Just because new shiny hits doesn't mean we all have to upgrade

kayomani · 8 Aug 2015 at 18:59

Buying a dead socket is a waste of money unless you need the threads in the short term because your a video encoder or w/e as it'll not help with games.

easyrider · 8 Aug 2015 at 19:26

kayomani said:
Buying a dead socket is a waste of money unless you need the threads in the short term because your a video encoder or w/e as it'll not help with games.

HT in modern games is where it's at ...significant boosts in crisis 3 for example

harz_marz · 8 Aug 2015 at 19:33

I really cannot make my mind up! Had ordered an i7 4790K but have kinda changed my mind and want the i7 5820K.

Yes I know there is the added expense of RAM etc but my existing stuff is showing it's age and DDR4 will be more useful in the future.

Am I crazy? The way I see it... I have splashed out on a 980 ti so I may as well go the whole way!

C64 · 8 Aug 2015 at 19:36

harz_marz said:
I really cannot make my mind up! Had ordered an i7 4790K but have kinda changed my mind and want the i7 5820K.

Yes I know there is the added expense of RAM etc but my existing stuff is showing it's age and DDR4 will be more useful in the future.

Am I crazy? The way I see it... I have splashed out on a 980 ti so I may as well go the whole way!

which setup do you currently have as long as the 980 isn't bottlenecking why bother ?

For all we know socket 1151 will be a dead socket too and I think it will because of the small gains made over haswell.
I wont be shocked if there's socket 1152 in a year and a half.

jjgreenwood · 8 Aug 2015 at 19:37

easyrider said:
HT in modern games is where it's at ...significant boosts in crisis 3 for example

Crysis 3 is 2 and a half years old, hardly modern.

Devils Canyon is faster than X99 in Crysis 3:
http://uk.hardware.info/reviews/559...-benchmarks-hd-7970-crysis-3n1920x1080-medium

Not that there is any meaningful difference between any of the CPU's

Threepwood · 8 Aug 2015 at 19:49

jjgreenwood said:
Crysis 3 is 2 and a half years old, hardly modern.

Devils Canyon is faster than X99 in Crysis 3:
http://uk.hardware.info/reviews/559...-benchmarks-hd-7970-crysis-3n1920x1080-medium

Not that there is any meaningful difference between any of the CPU's

It as if the test was done with all those CPU's at stock speeds or something.

Hmmm, what is the stock speed of devils canyon, lets see... 4GHz!

What is the stock speed of the the other CPU's you mentioned again?

haszek · 8 Aug 2015 at 19:54

Threepwood said:
It as if the test was done with all those CPU's at stock speeds or something.

Hmmm, what is the stock speed of devils canyon, lets see... 4GHz!

What is the stock speed of the the other CPU's you mentioned again?

But then again people run DC let's see at 4.6-4.8 and others run x99 at? what's the average oc speed?

harz_marz · 8 Aug 2015 at 20:05

C64 said:
which setup do you currently have as long as the 980 isn't bottlenecking why bother ?

For all we know socket 1151 will be a dead socket too and I think it will because of the small gains made over haswell.
I wont be shocked if there's socket 1152 in a year and a half.

I have an i5 2500 (non k). In most modern games at 1440 I've cpu usage at 100%. Games like Crysis 3, Ryse, Project Cars.

I basically need a nice new cpu to keep me going for the 5 years my i5 2500 has. And I just can't make my mind up!!

Threepwood · 8 Aug 2015 at 20:09

haszek said:
But then again people run DC let's see at 4.6-4.8 and others run x99 at? what's the average oc speed?

Idk, but I imagine one could disable a couple of cores on the 5820K and hit the same, if not more... all without having to faff about delidding - like with the 4790K.

That would be a 'fair' test, no?

Worth noting that with the 5820k you can choose exactly which cores to keep on or off as well.

easyrider · 8 Aug 2015 at 20:12

I ran my 5930k at 4.4ghz

haszek · 8 Aug 2015 at 20:17

I ran my 5930k at 4.4ghz

There you go, I'm pretty sure 0.4ghz can make up for lost cores (only in games which use multiple cores, in others it's worse). Still no reason to pay more.

easyrider · 8 Aug 2015 at 20:25

haszek said:
There you go, I'm pretty sure 0.4ghz can make up for lost cores (only in games which use multiple cores, in others it's worse). Still no reason to pay more.

The turbo on my 4790k is 4.4 ghz at stock

jjgreenwood · 8 Aug 2015 at 20:31

Threepwood said:
It as if the test was done with all those CPU's at stock speeds or something.

Hmmm, what is the stock speed of devils canyon, lets see... 4GHz!

What is the stock speed of the the other CPU's you mentioned again?

