Moar threads

[frozentundra123456]

Just another shot in the continuing battle of i5 vs i7:

game.gpu Total War Arena

30% gain from 4670k to 4770k. Also of course, hex cores do well. Surprisingly for a Total War game (which I thought likes single core IPC), FX does pretty well too. I think we really need to re-examine the recommendation of an i5 for anything over a midrange system. (Edit: even a two generation old 2600k beats 4670k.)

Granted it is a beta, but if you look at most recent games, 4770k is usually a few percent or more faster, and in some games like this, a lot faster. I hope they add Skylake with fast ram to their test suite pretty soon.

 

[ShintaiDK]

Network concurrency. Its a MP only game.

Looking on the CPU graphs, it seems to be a game more about cache than cores.

 

[frozentundra123456]

Network concurrency. Its a MP only game.

Well there are a lot of multiplayer games. Would this be a common effect of more cores/threads then?

I know you are an "i5 guy", and I am perfectly happy with my Sandy i5, but really in a new thousand dollar or up system with the slow progress in cpus I am starting to agree that the extra hundred dollars for hyperthreading or 5820k is worth it for a heavy duty gamer, since he probably wont upgrade for 3 to 5 years.

 

[cbn]

(Edit: even a two generation old 2600k beats 4670k.)

In Witcher 3 i7 2600K also beats i5 4670K:

 

[ShintaiDK]

Well there are a lot of multiplayer games. Would this be a common effect of more cores/threads then?

I know you are an "i5 guy", and I am perfectly happy with my Sandy i5, but really in a new thousand dollar or up system with the slow progress in cpus I am starting to agree that the extra hundred dollars for hyperthreading or 5820k is worth it for a heavy duty gamer, since he probably wont upgrade for 3 to 5 years.

I would bet you that a 5675C would beat a 4770K with ease in this game.

 

[Absolute0]

Just another shot in the continuing battle of i5 vs i7:

Well there are a lot of multiplayer games. Would this be a common effect of more cores/threads then?

I know you are an "i5 guy", and I am perfectly happy with my Sandy i5, but really in a new thousand dollar or up system with the slow progress in cpus I am starting to agree that the extra hundred dollars for hyperthreading or 5820k is worth it for a heavy duty gamer, since he probably wont upgrade for 3 to 5 years.

This thread has everything to do with Tovarisc's posts in the Skylake thread, where he was repeatedly criticized for his purchase of an i5 over an i7. I declined to comment there but making a new thread to continue the discussion is inviting a comment...

It's one thing to give advice when a guy is asking for it, it's quite another thing to criticize someone's choice of parts right after they've bought everything and are excitedly about to put it together.

I'm glad to see hyperthreading is improving game performance, really I am - for years, dual core and quad core meant nothing for gaming. But if someone wants to game, you can't fault the decision to put the $ into a 980Ti, that card is BEAST.

Every part can be sold, upgraded, whatever. It's a hobby that's about having FUN with hardware/building.

Edit: I felt bad for for this guy, who had just posted a picture of his parts, and what promises to be a badass build:

Kinda regretting that I posted anything about my upcoming rig, didn't realize how much I would get criticized for getting i5 over i7. That said last time I build was early 2010 and took i5

Edit: Also I know how ridiculous this may sound, but going even that lousy 100€ over budget wasn't option.



Oh wow, condescending much?

 

[ShintaiDK]

In Witcher 3 i7 2600K also beats i5 4670K:

But is it due to threads?

A lower clocked 5675C beating 2600K and 3770K while basicly matching a 4770K.

Its the same reason why a 5960X doesnt completely stink in 1-4 threaded games with its 3.5Ghz max turbo.

 

[SlowSpyder]

I would bet you that a 5675C would beat a 4770K with ease in this game.

I know it's hard to compare across different architectures, but the jump from an FX 6300 to the FX 8350 is close to 37%. The clock speed difference is only 14%. Both have 2MB L2 per module and 8MB L3. Cache may certainly make a difference, but I have to think there has to be some advantage from the extra cores in this based on the AMD numbers.

 

[ShintaiDK]

I know it's hard to compare across different architectures, but the jump from an FX 6300 to the FX 8350 is close to 37%. The clock speed difference is only 14%. Both have 2MB L2 per module and 8MB L3. Cache may certainly make a difference, but I have to think there has to be some advantage from the extra cores in this based on the AMD numbers.

Its not utilization rate:

Remember how AMDs cache system works on the FX.

 

[Flash831]

In both these examples, going up in threads seems to be improving gameplay, as well does adding L4-cache.

 

[Yuriman]

If this is the case, it's good to see that there are finally a few games where FX chips are not walked all over, and where an i7 might actually make sense over an i5. I was skeptical that we'd start to see many highly threaded games before the FX chips were completely obsolete.

Once this becomes more common I'll likely swap my i5 out for an i7, but as-is both benches listed will still be over 60fps minimums with an i5, and I don't have a 120hz panel.

 

[ShintaiDK]

In both these examples, going up in threads seems to be improving gameplay, as well does adding L4-cache.

In case of i5 to i7 its also 33% more L3 cache.

 

[SlowSpyder]

Like I said, hard to compare across architectures. Just thinking out loud, maybe the 3x2MB cache has to hold more data than the 4x2MB cache in the 8350 as there are less modules/cores/ and it tops out. Intel's fast L3 can make up for the i5 and i7's smaller L2... so certainly cache could make a difference.

 

[frozentundra123456]

This thread has everything to do with Tovarisc's posts in the Skylake thread, where he was repeatedly criticized for his purchase of an i5 over an i7. I declined to comment there but making a new thread to continue the discussion is inviting a comment...

It's one thing to give advice when a guy is asking for it, it's quite another thing to criticize someone's choice of parts right after they've bought everything and are excitedly about to put it together.

I'm glad to see hyperthreading is improving game performance, really I am - for years, dual core and quad core meant nothing for gaming. But if someone wants to game, you can't fault the decision to put the $ into a 980Ti, that card is BEAST.

Every part can be sold, upgraded, whatever. It's a hobby that's about having FUN with hardware/building.

Edit: I felt bad for for this guy, who had just posted a picture of his parts, and what promises to be a badass build:

There have been repeated threads about this for a long time. I started a new thread because I specifically did not intend it to be directed towards him. I also think it is important because the standard recommendation in general hardware I believe is an i5.

 

[richierich1212]

This video has a pretty good comparison: http://www.youtube.com/watch?v=EhaB1dqYv_I

You can see where the 6600K dips much lower during gameplay in each of the games.
I would spend the extra difference for i7, especially when you know you won't be upgrading the CPU for several years.

 

[TheELF]

which I thought likes single core IPC

It does and it shows in the "load core cpu intel % " picture they posted.
Lots of cores/threads allows you to display more graphics (soldier units) on screen faster but it does not influence the speed at which the game runs.
A fast core will make the game respond quicker/finish the turn faster* and so on.


*For Civ,where we have the same story since they added mantle to fool people into thinking that their game will run faster now.