Which one Overclocks the best: Phenom II X4 or Corei5 750?

[ElMoIsEviL]

I have heard many claims being made by folks on both sides of this debate. I wanted to end this debate once and for all but there was a problem. It had come to my attention that the samples sent to most reviewers were handpicked and not necessarily representative of what one could expect of seeing with a retail sample.

At first I was a tad perplexed as to how I should tackle this question and then it hit me. I should go to the place that is the most recognized symbol of overclocking prowess and poll its users.

Here are the findings:

Phenom II X4 *:

Corei5 750:

* Worth mentioning that 4 of the claimed 4GHz+ systems for the Phenom II X4 were in fact watercooled and not aircooled results as requested. Have a look for yourself in the thread here: http://www.xtremesystems.org/forums/showthread.php?t=244754

For those who are curious the Intel thread is available here.

Peace.

 

[ulysses35]

On air the i7 should be the better overclocker. Though you didnt state if you fancied water cooling your CPU.

AMD are lagging behind Intel at the moment - the current Phenoms are barely up the mid / high end C2D / C2Q - let alone i5/ i7

 

[jennyh]

Interesting...at least seeing so many i5's that didn't get to 4ghz is. 33% of i5's don't get to 4ghz? That's higher than most would have thought.

I would assume almost all of those Phenoms that didn't get to 4ghz were the older C2 stepping.

If you had limited it to C3 stepping only, I'm sure the number of them failing to get to 4ghz and beyond would be a lot less than 1 in 3.

 

[bige420]

I dont think that 33% was accurate, that might have been the people that just didnt want to go any further or didnt know how. The overwhelming majority are 4+ghz for the i5's, something you just dont see with the PII's.

 

[keithlm]

* Worth mentioning that 4 of the claimed 4GHz+ systems for the Phenom II X4 were in fact watercooled and not aircooled results as requested.

Let's review a few minor issues with your "scientific approach":

1. The "unspecified" versus "air only" poll. Something more to skew the results. And not something an unbiased observer would do.
2. As Jenny pointed out, not specifying the stepping skews the results. Unless you wanted to compare a new stepping of one brand with a competitors old and new steppings. That not something an unbiased observer would do but whatever.
3. You actually skewed the way you asked the poll questions. (Definite FAIL for being unbiased.)

To be more specific:

You gave the Intel group the following 'range' to answer:
2.8-3GHz 0 0%
3.1-3.4GHz 0 0%
3.5-3.8GHz 3 12.50%
3.9-4.0GHz 5 20.83%
4GHz+ 16 66.67%

You gave the AMD group this "range" of answers: (The results are different than last night.)
3.4GHz - 3.6GHz 0 0%
3.6GHz - 3.7GHz 3 7.50%
3.7GHz - 3.8GHz 3 7.50%
3.8GHz - 3.9GHz 15 37.50%
4GHz+ 19 47.50%

If somebody got a result between 3.9Ghz and 4.0Ghz then if they were being honest they would answer 3.9-4.0Ghz in the first set of ranges but they would have to answer 3.8Ghz-3.9Ghz given the second set of ranges since it would be dishonest to claim 4Gz+. Regardless.... your ranges are completely skewed.


(QUESTION: Are you still pretending to be unbiased and "not a fanboy" or did you finally give that hopeless claim up? Rest assured that when you post things like this you reveal your true bias.)

 

[jennyh]

I dont think that 33% was accurate, that might have been the people that just didnt want to go any further or didnt know how. The overwhelming majority are 4+ghz for the i5's, something you just dont see with the PII's.

So we're just ignoring elmo's little sample? Ok lol.

 

[jennyh]

Also I have to agree strongly with Keith on this one elmo.

Elmo, you are good at finding the result you want from this sort of thing (as am I, clearly). However, you have to admit that Keith is correct in saying that your methods aren't unbiased. My methods aren't unbiased either, but if you just take it totally unbiased like Keith is doing here, you will have to conclude that he is correct.

 

[jennyh]

You gave the Intel group the following 'range' to answer:
2.8-3GHz 0 0%
3.1-3.4GHz 0 0%
3.5-3.8GHz 3 12.50%
3.9-4.0GHz 5 20.83%
4GHz+ 16 66.67%

You gave the AMD group this "range" of answers: (The results are different than last night.)
3.4GHz - 3.6GHz 0 0%
3.6GHz - 3.7GHz 3 7.50%
3.7GHz - 3.8GHz 3 7.50%
3.8GHz - 3.9GHz 15 37.50%
4GHz+ 19 47.50%

Says it all really. You know that very few C2 stepping Phenoms reach 4ghz Elmo.

 

[keithlm]

but if you just take it totally unbiased like Keith is doing here

I am completely biased and would never attempt to claim otherwise.

However, often when people claim to be neutral they are really saying: "I'm completely neutral, I'm just attempting to force MY version of the truth onto you."

(And they REALLY hate having somebody point out their obvious bias.)

 

[ElMoIsEviL]

Also I have to agree strongly with Keith on this one elmo.

Elmo, you are good at finding the result you want from this sort of thing (as am I, clearly). However, you have to admit that Keith is correct in saying that your methods aren't unbiased. My methods aren't unbiased either, but if you just take it totally unbiased like Keith is doing here, you will have to conclude that he is correct.

Look at the dates both polls were done. One was done way before the other. The suggestion results are the same. People who hit 4GHz or more (on the AMD poll) got to vote for 4GHz+. So it was an advantage for AMD and not a disadvantage.

Subsequently I know which users are voting in these polls and it was nice to see an influx of false votes (after I published the results) come in from AMDZone. It went from 13 to 21 (4GHz+) with new users voting. Interesting no?

A bunch of new people registered just to vote up AMDs 4GHz+ figure.

None of that was witnessed with the Intel figures which have remained the same.

 

[jennyh]

Oh right it's all an AMDZone conspiracy lol.