Too hot on water. Why?

[radrok]

It could peak at 90c with 1.55v and it doesn't get hotter than 60c with 1.42v/1.45v

Ambient is always 20/22c

I have a lot of radiator surface though, two 420 and a 1080 rad.

I'm pretty much at waterblock-to water heat transfer limits.

 

[reppel]

jeeezus, that's a lot of stuff. You think I could get 2-3 degrees off if I added another 240 rad?

Isn't the waterblock plus the speed of the water running the biggest factor, at least in the first minutes of running it stressed?

 

[radrok]

I skimmed through the thread, are you only cooling the CPU with that radiator?

 

[reppel]

Yeah, I thought it would be enough from the reviews, 60-70 with 3930k @4600+
I'm starting to think 3930k is much cooler then the 4930k.

Also, the amount of radiators shouldnt make a difference in the first couple minutes, right?
I mean, if I start my pc, and run prime95 right away, it jumps immediately to 85 degrees on 1,3v, I suppose my water at that point is as cold as your water, right?

 

[radrok]

What is your default voltage for stock frequencies?

Could be a very leaky sample because it isn't supposed to jump like that.

Anyway yes a 240 is plenty for a 130W CPU.

 

[the54thvoid]

Yeah, I thought it would be enough from the reviews, 60-70 with 3930k @4600+
I'm starting to think 3930k is much cooler then the 4930k.

Also, the amount of radiators shouldnt make a difference in the first couple minutes, right?
I mean, if I start my pc, and run prime95 right away, it jumps immediately to 85 degrees on 1,3v, I suppose my water at that point is as cold as your water, right?

I've remembered that I had a dual loop before my current single loop and my CPU ran off a single 240 rad. I had to downclock my 3930k speeds to 4.2 for everyday use. Ten minutes in Prime would start to push 70 degrees.
When i put the cpu into a new single loop (with two 360 rads for gpu and cpu) my temps dropped significantly and volts and clocks could be raised comfortably. I could easily push 4.5-4.6GHz with a little more juice for everyday use. Would probably only push temps to high 60's low 70's.

Also, it can make a difference quite quickly (a few mins) with a smaller loop. Your CPU heats up fast so if you run on a smaller rad loop, there is less water volume to heat - so it gets hotter quicker. On Radrok's loop, it would take a lot longer to get that volume of water equally heated. If your loop went into a bathtub of water, you wouldn't even need fans - the volume of water would cool faster than the cpu could heat it (though the water would slowly evaporate).

You need a 1200W source (kettle) to pretty much boil water in a few minutes so a 140W source of heat theoretically couldn't heat a specifically sized volume of liquid. All that said, if your temps rocket from cold past 70 in a minute or so - something is very wrong.

 

[reppel]

What is your default voltage for stock frequencies?

Could be a very leaky sample because it isn't supposed to jump like that.

Anyway yes a 240 is plenty for a 130W CPU.

I've remembered that I had a dual loop before my current single loop and my CPU ran off a single 240 rad. I had to downclock my 3930k speeds to 4.2 for everyday use. Ten minutes in Prime would start to push 70 degrees.
When i put the cpu into a new single loop (with two 360 rads for gpu and cpu) my temps dropped significantly and volts and clocks could be raised comfortably. I could easily push 4.5-4.6GHz with a little more juice for everyday use. Would probably only push temps to high 60's low 70's.

Also, it can make a difference quite quickly (a few mins) with a smaller loop. Your CPU heats up fast so if you run on a smaller rad loop, there is less water volume to heat - so it gets hotter quicker. On Radrok's loop, it would take a lot longer to get that volume of water equally heated. If your loop went into a bathtub of water, you wouldn't even need fans - the volume of water would cool faster than the cpu could heat it (though the water would slowly evaporate).

You need a 1200W source (kettle) to pretty much boil water in a few minutes so a 140W source of heat theoretically couldn't heat a specifically sized volume of liquid. All that said, if your temps rocket from cold past 70 in a minute or so - something is very wrong.

