Cores vs. Clock speed Comparison

[IntelUser2000]

Digit-life has good reviews comparing different cores and different frequency.

1-4 cores:
http://ixbtlabs.com/articles3/cpu/archspeed-2009-5-p9.html

1.86-3.06GHz:
http://ixbtlabs.com/articles3/cpu/archspeed-2009-6-p1.html

 

[cbn]

Digit-life has good reviews comparing different cores and different frequency.

1-4 cores:
http://ixbtlabs.com/articles3/cpu/archspeed-2009-5-p9.html

1.86-3.06GHz:
http://ixbtlabs.com/articles3/cpu/archspeed-2009-6-p1.html

http://ixbtlabs.com/articles3/cpu/archspeed-2009-5-p8.html

I'm actually suprised to see some games on the bottom of that page get a small benefit from Core i7 with HT on.

 

[IntelUser2000]

These are the better reviews for Hyperthreading comparison:

4 cores + HT:
http://ixbtlabs.com/articles3/cpu/archspeed-2009-3-p3.html

1 core + HT:
http://ixbtlabs.com/articles3/cpu/archspeed-2009-4-p2.html

Original post is about high clock dual core vs low clock quad core.

The Core i5 670 is a dual core with 3.46GHz frequency and HT
Core i5 750 is a quad core with 2.66GHz frequency and no HT

I can see how Intel came up with similar pricing for both.

2 to 4 cores: 30%
Hyperthreading: 10%
1.86-3.06GHz: 51%(80% scaling)

3.6GHz Turbo Mode on 670, vs 2.8GHz on 750 = 29%(this is comparing on HIGHEST thread count)

0.8% of 29%, 26%

1.26 x 1.10(Hyperthreading) = roughly equal to 2 to 4 cores gain

 

[aigomorla]

I'm actually suprised to see some games on the bottom of that page get a small benefit from Core i7 with HT on.

thats because you wont see many games that can utilize more then 2 working threads.
In some games, that virtual core can also hurt you, depending on clock speed.

I love the editors comment at the bottom, because its hogbosh.
I have yet to see a PHII cpu scale as high as i7 920 D0, minus extreme cooling.

 

[IntelUser2000]

thats because you wont see many games that can utilize more then 2 working threads.
In some games, that virtual core can also hurt you, depending on clock speed.

I love the editors comment at the bottom, because its hogbosh.
I have yet to see a PHII cpu scale as high as i7 920 D0, minus extreme cooling.

I think by "scaling" they don't mean overclocking.

 

[jvroig]

I think by "scaling" they don't mean overclocking.

Aigo must have been sleepy at the time; as I understand it, he's a non-native English speaker (I might be wrong) so he gets a free pass

I cannot read the links, it keeps timing out on me. Is it really "ixbtlabs" instead of "xbitlabs"?

 

[aigomorla]

I think by "scaling" they don't mean overclocking.

I kinda forgot to say Multi GPU platform scaling.

The only time i saw an AMD beat an intel was TWEEKER + LN2. :\

 

[nyker96]

interesting but I think the scaling effect is due to intel's shared L3 vs amd's non shared cache. just different architecture here. his experiment only verified the cache architecture of the two system is different not much else.

 

[cbn]

thats because you wont see many games that can utilize more then 2 working threads.
In some games, that virtual core can also hurt you, depending on clock speed.

I realize most games only use two threads. That is why I am suprised to some tiny positive results listed with what they call "8 cores" (see the very bottom of that page I linked).

Normally I would think turning on HT w/ i7 would actually make FPS worse.

 

[IntelUser2000]

interesting but I think the scaling effect is due to intel's shared L3 vs amd's non shared cache. just different architecture here. his experiment only verified the cache architecture of the two system is different not much else.

The less scaling shown by Core i7 could be because the performance is much higher than AMD Phenom II and hits a bottleneck in other systems faster(like I/O and GPU).