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 .
SomeGuy133 said:
it depends on timings.
Click to expand...

When you note that some apps are latency sensitive and some are bandwidth dependent, one of the reasons for this is that 'latency' can also be measured for particular access magnitudes. The latency which you are calculating (which I agree with) is the instantaneous latency, however, as the magnitude of data that is needed increases, bandwidth affects effective latency to greater degrees.

This is why high bandwidth with relatively tight timings is important to consider in addition to low instantaneous latency. So I agree with you here as well:

SomeGuy133 said:
Just find a ram you can afford that has good latency and BW.
Click to expand...
 
IdiotInCharge said:
When you note that some apps are latency sensitive and some are bandwidth dependent, one of the reasons for this is that 'latency' can also be measured for particular access magnitudes. The latency which you are calculating (which I agree with) is the instantaneous latency, however, as the magnitude of data that is needed increases, bandwidth affects effective latency to greater degrees.

This is why high bandwidth with relatively tight timings is important to consider in addition to low instantaneous latency. So I agree with you here as well:
Click to expand...
my excel also go to 8th word which isnt very deep in regards to large files but gives you a little feel on how latency and BW interact and pros and cons of BW vs latency.

I learn towards latency because I care about snappiness of my total system (OS/program responsiveness) so I go with the highest BW that i can get while getting the lowest latency which was 3200 14-14-14-34 at the time. There is one that is slightly better now over all than my current RAM.
 
We are planning to update Tornado F5 to Z390 chipset supporting 8C/16T CPUs coming in H2/18. We will launch F7 at the same time too. We will skip z370 chipset. Meantime we added support for Quadro P5000 and P3000.
Click to expand...

looks like intel CNL will have 8 cores by 2018. but intel did say CNL wont have desktop CPU it'll just be CFL6c or possibly tigerlake 8c. time to wait and find out
 
Chips are popping up everywhere in shops. I suspect Intel announces this upcoming week, with pre-orders going live.

The mysteries around Z390 are certainly curious.
 
Looks like Intel is going to pull another "7700K" on the 8700K.
Guess we shouldn't be mad at innovation/competition?
 
It depends on how its priced I guess, I assume Z390 will also support Coffee lake (as shown in the leaked slide), so it isn't entirely the same thing. Though if it supports Coffee, then Icelake, surely it should support Tiger. 3 gens, isn't that a big complaint about Intel? Their chipset/socket life? So 3 gens should be nice I suppose
 
Ring or mesh though? Mesh has more potential and will probably be the main topology. They just gotta lower the latency and offer more L3 cache like those gold Xeons lol.
 


2:50~

"I won't be able to show you the 2nd system for another month or so." Coffee lake? :ROFLMAO:
 
Eurocom: "We are planning to update Tornado F5 to Z390 chipset supporting 8C/16T CPUs coming in H2/18. We will launch F7 at the same time too. We will skip z370 chipset. Meantime we added support for Quadro P5000 and P3000."
Click to expand...
Let me speculate on this since two things jump out at me.

1. The chipset is not Z470. Which makes me think it's not Icelake, ie: 9000 Series CPUs.
2. In the recent roadmap, the Z390 chipset still falls under Coffee Lake-S. Also, Icelake isn't listed at all in 2H. It's also odd to plan a chipset launch w/o a CPU alongside it.

Update:
3. The Icelake info a few posts down indicates Icelake PCH whereas Z390 is Cannonlake PCH (from previous leaks).

Based on this info, I'm going to say Z390 isn't Icelake and it's also not "Coffee Lake 2" / Refresh. Instead, I think Intel is releasing a new tier of premium 8C chips in the mainstream market paired with the Z390. The chipset will also be backwards compatible with Coffee Lake, so CFL-S owners can get the onboard Wi-Fi, etc etc. Not sure if the CPUs themselves would be backwards compatible, probably not but who knows. I think 10nm+ desktop is being delayed and they're planning ahead to respond to R+.

Also a reminder, the Z370 vs "Z390" slide we got a few months ago doesn't mention 8-core at all.
https://i.imgur.com/DgL9bkw.jpg

Which means 3 possibilites.
Option 1: "CNL 300 Series" is not Z390.
Option 2: The 8C is a new addition to Z390, added after the slide was made.
Option 3: Eurocom is wrong.

I'm hoping during CFL-S's launch in a few weeks, Intel will release info about the Z390 / 8C as part of the press packet. We already know CFL-S samples are in the wild, perhaps this is how Eurocom found out about it and leaked the info early.
 
Last edited:
I don't think it would make sense for Intel to release a new high tier chip on Z390, In H2 2018 when its supposedly launching. Awfully close to the Icelake supposed launch don't you think? If it is not Icelake I smell another delay with Intels 10nm process, which would be highly detrimental to Intel's business.
 
