i5 2500K to ii7 4790k?

[crashtech]

Does clock speed really affect minimum's in single player games a lot? Like 20%+? I really want to see a 5820K @ 3.3GHz vs 3.6GHz MCE vs 4.0GHz vs max OC against a SNB at similar clocks vs a Haswell i5 @ 3.5GHz / max turbo / 4.0GHz / max OC in the last 5 or 6 major 2014/2015 AAA titles @ 1080p and above. Really curious.

Minimums seem to be affected by a blend of cores, clock, and IPC, don't you think? There's no perfectly accurate way to quantify it because each game looks for something a little different in terms of resources. Some games need a fast clocked main thread as well as lots of multithread capability, others lean toward one or the other. That's why high clocked Intel hexacores are such a safe bet, no matter the need the resource is just there.

 

[Dave3000]

Minimums seem to be affected by a blend of cores, clock, and IPC, don't you think? There's no perfectly accurate way to quantify it because each game looks for something a little different in terms of resources. Some games need a fast clocked main thread as well as lots of multithread capability, others lean toward one or the other. That's why high clocked Intel hexacores are such a safe bet, no matter the need the resource is just there.

What about a quad core i7 such as the 4790k? The clock speed and IPC is there and it has hyperthreading. There is no hex core out right now that matches or beats a 4790k in terms of clock speed and single-to-quad threaded performance. Most games don't take advantage of more than 4 cores and the few that do, it's not a big difference based on gaming benchmarks posted on the internet, maybe because of the higher clock speed of the 4790k making up for it.

 

[crashtech]

What about a quad core i7 such as the 4790k? The clock speed and IPC is there and it has hyperthreading. There is no hex core out right now that matches or beats a 4790k in terms of clock speed and single-to-quad threaded performance. Most games don't take advantage of more than 4 cores and the few that do, it's not a big difference based on gaming benchmarks posted on the internet, maybe because of the higher clock speed of the 4790k making up for it.

There have been games that show performance regressions with HT enabled on the i7. I think it is becoming less true as all four cores become saturated by the game engine. What was happening in some instances is that Windows would allocate CPU intensive loads to the logical (HT) core of a physical core that was already loaded instead of moving it to a less used core, resulting in performance degradation. This is not an issue when the load is very heavy and parallel, but can be a problem when the load is variable.

 

[lopri]

There are those who can actually utilize HT in heavy loads, and for those few, going from 4C/4T to 4C/8T plus around 10% IPC increase can make a significant difference. Most of us talk in terms of games since that is the most demanding thing users commonly run, but we ought to be adding caveats with statements like yours instead of asserting them as a reality for all usage cases. Many games don't use HT on the i7 well, even registering performance regressions, but many other apps do not have this problem, like rendering and transcoding, for example.

6C/6T (my other CPU, 1045T) does quite well against the 4C/8T when it comes to throughput computing and multitasking, which is why it is running my main system and the 2500K is running my gaming system.