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 .
Looks like the mobos are ready and CPUs too.... might not be that much time for NDA to lift for reviews (if they are not on 21st August) and a short time till shelf launch.
 
TaintedSquirrel said:
This launch has made me realize what a great investment the 5820K would've been.
6C CPU mostly capable of 4.6~4.7 GHz, the first (?) DDR4 high end platform. It would've cost an extra $100-$200 over Coffee Lake but you would've had it a little over 3 years ahead of time. Also, soldered, and actually better than it's replacement 2 YEARS later (6800K).
Click to expand...

I've been very happy with mine at 4.5 in my Ncase :)
 
The 8700K looks like a decent chip for sure, but I wonder about pricing. I would not put it past Intel to delay the launch until next year which gives them time to sell the 7800X. Then, they will drop the six core from the HEDT lineup and and make the 8700K the flagship part on mainstream at $400. Hopefully not the case though.
 
Revenant_Knight said:
5820K is an unsung hero of the CPU world. If more games could take advantage of it, we'd be calling it the new Q6600 of its time.
Click to expand...
The recent PUBG updated added 6 core support and actually did noticeably increase my frame minimums so hopefully (thanks to Ryzen) 6 core support becomes more commonplace
 
I am really excited about this arms race between blue and red. Hope it drives up quality and down cost!
 
TaintedSquirrel said:
What if they announce on the 21st but don't launch until 5 months later? Long wait.
Click to expand...

MorgothPl said:
4,3/4,7 meaning that it can go up 4o 4,7 boost? If so, it might be a beasty thing :)
Click to expand...

fk better not be paper launch. i want 8700k to go into a laptop asap i need to upgrade from my 1680v2 laptop. 8700k 6c at 4.9-5ghz will do nicely, beating single threaded performance ivy at 4.3ghz and matching almost closely to it in multi threaded performance, along with bunch of new features like NVME SSDs, DMI 3.0, USB 3 and 3.1 etc.

all that so i can play thief gold, lemmings and total annihilation
 
TaintedSquirrel said:
We're starting to get Z370 board leaks so it's probably not going to be too long. Also one of the Intel materials mentions the new CPUs being available prior to the holiday season, meaning Sept/Oct most likely.
Click to expand...

thank god. had windows and bunch of all other junk software are as optimized as CB15 then i'd have gone TR without 2nd thought. 80% of the software i use are still legacy software which only uses ST and will never get the optimization or update. to make it fast as possible and as snappy as possible we need intel's high 5ghz high IPC so sad
 
oleNBR said:
thank god. had windows and bunch of all other junk software are as optimized as CB15 then i'd have gone TR without 2nd thought. 80% of the software i use are still legacy software which only uses ST and will never get the optimization or update. to make it fast as possible and as snappy as possible we need intel's high 5ghz high IPC so sad
Click to expand...

CB15 IS legacy. Its 5 years behind their Cinema4D engine that's now on version 19. And CB15 is made without supporting any form of newer instruction types. Its as legacy as it can be. You can run it on a K6/P3 and still get no performance penalty.

And CB with its tile based rendering cant be used to compare any form of normal software. Its like claiming the GPU could run all software better and faster than the CPU. Same reason nobody is using CPU to do the rendering in CB in the first place.

And then there is the fun when multithreading turns into negative performance.
https://blogs.msdn.microsoft.com/dd...ilds-play-part-2-amdahls-law-vs-gunthers-law/

Examples:
pic_disp.php

pic_disp.php

pic_disp.php
 
Last edited:
Me thinks all 6 cores @ 4.8 air cooled and all 6 cores @ 5.0 water/AIO. That would rock.
 
Shintai said:
CB15 IS legacy. Its 5 years behind their Cinema4D engine that's now on version 19. And CB15 is made without supporting any form of newer instruction types. Its as legacy as it can be. You can run it on a K6/P3 and still get no performance penalty.

And CB with its tile based rendering cant be used to compare any form of normal software. Its like claiming the GPU could run all software better and faster than the CPU. Same reason nobody is using CPU to do the rendering in CB in the first place.

And then there is the fun when multithreading turns into negative performance.
https://blogs.msdn.microsoft.com/dd...ilds-play-part-2-amdahls-law-vs-gunthers-law/

Examples:
pic_disp.php

pic_disp.php

pic_disp.php
Click to expand...


yes i understand CB is considered as "legacy" compare to some of other software like dophin and other enterprise software which uses avx2 or 512 the newer extension. however CB uses AVX and is multi threaded and very well optimized in that regard, where as majority of consumer software aren't even close to CB15. of the pictures you posted, i use none of them and actually never heard of them at all rofl, also those effiency after CPU1 or # of cores, wouldnt TSX support in those software help?
 
oleNBR said:
yes i understand CB is considered as "legacy" compare to some of other software like dophin and other enterprise software which uses avx2 or 512 the newer extension. however CB uses AVX and is multi threaded and very well optimized in that regard, where as majority of consumer software aren't even close to CB15. of the pictures you posted, i use none of them and actually never heard of them at all rofl, also those effiency after CPU1 or # of cores, wouldnt TSX support in those software help?
Click to expand...

CB15 benchmark doesn't use AVX. Cinema4D does.

And using CB15 as an example of multithreading that should be applied to everything else couldn't be more wrong. Also quad 8180 "breaks" CB15 ;)



TSX only work in somewhat limited cases, SQL being one of them. But it doesn't magically solve the issue. And that's in a case where you can scale concurrency.

