Haswell Vs Coffeelake Benchmarks @ 4.3Ghz

geordiejc · 5 Oct 2017 at 23:58

4 core 4770k @ 4.3Ghz (O/C) DDR3 1600 ROG 1080 Ti OC edition STOC
vs
6 core 8700k @ 4.3Ghz (Stoc) DDR4 3200 ROG 1080 Ti OC edition STOC
Rise of the Tomb Raider : DX12 Very High preset FXAA 1080p

Coffee Lake Win @ +0.79fps

1440p

Haswell win @ +9.71fps

Weird but interesting results, i'm going to do a few more benchmarks tomorrow on other games myself and reviewers own using the 1080 ti and i'll post the results.

Panos · 6 Oct 2017 at 00:54

geordiejc said:
4 core 4770k @ 4.3Ghz (O/C) DDR3 1600 ROG 1080 Ti OC edition STOC
vs
6 core 8700k @ 4.3Ghz (Stoc) DDR4 3200 ROG 1080 Ti OC edition STOC
Rise of the Tomb Raider : DX12 Very High preset FXAA 1080p

Coffee Lake Win @ +0.79fps

1440p

Haswell win @ +9.71

Weird but interesting results, i'm going to do a few more benchmarks tomorrow on other games myself and reviewers own using the 1080 ti and i'll post the results.

Interesting. Saw Paul's Hardware 8700K review and decided to hold on my 6800K.
At 2560x1440 a 4ghx 6800K is just 1fps slower than the 1600X/1800X/7700K/8400/8700K with the 1080ti.
Considering the 8700K has 10-15% higher clocks, aint worth. Especially since I will have to sacrifice my second 960Pro and not use it, as it would be bottlenecked.

AMD Zen+ and x390 in Feb feels proper upgrade. maybe grab a 6900k/6950x cheaply

geordiejc · 6 Oct 2017 at 09:11

There must have been something not right with Paul's system. There should be a c.10% IPC benefit (minimum) between these architectures. Haswell to Skylake was c.7.5% right?

Obviously this was only 1 game. I will test some more tonight when i get home.

Sasahara · 6 Oct 2017 at 10:11

Or the GPU is the bottleneck in that benchmark?

Hagar · 6 Oct 2017 at 10:22

I think that once you get +/_ 10% frame results in any testing between architectures, it is too close to call. You get into the realms of drivers, motherboards, bios settings, memory efficiencies and last but not least differences in software.

geordiejc · 6 Oct 2017 at 16:03

Sasahara said:
Or the GPU is the bottleneck in that benchmark?

I think in the vast majority the gpu will be the bottleneck - I'm no expert I could be wrong.

jigger · 6 Oct 2017 at 16:13

geordiejc said:
I think in the vast majority the gpu will be the bottleneck - I'm no expert I could be wrong.

Increase resolution untill the frame rate levels out. Then you're limited by something other than the GPU not including VRAM.

geordiejc · 6 Oct 2017 at 16:29

@jigger if i increase the resolution does this not become more gpu bound? I'd have thought it was the other way round and 720p would mean cpu?

jigger · 6 Oct 2017 at 17:04

If you increase resolution and frame rates drop then you're leaning on the GPU. Or API depending on how you look at it.

geordiejc · 6 Oct 2017 at 18:15

@jigger i'm confused.

Are you saying increasing resolution you are limited by the GPU or not (Posts 7 & 9 seem to contradict) Post 7 you say increasing resolution you are not limited by GPU, but in post 9 you say you are leaning on GPU. Which post is correct?

AndreiD · 6 Oct 2017 at 18:25

You can't really compare scores like that, find a review that includes a 4770K/4790K and go with that.
And the 8700K has pretty aggressive turbos, all core turbo is 4.3Ghz but it goes 4.7 for 1 core, 4.6 for 2 and 4.4 on 4 cores.

TechReport have a 4790K in their benches, but it's stock: http://techreport.com/review/32642/intel-core-i7-8700k-cpu-reviewed/6
The stock 8700K will still turbo a bit higher than a 4790K, but it's at least a better comparison.

geordiejc · 6 Oct 2017 at 18:35

@AndreiD

Thanks for the link.

