Intel "Coffee Lake" (2017/2018, 14 nm)

There is zero doubt that perf/price improved massively as long as the task concerned is parallelizable.

And unlike Ryzen, where the huge improvement was at high prices & high core counts, the most massive improvement with CoffeeLake is at low core counts & low price points.
No love for the 1600? unlocked and the Same price as an i5 7600.
 
No love for the 1600? unlocked and the Same price as an i5 7600.
I count it at the expensive end. And in any case, 1600 did not make my future PC cheaper. Quad i3 did (though yes, I would like a cheaper motherboard; but unlike Ryzen, the i3 does not need a GPU).
 
there is no proof of any technical limitation
Looks like 8000-series Coffee Lake processors and 300-series chipsets do have a new socket, LGA 1151v2, which re-designates previously reserved pins to VCC (+5V) an VSS (ground), and PROCC_DETECT#/SKTOCC# pin is moved to a new location (AC38 instead of AB35), so new processors and chipsets are not compatible with existing ones based on current LGA 1151.

http://wccftech.com/intel-coffee-lake-lga-1151-pin-configuration-detailed/

https://www.intel.com/content/www/u...cessor-family-s-platform-datasheet-vol-1.html
https://www.intel.com/content/www/u...ktop-6th-gen-core-family-datasheet-vol-1.html

Coffee Lake LGA 1151v2 vs. Kaby Lake LGA 1151

3d2FYWy.png


imgur.com/a/Af8DI
 
Last edited:
Looks like 8000-series Coffee Lake processors and 300-series chipsets do have a new socket, LGA 1151v2, which re-designates previously reserved pins to VCC (+5V) an VSS (ground). So both are not compatible with existing chipsets and CPUs based on current LGA 1151.

http://wccftech.com/intel-coffee-lake-lga-1151-pin-configuration-detailed/


Coffee Lake LGA 1151v2 vs. Kaby Lake LGA 1151

r8UcAYE.png

https://imgur.com/a/F2sNj

Then they should just change the socket and/or just tell the difference since the beginning and avoid the "confusion"
 
Intel won't make a new socket - they already made a new incompatible revision. As for 'confusion', Intel already did this with LGA 2011 socket, which had three different incompatible revisions during the course of production, and nobody cared. It's just:

1) repurposing some of the pins is simpler than changing the mechanicals and introducing a new socket - and cheaper for Intel and its OEMs;
2) the PC market is shrinking and very few people end up buying a desktop PC - most new users buy notebooks or Android/Apple tablets, none of which are CPU upgradeable;
3) of those who still buy traditional desktops, very few ended up upgrading their CPU or GPU - mostly because mid-range products practically stalled for the last 7 years.
 
3) of those who still buy traditional desktops, very few ended up upgrading their CPU or GPU - mostly because mid-range products practically stalled for the last 7 years.

Of course. And such practices contribute to them being very few.
 
such practices contribute to them being very few.

AMD did keep their CPU sockets for like 10 years, and their market share was still eaten by Intel - until they dramatically improved their performance with Ryzen, and replaced their sockets as well :)


Case in question. I still use Intel Core i5-2500K, a LGA 1155 part from 2011, because I couldn't make a good reason for replacing it with any recent mid-range CPU.

I replaced my GPU twice during this time, and each new graphics card offered at least 2x real-world performance gain - though these were high-end cards, HD6950 and R9 290X. To achieve a modest 2x gain on the CPU side in very select workloads, I would need to buy top-end 6- and 8-core LGA 2011 parts - everything else would be essentially the same in real-word performance. I couldn't justify such an upgrade, even without considering the cost of a new motherboard and memory.

So would it really help if all successive generations of Intel Core processors used the same socket? I don't think so.


6-core Coffee Lake CPUs look interesting, but they still barely offer a 2x real-world performance gain - I would still need insanely priced top-end 8-core or 10-core i7-X or i9-X series parts to break it, and it would only show in quite a few very specific workloads optimized for many-core processors.

http://cpu.userbenchmark.com/Compare/Intel-Core-i7-8700K-vs-Intel-Core-i5-2500K/3937vs619
http://cpu.userbenchmark.com/Compare/Intel-Core-i9-7900X-vs-Intel-Core-i5-2500K/3936vs619
 
Last edited:
AMD did keep their CPU sockets for like 10 years, and their market share was still eaten by Intel - until they dramatically improved their performance with Ryzen, and replaced their sockets as well :)

I'm left speechless by the collosal strength of this argument



Anyway, mid-end skylake owners would get quite a bit of value by upgrading to 8700K.

