QUick question about Ram speed vs. Cpu clock

[ITSTHINKING]

I am overclocking my e8500 and for some reason it is hard for me to get it stable when I push my ram speed to a higher level.

I have reached a sweetspot and got the processor stable at 3.6ghz but the ram is at 380 in cpu-z (760mhz)

Now here's my question, it's stable at 3.6 but with a slight reduction of ram speed. If you had to choose between say, 3.3 ghz with ram @800 or 3.6ghz with ram at 760, which would be the better choice?

I am still working on the overclock so maybe when all is said and done I won't have to settle with less out of my ram but I was just wondering how bad having the ram speed 40mhz slower is going to hurt me.

Thanks, hope I make some sense

 

[jimbo6846]

Prioritize CPU clock speed over RAM clock speed. You'll see much more benefit from the CPU clock increase vs. the RAM clock increase.

 

[Arcygenical]

RAM speeds, these days, make very little difference in any real world situation... In-Fact, anything over DDR2-800 is vastly useless in most normal scenarios.

 

[ITSTHINKING]

Thanks a lot for the quick response. I appreciate it

 

[Nenu]

Before clocking your CPU, find the max speed of your motherboard and then the max speed of your memory.
Reduce the speed of your memory and CPU to find the max motherboard FSB speed.
Reduce the speed of the CPU (and maybe the motherboard) to find the max memory speed.
Then you can know the limits before you start to clock your CPU and will be able to tell which component is at its limit.

 

[GonzoP]

Solid advice above. Yeah 3.6 is a world better than 3.3 if its only 20mhz on the ram. Shoot I'd take 3.6 over 3.5 with the same difference in ram speed. BTW, 3.6Ghz from that Duo is pretty nice

 

[ITSTHINKING]

Well I'll admit it was done rather hastily(the oc), I ran prime95 for 2 hours and ran some superpi.
I'll take more time later on to really get the most out of it. Here's what I got so far.

 

[ITSTHINKING]

Question, i know prime95 is better but when you use superpi to test what do you use to test? 1m? 2m?

 

[jimbo6846]

SuperPi is more of a benchmark than a stress test, as far as I know. For benchmarking, people usually use the 1m test. 13s or below is great, I think.

 

[Zero82z]

Question, i know prime95 is better but when you use superpi to test what do you use to test? 1m? 2m?

SuperPi is not by any stretch of the imagination a good stress-testing program. Even the 32m test wouldn't be enough for that purpose. I suggest you use IntelBurnTest to check your stability.

 

[jimbo6846]

I wouldn't recommend IBT by itself. I've had it run 10 tests and pass, and then turn around and run Prime95 Blend and it crashes and reboots the PC.

 

[Zero82z]

IBT is better for checking the reliability of your CPU rather than overall stability (meaning that it'll tell you if your CPU is producing incorrect results, but won't necessarily crash the PC). I agree that Prime95 is very good for doing that.

 

[jimbo6846]

IBT is better for checking the reliability of your CPU rather than overall stability (meaning that it'll tell you if your CPU is producing incorrect results, but won't necessarily crash the PC). I agree that Prime95 is very good for doing that.

I'll keep that in mind. I can't use it for the higher OC's I'm doing, it raises my core temps WAY too much.

 

[Zero82z]

I'll keep that in mind. I can't use it for the higher OC's I'm doing, it raises my core temps WAY too much.

Nothing to worry about unless you hit Tjmax. IBT makes my CPU hit 93C at my current overclock.

 

[jimbo6846]

I'm not worried about it destroying the PC outright, just slowly degrading the lifespan of the chip. The hotter they get, the shorter they last.

Also, 93C seems a bit excessive even for IBT. My Q9550 doesn't hit that high with IBT, even at 4Ghz.