Overclocking Core i7: A Tale of Two Retail Processors…

[geokilla]

http://www.anandtech.com/cpuchipsets/in ... spx?i=3502

I stopped reading after halfway through the 2nd page, but damn. Those are some HUGE differences in terms of voltages and CPU speed and stuff. Maybe Nehalem was released a tad too early...

[FZ1]

There is always variation in the production process. That's why they bin cpu's, RAM, etc, etc. I started reading it this morning and had to laugh as he was actually surprised some chips clock better than others. How long has he been doing this?

[martini161]

that has to be one of the stupidist articles ive ever read, especially the second page. all he did was say this in 1,119 words: Results will vary.

[geokilla]

It's true that results will vary, but that big of a difference is just unbelievable. In my opinion, Intel needs to do something about that and make that difference a bit smaller.

[martini161]

It's true that results will vary, but that big of a difference is just unbelievable. In my opinion, Intel needs to do something about that and make that difference a bit smaller.

any suggestions? they cant. no two peices of silicon are the same. when your making things this precise, the tolerances are so small, that if it works, you bin it. if they went around taking out all the bad overclockers, the price would skyrocket

[Alathald]

If you want a guarantee that a cpu will hit a certain frequency, buy one rated for that frequency. Short of that, it's a crap shoot.

[geokilla]

Guess my two years of being in the AnandTech and Legit Reviews community hasn't taught me some obvious stuff like this. I shoulda known bout this.

Where is Anand anyways....God I miss his reviews and same with Gary's reviews.

[Apoptosis]

Guess my two years of being in the AnandTech and Legit Reviews community hasn't taught me some obvious stuff like this. I shoulda known bout this.

Where is Anand anyways....God I miss his reviews and same with Gary's reviews.

Anand is still around... saw him at CES... Looking good and he's just doing the business side of the site if you ask me more than anything.

[Kougar]

Anand was saying there is a huge difference in memory controllers, not the actual core silicon, that is behind the widely varying results. But this was to be expected, several people wrote articles that they were concerned about IMC stability when packed in next to four hot CPU cores.

He just spent two pages stating that ut just means Core i7 is not as certain to hit specific overclocks as any Quadcore Penryn was.

[geokilla]

Anand was saying there is a huge difference in memory controllers, not the actual core silicon, that is behind the widely varying results. But this was to be expected, several people wrote articles that they were concerned about IMC stability when packed in next to four hot CPU cores.

He just spent two pages stating that ut just means Core i7 is not as certain to hit specific overclocks as any Quadcore Penryn was.

So it could be the memory controllers which are causing all these weird and "unstable" overclocks as compared to Penryn? Maybe Intel moved a bit too quickly with this direct connect thing.