4670k vs 3570k

[AnyOldName3]

It just happened that Dolphin really likes the small improvements Intel makes, whereas other applications aren't really affected hugely.

[xemnas]

That is why I wonder. If other applications are not a lot faster but WW on Dolphin is then it means the big improvement comes from either the way WW is emulated or the way Dolphin works. I wanted to know which one.

I wouldn't have asked if all other applications were significantly faster on 4670K.

[LordVador]

(10-04-2013, 10:55 PM)ScorpionKing Wrote: I understand that the 4670k is faster than the 3570k but my question is that since they're both so fast, would it make a difference?

Currently 3570K is enough for most games (assuming you overclock it). They will work just fine.
But since Dolphin is getting more and more demanding 4670K might be helpful soon

[pauldacheez]

(10-05-2013, 01:13 AM)rokclimb15 Wrote: You can't run certain games in HLE (such as LoZ: TP) without nasty glitches and some issues that could halt progress. Zelda ucode is often used to trigger events within the game and vice-versa, so LLE is the only option for TP. If you can't get full speed with LLE, then your audio lags, making things worse.

Clarification: the game often uses the ends of audio streams to trigger events, e.g. door transitions – the problem is that the DSP HLE for the Zelda µcode processes audio asynchronously, so it often fucks up the timing for those, thus the game receives them too early or too late and it sits there waiting forever for shit to happen.

[NaturalViolence]

AnyOldName3 Wrote:It just happened that Dolphin really likes the small improvements Intel makes, whereas other applications aren't really affected hugely.

Not true. Ivy bridge improved performance by 10% on average in most applications. Dolphin saw a 0% improvement. Dolphin just seems to have a very unique workload. It's difficult to predict what microarchitecture changes will and will not affect dolphins performance based on how they effect other applications. So far I cannot discern any pattern between architectural changes and performance increases with dolphin.

[AnyOldName3]

(10-07-2013, 06:51 AM)NaturalViolence Wrote:

AnyOldName3 Wrote:It just happened that Dolphin really likes the small improvements Intel makes, whereas other applications aren't really affected hugely.

Not true. Ivy bridge improved performance by 10% on average in most applications. Dolphin saw a 0% improvement. Dolphin just seems to have a very unique workload. It's difficult to predict what microarchitecture changes will and will not affect dolphins performance based on how they effect other applications. So far I cannot discern any pattern between architectural changes and performance increases with dolphin.

I was responding to

Quote:According to other benchmarks, 4670K is only a little bit faster than 3570K.

Between Ivy and Haswell, this was what happened

[xemnas]

I asked for another benchmark because the pattern was so strange on Dolphin. So nobody benchmarks the performance on other games? I just suspect that maybe WW is an exceptional case because of the way it's emulated.

[Garteal]

We'll surely have another benchmark now that Dolphin 4.0 is released. The question is just when it'll happen. Everyone is pretty busy nowadays, but hopefully someone will take the responsibility for it. We need to discuss what game, etc we'll use before we can get to it though.

[NaturalViolence]

Nobody has half a dozen desktops lying around with each generation of Intel/AMD cpus to benchmark with. And that's what you would need to do a proper comparison.

I highly doubt WW is an exception. That would have to be such a massive coincidence that it just wouldn't make any sense.

[AnyOldName3]

Also it wouldn't explain why I can do SMG at stock clocks.