BonzaiDuck
Lifer
- Jun 30, 2004
- 15,785
- 1,500
- 126
Probably a useful exercise, if you thought it was worth it.
I would think 1.44V is abusively excessive for those cores. It's a 22nm lithography, so you could risk an educated guess that the limit could be the pervasive opinion about the Ivy Bridge cores. People were speculating that it was around 1.30V, and that sounds reasonable to me.
I'd have to see firsthand how voltage and temperatures vary.
From your project, I conclude that a single-fan AiO is sufficient to cool the 4790K @4.6 to keep the core temperatures at or below the TCASE spec -- for what the spec might be worth. Some folks think that the arcane spec is irrelevant to anything, but it's a useful comparison.
It also seems obvious -- you say so -- that squeezing another 200 Mhz isn't worth the trouble except to just say "you can."
The only thing I don't like is the AiO configuration in general with a single pump. If the computer shuts down at the throttling temperature limit, I suppose that a failed pump wouldn't kill the whole monster.
Somebody else noted that the quality of AiO pumps had quickly improved from the time of their introduction. But I'm also wondering what clock-speed and thermal wattage would be under stress load by putting my ACX cooler on a 4790K.
It certainly proves that you can keep Devils Canyon pretty chill under stock speeds that push the limits of the stock cooler.
With a Sandy Bridge and the stock Intel cooler, the lead core temperature exceeds 87C when just overclocking to 4.2Ghz on a BIOS "auto" voltage setting. I'm at about 74C at 4.8 Ghz. I don't like the voltage at 4.8, and 4.7 runs as cool as your system at 4.6. Obviously, the voltages are higher on the SB-K chip.
I'm stunned at the thermal wattage you reported for the 4790K. I guess it proves my point about the Haswell die-shrink, if it's a point that needs proving.
I would think 1.44V is abusively excessive for those cores. It's a 22nm lithography, so you could risk an educated guess that the limit could be the pervasive opinion about the Ivy Bridge cores. People were speculating that it was around 1.30V, and that sounds reasonable to me.
I'd have to see firsthand how voltage and temperatures vary.
From your project, I conclude that a single-fan AiO is sufficient to cool the 4790K @4.6 to keep the core temperatures at or below the TCASE spec -- for what the spec might be worth. Some folks think that the arcane spec is irrelevant to anything, but it's a useful comparison.
It also seems obvious -- you say so -- that squeezing another 200 Mhz isn't worth the trouble except to just say "you can."
The only thing I don't like is the AiO configuration in general with a single pump. If the computer shuts down at the throttling temperature limit, I suppose that a failed pump wouldn't kill the whole monster.
Somebody else noted that the quality of AiO pumps had quickly improved from the time of their introduction. But I'm also wondering what clock-speed and thermal wattage would be under stress load by putting my ACX cooler on a 4790K.
It certainly proves that you can keep Devils Canyon pretty chill under stock speeds that push the limits of the stock cooler.
With a Sandy Bridge and the stock Intel cooler, the lead core temperature exceeds 87C when just overclocking to 4.2Ghz on a BIOS "auto" voltage setting. I'm at about 74C at 4.8 Ghz. I don't like the voltage at 4.8, and 4.7 runs as cool as your system at 4.6. Obviously, the voltages are higher on the SB-K chip.
I'm stunned at the thermal wattage you reported for the 4790K. I guess it proves my point about the Haswell die-shrink, if it's a point that needs proving.
Last edited: