Not in a sealed system with appropriately formulated coolant.
The reason it's bad in regular open loops is because you have much less control over EXACTLY which metals and which grade of that metal, is used to make your loop. You can have different copper, different grades of aluminium, etc.
Asetek's coolers are all mixed-metal and work fine. I still wouldn't do it in an open loop, but there's no problem with it in AIOs.
I have to disagree here. Water loops is not a new science.... it's been "a thing" since the the first engine was built. Water loops are used in power plants, building heating and cooling and ... when water cooling was adopted for PCs, they didn't create new science .... they just adopted from there. In any such system of significant cost, building owners and power plant operators have the coolant sampled and analyzed at worse, every three months. Adjustments are made with the chemistry tweaked to provide optimum cooling as well as protection of all the system components. Having "been there and done that" as part of my job responsibilities, I can tell you it's a serious thing. When the chemistry gets out of whack, if ignored, you're talking millions of dollars in corrective measures.
These additives have a useful life of, at max, 18 to 24 months. There's no "magic formula" available to CLC builders that suspends the laws of chemistry. The useful life of corrosion inhibitors, bacteriacides, algacides and even surfactants is limited.
And yes, you have absolute control over what's in you loop when doing custom ... you get to pick the components and you get to ask questions. With CLCs the driving factor is cost ... the warranty is an insurance policy, basically a bet that most of them won't fail or the user will buy a new system before warranty expires. But that's true of every warranty. Now to the science.
https://martinsliquidlab.wordpress.com/2012/01/24/corrosion-explored/
As we can see here ....
Aluminum is 0.75 to 0.95 .... copper and brass is 0.35 and 0.40, The amount of corrosion potential is based upon the difference in those numbers. So copper and aluminum is at best 0.40 .... in an all copper / brass (brass fittings are painted) system it is 0.05. No "magic formula" will ever change that. With aluminum present, the risk factor is 8 times greater. It's the reason we change the coolant even in our cars ... and there, the presence of toxic chemicals (anti-freeze) and hence toxic to little buggies. However, glycols and other anti-freeze like products can be damaging to certain plastics (acrylics) used in reservoirs and other components. In the CLC, you are stuck with that 0.40 .... at least 8 times the rate that you can have when choosing your custom loop components. Of course,when doing a custom loop, avoid want to choose that suspiciously inexpensive radiator which may have used tin solder.... make sure it's quality silver solder. I'm not sure why one would think Swiftech has any less control than Asetek does. But Swiftech has name recognition, few know Asetek, they bought"a Corsair". So Swiftech will be more concerened with reputation than Astek would be. That's why Asus spun off AsRock originally as they wanted to sell low cost products to the "builder" market w/o damaging the rep of their brand name.
My biggest problem with CLCs however is they have no "Raison d'être" (reason for being); biggest attraction seems to be folks telling their friends "I have water cooling" . They don't really do anything better than air coolers which are a fraction of the cost and far quieter. Let's look at the $45 Scythe Fuma ....
https://tpucdn.com/reviews/Scythe/Fuma/images/temp_oc_aida64.png
It's gets beat by 5 coolers in TPUs test ... 3 of them are Swiftech units, 1 is a EK Swiftech like design and 1 is a $132 (293% of Fuma's cost) Cryorig AIO that needs to be 50% louder to get that edge. Looking at the latest comparison. The H100i pro matches the Swiftech's temps at the same price, but Swiftech has far better componentry, a reservoir, no galvanic corrosion issues, 10 times the pump flow and is expandable to add GFX card and other components .... and is 62% louder than the Swiftech. The $170 H115 is 32% louder than the Swiftech ... why ? Aluminum has much lower heat transfer capability and therefore needs higher speed fans.
Simply put, less than a handful of CLCs can themally outperform the $45 Swiftech ... spending 3 times as much for a CLC will net you 3C (well in just 1 instance) but at a prohibitive penalty of 50% noise increase. If you **need** better than that, then a OLC like the Swiftech / EK models offer far better quality components at the same price or less but more importantly, allow you to sit in the same room w/o being assaulted by noise. So the question one should consider is "What am I getting for 3 x the cost of an air cooler ? And at what cost in noise levels ? When those questions are answered, it's real hard, for me at least, to make a case for a CLC.
Arguments for CLCs would include:
- I wear headphones and don't care about noise .... Ok , but, **for me"", that's kinda like going to a football game and not caring about rain cause a) they still gonna play and b) you have a rain coat. The raincoat addresses the problem but I'd just rather it didn't rain.
- I don't want a 2 pound weight hanging off my MoBo, could damage my MoBo ... OK, but no concern about the 60 - 70 pounds of clamping force from the mounting mechanism ? ... However, if your routinely ship your PC, this is a valid issue given my experience w/ FedEx ground. My son took his PC back and forth to college in his off road vehicle's roof rack ... was no issue laid on its side but you won't get such care from commercial carriers. I must commend these carriers for the forgiving nature of their hiring policies ... it would seem simians are handling much of the stuff.
- I don't like looking at a big shiny metal thing. Well that's a personal thing, I think they look kinda cool, especially the Phanteks ones color matched to your case theme, but if that's the deciding fator, it's personal preference.
Custom loops can provide more of everything ... more cooling .... drastically lower noise levels (aka "completely inaudible") tho it's gonna cost you for that silence. Yes, obviously I have a personal bias here because of my profession (nerd alert => systems engineer) but the numbers for CLCs just don't work for me. In the rare instances they edge an air cooler, the additional noise and cost are prohibitive.
As for delidding, since Sandy Bridge the only builds we considered delidding was Ivy Bridge. On SB and everything since Ivy, we have hit the voltage wall well before the temperature wall on our overclocks. The goal of an OC to our eyes is to run all your apps and games at the highest possible settings within thermal and voltage limits that one can have a reasonable level of comfort. On production boxes, I don't like to exceed 1.4v in BIOS (1.5v momentary peaks w/ AVX present) and 80C.
Of course, if I choose to run synthetics, Im going to exceed those temp limits, but again, since Sandy bridge, I can't think of a "Raison d'être". If I was building a box as an enthusiast and doing it for fun and looking to get my name on a web site's "leader board". But as staying under those temp figures would involve lowering the OC, why have to consider whether it's worth doing that to accommodate a utility which I will never run again on that box ? .... If it's between RoG Real bench stable at 5.1 Ghz or 4.9 GHz P95 stable, I'll take RB .. especially when we have had 24 hour P95 stable OCs fail under a multitasking benchmark like RB
