Which CPU best for an EASY OC? (without uber ram)

[aigomorla]

Originally posted by: SerpentRoyal
E4400 or E4500. High quality JEDEC 1.8V DDR2 667 should be good enough to take a capable chip up to 3.4GHz.

^

totally agree with this. It has the highest multi you can possibly get on any cpu besides the top end and the Extreme series.

 

[aigomorla]

Originally posted by: Idontcare

Originally posted by: Deleted member 4644
You are wrong about the dividers. I KNOW that motherboards can divide the FSB speed to get lower mem speeds.

Originally posted by: XBoxLPU
You are wrong

Two things you both have in common - neither one of you posted anything that anyone could use.

If you KNOW of a mobo then list it or post a godam link to it on the web.

If you THINK someone is wrong then post some godam proof...calling people a liar does no one any good as it is your posted word against their's...and if that is all you are going to bother posting to accomplish then why bother?

LOL...

If your talking about the NVIDIA statement, the 680, and 780i are the only LGA775 chipset boards to offer a independant Memory overclocking feature.

That means your FSB/MEM ratio doesnt apply, the board automatically scales to the ratio closest to your desired MEM Speed.

However, the 680i is a total POS which is worth just about as much as my dog's dookie. And i have one sitting all nicely in a box waiting in line for my stepup to 780i, which im worried is going to be the same doggie dookie my dog makes.


Edit: @Idontcare: sorry to quote you, it was easier grabbing both of them this way.

 

[Pain999]

E4600 has a 12x.

 

[harpoon84]

Originally posted by: betasub
AT articles also confirm that the Core2 architecture benefits more from higher memory bandwidth than the A64 architecture (i.e. is more sensitive to memory speed).

Wrong, very wrong. K8 is much more sensitive to bandwith/latency than Core2. I'm not sure which Anandtech article you read, but it sure as hell ain't this one http://www.anandtech.com/memory/showdoc.aspx?i=2800&p=1 since it totally contradicts your view.

 

[VirtualLarry]

That's because the large L2 cache of C2D's has a insulating factor relative to memory bandwidth/latency.

 

[HopJokey]

Originally posted by: Deleted member 4644

Originally posted by: KIAman
Dividers go from 1:1 to increasing the Memory Speed relative to FSB. There's nothing to make memory speed lower than FSB.

hmm my nforce 4 AMD board can do a mem clock divider.. ?

Apples to oranges, AMD system does not use a FSB and the memory controller is different.

Intel system uses a different memory controller and passes the data over the FSB.

 

[betasub]

Originally posted by: harpoon84
I'm not sure which Anandtech article you read, but it sure as hell ain't this one http://www.anandtech.com/memory/showdoc.aspx?i=2800&p=1 since it totally contradicts your view.

Thanks for correcting me and providing that link. I was looking at an earlier (superceded?) article, where large increases in memory bandwidth only gave minor improvements in A64 performance.
Your link provides better (and more recent) evidence.

 

[Zap]

Originally posted by: aigomorla
If your talking about the NVIDIA statement, the 680, and 780i are the only LGA775 chipset boards to offer a independant Memory overclocking feature.

The 650i chipset also supports it. I had an MSI 650i SLI FI board that I basically set the memory to 800MHz and then was free to play around with the CPU speed (got around 3.5GHz out of an E4300).

For easy overclocking, with DDR2-667 you want at least a 9x multiplier for around 3GHz before overclocking RAM. With DDR2-800 an 8x multiplier will let you hit 3.2GHz before pushing the RAM. This is all assuming the board gives your particular chip a 1:1 memory ratio.