Hi, guys.
I need your help debugging an issue that I think is heat related. Given that reading some of the excellent posts here is what drove me to upgrade in the first place, I think it's only fair to give you guys a crack at analyzing this.
Please forgive the wall of text; I'm trying to include as much info as possible to help debug this.
First, some history. This PC originated as a Core 2 Duo E8400, which I had overclocked to about 3.3GHz. About a year ago, I upgraded it to a Q9550. I was able to push it to 3.2GHz, but I wasn't happy with the temperatures. They approached 75C at load, and I was hoping to keep it under 70C. I figured it was either the heat sink not installed right, or me not knowing how to overclock properly. However, the computer was stable even with those temperatures, so I ignored the issue. In retrospect, I'm wondering if this might have been the same issue that's now resurfacing.
A couple of weeks ago, the upgrade itch struck again, so I picked up a Core i7 930, an Asus P5X58D Deluxe motherboard, and six sticks of 2GB OCZ Gold DDR3 1600 RAM (which the BIOS runs at 1066MHz by default), along with a 5870 and a Corsair H50 cooling system. I reused the case, SSD, hard disk, DVD drive, and power supply (PC Power and Cooling Silencer 610W). I also wiped the system partition and reinstalled Windows 7 64-bit from scratch.
However, this build has been giving me some trouble. When I built the system, it booted fine the first time (ignoring the one attempt where I forgot to actually switch on the power supply ... a true "D'oh!" moment if there ever was one). I ran the build at stock settings for everything, just to shake it out and make sure everything worked OK. Prime 95 was stable for 12 hours, although the CPU temperature got to a surprisingly high 85C.
The next test I did was to run MemTest86+ overnight. This failed with a couple of memory errors. The next day, I tested each stick individually in the motherboard, and allowed MemTest to run for two full passes on each one. No errors. Curious ... could it be the motherboard? I tested one stick in each slot successively; other than the non-booting slots (which are documented in the motherboard manual), it all worked fine. I started testing various combinations of 2 and 3 at a time, and all ran fine.
Finally I hit on the idea that heat might be the problem. I jury-rigged a stand for a spare 120mm fan I had and pointed it at the RAM. I stuck all six sticks back in and ran MemTest overnight -- no errors. However, if I bump the PC, the fan tips over, and the computer bluescreens shortly thereafter, complaining of a memory error.
I'm also concerned about my video card. With the default driver settings, my 5870 idles at about 63C, and under load (I tested with World of Warcraft and Dragon Age: Origins, the two games I currently play most) it tops out at 88C. I do notice the fan spinning up occasionally, but not to full volume. However, after an extended play session, I tend to get a visual errors -- tearing in the display for a split second, or flickering in the screen. This builds up to a crash. The entire screen goes into a brown patchwork-like display, but the PC keeps working a moment longer -- I can hear audio from the game indicating stuff is happening. A few seconds later, the entire PC locks up, and only a hard reset starts it back up.
On a related note, making any change at all in the ATI Overdrive panel, even just to increase fan speed while leaving the overclock settings at stock, causes visual artifacting. The 5870 idles at about 65C.
Here's a picture of my PC's innards before I cleaned up the wiring some; this was just after I completed the initial build. I highlighted the fan locations in green, and I put in blue (cool) and red (hot) arrows to show airflow. The case is a Coolermaster Cosmos 1000 -- no side window and no way to add a side-mounted fan, short of modding the case -- and that's definitely not in the cards for me. (And yes, I realized that the H50 radiator is installed upside-down in the photo. I don't know if that matters, but I did flip it right-side up as shown in the H50 documentation.)
My questions:
1) Could RAM really get so hot it fails at stock speeds?
2) I'm concerned about the 5870's temperatures, but I can't find a reliable guide anywhere that says what is normal for this chip. Any ideas? Is 88C under load normal?
3) Similarly, is 85C under load normal for the CPU with this cooler? I don't think this high of a temperature at stock speeds leaves much room for overclocking.
4) Any ideas on how to mount a fan pointing at the RAM? Not sure what to attach it to.
5) I'm going to add a second fan to the H50 radiator. Given my case layout, would you advise that these fans blow in or out? Bear in mind, if they blow in, then hot air is entering the case; if they blow out, only the single bottom fan is acting as an intake.
6) The metal back plate (top side as installed) on the 5870 gets very hot under load. Would sticking some (real) heat sinks on the back of the card help with the heat dissipation? It would also give me someplace permanent to mount the fan. However, it would effectively move more heat up towards the RAM. This would likely exacerbate the RAM issue.
Many thanks in advance for your advice!
