- Joined
- Oct 27, 2014
- Messages
- 46
- Motherboard
- GA-H97N-WiFi
- CPU
- i5-4690
- Graphics
- HD4600
- Mac
- Classic Mac
- Mobile Phone
After making due with disabling sleep on my system almost since building it last fall, I finally decided to try replacing RAM. I had been using the Crucial Ballistix Sport 16GB kit (BLS2KIT8G3D1609DS1S00) that I'd bought following the October 2014 Buyer's Guide. I replaced it with G.SKILL Ripjaws X Series 16GB kit (F3-12800CL10D-16GBXL) that's on the memory compatibility list for my board. In order to get the new RAM recognized at 1600MHz I had to enable XMPDetection and Trust in my Clover Config.plist. After five days without shutting down and multiple sleep/wake cycles each day: no crashes.
I haven't tested running my old Crucial RAM with the changes to XMPDetection and Trust in Clover Config.plist—I don't think I will—but those changes might be worth trying for anyone else experience this problem, just in case.
Following up: After running with the G.SKILL RAM for a while, my system is much more stable running with sleep than with the Crucial RAM, but unfortunately there is some lingering instability, though the symptoms are a bit different than previously.
With the Crucial RAM I was having crashes after sleep daily (or more), I've regularly been running with the G.SKILL RAM for 5–6 days through many sleep/wake cycles without crashes. I didn't experience my first crash until almost two weeks after installing the new RAM. All together I've experienced three (maybe four) spontaneous reboots (similar to what I had with the Crucial RAM) since installing the G.SKILL RAM; I've also had three or so crashes/reboots when shutting down. But I'm not longer having the crashes at boot than I sometimes had running the Crucial RAM.
Overall, with the G.SKILL RAM, crashes have been infrequent enough to allow me to keep sleep enabled, which is a huge improvement for me over the Crucial RAM. I haven't had time to engage in further troubleshooting, but my experience certainly suggests that there is some hardware dimension to this problem.