4.3Ghz on all 6 cores according his review, i wish i had more games to compare.

I think when it comes down to it for games, the GPU is always going to be the bottleneck and anything from Sandy bridge and upwards can do 1080/1440 with current gen GPU's

jigger · 6 Oct 2017 at 18:39

geordiejc said:
@jigger i'm confused.

Are you saying increasing resolution you are limited by the GPU or not (Posts 7 & 9 seem to contradict) Post 7 you say increasing resolution you are not limited by GPU, but in post 9 you say you are leaning on GPU. Which post is correct?

Increase the resolution to the point the frame rate stops dropping. Then something other than the GPU is holding back performance. If you get the same performance at 2560x1440 as you do at 3880x2160 then you are bottlenecked.

geordiejc · 6 Oct 2017 at 18:56

Thanks I understand if u can edit your post 7 that would help as it's not correct

Scougar · 6 Oct 2017 at 19:55

What a bizarrely low result for the 1600X btw.

Seem like some nice improvements for the 8xxx series, even if there are some quirks.

Panos · 6 Oct 2017 at 20:24

geordiejc said:
I think in the vast majority the gpu will be the bottleneck - I'm no expert I could be wrong.

Yes it is. The 8700K makes sense only for a very narrow spec. You should have 1080/1080ti, at 1080p with 144hz-200hz monitor without gsync, wanting 10fps over and above 100-120fps mark.
On that strict spec is possibly the best CPU to get.

The moment you deviate for this exact spec, makes no sense to buy it over far cheaper or similarly priced products. And if someone is on budget, going to be better off with 1600X/1700 or even 6700K/7700K (they are bellow the £275 point "new" both) and GTX1080ti than 8700K and GTX1070 even at 1080p.

geordiejc · 6 Oct 2017 at 20:29

You are spot on Panos i agree totally.

so 8700k is best for cs:go pro's with 240hz monitors 1024 x 768 black bars

I personally game at 1440p / 165hz with gsync with a 1080 ti and use ingame frame limiters so gsync is always working.

jigger · 6 Oct 2017 at 20:37

Panos said:
Yes it is. The 8700K makes sense only for a very narrow spec. You should have 1080/1080ti, at 1080p with 144hz-200hz monitor without gsync, wanting 10fps over and above 100-120fps mark.
On that strict spec is possibly the best CPU to get.

The moment you deviate for this exact spec, makes no sense to buy it over far cheaper or similarly priced products. And if someone is on budget, going to be better off with 1600X/1700 or even 6700K/7700K (they are bellow the £275 point "new" both) and GTX1080ti than 8700K and GTX1070 even at 1080p.

If you really are wanting to go full retard I'd go with a volt modded Titan rhen ask 8 pack for a 5.4Ghz 6700/7700k all wrapped up in the best water loop. That's the only way to cut above anyone else and garrenteed 144Hz+ gaming for a while. Just don't try and stream at the same time.

jigger · 6 Oct 2017 at 20:39

geordiejc said:
You are spot on Panos i agree totally.

so 8700k is best for cs:go pro's with 240hz monitors 1024 x 768 black bars

I personally game at 1440p / 165hz with gsync with a 1080 ti and use ingame frame limiters so gsync is always working.

Most pros will be using 4x3 CRTs around 100Hz and Pretty low end graphics cards.

Panos · 6 Oct 2017 at 20:40

geordiejc said:
You are spot on Panos i agree totally.

so 8700k is best for cs:go pro's with 240hz monitors 1024 x 768 black bars

I personally game at 1440p / 165hz with gsync with a 1080 ti and use ingame frame limiters so gsync is always working.

I was going to buy the 8700K but then saw the 2560x1440 reviews with GTX1080ti, and compared the same games with the 6800K @ 4Ghz and 1080ti I have, ain't going to bother.
At best the difference is 1fps!!!!!!!!!!!

So yes I am GPU limited, oo poor shame on me, a 2Ghz clocked GTX1080ti is bottlenecking me.