what the % CPU usage. Electronics don't have ware and tare the way mechanical parts do, because of friction. They have issues due to heating and current draw mostly damage due to excessive heat - there is a level of heat that is non-damaging. Damage due to current draw is what you see with high voltage damage you are forcing more electrons through a part of the chip than it can handle - i.e., there is enough potential difference that more electrons can flow than the material can handle (high current) which will damage the components faster than heat but just as much.

 

So yes high voltage and high heat will damage the CPU, no just running at 100% CPU for long periods of time will not cause this. With the assumptions that 1) you have not over clocked and 2) you have the correct cooler for the TDP.

 

I totally accept that your explanation is correct on paper.  I guess I just feel like I've seen enough real-world examples of all kinds of computer parts failing to think that there's a difference between 'theory on paper' and 'the real world'.  In "theory", there's nothing wrong with stressing a circuit right up to 100% of its specification threshold for extended periods of time, but in reality... there's a reason nobody wants to buy GPU's that have been used to mine crypto.  The mass perception (which I'm inclined to believe is based in reality) is that these parts are significantly more prone to failure, having been so heavily used.

 

As with most things in life, the truth is probably somewhere in the muddy middle.

[knightslugger]

8 minutes ago, Vantage9 said:

there's a reason nobody wants to buy GPU's that have been used to mine crypto. 

probably due to the incorrect assumption that they've been "abused" somehow.

 

The reality of crypto is that for the most p[art people just plug them in an go. Overclocking will get you better results, but WHAT you overclock will depend on the algo. For instance on nVidea cards the better cyptos to work are in the equihash family, which generally benefits from core clock speed. However the gains you get from overclocking are inefficient and the Solutions per Second suffers greatly when going over 80% power max. I have both a 1070 and a 1080ti that are forced to a 70-80% power max because if i go to 100% OR 120%, while i get more sol/s, the sols/w (that's solutions per watt consumed) drops to just over 2 sols/w. That's inefficient, and for the extra gains, it's not ideal. ideal is around 3-4 sols/w. that way heat generated is low, power consumption is low (helps on the electricity bill), and it doesn't effect your payout much.

 

efficiency is the name of the game in crypto mining.