Yeah, was wondering if that is normal?
Like start the pc, let's say 4.5ghz at 1.3V, in the 40s degrees. As soon as I click start on prime95, it jumps to 80s in half a second, then it starts increasing slowly, degree by degree until stabilized. Doesn't this jumping happen to everyone?

 

[radrok]

Man that's a very leaky chip if it needs that much voltage to run stock

 

[the54thvoid]

Yeah, was wondering if that is normal?
Like start the pc, let's say 4.5ghz at 1.3V, in the 40s degrees. As soon as I click start on prime95, it jumps to 80s in half a second, then it starts increasing slowly, degree by degree until stabilized. Doesn't this jumping happen to everyone?

That doesn't seem right at all. Could the IHS be poorly placed? That's beyond my technical merits - I've never removed one but if your loop is carrying cool water over the CPU, it shouldn't jump so high unless it is (A) poorly constructed, (B) your loop isn't pushing water over it - airlock or something, or (C) it's a very poor chip

 

[MxPhenom 216]

Yeah, was wondering if that is normal?
Like start the pc, let's say 4.5ghz at 1.3V, in the 40s degrees. As soon as I click start on prime95, it jumps to 80s in half a second, then it starts increasing slowly, degree by degree until stabilized. Doesn't this jumping happen to everyone?

I think we have all said it multiple times. Every chip is different, thus everyone will have different experiences. i think you should try a different thermal paste like MX 4. The picture you showed of the IC Diamond didnt look like it spread well. But, thats just my thought. if you reset your cmos and run prime what do the temps get too?

@the54thvoid He cannot remove the IHS. It is soldered.

 

[reppel]

I think we have all said it multiple times. Every chip is different, thus everyone will have different experiences. i think you should try a different thermal paste like MX 4. The picture you showed of the IC Diamond didnt look like it spread well. But, thats just my thought. if you reset your cmos and run prime what do the temps get too?

@the54thvoid He cannot remove the IHS. It is soldered.

That picture with the temps at stock are right after doing just that. I wish I had another cooler, could see right away if it's the water system, cpu, or even both.

That doesn't seem right at all. Could the IHS be poorly placed? That's beyond my technical merits - I've never removed one but if your loop is carrying cool water over the CPU, it shouldn't jump so high unless it is (A) poorly constructed, (B) your loop isn't pushing water over it - airlock or something, or (C) it's a very poor chip

Any smart way to remove airlock, if it's that? I've done some turning and shaking but don't think it did anything.

 

[Vario]

That doesn't seem right at all. Could the IHS be poorly placed? That's beyond my technical merits - I've never removed one but if your loop is carrying cool water over the CPU, it shouldn't jump so high unless it is (A) poorly constructed, (B) your loop isn't pushing water over it - airlock or something, or (C) it's a very poor chip

I've owned 3 ivy bridge (non extreme) that were worse than that one. One 3570k required 1.4v for 4.5ghz and could never stabilize due to temperatures. The first 3770k required 1.28v for 4.3ghz and the second 3770k required 1.28v for 4.2ghz (re-rolled and got worse silicon). I delidded the first 3570k and that reduced temperatures 15-20*C but still didn't stabilize 4.5ghz and required enormous voltages.

I think the later ivy stuff sucks. My current i7 is binned and was bought used. Its an older serial number iirc. I also had a few xeons I played with. I live right next to microcenter and can sell them used on ebay for a $10 net cost or return to store ( don't like to abuse that). The xeons undervolted well, one ran 3.5ghz at only .880v! I wish it had an unlocked multiplier...

If it helps I've been running 4,8GHz at 1.42v for a little over a year with no issues but I don't know if Ivy-E is more delicate than SB-E.

This is literally 24/7 because I can't remember how many nights I left this PC rendering projects.

Benched 5,3 GHz at 1.55v several times too!

A lot of people say to run ivy at lower voltages than sandy. Not sure if true.