More than the question of "would anyone actually upgrade ?" , there's the fact that w/ Cofee the socket change was uncalled for and artificial. Intel could have decided to help the consumers , but actively went against them instead (I guess mb vendors stand to benefit)
 
The fact that some reactions/statements are repeated regularly does not make them wrong, I'm sorry. Or right. So it's irelevant wether I'm playing bingo or not.

Also I said that that Intel is moving against consumers. That doesn't mean consumers will stop buying Intel ( or that they'll start to..) . It's a strech (pointless, for now I'd say) to jump straight from a socket change to market share trends. Not sure why you did that
 
6-core Coffee Lake CPUs look interesting, but they still barely offer a 2x real-world performance gain -
It's actually faster than that, gaming and otherwise, 8700K is practically on the heels of Ryzen 7 1800X.

Although Intel took its sweet time in delivering the coup de grace, the i7-8700K dramatically puts the final nail in the coffin of 2011's Core i7-2600K. Clock for clock, the i7-8700K delivers 2.5 times the performance of Sandy Bridge's finest four-core avatar in our productivity index.
http://techreport.com/review/32642/intel-core-i7-8700k-cpu-reviewed/16
 
8700K is more interesting than Ryzen 7, but I'm still disappointed as it's roughly just a Skylake CPU with two more cores.
Intel CPUs have had the same IPC for three "generations" now.
 
It's actually faster than that, gaming and otherwise, 8700K is practically on the heels of Ryzen 7 1800X.
Clock for clock, the i7-8700K delivers 2.5 times the performance of Sandy Bridge's finest four-core avatar in our productivity index.
They assume i7-8700K is 2.5 times faster than i7-2600K, but that's just the bias of their weighting formula.

It's certainly not that faster "clock-for-clock", i.e. clocked on the same frequency - the last three major generations of Intel processors (i.e. Sandy/Ivy Bridge, Haswell/Broadwell, and Skylake/Kaby Lake/Coffee Lake) have very similar IPC , and real-world performance gains come from increased clocks and cache memory, made possible by process node improvements.

So performance gain can be roughly estimated using turbo clocks and number of cores: for Core i5-2500K and Core i7-8700K, that would be (4.3*6)/(3.3*4) = 1.9, or 90% faster - assuming workloads that scale very efficiently with many-cores, such as video encoding.
If you count in HyperThreading in Corei7, that could give you additional 20-30% over Core i5 in ideal conditions.

So in the end, 8700K would be about 2x (or 100%) faster, but not by a wide margin, in many-core workloads. For single-threaded workloads, it should be about 30% faster (4.7/3.7=1.3).
 
It's a strech (pointless, for now I'd say) to jump straight from a socket change to market share trends. Not sure why you did that.... I said that that Intel is moving against consumers
Because you are overlooking who are really Intel's customers.
Processor sockets do not exist to provide an easy upgrade path for the end-user - though it could be advertised as such for the enthusiasts. It's rather for the convenience of desktop PC builders, to give them flexibility with custom-built configurations and support/repairs.
They don't really care for cross-compatibilty between CPU generations, as they can order any part through the OEM channels.

The fact that some reactions/statements are repeated regularly does not make them wrong, I'm sorry.
It's not the quantity of these statements, it's their quality.
 
Because you are overlooking who are really Intel's customers.
Processor sockets do not exist to provide an easy upgrade path for the end-user - though it could be advertised as such for the enthusiasts. It's rather for the convenience of desktop PC builders, to give them flexibility with custom-built configurations and support/repairs.
They don't really care for cross-compatibilty between CPU generations, as they can order any part through the OEM channels..

Okay, (in an attempt) to remove all ambiguity , by customer/consumer I meant end users. So you now say Intel doesn't care about end users, with regards to sockets at least. We're in agreement then; in this instance at least Intel behaved as if it would not care.

It's not the quantity of these statements, it's their quality.

Then the quality was just dismissed by you considering them similar to that bingo thinggie, not addressed.
 
Back
Top