Yup, more to the point at the settings that the game is likely to be played at the overclock makes no difference

http://www.techspot.com/articles-info/1041/bench/dGPU_05.png

May as well leave it at stock. Certainly in this title there is no reason to have six cores. If we are using this as the example then any money spent on X99 over any other platform is a waste.

Maybe the picture might be different if multi gpu were a factor, beyond 2 there is rapidly diminishing returns there as well though.

r7slayer · 8 Aug 2015 at 20:44

So most would say for me going from a ivybridge i5 3750k at 4.2Ghz OC

to a 6700k 3.9Ghz (4.2Ghz Boost) is not worth it over going x99 5820k?

the x99 route is the better option??

Also is 4.2 achievable on stock volts on the 5820k??

CAT-THE-FIFTH · 8 Aug 2015 at 20:45

easyrider said:
I know ...

Its another case of old=crap and new=bestest syndrome.

With DX12 and Vulkan promoting better multi-threading and less single thread bottlenecks,unless you are doing a mini-ITX build,it just makes more sense to get the Core i7 5820K over a Core i7 6700K IMHO. Plus on top of that you can even get mini-ITX X99 motherboards although they are very pricey.

CAT-THE-FIFTH · 8 Aug 2015 at 20:47

jjgreenwood said:
Yup, more to the point at the settings that the game is likely to be played at the overclock makes no difference

http://www.techspot.com/articles-info/1041/bench/dGPU_05.png

May as well leave it at stock. Certainly in this title there is no reason to have six cores. If we are using this as the example then any money spent on X99 over any other platform is a waste.

Maybe the picture might be different if multi gpu were a factor, beyond 2 there is rapidly diminishing returns there as well though.

They obviously have not tested a more intensive part of Crysis3 and if you played it you would know what parts are more intensive,since they use animation of the foliage.

Pcgameshardware tested the Core i7 against the Core i5 and HT added a decent amount of performance.

Edit!!

Even Intel has bumpf showing how DX12 promotes better multi-threading using its own CPUs as examples.

haszek · 8 Aug 2015 at 20:50

8pack compared 4.2 (5820) to 4.5 (6700k) and fps was slightly higher for skylake.

Sorry but when I see someone who wants to change platform from 2500(nonK) and people on forum recommend x99 because of the hype on forum, this is silly to me. There is no reason to pay more if skylake is as fast (slightly faster).

From my point of view only fans x99 are saying it's better (not providing any proof for this) while new platform fans are saying it's equal just cheaper, so no point to spend more to feel better (with more cores).

CAT-THE-FIFTH · 8 Aug 2015 at 20:57

haszek said:
8pack compared 4.2 (5820) to 4.5 (6700k) and fps was slightly higher for skylake.

Sorry but when I see someone who wants to change platform from 2500(nonK) and people on forum recommend x99 because of the hype on forum, this is silly to me. There is no reason to pay more if skylake is as fast (slightly faster).

From my point of view only fans x99 are saying it's better (not providing any proof for this) while new platform fans are saying it's equal just cheaper, so no point to spend more to feel better (with more cores).

Its nothing to do with fans,Intel themselves have bumpf showing how DX12 REDUCES single thread bottlenecks on its own CPUs and promotes better multi-threading.

Vulkan does exactly the same thing - people need to stop sticking their head in the sand and actually read what even AMD and Nvidia themselves are talking about.

People are obssessed with the new=better and the old=crap syndrome on forums.

Yet,someone who invests in a Core i7 5820K will most likely be still using their setup in 4 to 5 years whilst someone who gets a Core i7 6700K will have probably ended up having to upgrade sooner rather than latter.

At least if you can get a Core i5 6600K for under £190,at least there is a cost argument,but with the Core i7 6700K??

haszek · 8 Aug 2015 at 21:01

CAT-THE-FIFTH said:
I
People are obssessed with the new=better and the old=crap syndrome on forums.

You keep repeating it and who's saying old=crap?

Sorry must've missed info that dx12 games are coming out shortly.

jjgreenwood · 8 Aug 2015 at 21:03

CAT-THE-FIFTH said:
They obviously have not tested a more intensive part of Crysis3 and if you played it you would know what parts are more intensive,since they use animation of the foliage.

Pcgameshardware tested the Core i7 against the Core i5 and HT added a decent amount of performance.

Edit!!

Even Intel has bumpf showing how DX12 promotes better multi-threading using its own CPUs as examples.

I think the test was welcome to the jungle. Been a while since I played it so can't remember which part that is.

I hope this dx12 takes off sooner rather than later. I'm surprised more games aren't developed to take advantage of modern processing power. I'm old enough to remember when civilisation needed a decent processor to process all its units etc. It does seem that rts and strategy in general isn't popular anymore.