I need your help debugging an issue that I think is heat related. Given that reading some of the excellent posts here is what drove me to upgrade in the first place, I think it's only fair to give you guys a crack at analyzing this.
Please forgive the wall of text; I'm trying to include as much info as possible to help debug this.First, some history. This PC originated as a Core 2 Duo E8400, which I had overclocked to about 3.3GHz. About a year ago, I upgraded it to a Q9550. I was able to push it to 3.2GHz, but I wasn't happy with the temperatures. They approached 75C at load, and I was hoping to keep it under 70C. I figured it was either the heat sink not installed right, or me not knowing how to overclock properly. However, the computer was stable even with those temperatures, so I ignored the issue. In retrospect, I'm wondering if this might have been the same issue that's now resurfacing.
A couple of weeks ago, the upgrade itch struck again, so I picked up a Core i7 930, an Asus P5X58D Deluxe motherboard, and six sticks of 2GB OCZ Gold DDR3 1600 RAM (which the BIOS runs at 1066MHz by default), along with a 5870 and a Corsair H50 cooling system. I reused the case, SSD, hard disk, DVD drive, and power supply (PC Power and Cooling Silencer 610W). I also wiped the system partition and reinstalled Windows 7 64-bit from scratch.
However, this build has been giving me some trouble. When I built the system, it booted fine the first time (ignoring the one attempt where I forgot to actually switch on the power supply ... a true "D'oh!" moment if there ever was one). I ran the build at stock settings for everything, just to shake it out and make sure everything worked OK. Prime 95 was stable for 12 hours, although the CPU temperature got to a surprisingly high 85C.
The next test I did was to run MemTest86+ overnight. This failed with a couple of memory errors. The next day, I tested each stick individually in the motherboard, and allowed MemTest to run for two full passes on each one. No errors. Curious ... could it be the motherboard? I tested one stick in each slot successively; other than the non-booting slots (which are documented in the motherboard manual), it all worked fine. I started testing various combinations of 2 and 3 at a time, and all ran fine.
Finally I hit on the idea that heat might be the problem. I jury-rigged a stand for a spare 120mm fan I had and pointed it at the RAM. I stuck all six sticks back in and ran MemTest overnight -- no errors. However, if I bump the PC, the fan tips over, and the computer bluescreens shortly thereafter, complaining of a memory error.
I'm also concerned about my video card. With the default driver settings, my 5870 idles at about 63C, and under load (I tested with World of Warcraft and Dragon Age: Origins, the two games I currently play most) it tops out at 88C. I do notice the fan spinning up occasionally, but not to full volume. However, after an extended play session, I tend to get a visual errors -- tearing in the display for a split second, or flickering in the screen. This builds up to a crash. The entire screen goes into a brown patchwork-like display, but the PC keeps working a moment longer -- I can hear audio from the game indicating stuff is happening. A few seconds later, the entire PC locks up, and only a hard reset starts it back up.
On a related note, making any change at all in the ATI Overdrive panel, even just to increase fan speed while leaving the overclock settings at stock, causes visual artifacting. The 5870 idles at about 65C.
Here's a picture of my PC's innards before I cleaned up the wiring some; this was just after I completed the initial build. I highlighted the fan locations in green, and I put in blue (cool) and red (hot) arrows to show airflow. The case is a Coolermaster Cosmos 1000 -- no side window and no way to add a side-mounted fan, short of modding the case -- and that's definitely not in the cards for me. (And yes, I realized that the H50 radiator is installed upside-down in the photo. I don't know if that matters, but I did flip it right-side up as shown in the H50 documentation.)
My questions:
1) Could RAM really get so hot it fails at stock speeds?
2) I'm concerned about the 5870's temperatures, but I can't find a reliable guide anywhere that says what is normal for this chip. Any ideas? Is 88C under load normal?
3) Similarly, is 85C under load normal for the CPU with this cooler? I don't think this high of a temperature at stock speeds leaves much room for overclocking.
4) Any ideas on how to mount a fan pointing at the RAM? Not sure what to attach it to.
5) I'm going to add a second fan to the H50 radiator. Given my case layout, would you advise that these fans blow in or out? Bear in mind, if they blow in, then hot air is entering the case; if they blow out, only the single bottom fan is acting as an intake.
6) The metal back plate (top side as installed) on the 5870 gets very hot under load. Would sticking some (real) heat sinks on the back of the card help with the heat dissipation? It would also give me someplace permanent to mount the fan. However, it would effectively move more heat up towards the RAM. This would likely exacerbate the RAM issue.
Many thanks in advance for your advice!
Facebook
Twitter