Too big for SS?

[[XC] gomeler]

I'm going to have to do a little research into this to see how much power Core i7 is actually pulling. I have a Core i7 920 but it's limited to ~4100MHz so I can't push it very hard under my single-stage systems. I find my Phenom II to run hotter as it requires 1.6v for 4200MHz(it's a bum chip ).

[sdumper]

looks more and more like 10 feet of .042 to me chris....

[TheKarmakazi]

Well then 230W ain't nearly enough. I don't know a lot about phase building, but I do know my unit is tuned for 280W and I can collapse it rather easily with the i7 running prime at very high speeds and >1,5V Vcore.

Yeah a friend of mine has a unit tuned for 250w and it cant hold i7 any higher than 4.5/4.6ghz for more than a few minutes.

[sdumper]

Shipping to anywhere in the US is very cheap:
Priority mail large flat rate box shipping under $14.00

http://postcalc.usps.gov/MailService...lingTime=17:00

[aenigma]

All this CPU wattage talk reminds me of when I was doing all of this back in '04 and everybody was talking about CPU wattage being over 200 watts, when it was closer to 100 watts. The only thing I can be sure of is that nobody knows what they are talking about, it's all speculation.

[jcool]

All this CPU wattage talk reminds me of when I was doing all of this back in '04 and everybody was talking about CPU wattage being over 200 watts, when it was closer to 100 watts. The only thing I can be sure of is that nobody knows what they are talking about, it's all speculation.

Well, not quite.. back to the Q6600, I used to have Abit boards and they could display the current the CPU was drawing. So using the simple P=U*I formula you could easily get a pretty good idea of what your CPU was consuming at any given time.
Right now I am running an i7 920 at 4,4Ghz with 1,52V, HT enabled. Running BOINC it pushes my SS down to -26C evap temp at 21C ambient, which means I am effing close to the 280W barrier.
Probably 250-260W right now.

[rge]

I was just looking at SS/phase cooling, getting tired of only being able to bench when it gets really cold outside. And even that is limited. But with core i7, based on Jcool's post not sure if I could bench much at 4.6ghz with 1.6+ vcore needed and keep temps still low, unless using a monster cooler. Kind of wanted something that was relatively quiet and cools 350W

Regarding TDP in watts, since turbo engaging/disengaging is dependent on accurate TDP information from core i7, it can be read. Everest is one program that reads it. On stock settings TDP at full prime load is around 76 watts on mine. But overclocking to 1.55vcore, at full load TDP is 263W (pic). I did this test with linx a while back to illustrated this F4beta bios has disabled turbo TDP limits, versus the F3 bios (watching everest) turbo kicks out at around 110-120W TDP.

[jcool]

Poor CPU
Neat idea with everest tho, gonna see what mine says to that. Figure sounds really plausible to me.

[sdumper]

Very hot chips but tuning to 350 watts on a SS is not very realistic unless you dont mind the sound of a lawn mower in your house ...

[rge]

Yeah kind of what I figured regarding SS. Also, I looked at the 180W vapochill unit, fairly quiet per sound testing on website, would have considered that, but not going to touch my i7 for benching. Need to do a lot more reading to learn possibilities for cooling it and then if I am willing to do any.

[sjg0]

Very hot chips but tuning to 350 watts on a SS is not very realistic unless you dont mind the sound of a lawn mower in your house ...

Bose noise canceling headphones ftw.

[[XC] gomeler]

It all depends on the load and the temperatures. With a 1/8hp compressor you can hold 350w but the evap might be at 15 Celsius. You can size the capillary tubing on a 1/3hp compressor to hold 300w also, but the problem occurs at idle where it may drop to what.. 50w? Without either maintaining a constant load, having a suctionline heat exchanger and accumulator, or a TXV you'll have massive floodback as the compressor will more or less pump a specific amount of refrigerant. This is a huge problem I ran into and which caused me to stop building single-stages, meeting CPU demands while building within case constraints, and achieving temperatures desired. I've only used a TXV on a chiller that always had a constant load and never suffered from floodback but I imagine it would be a way to reduce the issues caused by high-capacity low-displacement builds.

[sjg0]

It all depends on the load and the temperatures. With a 1/8hp compressor you can hold 350w but the evap might be at 15 Celsius.

Uhh.... 1hp = ~750W
so
1/8 x 750 = ~95W

A compressor that consumes 95W can remove 350W of heat?
Conservation of energy comes to mind......

