Intel's 8th Generation Core Family - Coffee Lake (LGA 1151, 6C/12T)

Where do you expect Core i7-8700K's Turbo to land?

  • 3.8/3.9 GHz

    Votes: 0 0.0%

  • 4.0/4.1 GHz

    Votes: 3 23.1%

  • 4.2/4.3 GHz

    Votes: 6 46.2%

  • 4.4/4.5 GHz

    Votes: 3 23.1%

  • 4.6/4.7 GHz

    Votes: 1 7.7%
  • Total voters
    13
  • Poll closed .
AbRASiON said:
So the next desktop chip after October 5 is when and what exactly?
Click to expand...

The next desktop chips are the rest of the CFL line that comes in Q1. Else you are looking at Q3 or Q4 2018 for ICL-S with Z390 and another early line of around 6 SKUs.
 
Shintai said:
TSMC just renamed a process, cant be left behind in that game :p
Click to expand...

Global-Foundaries-12LP.jpg


So this Glofo 12nm node is a 15nm node per TSMC standards, aka a real 18nm node :whistle:
 
Wait! Because it can be even more hilarious.

Glofo is using TSMC as baseline for marketing claims. I did some computations and I get that this 12LP would bring about 5% density improvements over current 14LPP. This added to the claim made on the slide "12LP builds on the GF 14LPP platform" makes me believe this 12LP node is just a rename for the older 14nm+ node :eek:

aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS9HL0cvNjc3MTA0L29yaWdpbmFsLzE1LkpQRw==


So we have Zen on 14nm this year. Zen on 14nm+ (renamed to '12nm') in second half of 2018 and Zen2 on '7nm' somewhat in 2019. On Intel side we have 10nm Canonlake mobile and 14nm++ CoffeLake desktop this year and 10nm+ Icelake next year. So 6-core CoffeLake will humiliate compete with Summit Ridge, and 8-core Icelake will kill compete with Pinnacle Ridge. Is this resume more or less correct?
 
Last edited:
juanrga said:
So this Glofo 12nm node is a 15nm node per TSMC standards, aka a real 18nm node :whistle:
Click to expand...

This is why I'm confused at the razzing people want to give Intel over their 10nm roadmap and progress. There is a lot you can give Intel legit crap about, but their fab tech is really not one of those things.

Fun side note, this density and feature size is actually well beyond what really smart people thought was going to be possible with silicon when I was in college. But I did have one prof who indicated prophetically "The future is GaAs, and always will be." We just keep getting better at wringing blood from a stone (silicon!).

Now of course, this was back in the "should we develop The Wheel, or go for the Granary" level of tech, since I'm an old fart. But it's still a fun memory.
 
Shintai said:
8700K vs the 1600? Hahahaha that's a good one. :D

3.2Ghz 6 core Ryzen with ST? 3.6Ghz boost and more than not SB IPC, against a 8700K that clocks 4.3Ghz on all cores with turbo and 4.7Ghz ST.

MT wise the 8700K is around the 1800X. In less than optimal threads for the 1800X the 8700K runs in circles around it. Expect big price cuts on the entire Ryzen lineup.
Click to expand...

If Ryzen = SB ipc, Intel has really drug their feet..
http://www.tomshardware.com/reviews/amd-ryzen-threadripper-1950x-cpu,5167-11.html
 
juanrga said:
It is not = but ~ and we are talking about the average IPC, not only on Blender, Luxrender, 3DS Max, and other GPU-like tasks, where the throughut-oriented Zen muarch does well.
Click to expand...
??? IIRC overall ryzen is haswell on average give or take the metric. Anyone who says more or less one way is a shill ¯\_(ツ)_/¯
 
SomeGuy133 said:
??? IIRC overall ryzen is haswell on average give or take the metric. Anyone who says more or less one way is a shill ¯\_(ツ)_/¯
Click to expand...

But it isn't. In some cases it is, in more often than not its far from it.
 
TaintedSquirrel said:
Icelake is 10nm+
Click to expand...

the thing is, although 10nm is delayed to end of 2018 possibly early 2019 like originally planned, eurocom brought up the 8c on 2H2018 which could literally mean end of 2018. i was thinking along the line is if that 8c is around sept release then it'll just be 14nm++ a refresh of CFL except 8core.
 
oleNBR said:
the thing is, although 10nm is delayed to end of 2018 possibly early 2019 like originally planned, eurocom brought up the 8c on 2H2018 which could literally mean end of 2018. i was thinking along the line is if that 8c is around sept release then it'll just be 14nm++ a refresh of CFL except 8core.
Click to expand...

The delay got debunked.
 
