Swiftech's new waterblock test results.
 

I recently got the Swiftech MCW 462 U and running my rig with it now.

Epox 8KHA+
Athlon XP 1800+

Right now its at full load with 33 C, running Seti. CPU is O/Ced to 1.7 ghz with 1.85 V Core. Running Seti without O/C and default voltage temps dont go above 30 C. Ambient Temperature is 22 C.

I am using the Perma Cool radiator (you can find a pic at crazypc.com) and the Eheim 1048 pump. Tubing setup was ahh interesting considering one part of the pump is 1/2 ID, waterblock 1/4 ID, rad 3/8 ID, and the other part of the pump 3/8 ID.

Radiator is being cooled by a YS-Tech 120 X 38 fan, which is suppose to put out 130 CFM. Its running on a volt modded line though (overclockers.com has an article on how to mod the wiring), so its less airflow (and quieter).

Temp readings arent accurate since they are from MBM and not a temp probe.

I know this isnt a professional review or anything but its just supposed to give you an idea :)

Only gripe i have about this block is the quick connect fitings for the tubing. If you dont push hard enough its gonna leak, big time.

So far so good.

Re: Swiftech's new waterblock test results.
 

Sounds good. Don't rule out MBM temps all together. The Epox boards seem to do pretty good telling the temp. I did a test on the Epox 8K7A on my site. I posted it here a long time ago. I milled a channel in my water block and put a probe touching the core itself and tested it. It was 1C warmer than the onboard probe with MBM as the monitor. One big thing to watch out for though is airflow behind the mobo, that will give you almost room temp reading!!!

Good work on the setup!!!

sounds like a fairly good result to me, although a bit too optimistic I think, looking at results elsewhere with other blocks, you'd expect 38 - 40C

Epox 8KHA+ temp sensor isn't touching cpu at all, so you really can't trust those temp. I have the same board and I have Senfu sensor on touching cpu core, epox sensor is showing few degrees less at full load and few more at idle, and somethimes it's showing right, so you can just speculate what your real temp. are.

Yes, it is set up exactly like the 8K7A, not touching the bottom. The onboard dosn't change temps as fast as my external did but they where very close. One thing that will really throw it off is airflow behind the mobo.

Air flow behind the motherboard is non existant. I dont have any case fans, except the YS-Tech that cools the radiator. I have it blowing directly at the waterblock but thats all.

external probe on the side of CPU and the onboard thermal probe are both going to read low (unless you use an Asus then it may either be low or high depending on your cooling quality. If you are using a TBird to test blocks then I suggest you follow Joe's guide to epoxying a thermocouple under the center of the core. AMD has a technical pdf with an equation to relate that temp reading back to internal CPU temp:

Tdie = (1.209 * (Tceramic-Tamb)-1.3778) + Tamb

http://www.amd.com/us-en/assets/cont...docs/24228.pdf

Other than diode or the under socket thermocouple I don't think there is a good way to know CPU temp personally.

Mine was NOT on the side of the CPU, it was on the side of the CORE of the CPU.

http://www.custom-cooling.com/after.JPG
http://www.custom-cooling.com/probe1.JPG
Shows where the thermal compound of the core contacting the WB.

Everyones point exactly!!

yeah we really can't pay attention to any of the results given until we can get our hands on the new boards coming out w/ the circuit to read the thermal diode. Reading the bottom of an Athlon XP is even less accurate than reading the bottom of an Athlon Regular due to the circuitry on the back of the chip.
-Kev

or you can tap into the smbus and do that mod. pHaestus has done it on quite a few boards now, he'll be able to help you

Err I am selling this combo soon and getting a P4 1.8 Northwood. Ill post temp results from that, should be more accurate since it will be from the internal diode right?

yes, they will be from the internal diode

have you decided on a mobo yet? I like the abit, asus, epox and soyo sis645 boards, and for 845D stuff, the abit bd7 is really nice

Leaning heavily on the Abit BD7 right now. Its got nice Vcore (up to 2.2) and its resonably priced. Of course that is too much voltage, but it might be nice for that one time high overclock for benchmarking. I dont know though, some people say avoid the 845 chipset like the black plague, but they migth be reffering to the SDRAM versions....

645 seems to win at some test reports and is cheaper, but its DDR is running at 333 so its not really a fair comparison.

I don't like when people personally recommend boards that they've never even touched...

neither do i

huh

Why buy a whole new motherboard and CPU to get internal temps? Roll your own XP diode reader, plug it into the SMBus header and solder it to 2 pins on the socket backside and all is well. On my current Epox motherboard, I don't even have a SMBus header so I soldered onto 2 pins of a DIMM slot to get SMBus capability :)

Actually the whole reason im changing the waterblock was because of switching systems. My current setup will be the same speed as a 1.8ghz Northwood O/Ced. Dont really know why I am doing this lol :D

Just wanted a change i suppose, but i know i will be kicking myself when the t-bred are released.

I was really scared doing the Vmod on my epox, dont think i can handle that.

It's only scary once; if you have the skill to make the diode reader then a vmod is nothing :)

the 845 is shit when running with pc133, but with ddr it is quite good. The sis645 is a little slower when running at 266, but when at 333 it is quite a bit infront

Hey pHaestus could you link to your tutorial on hooking up that internal diode?


Thanks