Intel to launch 6 core Coffee Lake-S CPUs & Z370 chipset 5 October 2017

[DarkHorizon472]

That result booms linked looks like the ones Ryan Strout already said is wrong.
Also waiting for results until its released and we get proper reviews i dont see the point in writing it off

The pricing and performance will be key. Most people simply want the best combination of components for their needs in their budget. Once proper testing and evaluation starts we will see. Initially it is likely to be infomercial style reviews rather than real tests and comparison that provide useful information.

 

[Jono8]

From watching that HP omen youtube video, CPUZ seems to be stuck at 3.7ghz (doesn't change from that the whole time he is displaying it). Also, it is showing some crazy 154 x 23.98 multi which is weird.

That multi core CB r15 score does make a lot more sense if it all cores were at 3.7ghz.

I don't see how it can only score 1230 with 6 cores and 12 threads at 4.3ghz as that would display some serious ipc regression!

 

[Panos]

From watching that HP omen youtube video, CPUZ seems to be stuck at 3.7ghz (doesn't change from that the whole time he is displaying it). Also, it is showing some crazy 154 x 23.98 multi which is weird.

That multi core CB r15 score does make a lot more sense if it all cores were at 3.7ghz.

I don't see how it can only score 1230 with 6 cores and 12 threads at 4.3ghz as that would display some serious ipc regression!

Actually the score is right if the 8700k using Mesh not Ring topology, since its similar to 7800X

 

[Jono8]

Actually the score is right if the 8700k using Mesh not Ring topology, since its similar to 7800X

It really isn't. The 7800x scores ~1333 (http://www.anandtech.com/show/11550...core-i9-7900x-i7-7820x-and-i7-7800x-tested/17), and that is on the skylake arch.

Also, the 7800x only boosts to 4.0ghz where as this is supposed to boost to 4.3ghz.

This thing should be in the 1400s easily.

 

[jigger]

I think the only hope for this platform is if z370 allows for BCLK overclocking.

 

[Hagar]

I don't see how it can only score 1230 with 6 cores and 12 threads at 4.3ghz as that would display some serious ipc regression!

8700k 1230@12t 3.7GHz
R1700 1600@16t 3.6GHz all cores

Similar multicore IPC to mine if at 3.7

 

[gavinh87]

8700k 1230@12t 3.7GHz
R1700 1600@16t 3.6GHz all cores

Similar multicore IPC to mine if at 3.7

It's been said the scores are too low by pcper.
It's been shown already that kaby has 10% IPC increase over ryzen, add the extra clockspeed onto that and it's clear to see the winner here.
As for your comment on my ram on ryzen i suggest you read back further to see my issue. I'm not attempting to run 4266, we all know that isn't happening. However as I've shown it does have an issue with booting at 3200 or higher despite being perfectly stable when booted.
These are the issues I'll be glad to be rid of when coffee arrives.

 

[jigger]

It's been said the scores are too low by pcper.
It's been shown already that kaby has 10% IPC

Very debatable.

 

[gavinh87]

https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/

This is a good read for IPC comparison.

 

[Hagar]

It's been said the scores are too low by pcper.

The CB benchmark renders a C4D file located at “C:\Program Files\CINEBENCH_R15\plugins\bench\cpu\cpu.c4d”. Default resolution is 896x640px.

AFAIK, This file has not changed with the cinebench R15 revisions it can use 224 threads. I can be wrong though. One thing I did note was that Guru3D were using win 8 64 bit as the OS

 

[gavinh87]

The CB benchmark renders a C4D file located at “C:\Program Files\CINEBENCH_R15\plugins\bench\cpu\cpu.c4d”. Default resolution is 896x640px.

AFAIK, This file has not changed with the cinebench R15 revisions it can use 224 threads. I can be wrong though. One thing I did note was that Guru3D were using win 8 64 bit as the OS

Its more down to clockspeed going from his tweets:
https://twitter.com/ryanshrout/status/907280680207503361

 

[jigger]

7700k is 20% faster in clockspeeds.

 

[gavinh87]

7700k is 20% faster in clockspeeds.

Which makes no sense, probably why he tweeted that.

 

[humbug]

Its more down to clockspeed going from his tweets:
https://twitter.com/ryanshrout/status/907280680207503361

Ay, @ 3.7Ghz the numbers are probably right but it probably also has an all core boost that will clock up as high as thermals and power levels will allow, Much like nVidia's GPU Boost 3, so with good cooling it would overclock its self.

Should be interesting.

These scores are much higher and i think probably right when boosting to way over 4Ghz...

I think AMD's Ryzen 5 and 7 chips are about to get very cheap.

 

[Hagar]

I think AMD's Ryzen 5 and 7 chips are about to get very cheap.

https://forums.overclockers.co.uk/t...ly-289-99-with-free-quake-champions.18792886/

Just did

 

[humbug]

Others too ^^^ i'm thinking the Ryzen 5 1600 might come down to £150.

Ay, @ 3.7Ghz the numbers are probably right but it probably also has an all core boost that will clock up as high as thermals and power levels will allow, Much like nVidia's GPU Boost 3, so with good cooling it would overclock its self.

Should be interesting.

These scores are much higher and i think probably right when boosting to way over 4Ghz...

I think AMD's Ryzen 5 and 7 chips are about to get very cheap.

Actually we can work that out.

Apparently scoring 1230 points @ 3.7Ghz, on the chart i posted its scoring 1523 points, a difference of +25%, so 3.7Ghz + 25% = 4.625Ghz.

The chart says 3.7 - 4.7Ghz, is that 4.7Ghz an all core Boost? the score would suggest it is, the numbers add up.

 

[Evian]

Glad I waited for Intel.

 

[humbug]

Glad I waited for Intel.

Its not going to be a cheap CPU, over £350 for the 8700K maybe way over and its not going to do any better in games than the 7700K.

But if you are looking for Intel only and better productivity work than the 7700K for about the same money, yeah this is it.

 

[KompuKare]

I'd be pretty certain that the bus is not mesh.
Reason being that CoffeeLake is basically a 'quick port' of the Skylake architecture to a new 6C die and Intel will have changed as little as possible to get it to work. A change from ring to mesh would be a major change to the non-core part of the chip.
Basically, Codelake is a new revision of Skylake/Kabylake but Intel will now produce three dies (2C/4T, 4T/8T, and 6C/12T) instead of the two (2C/4T and 4C/8T) they previously did.
As for IPC*, at the same clock it should be the same unless the cache has changed or increased in size.

*Little IPC rant follows:
Skylake, Kabylake and Coffeelake are different implementations of the same architecture so IPC comparisons are an easy concept.
However, when comparing totally different architectures like Skylake vs Zen or Skylake vs Atom etc., things quickly get far more complicated and this is often missed when someone says "X is 15% faster/slower than Y", since what they really mean to say is "at running task T1, CPU-X is 15% faster/slower than CPU-Y". And of course that says nothing about task T2 or task T3 since it is possible and even likely that those tasks will not have the same ratio as task T1.
Actually, different implementations of the same architecture can have different IPC too as most recently witnessed with Skylake-X

 

[gavinh87]

Ay, @ 3.7Ghz the numbers are probably right but it probably also has an all core boost that will clock up as high as thermals and power levels will allow, Much like nVidia's GPU Boost 3, so with good cooling it would overclock its self.

Should be interesting.

These scores are much higher and i think probably right when boosting to way over 4Ghz...

I think AMD's Ryzen 5 and 7 chips are about to get very cheap.

Indeed, let the CPU wars begin!