At what load level does a quad start to overload current SS phase solutions?

[Speederlander]

Title. I think a lot of the problems running quad on phase occur with the more extreme overclocks. But what about SS phase in the vcore range of 1.3v to 1.5v? When voltage is 1.5v or below the heat load would be low enough not to have to overloading existing solutions, yes? Or if not, what is the "magic number" where concern starts to mount?

[madgravity34]

i think it doesnt just depend on a given vcore, it also depends on a given clock speed the chip is running at. In most cases, the better clocking chips (run higher on lower volts) also tend to run hotter (from what I have seen) because they are, essentially, making better use of the voltage fed to them to achieve given clocks. Its because of this that each chip will have varying levels of clock/voltage where they will breach an SS's power-removing ability.

I dont know too much about the phase part of things so I will leave that to the experienced guys here, but I thought I'd offer an explanation for the chip side.

[[XC] gomeler]

There are too many variables to take into account for such a general question. If you are asking about the commercial units by Vapochill and Prometia then I can almost guarantee you'll kill those units with a quad. Custom build units tend to be tuned for higher heatloads and if you have the system built for yourself then you can request the builder to tune it. A system tuned for an AMD X2 would most likely choke under the load of a high-clocked Q6600 just like a system tuned for a Q6600 has the possibility of being slugged to death by liquid feedback on a weaker processor if it isn't properly designed correctly. There were two enormous threads about this subject, I suggest using the search function to get a more clear understanding.

[Speederlander]

gomeler;2099239']There are too many variables to take into account for such a general question.
Not so much. The idea is to get a general idea of the heat produced. If the question couldn't be answered within a reasonable degree of error a lot of the finer tuning people do to their units would be fairly useless.


If you are asking about the commercial units by Vapochill and Prometia then I can almost guarantee you'll kill those units with a quad. Custom build units tend to be tuned for higher heatloads and if you have the system built for yourself then you can request the builder to tune it. A system tuned for an AMD X2 would most likely choke under the load of a high-clocked Q6600 just like a system tuned for a Q6600 has the possibility of being slugged to death by liquid feedback on a weaker processor if it isn't properly designed correctly. There were two enormous threads about this subject, I suggest using the search function to get a more clear understanding.
I'm not looking for phase 101. ;) I say again, I am talking about lower voltages, not the massive heat producing levels from extreme overclocks. A QX6700 running prime load at 1.3v is not going to overwhelm a normal SS. I pull 60 C or so with a Tuniq at load with this voltage. I pulled higher temps with my 6600 for its 24/7 overclock under load (when I had 1.6 volts). So, I am trying to get an understanding of where this cut-off starts to exhibit. If someone can do a 24/7 quad machine getting say 3.7 or 3.8 GHz @1.4 vcore w/o all the issues associated with a benching machine that shoots for 4.2 or 4.3GHz or more, you would have dealt with several concerns. You don't need the louder system. You don't have to worry about reduced lifespan or damage to your phase system as long as you stay below the ballpark voltage safety level. And you can get a nice 24/7 overclock that is otherwise not obtainable through other methods.

[TopherTony]

Okay. The simple use of an accumulator and slhx will most likely solve your problems. The accumulator with prevent liquid slugback that would kill your comp. And the slhx will give you better temps. I imagine that a stock prometia or otherwise will benefit from one of these and be able to manage your 1.5V+ system on a 24/7 system. Its a careful balance that will determine your system's stability.

Read this thread: http://www.xtremesystems.org/forums/showthread.php?t=135385

PM NoL, wdrzal, xeon, or any of the phase kings. They will tell you straight up how to get your system quad stable at high overclock.

[Speederlander]

Okay. The simple use of an accumulator and slhx will most likely solve your problems. The accumulator with prevent liquid slugback that would kill your comp. And the slhx will give you better temps. I imagine that a stock prometia or otherwise will benefit from one of these and be able to manage your 1.5V+ system on a 24/7 system. Its a careful balance that will determine your system's stability.

Read this thread: http://www.xtremesystems.org/forums/showthread.php?t=135385

PM NoL, wdrzal, xeon, or any of the phase kings. They will tell you straight up how to get your system quad stable at high overclock.