TaintedSquirrel said:
Let me speculate on this since two things jump out at me.

1. The chipset is not Z470. Which makes me think it's not Icelake, ie: 9000 Series CPUs.
2. In the recent roadmap, the Z390 chipset still falls under Coffee Lake-S. Also, Icelake isn't listed at all in 2H. It's also odd to plan a chipset launch w/o a CPU alongside it.

Based on this info, I'm going to say Z390 isn't Icelake and it's also not "Coffee Lake 2" / Refresh. Instead, I think Intel is releasing a new tier of premium 8C chips in the mainstream market paired with the Z390. The chipset will also be backwards compatible with Coffee Lake, so CFL-S owners can get the onboard Wi-Fi, etc etc. Not sure if the CPUs themselves would be backwards compatible, probably not but who knows. I think 10nm+ desktop is being delayed and they're planning ahead to respond to R+.

Also a reminder, the Z370 vs "Z390" slide we got a few months ago doesn't mention 8-core at all.
https://i.imgur.com/DgL9bkw.jpg

Which means 3 possibilites.
Option 1: "CNL 300 Series" is not Z390.
Option 2: The 8C is a new addition to Z390, added after the slide was made.
Option 3: Eurocom is wrong.

I'm hoping during CFL-S's launch in a few weeks, Intel will release info about the Z390 / 8C as part of the press packet. We already know CFL-S samples are in the wild, perhaps this is how Eurocom found out about it and leaked the info early.
Click to expand...

got a few more concerns

first was that originally, intel never meant to have CNL 10nm desktop it was only for low end mobile and mainstream mobile, and higher end mobile + desktop will have CFL 14nm++ instead.

assuming if things changed and intel decided to release 8c onto desktop mainstream, 2018 would be icelake? or still cannonlake?

last is assuming that it is going to be 10nm would it be first gen or 2nd gen 10nm, possibly 2nd if i had to guess. 1st gen 14nm was junk, look at broadwell silicon overclocking was trashy compare to SKL/KBL.
 
TaintedSquirrel said:
I'm hoping during CFL-S's launch in a few weeks, Intel will release info about the Z390 / 8C as part of the press packet. We already know CFL-S samples are in the wild, perhaps this is how Eurocom found out about it and leaked the info early.
Click to expand...

They found it because Intel already released the specs and platform info about ICL-S to motherboard manufacturers.

PMdMTDC.png


2018 launch is looking like a real possibility now. Desktop first, just like Skylake.
 
realworld said:
Ring or mesh though? Mesh has more potential and will probably be the main topology. They just gotta lower the latency and offer more L3 cache like those gold Xeons lol.
Click to expand...

I expect meshes to be the main topology in the long run with higher core counts. Even Nvidia has a long-term project for heterogeneous SoC around mesh.
 
TaintedSquirrel said:
Let me speculate on this since two things jump out at me.

1. The chipset is not Z470. Which makes me think it's not Icelake, ie: 9000 Series CPUs.
2. In the recent roadmap, the Z390 chipset still falls under Coffee Lake-S. Also, Icelake isn't listed at all in 2H. It's also odd to plan a chipset launch w/o a CPU alongside it.
Click to expand...

You are mixing chipsets and CPUs. H110 for example is also the chipset for Kaby Lake.
 
drescherjm said:
Depends when DDR5 and PCIe 4.0 arrive. I mean will Ice Lake have these??
Click to expand...
Intel are extremely conservative on this kind of stuff, maybe for our betterment, maybe not.

They adopted SATA 1/2/3 and USB 1/2/3 slowly on their chipsets. We got them in the end, but it was always stingy - start with 1 or 2 ports, next chipset, 4 ports, next chipset 6 and so on.
PCI-e - you can be almost entirely positive you won't see DDR5 or PCI-e 4.0 support for the ice lake chipset, MAYBE the one after.
 
bigdogchris said:
I can't believe how much more the 8700k is going to by than Ryzen 6c chip.
Click to expand...

If I got an AMD CPU, I'd be forced to spend another $60 US just because they forgot to put a simple GPU in there, like Intel did. (no really, some of us don't need video cards)
They also forgot to put some decent clock frequencies and IPC in there.

If they want my business, with that product? At least 25% cheaper. Then I'd consider that.

Edit: I'm serious too, the 1600 for say 140 us, total instant buy for me, even with the awful itx options for it.
 
Last edited:
AbRASiON said:
If I got an AMD CPU, I'd be forced to spend another $60 US just because they forgot to put a simple GPU in there, like Intel did. (no really, some of us don't need video cards)
They also forgot to put some decent clock frequencies and IPC in there.

If they want my business, with that product? At least 25% cheaper. Then I'd consider that.
Click to expand...
Classic AMD, cutting corners to save a few pennies.
 
