Greg,, Thanks for the thoughtful and informative post. Being a complete newbie to water cooling, I went with the Swiftech H20-220-APEX Ultima CPU Liquid cooling kit with Apogee XT water block. It includes the MCRES-Micro Revision2 reservoir, MCP655-B pump and MCR220 dual 120mm radiator. I am running a core i7 920 (oc'd to 3.0 GHz) on an ASUS P6T6 WS motherboard.
Your post confirms what I was thinking about taking the assembly apart to check the TIM. Your results are more what I was expecting (but jealous of your 6 cores!). Thanks for all the suggestions. Jim > -----Original Message----- > From: Greg Sevart > Sent: Saturday, July 09, 2011 11:36 AM > To: [email protected] > Subject: Re: [H] Water Cooling and ambient temperature > > What processor and what water components (kit, or each component)? > > I've been watercooled for several years now and can't imagine going back. > I'm running a full custom loop with mostly Swiftech parts, including the > MCP655 variable pump, Apogee XT block, and MCR320 radiator. Ambient > room temp is about 20C, and my Core i7 970 (32nm 6-core) at stock speeds > runs about 22c idle, 42c load. When I'm done overclocking, I expect to be > running in the mid 70's under LinX load. > > > I think something is definitely wrong, and my first bet would be a bad > mount/TIM application. It happens, especially with water blocks. They > typically have bulky bolt-though mounting mechanisms that sometimes don't > line up perfectly flush on the processor. Further, the lack of a fan and its > associated vibration means that the TIM often doesn't spread as well--or as > quickly--as it would with an air cooler. I would consider using a different TIM- > -I use Shin Etsu X23-7783D for my main system, and keep some Arctic Cooling > MX-4 for bulk. Either is going to do better than AS5 or Ceramique. > At the very least, you should pull off the block and look at the spread of the > TIM to see if you should try adding more or using less with a second mount > attempt. > > Depending on what CPU you have though, while 80C is hotter than I'd like, it > isn't likely to make it explode. Most recent Intel chips (back to the 4C 45nm > Bloomfield Ci7's) have a Tjmax of around 100C. Further, they will self-throttle > when they get too hot.