Ack. It's exactly not a 1.5v+ system. I am asking the question for lower voltage systems (like 1.4v or even slightly less). Do the typical concerns raised over quads come into play at low volts? The heat output is dramatically reduced. What is a rule-of-thumb voltage cut-off before you have to start worrying about the typical quad issues and ensuing mods?

[n00b 0f l337]

Whenever your wattage exceeds about 190, its time to upgrade the unit.
How you manage to see when your chip puts out 190, thats a good luck to you affair, because honestly, we cant calculate it.
As for an accumulator, I need to test a proper and small one, but I think actually moving to smaller and quieter rotarys will be a be all possibly end all solution to this problem.

[harvshark]

Ack. It's exactly not a 1.5v+ system. I am asking the question for lower voltage systems (like 1.4v or even slightly less). Do the typical concerns raised over quads come into play at low volts? The heat output is dramatically reduced. What is a rule-of-thumb voltage cut-off before you have to start worrying about the typical quad issues and ensuing mods?

If you don't wanna worry about modding a phase for a quad, don't do it. That simple.

At least that's my experience. I am running my quad right now at 55c, 3.4ghz, 1.42vcore (probe II reading) on air. If you are gonna go for more than then you are gonna need a phase unit tuned for around 275 watts or more. If you don't wanna do that, what's the use of going to phase?? Just to say you are phase cooling?
Please don't flame me for this post, 'cuz I am speaking from experience. If you need to phase, you need to be xtreme. Otherwise, you needn't bother.
It's a rough ride, depending on where you really wanna go. Phase will get you there, but you gotta be willing to pay the ferry man.

[Speederlander]

If you don't wanna worry about modding a phase for a quad, don't do it. That simple.
Edit: Ok, i should have worded that last one better. It could be misconstrued. "Worried" in the theoretical sense, not the literal. ;)


At least that's my experience. I am running my quad right now at 55c, 3.4ghz, 1.42vcore (probe II reading) on air. If you are gonna go for more than then you are gonna need a phase unit tuned for around 275 watts or more. If you don't wanna do that, what's the use of going to phase?? Just to say you are phase cooling?
Please don't flame me for this post, 'cuz I am speaking from experience. If you need to phase, you need to be xtreme. Otherwise, you needn't bother.
It's a rough ride, depending on where you really wanna go. Phase will get you there, but you gotta be willing to pay the ferry man.
I am asking a theoretical question with practical implications. I am not trying to solicit a lecture on "being xtreme". If I can get 3.6 GHz with 1.375 vcore on my Tuniq, what can I get with the same voltage on phase? Probably into the 4's, all with low volts and much lower heatload than what is experienced at benching levels. Could I even notch down my voltage a bit while taking the OC to the 3.8 GHz range? Thus, I am trying to explore this issue because I am CURIOUS.

[Gray Mole]

It IS actually a fair question.

While commercial AND custom units tuned for 200w +/- a bit may not be able to push quads to the limit, they may still be able to push them further than air or water.

From what I have seen firsthand though, even at lower vcore the quads pull hard. The only way to know for sure how far you can go with a cpu on a cooler is to push it and then prime it.

The real problem though, is if you run the cooler on the edge for too long, it can burn itself out even if it's still working. Generally speaking I like to see a tolerance of at least 5-10% to make sure there's something left to cool the compressor.

I've already had my views on Quad tuning made clear, but there's 4 easy steps.

1. Ssilencer/@itor evap. makes high load tuning easy.
2. Accumulator. Can be substituted for a significantly larger run of pipework.
3. SLHX. Not necessarily in the traditional sense, but increased heat exchange on the suction return makes a huge difference to tuning for higher loads.
4. Greater condensor capacity.

The units I've tuned for Quad core capacity, around 250-260w, have worked. There's no 'secret' to it, it just has to be built keeping these ideas in mind.

The last one I built for high load, was made with one of Kayl's heads. Nice head, and while it's not as easy to tune for higher loads it still did the job very well. It just means that you need to compensate for a little more refrigerant flooding of the evap, by ensuring that much more volume and heat exchange are happening in the suction return.

Anyways, I'm not going to get into it again, just wanted to cover that for those who are still a bit mystified by high load tuning on SS. It doesn't have to be difficult, it also doesn't have to mean crazy rotary compressors and such, it just means the pipework on SS made for high capacity have to be better engineered for the purpose.