-Sweeper_ said:
They found it because Intel already released the specs and platform info about ICL-S to motherboard manufacturers.

PMdMTDC.png


2018 launch is looking like a real possibility now. Desktop first, just like Skylake.
Click to expand...

so intel milked 4 cores for almost a decade, then jump to mainstream 6 cores then onto mainstream 8 cores all under 2 years. thank you AMD
 
One company cuts corners to make a dollar, another simply overcharged for tech we had the last 8 years or so. Both are set to make a profit so let's not debate who is the "good company".
 
Nightfire said:
One company cuts corners to make a dollar, another simply overcharged for tech we had the last 8 years or so. Both are set to make a profit so let's not debate who is the "good company".
Click to expand...

Price is not a linear function of performance. The company that choses lower performance targets almost automatically has an performance/price advantage.
 
I was trying to quench a fire before it spread, but you have an answer for everything. Um how do you know what the profit margins are for each chip? Do you really think it costs 5x as much to make a 7900x compared to an i5? If anything, the profit margins are higher with the top end product.
 
Nightfire said:
I was trying to quench a fire before it spread, but you have an answer for everything. Um how do you know what the profit margins are for each chip? Do you really think it costs 5x as much to make a 7900x compared to an i5? If anything, the profit margins are higher with the top end product.
Click to expand...

I don't have answer for everything, but I know some things.

* I know that IPC is not a linear function of number of transistors. Thus getting 10% higher IPC doesn't cost 10% more.
* I know that 512bit FMAC units and datapaths require much more transistors than 256bit or 128bit, and that extra space adds to the cost, but that extra cost is not counted when reviews only check serial x86 workloads or 128bit vector code.
* I know yields aren't linear. So dies with higher core counts cost much more than dies with lower core count.
* I know that developing a process node that can hit 5GHz costs much more than developing a process node that can hit 4GHz.
* I know that developing a process node with HD cell density of 0.0499μm² costs much more than developing a process node with 0.064μm² density.

Summing all, we find that the CPU that targets higher performance will cost much more and will have a lower performance/price ratio that the CPU that targets lower performance. And all that without even considering production volumes. Mainstream products fabricated in larger volumes are cheaper than enthusiast products fabricated in low volume, as follows from a simple application of economic laws of scale and serial fabrication.

That is why Intel charges more for its chips than AMD. Some of the above reasons also apply when comparing i9 vs i5.

Note that AMD charges $500 for the 1800X and $300 for the 1700, despite both are same microarchitecture, same core count, same die, same process node... I can sure you that if AMD ported the 1800X to Glofo 14HP node, added two extra cores, and updated to 512bit, the resulting 10 core chip wouldn't cost $600, but more than $1000.
 
Last edited:
Ah. And can you explain why the 6950x cost $1700 when it was released and now the 2 extra cores of the 8700k are "magically" free or at least a small increase??
 
Nightfire said:
Ah. And can you explain why the 6950x cost $1700 when it was released and now the 2 extra cores of the 8700k are "magically" free or at least a small increase??
Click to expand...
6950x was their flagship halo CPU. Just like the the Titan is $1200 when the next step down is almost half the price for what 5% less performance. Halo products are always like that. The 8700 is just their mainstream CPU.
 
Spoon-feeding paid AMD shills is pointless, just block/ignore every account that doesn't know what it's talking about that conveniently registers their alt account recently.

AMD's paying of reviewer shills slowed down after launch, but their paying of viral marketing shills clearly hasn't lowered by as much.
 
Nightfire said:
Ah. And can you explain why the 6950x cost $1700 when it was released and now the 2 extra cores of the 8700k are "magically" free or at least a small increase??
Click to expand...
HEDT has quad channel (or is it 6 channel now?) and more PCIe and has other more advance instruction sets...thats biggest reasons why it costs more.

there now stop posting off topic
 
juanrga said:
Intel planned 4/6/8 core for mainstream on 2014--2015. On the other hand ThreadRipper wasn't initially planned by AMD. Thank you Intel.
Click to expand...

did it come out 2014-2015? yea its hardcore m!lking. thank you AMD for pushing intel, now i can get tiger/icelake w/e lake it is called 8 cores in a laptop.
 
Nightfire said:
I was trying to quench a fire before it spread, but you have an answer for everything. Um how do you know what the profit margins are for each chip? Do you really think it costs 5x as much to make a 7900x compared to an i5? If anything, the profit margins are higher with the top end product.
Click to expand...

Do you think a 1800X cost 4x as much as a 1200 to make? Or a 1950X 9x?

If you dont understand simple business segmentation then the question is plain silly. It´s so with all products. Premium products got premium prices.
 
You must log in or register to reply here.
Back
Top