Liquid Ultra

[Intervention]

Anyone else seeing these results?

 

[xvi]

If I'm not mistaken, Liquid Ultra works well on bare die. You do have to make sure it's not going to touch aluminum though. With Skylake CPUs, people are delidding their processors with temperature drops in the 10-20c range when switching to Liquid Ultra. I think it's not nearly as good when it's on the outside of a heatshield though.

 

[Intervention]

If I'm not mistaken, Liquid Ultra works well on bare die. You do have to make sure it's not going to touch aluminum though. With Skylake CPUs, people are delidding their processors with temperature drops in the 10-20c range when switching to Liquid Ultra. I think it's not nearly as good when it's on the outside of a heatshield though.

I have used it on my 4790k. It dropped my temps by 20 degrees. So yes, it works.

 

[CAPSLOCKSTUCK]

It dropped my temps by 20 degrees.

Remarkable.......any evidence?

 

[Intervention]

Remarkable.......any evidence?

Before delid and replacement of old thermal paste, during this test my temps would reach high 80's. Now, with liquid ultra, this is what it looks like. 1024M test.

 

[CAPSLOCKSTUCK]

well well, you learn something new everyday, i didnt think those chips would get close to the high 80.s without throttling



check the t case figure from Intel
http://ark.intel.com/products/80807/Intel-Core-i7-4790K-Processor-8M-Cache-up-to-4_40-GHz

 

[Bill_Bright]

I hope everyone noticed at about 23:43 he says if applied correctly, "you never have to do it again!" This is so true for really every TIM which is contrary to what many believe who claim TIM needs to be replaced periodically. It doesn't, unless the bond is broken.

i didnt think those chips would get close to the high 80.s without throttling

Note those were GPU temps in the video, not CPU. GPUs do normally run hotter.

 

[Intervention]

well well, you learn something new everyday, i didnt think those chips would get close to the high 80.s without throttling



check the t case figure from Intel
http://ark.intel.com/products/80807/Intel-Core-i7-4790K-Processor-8M-Cache-up-to-4_40-GHz

Yes, mine would throttle back to 4.0 when in the 80's. When AVX would kick in during prime95, temps would reach the 90's at which point I would stop the test. During WPrime, however, temps would get as high as 82-83.

 

[CAPSLOCKSTUCK]

ote those were GPU temps in the video, not CPU

i was referring to the HWMon snip

Nice to see you back @Bill_Bright

 

[Bill_Bright]

i was referring to the HWMon snip

Ah! Thanks for clarifying. Yeah, his comment does say "high 80s" as you note so that is good if it stayed at full clock speeds.

That said, I sure wish the processor industry (Intel, AMD, NVIDIA, Motorola, Via and all the rest) would get together and come up with an industry standard on sensor placement and terminology. And then I wish all motherboard/chips makers and all the hardware monitor software programs would agree to follow those standards too.

I note the ARK says under Package Specifications for that CPU 74.04°C for Tcase but HWMon reports the "core" temps which often are higher than Tcase. But without a direct comparison/sensor reading, it is hard to really know how close those "high 80s" really are to the throttling thresholds.

Even among Intel processors, they are inconsistent. Note with the i7-3612QM, the ARK makes no mention of Tcase or Core temps under Package Specifications but instead, shows a smoking hot 105°C for the Tjunction temp - which is measured "at the processor die" somewhere.

At any rate, a 20°C drop is significant - assuming the previous application of TIM was still intact.

 

[CAPSLOCKSTUCK]

A "before and after" would be very useful.

I suspect the OP's AIO was fitted incorrectly to start with, though i may be wrong.

 

[Intervention]

Ah! Thanks for clarifying. Yeah, his comment does say "high 80s" as you note so that is good if it stayed at full clock speeds.

That said, I sure wish the processor industry (Intel, AMD, NVIDIA, Motorola, Via and all the rest) would get together and come up with an industry standard on sensor placement and terminology. And then I wish all motherboard/chips makers and all the hardware monitor software programs would agree to follow those standards too.

I note the ARK says under Package Specifications for that CPU 74.04°C for Tcase but HWMon reports the "core" temps which often are higher than Tcase. But without a direct comparison/sensor reading, it is hard to really know how close those "high 80s" really are to the throttling thresholds.

Even among Intel processors, they are inconsistent. Note with the i7-3612QM, the ARK makes no mention of Tcase or Core temps under Package Specifications but instead, shows a smoking hot 105°C for the Tjunction temp - which is measured "at the processor die" somewhere.

At any rate, a 20°C drop is significant - assuming the previous application of TIM was still intact.

Yes, the old TIM was still intact during the tests. I bought this chip because I thought the "devils canyon" revision would have better thermal compound and thus better temps than the 4770k. But to my surprise, the temps were not so great after overclocking. And yes, 1.3 volts does not help. But that is what I had to go up to to get a stable 4.6 out of this thing. Using prime95 small ffts during avx the wattage usage would go up to 150 watts, and temps would reach the 90's, causing the cpu to throttle back down to 4.0. After delid and application of liquid ultra, my temps do not go pass 62-63. This stuff works for me atleast

 

[Intervention]

A "before and after" would be very useful.

I suspect the OP's AIO was fitted incorrectly to start with, though i may be wrong.

No, I installed and uninstalled my H110 GT several times thinking it was an issue with the back plate or how it was sitting on the IHS. But after noticing the thermal compound even spread between the IHS and pump, I was convinced mounting was not the problem. So I delided the cpu and added liquid ultra on die and IHS. Between IHS and pump I used IC diamond. Problem solved!