Cheers

Gray

[bazx]

dont no about ss but with a cascade

around 1.75v@4.5ghz

i can hold a quad and bench it @ 4.5ghz with 1.7v (32m or cpc)

just leaving the mhz @ 4500 and upping the vcore to 1.8/5 crushes my cascade

so i would say vcore is the key to the quad oc

you need to find the lowest vcore to mhz balance to be a winner here

[wdrzal]

Title. I think a lot of the problems running quad on phase occur with the more extreme overclocks. But what about SS phase in the vcore range of 1.3v to 1.5v? When voltage is 1.5v or below the heat load would be low enough not to have to overloading existing solutions, yes? Or if not, what is the "magic number" where concern starts to mount?

How can someone give you a "magic number" when no one seems to really know there stock output and I don't follow oc'ing closely and you don't know what your rig is built and tuned for.

BTW in refrigeration there usually not to many "magic numbers" as its a game of give and take.The only "magic numbers" is possible the points ,like the triple point of a substance,critical pressure and critical temperature.You may want to add the numbers on a pressure/temperature chart.

Their are quite a few thermodynamic laws and gas laws that govern the known boundaries as we understand them today.

[Gray Mole]

Nice one Baz, your cascades are just getting better and better.

The big help of a cascade over SS, funny enough, is space.

Trying to stuff a bunch of extra pipework into a small case is a real headache. The use of fullsize slhx, room for decent size suction return, oversize condensors, etc can make cascades a lot more versitile especially where higher load is concerned.

The fact that your cascade can hold a good 300w really shows what you've achieved though :toast:

Balancing minimal volts with maximum overclock has always been the key to getting the best out of a chip, and when using a cooling system that's not able to deal with the full output of a chip, finding the best you can get with what you already have can be the only way to put off an expensive upgrade.

However, I've seen posts about both mach2 systems and VapoLS systems 'burning out' over time even though they were still operating, which concerns me about running the system too close to the mark in terms of capacity. I suppose a basic regass can help, but altering the entire system to deal with the load is the best bet.

When it comes to using a cooler on a cpu capable of producing more heatload than the cooler can work with, there's really only one way to be sure about not killing it.

1. Test it to max capacity/failure of cooling
2. Lower the vcore and the overclock by 5-10%
3. Retest the system, under load
4. Calculate the results and estimate for 10% minimum tolerance
5. Increase/Decrease your overclock and voltage until an acceptable tolerance is achieved

This way, you will have it covered, with enough refrigerant moving through the system to keep the oil moving well, and the compressor reasonably cooled, ensuring a reasonable lifespan for the cooling system used.

Cheers

Gray

[Speederlander]

How can some give you a "magic number" when know seems to really know there stock output and I don't follow oc'ing closely and you don't know what your rig is built and tuned for.

BTW in refrigeration there usually not to many "magic number" as its a game of give and take.The only "magic numbers" is possible the points ,like the triple point of a substance,critical pressure and critical temperature.You may want to add the numbers on a pressure/temperature chart.

Their are quite a few thermodynamic laws and gas laws that govern the known boundaries as we understand them today.


But if you read what I said, the "magic number" is nothing more than a general idea of where things start to get ugly, nothing more than a general guideline meant to inform (both potential buyers and potential builders). Honestly, since quads can run the whole gamut from voltages in the 1.3's to short stints in the 1.7+ range, any one of which could yield satisfactory overclocks for their owner, it seems that a general sense of where a person will be running their chip along with a very general idea of the relation between high-load vcore and heat load is necessary to properly design and tune that person's machine, yes? I mean, the difference in output between a quad running 1.35v and a quad running 1.65v is massive, to say the least. I would think that the individual building a phase system for that person would want to take this into account.

I dunno, maybe I'm just totally and completely off base here. But I seem to recall all kinds of speculation about what quads were actually putting out a couple months ago. I have to believe that 1) such discussions weren't without purpose and 2) quad output under full load (Prime95 as the gold standard?) is a function of vcore more than anything else and as such, the final top-end vcore a machine will hit full load at for reasonably long periods of time is critical to the key design considerations and proper tuning of the machine.