SomeGuy133 said:
??? IIRC overall ryzen is haswell on average give or take the metric. Anyone who says more or less one way is a shill ¯\_(ツ)_/¯
Click to expand...

About 10--20% behind Broadwell is not Haswell. There are a pair of benches (usually rendering where Zen IPC is above Broadwell, but overall it is behind).

About 10% behind on applications

getgraphimg.php


About 20% behind in games

getgraphimg.php
 
Random unlinked fruity pebble charts brought to you by Juangra.

It isn't always just the bios. Ages 1006 brought much higher memory capability as well.

Like I say, show me where SB is even close to TR, since those reviews always seem to have descent memory and a new bios, and I will submit.
 
You really love that Russky chart lol!

As for the pc gamer link, I saw no mention of the software/ram for the TR.

But still, the IDF will put more weight on this single, clearly unoptimized, game over a 30 game roundup from techspot.

At least you got your 'like' from sweeper.
 
Right, you'd rather believe the single website that managed to put up fairy tale results where a 4.9 GHz 7700K is only on par with Ryzen 5. Cherry picked GPU-bound results are a no-no to people who read multiple reviews and follow GameGPU's testing. There's plenty of numbers above showing a significant advantage for SKL core based CPUs in multiple titles, and, hint, they won't disappear if you ignore them.

Interesting to see how many new regs are posting here right before CFL-S launch (I bet the one above is from AT). Keep doing damage control guys, very soon a ~$350 all-arounder from Intel will manage to match the more expensive 1800X on average in MT apps while murdering it in games. I guess there's always Techspot when you need to feel good about your purchases. :D
 
Last edited:
It doesn't seem o be patched for CORES since the R3/i3 hold their own.

$350? We will see. It will still be way above the 1700/x which most get anyway.

The 8700k will not compete with R7 directly. More so, the 8700 will.
 
AGESA 1006 coupled with overclocked samsung b-die can bring Ryzen systems much closer to intel systems in most games but the problem lies with the fact that A) you have to buy 200+ dollar ram to make this happen and B) the more you push the IMC on Ryzen the harder it gets to maintain a good core overclock at or near 4 ghz which is also important to catch a 7600k or 7700k at stock speeds in most games. once you overclock either intel K SKU's they are easily back out ahead in the vast majority of games and they don't need expensive B-die to do it because they don't have infinity fabric tied directly to the DDR4 clock speed hampering performance.

A LOT of my friends are on Ryzen now and they love it, I even considered it for a time but simply could not stand the thought of regressing in gaming performance so decided to hold off and see what Intel does. its highly likely i'll be buying a 8700k and Maximus X Apex next month once they are available esp. if daily OC's of 4.7+ ghz are the norm i'm definitely all in with CFL, its easily best all around choice at that point and perhaps will even be a 7700k Slayer in gaming.

another thing to note is that even though Ryzen produces really great physics scores in 3dmark it suffers in combine and graphics scores just like real games do and thus in most cases ryzen isn't even competitive on HWBOT with Intel cpus even though it looks like it should be with how well it excels at 3dmark physics tests, the other main reason its not competitive is because its a pain to overclock past 4 ghz even on ln2 due to CB and CBB.
 
12nm LP from GloFo is risk production Q1 2018, if it's used for Zen + surely that means its Q2-Q3, then Zen 2 on GloFos 7nm would be Q2 2019 or later, anyone else think their 7nm got delayed? 12nm LP came out of no where and delays with Glofo aren't exactly unheard off.
 
Dayman said:
12nm LP from GloFo is risk production Q1 2018, if it's used for Zen + surely that means its Q2-Q3, then Zen 2 on GloFos 7nm would be Q2 2019 or later, anyone else think their 7nm got delayed? 12nm LP came out of no where and delays with Glofo aren't exactly unheard off.
Click to expand...

12nm is actually H2 2018 for any kind of production. "7nm"? Likely 2020.
 
Dayman said:
12nm LP from GloFo is risk production Q1 2018, if it's used for Zen + surely that means its Q2-Q3, then Zen 2 on GloFos 7nm would be Q2 2019 or later, anyone else think their 7nm got delayed? 12nm LP came out of no where and delays with Glofo aren't exactly unheard off.
Click to expand...

7nm at GF is in risk production but with the current way they make it it's not ready for mass production. EUV machines have to be installed for mass production, depending on how smooth that goes is when 7nm chips will be available. The original plan is 2018 Q4 for 7nm, only AMD knows if that will happen or not, the 12nm is more or less just a improved 14nm process which might allow slightly better clocks but nothing earth shattering which may allow for a 1900X version to be released.
 
You must log in or register to reply here.
Back
Top