Swiftech's new CPU waterblock MCW5000

[Brad]

Welcome Gabe!

I think the fact it can be used without using the motherboard is great, but at the same time it doesn't use the amd socket holes. Which is something a lot of people won't like.

What about a central inlet in the middle of the block, with the two outlets where they are now?

Do you consider this your new flagship block?

[bigben2k]

Given that the socket holes are no longer a requirement, it makes sense to drop them: not all mobo manufacturers are going to keep them.

[Gabriel Rouchon]

with AMD dropping the specs, there is not much we can do. Look at the comments of Kyle [H] : everytime we receive an editor's choice or top honors for our heatsinks, we get something like "sure it's great, be beware the 4 mounting holes is disappearing, inferring "wait before you buy"...

With such comments from the press, and AMD dragging their feet to release their new specs, what choice do we have ?????

On the other hand, this block is pretty light compared to our H/S, so the socket retention latches are OK for this.

Sorry guys, I can't start a discussion on this, I do have a lot of work ..

Gabe

[Gabriel Rouchon]

forgot to respond to one of your questions, Brad: yes I definitely consider the MCW5000 our new flagship

Gabe

[Cathar]

Originally posted by bigben2k
Given that the socket holes are no longer a requirement, it makes sense to drop them: not all mobo manufacturers are going to keep them.

Yes, but socket tabs do break...

I wouldn't buy a mobo + water-cooling unless it had the through-motherboard mounts.

The 4 holes are now just optional. All top-end overclocking boards are keeping them, and yes, that even includes the Abit NF2 board which will be released with the 4 holes.

[bigben2k]

With that logic, 2 different mounts would be in order, which may make the block unecessarily complicated. The socket mount is standard.

Hey, I agree, I'm sticking to the Asus A7N8X, which has the holes. I hope that the Epox has them too.

[pHaestus]

I have used the Swiftech MCXC-370 heatsink before. It uses the same mounting, and I never had any issues. It was also probably heavier than this block. Most of the issues with socket lugs and blocks came from using a center screw to apply pressure to core. This block wouldn't have that problem.

My Innovatek block uses the lugs and has been trouble free for a good while and many remounts.

[Cathar]

It's not so much the weight of the block, but the added rotational torque provided by the water-filled tubing that hangs off the block. Of course every smart person will support their tubing properly.

The thing with socket tabs I've found is this. Maybe 19/20 times it'll be okay, but there'll always be that 1/20 case of a board where the tabs are just weaker through manufacturing variances, and you go to a LAN, hit a few bumps on the way, and "pop". It's happened to quite a few people.

Swiftech were the ones to really show the benefits of the 4 mounting holes by dropping a case off a building. Try that with a socket. Not that people go around throwing cases about, but accidents happen (slip while carrying the case, etc).

I personally view it like this. AMD have dropped the 4 mounting hole spec and left it in the hands of the mobo makers. Now the mobo makers all know that almost every man and his dog who overclocks uses the 4 holes for their cooling gear and so they'll continue to provide the holes as before on all their boards aimed at the enthusiast/overclocking crowd, and we can see that this is the case today. All serious overclocking focussed boards have the 4 mounting holes, despite the specification being dropped 4 months or so ago.

It really becomes an issue of market breadth. Making your product to sell on the widest variety of platforms, but I personally see it as targetting your market. I'd hazard a guess that 90% or more of the people who water-cool are using the 4 holes already, and if they plan to upgrade will be ensuring they buy a motherboard with the 4 holes.

I think it's just a bit of scaremongering by [H]ardOCP. Abit, Asus and Epox all know who their market is. They won't be cutting them off at the wrist. Heck, even Leadtek is releasing nForce 2 board with the 4 holes and they're specifically an enthusiast targetted brand.

When AMD release the specs for the Hammer's finally we'll all be a lot happier. Let just hope that they don't drop them again 12 months into the future and cause this crap again eh?

[BillA]

Originally posted by Cathar
. . . . Of course every smart person will support their tubing properly.
. . . .

quite a sense of humor you've got there Cathar
I like it

[DarkEdge]

Yeah but if you are Swiftech you need to have a block that will work with 99% of your palt forms. While I would never buy a block that uses the socket tabs I can certainly understand why they have chosen this route.

Cathar, are you selling your blocks personally or through a company. I kept meaning to ask you that.

[Cathar]

Originally posted by DarkEdge
Yeah but if you are Swiftech you need to have a block that will work with 99% of your palt forms. While I would never buy a block that uses the socket tabs I can certainly understand why they have chosen this route.

Oh I agree totally on that. What I object to is the attitude that some people take that no motherboard from now on will have the 4 mounting holes, when clearly that's not the case. The market just seems to have separated into budget boards without, and overclocking/enthusiast boards with, and I predict it'll stay that way until the Hammer takes over from the Socket A solutions, and in that event, all the waterblock makers can do is just offer a new mounting plate.

Cathar, are you selling your blocks personally or through a company. I kept meaning to ask you that.

Just personally at present.

[fyleow]

Originally posted by Gabriel Rouchon
superior to the MCW462-UH by at least 1.5C (all tests equipment/conditions being equal).

Gabe [/b]

Hmm the UH model has been shown to perform worse than the first generation U, so is this better than the standard U or not?

In any case I think this block is a smart move, I personally dislike the screw mounting for P4 as everytime I remove the processor I have to remove the motherboard as well.

