Does anyone else dislike water cooling and overclocking?

[Damascus]

7 minutes ago, emosun said:

how does it hurt an electrical component to run several times above it's specification for an extended period of time?

gee i dont know what a mystery. 

I'm amazed you don't see the correlation between a mining card running at 100% it's speed constantly vs a gaming card running at 140% it's speed for a few hours a day. amazing

Quick question for you, and if you can answer it seriously and convince me you are correct I'll admit you are right, tagging every person in this thread.

 

If my house OC of 1600mhz to 1980mhz is running it "way outside of spec" is a GPU (same unit, let's say both are 1070s with similar coolers) that run 1900mhz OUT OF THE BOX is that running it way out of spec and damaging to its lifespan?

 

 

[sazrocks]

2 minutes ago, emosun said:

how does it hurt an electrical component to run several times above it's specification for an extended period of time?

gee i dont know what a mystery. 

I don't appreciate the tone.

 

The specification is determined by the lowest common denominator, which an individual card is extremely unlikely to be at at stock. Also I honestly don't understand where you got "several times above it's specification." No one is running cards at 6GHz, or 4V.

3 minutes ago, emosun said:

I'm amazed you don't see the correlation between a mining card running at 100% it's speed constantly vs a gaming card running at 140% it's speed for a few hours a day. amazing

Let's do the math:

100% 24/7 for a year equals 8760 effective work hours per year.

120% (usually the max for power target) 4 hours a day average for a year equals 1752 effective work hours per year.

Using those numbers, the mining card is doing 5x the work of the normal overclocked gaming card. They are not comparable.

[PETRGangKing]

5 minutes ago, emosun said:

oh gee let me use a much more scientific argument like.....

much in the same way a light bulb is only 80% of it's brightness...... just bump up the wall voltage and watch it get brighter......

and everyone knows the gpu works best when its right on the edge of crashing.

 

ideally it should be pushed so hard that if it was pushed a single % more it would crash. nothing wrong with that. SUPER good for longevity. 

 

no , i agree with you. A+ methodology.

Why so snarky?

 

But realistically, running something just below it's theoretical limit, doesn't pose any significant risk to damaging it. When it comes to thermals, running closer to the ceiling can pose a risk, as you never know when another factor will come into okay and your temps will get too hot. When it comes simply to the silicon potential, it doesn't pose a risk, because no matter what, your chip is not going to bump to a higher voltage or clock speed than you have showed in the bios.

[emosun]

2 minutes ago, Damascus said:

Quick question for you, and if you can answer it seriously and convince me you are correct I'll admit you are right, tagging every person in this thread.

 

If my house OC of 1600mhz to 1980mhz is running it "way outside of spec" is a GPU (same unit, let's say both are 1070s with similar coolers) that run 1900mhz OUT OF THE BOX is that running it way out of spec and damaging to its lifespan?

well the question then is , are they both using the exact same pcb and is the pcb the same revision

[Damascus]

5 minutes ago, emosun said:

well the question then is , are they both using the exact same pcb and is the pcb the same revision

Yes and yes. Asus, Evga and Zotac do it all the time.

 

also, here's a kicker, the reference models from nvidia will self OC up to 2000mhz if you cool them better.  You do nothing but put on a better heat sink and it up clocks itself

[emosun]

4 minutes ago, sazrocks said:

Using those numbers, the mining card is doing 5x the work of the normal overclocked gaming card. They are not comparable.

that's provided the damage to both cards is linear , you assume that 120% would only damage the card 20% more when for all you and I know 120% it's capacity is twice as bad. with the amount of gpu's getting fried from OC vs stock clocks , it's pretty obviously not linear at all. So toss that one right in the can

[sazrocks]

1 minute ago, emosun said:

 for all you and I know 120% it's capacity is twice as bad. 

Appeal to the unknown.

1 minute ago, emosun said:

with the amount of gpu's getting fried from OC vs stock clocks

I have not heard of such an epidemic, do you have a source?

[PETRGangKing]

Just now, emosun said:

that's provided the damage to both cards is linear , you assume that 120% would only damage the card 20% more when for all you and I know 120% it's capacity is twice as bad. with the amount of gpu's getting fried from OC vs stock clocks , it's pretty obviously not linear at all. So toss that one right in the can

What?! That's not true at all. I run my card at 134% and because of my water cooling loop, my thermals are a little more than half what most graphics cards thermals generally are running stock. And the majority of my overclock was done automatically, by the graphics card with no input from me. I only added a +50mhz offset to the overclock it was doing completely on its own.

[emosun]

Just now, Damascus said:

Asus, Evga and Zotac do it all the time

lol every manufacturer makes it own pcb they don't all come from the pcb fairy.

maybe google what a pcb is , then google what a gpu is

[Damascus]

3 minutes ago, emosun said:

lol every manufacturer makes it own pcb they don't all come from the pcb fairy.

maybe google what a pcb is , then google what a gpu is

I'm quite aware of what a pcb is... You still refuse to answer my question and I consider you to be arguing in bad faith, consistently using logical fallacies.

 

Have a good one

[sazrocks]

1 minute ago, emosun said:

lol every manufacturer makes it own pcb they don't all come from the pcb fairy.

maybe google what a pcb is , then google what a gpu is

Strawman. Damascus was not saying that all those AIB partners use the same boards, rather each of them use the same bard multiple times within their own product stack.

 

And as for your second part, what exactly does a gpu itself dying have to do with the PCB?