[[XC] gomeler]

Yes, at say.. 15 Celsius with R134a. The movement of heat isn't a 1:1 ratio, it is a function of the gas used and compressor efficiency. In general, the lower the compression ratio and the higher the massflow(higher suction pressure resulting) the greater the heat handling ability but at the expense of evap temps. This is also why autocascades are often much better options than single stage systems. Autocascades achieve low discharge pressures, high suction pressures, excellent compressor cooling and efficiency when applied in similar instances to single-stages. There's a lot more than a simple power-in/power-out system

[sjg0]

Yes, at say.. 15 Celsius with R134a. The movement of heat isn't a 1:1 ratio, it is a function of the gas used and compressor efficiency. In general, the lower the compression ratio and the higher the massflow(higher suction pressure resulting) the greater the heat handling ability but at the expense of evap temps. This is also why autocascades are often much better options than single stage systems. Autocascades achieve low discharge pressures, high suction pressures, excellent compressor cooling and efficiency when applied in similar instances to single-stages. There's a lot more than a simple power-in/power-out system

Idk. I just find it extremely hard to believe a 1/8hp compressor will hold 350W load. Especially considering that is straining the compressor a lot, so it is generating lots more heat and dumping that into the system also.....
Yes movement of heat is never 1:1 there is always some heat lost which is an inefficiency that works against your argument.
Unless maybe is 134a a super easy gas to compress? nm. I think i get what you are saying. You would need a big ass condenser for that no?

[[XC] gomeler]

http://rc.danfoss.com/TechnicalInfo/...2_Cf52e222.pdf
TF4CLX, 4CC Danfoss compressor.

Holds 350w @ -12.2C under the ASHRAE LBP specifications (54.4C condensing temp, 32C suction gas temp, 32C ambient, 32C liquid temp). Consumes 239w, yielding a COP of 1.46.

That's with a 4cc compressor, tiny compared to what we are normally dealing with(9cc to 12cc).

edit: this was roughly a 1/4hp compressor but it's the lowest powered compressor I've ever used so I used it to illustrate my point

[quintus]

that 350w is not like our processor and that 350 might be only on papers unles there is a danfoss guy around here to spread some light on it.

@[XC] gomeler: lets say 25 or more not just 15

[[XC] gomeler]

The ASHRAE environment settings are actually harsher than what we run. Who runs a 54.4C condensing temperature, 32C suction gas temp, and doesn't sub-cool the refrigerant by wrapping it around the suction line? I'd say the values in the LBP spec are pretty close to correct, they are official documentation after all. Unless Danfoss fluffs up their capacities, I'd take them as true. A watt is a watt, a watt generated from silicon isn't any different than a watt generated from any other source. The only difference could be watt density, in which case the CPU has one of the highest densities known to man.

[sdumper]

Holding a evap at 15c doesnt really get you much....why even bother with the hassle of an SS for 15c evap and likely 30c cpu core.

[[XC] gomeler]

My sole point was that moving of heat isn't a 1:1 process. However, you don't necessarily need a -45C evap for ideal performance on processors. Intel's 45nm process yields chips that don't scale really well with cold. Just keeping the processor at 0 Celsius is a huge jump from the 60C we normally run them at. Fighting for another 30 Celsius won't yield you much as shown with Core i7. I however like colder temps as I tend to blast voltage through my chips so don't take this statement as me stating we should all strive for 0 Celsius evaps and such. I was just trying to show quintus and sjg0 the capacity in the compressors we use.

[quintus]

What you find on those pdf files doesn't happen with our system, it's never like it is on paper If i remember beter there is 54C for r404a but also for 134a compressor.

This is weird, sounds like paper only calculations to me.

Plus you are saying 15c, i will complete a bit by saying 15C evap but the cpu 45C or higher?? wtf!!!

[[XC] gomeler]

Egh, you are failing to look past the specific example I gave *facepalm* Let me try to be a bit more clear.

I specifically mentioned the 1/8hp compressor illustrate the point that amount of heat moved isn't just a function of compressor size but pressure ratios and gas choice factor in also. I'll meet you halfway and agree that the PDFs are more ideal than our cramped systems can achieve, but you can definitely get very close with a well engineered system.

[sdumper]

Gotcha...

[sjg0]

Well gomeler, after careful review of your specific example and convincing argument, I now see what you are saying. I believe you now, but its... idk.... this is actually quite difficult to explain, its like I understand it but I don't.

I hate not knowing how/why something works, but sometimes I have to just sit back and accept it is magic.

[[XC] gomeler]

I'd suggest just coming back to this in a day or two. I often see things much more clearly when I don't concentrate on them. I learned this with computer science classes, a professor would post some code that was ungodly simple and I'd fail to understand it as I was looking at it the wrong way.