[Mumrik]

So no one has tried this new block out yet?

I'm curious and concidering buying one...

[Voodoo Rufus]

Anyone know if the tube fittings are removable via unthreading? Or are they epoxied in place? I'd like to put some brass barbs on it instead.

[Cathar]

Originally posted by fyleow
Hmm the UH model has been shown to perform worse than the first generation U, so is this better than the standard U or not?

Taking from BillA's tests:



Let's use ~6lpm flow rates. The U is better than the UH by 0.025 C/W. Let's assume 50W real heat is making it from a CPU into the waterblock, 50 x 0.025 = 1.25C, so the U is better than the UH by 1.25C for a roughly average flow rate on a roughly average CPU.

[pHaestus]

Where this becomes confusing is what is required to push 6lpm through the -U vs. the -UH. Assuming that the HardOCP review's numbers were not just fabricated or random due to mounting variability, they showed that with Eheim pump and BIX and 1/2" tubing (can someone clarify which Eheim?) the -UH produced a slightly lower CPU temp. Certainly this could be in the margin of the test error, or it could also be true that with hobby pumps the -UH may result in lower temps when in a complete system. From a heat tranfer theory and from wb design standpoint, however, it is a poorer wb.

I would assume that you could extrapolate from the combination of pressure drop and C/W data from Bill that the -UH modified with the Swagelock fitting on the inlet would be the clear performance champion in a typical w/cer system.

[Voodoo Rufus]

That doesn't make sense to me. Why should a block with a higher ID tubing and barbs perform worse than one with a smaller tube?

[Cathar]

Originally posted by pHaestus
Where this becomes confusing is what is required to push 6lpm through the -U vs. the -UH. Assuming that the HardOCP review's numbers were not just fabricated or random due to mounting variability, they showed that with Eheim pump and BIX and 1/2" tubing (can someone clarify which Eheim?) the -UH produced a slightly lower CPU temp. Certainly this could be in the margin of the test error, or it could also be true that with hobby pumps the -UH may result in lower temps when in a complete system. From a heat tranfer theory and from wb design standpoint, however, it is a poorer wb.

I would assume that you could extrapolate from the combination of pressure drop and C/W data from Bill that the -UH modified with the Swagelock fitting on the inlet would be the clear performance champion in a typical w/cer system.

Well for arguments sake, let's pick any point on that graph as you wish. I don't see much deviated from a 0.025C/W difference even at 2lpm, or at 8lpm. We see about as low a 0.02C/W difference at 8lpm, and as high as 0.03+ at 2lpm, so between 1-1.5C.

I thought BillA had shown that the smaller ID setup was the better performer due to the higher velocity water flow impinging right above the CPU die area.

[BillA]

the performance of the Swiftech wbs (until this 5000) are dependant on the effect of coolant velocity to locally increase the convection rate over the die area
(the 'jet' reduces the boundary layer thickness)

if the flow rate is the same, or nearly so, increasing the connection size reduces the velocity
the article explains this and illustrates the effect of a smaller nozzle for the -UH

Bruce at CoolTechnica has offered to send me a 5000 (Gabe is strangely silent),
so inquiring minds should know 'how the cow et the cabbage' in a couple of weeks

and I have NO faith, or trust, in Steve's ability to design, perform, analyze the results of, or even report on a wb test program
his record is clear

the better graph,
Chart 4

[pHaestus]

Cathar: No I mean that if, in a system, the -U block resulted in a flow rate of ~2 lpm while the -UH block resulted in a flow rate of ~6 lpm, then the -UH would appear to be a superior block. It's illustrative of the difference between wb design and system design.

Voodoo: It is one of the fallacies of water cooling that big barbs and tubing means better temps. Forced convection is proportional to turbulence, and turbulence increases with water velocity. Think about the curves that Bill posted in those terms At any spot on the x axis, flow rate is the same for both blocks. NOT water velocity, however. For the -U block, the water is travelling at a higher velocity because it is going through smaller fittings and a restriction at the base of the inlet. The -UH block has big barbs and no restriction at the inlet so water velocity is decreased.

[Voodoo Rufus]

Ah, thank you for explaining. Now it makes more sense. So there's nothing wrong with having large tubing/connections as long as you put a reducer at the inlet on the UH model to get a higher velocity on the inlet. A little more bulky and probably not worth the effort, but it would work.

[BillA]

that Swagelok fitting referred to is quite different than the Ace Hardware variety
it does not have a counterbored NPT end,
rather it is solid with just a touch of a bevel at the ID
- a pretty fair nozzle

the alternative for a high flow system is to drill out the outlet of a -U

[pHaestus]

"drill out the outlet of the U"

How exactly are you modifying the U anyway Bill? Did I read somewhere that you had put a piece of 1/2" copper pipe into it? Does that entail removing the whole quick connect piece? Pliars? Drill? And how to seal it all up when done?

[BillA]

I have several of the old 462s, I call them "the king of the testing wbs"

one modded by mildhab has a JB Weld'ed compression fitting in the inlet, bored out to take a 1/2" Cu pipe
- I use it for nozzle testing < g >

to enlarge the outlet do the same, drill it out but put in a piece of Cu pipe directly
(trying to tap it and then get a seal WILL cause grief)

[bigdawginva]

Interesting mod to the 462-U. May have to try that on mine.

BTW Gabe. You may want to double-check the Option Fittings pics as they (to me) look backwards to the descriptions.