P67/2600K Vs X58/980X - Awesome Awesome Comparo

[Castiel]

http://www.xtremesystems.org/forums/showthread.php?t=264693

 

[RussianSensation]

?

This comparison makes no sense at all because no one in the world is going to run Core i7 980X with 4 cores and 8 threads after spending $1000 on that processor. He should have compared 4 core 8 threaded 2600k vs. 6 core 12 threaded 980X...

If he wanted to compare IPC to a Core i7 processor, he should have used 920-975 or 860-875.

 

[Shmee]

perhaps he wanted to simulate a 32nm i7, as all the 4 core i7's on x58/p55 are 45 nm? just a thought.

 

[Rifter]

?

This comparison makes no sense at all because no one in the world is going to run Core i7 980X with 4 cores and 8 threads after spending $1000 on that processor. He should have compared 4 core 8 threaded 2600k vs. 6 core 12 threaded 980X...

If he wanted to compare IPC to a Core i7 processor, he should have used 920-975 or 860-875.

could not agree more.

 

[Diogenes2]

perhaps he wanted to simulate a 32nm i7, as all the 4 core i7's on x58/p55 are 45 nm? just a thought.

Why simulate a non-product ?

That said, the price/performance advantage of a 2600k ( overclocked ) over the 980x is obvious.

 

[Shmee]

well, there was some talk of 32 nm 12 mb l3 cache xeons, but i don't know what happened to them. Obviously, that is some impressive performance

 

[Concillian]

Why simulate a non-product ?

To look at single threaded performance improvements the architecture is responsible for.

It's not that people are necessarily going to be deciding between a 2600K and a 980X with disabled cores, it's that you can look at this comparison and make pretty valid conclusions about the improvements in design.

You can see which applications are really cache dependent, as the 980X wins despite the 2600K winning or tying in just about everything else. You can see which applications are not so CPU dependent, because they tie. You can then extrapolate these conclusions to whatever CPU you have without having to find a specific benchmark comparison of <your CPU> vs. 2500K or 2600K.

IMO this is far more useful a benchmark than comparing an i7-950 to a 2600K or an i5-750 to a 2500K at stock speeds. Most here aren't going to run things at stock speeds.

Most people looking at SB are coming from a quad with less cache than a 980X, so will see gains at least as much as shown in this series of benchmarks (clock for clock,) then can also use the frequency scaling to estimate what any additional overclocking headroom will gain them.

Very useful information IMO.