The reason you have negative scaling is explained in the blog link. Not something you can magically fix.

People have to accept that we already have multithreading and its not going to change much. Multithreading isn't new, its been there for over 30 years in the professional segments.

In the old days speculative threading was attempted, but you had to "abuse" power to gain a benefit. Think a 4Ghz 16 core, being 20% faster than a 4Ghz quad, but using 300% more power/resources to do it.

mitosis1.jpg
mitosis2.jpg


This is from Adobe Premiere Pro CC 2015.
pic_disp.php

pic_disp.php

pic_disp.php

pic_disp.php


This is also why a 8700K is the new gaming king without any remote kind of competition.
 
Last edited:
Shintai said:
CB15 benchmark doesn't use AVX. Cinema4D does.

And using CB15 as an example of multithreading that should be applied to everything else couldn't be more wrong. Also quad 8180 "breaks" CB15 ;)



TSX only work in somewhat limited cases, SQL being one of them. But it doesn't magically solve the issue. And that's in a case where you can scale concurrency.

The reason you have negative scaling is explained in the blog link. Not something you can magically fix.

People have to accept that we already have multithreading and its not going to change much. Multithreading isn't new, its been there for over 30 years in the professional segments.

In the old days speculative threading was attempted, but you had to "abuse" power to gain a benefit. Think a 4Ghz 16 core, being 20% faster than a 4Ghz quad, but using 300% more power/resources to do it.

mitosis1.jpg
mitosis2.jpg


This is from Adobe Premiere Pro CC 2015.
pic_disp.php

pic_disp.php

pic_disp.php

pic_disp.php


This is also why a 8700K is the new gaming king without any remote kind of competition.
Click to expand...


wait what i thought CB15 uses avx, thats why sandy and beyond cpus does so much faster in ST vs first gen i7 the 980 and 990.
also i heard this from somewhere cant recall now donno if its true, that avx/avx2 are memory extensive so usually with only dual channel we see benefit going from non-avx to avx workload but from avx to avx2 theres little improvement. do u know anything about that

the thing i wanted with 8700k is the high ipc/core count without mesh/cache rework, as well as the IGPU which skylake-x didn't have.
 
oleNBR said:
wait what i thought CB15 uses avx, thats why sandy and beyond cpus does so much faster in ST vs first gen i7 the 980 and 990.
also i heard this from somewhere cant recall now donno if its true, that avx/avx2 are memory extensive so usually with only dual channel we see benefit going from non-avx to avx workload but from avx to avx2 theres little improvement. do u know anything about that

the thing i wanted with 8700k is the high ipc/core count without mesh/cache rework, as well as the IGPU which skylake-x didn't have.
Click to expand...

AVX isn't used and you can run it with it disabled if you got an AVX CPU and see no difference. Another obvious hint is the power consumption when running CB15. 55-60W on a 95W CPU.

cbpwr.png



i7 2600 gets ~600 at stock 3.4Ghz.
i7 860 gets ~565 at 3.46Ghz.
i7 980 gets ~790 at 3.33Ghz.

1415814040_991.png

1502096199_925.png

1428523747_396.jpg
 
Last edited:
Shintai said:
AVX isn't used and you can run it with it disabled if you got an AVX CPU and see no difference. Another obvious hint is the power consumption when running CB15. 55-60W on a 95W CPU.

cbpwr.png



i7 2600 gets ~600 at stock 3.4Ghz.
i7 860 gets ~565 at 3.46Ghz.
i7 980 gets ~790 at 3.33Ghz.

1415814040_991.png

1502096199_925.png

1428523747_396.jpg
Click to expand...

oh nice so its purely ipc improvement from first gen to sandy. so what software can bench and show difference with/without ipc
 
oleNBR said:
oh nice so its purely ipc improvement from first gen to sandy. so what software can bench and show difference with/without ipc
Click to expand...

The problem is how to measure IPC without the influence of cache, uncore etc. Chips with EDRAM vs chips without. Smaller and bigger caches vs one another etc.

Cache in one way or the other wins twice in this example. Not actual core performance.

pcars-fps.gif


And again here:
euler3d.gif
 
Last edited:
Shintai said:
The problem is how to measure IPC without the influence of cache, uncore etc. Chips with EDRAM vs chips without. Smaller and bigger caches vs one another etc.

Cache in one way or the other wins twice in this example. Not actual core performance.

pcars-fps.gif


And again here:
euler3d.gif
Click to expand...


sorry i meant to say "oh nice so its purely ipc improvement from first gen to sandy. so what software can bench and show difference with/without avx", not ipc

and yeah i get the part as it is software dependent. there are many opinions of what IPC should really be, some sugguest that its both hardware and software, however i believe that hardware itself is capable of something regardless of software. silly analogy would be saying race car has same "IPC" or in this case horse power as a crappy car simply because the road has a speed limit, however the argument from the other side is also true, assuming software does take full advantage of hardware.

in this case imho cache/uncore are all built into as part of the CPU shouldnt that be considered as part of IPC too unless one strictly testing the core design. of course i know some software uses cache/ram more than others, in those instances those software do not test "core" IPC well but a truely optimized software will take full advantage of the full cpu which includes cache/uncore etc so shouldnt that be the right way of calling measure of a CPU's IPC?

both are conflicting and both sounds correct, any thought?
 
Last edited:
Soo.....I know it's been asked but..... 3770K here for the last 5 plus years....Really been dying to upgrade and always keep holding back due to these continued new sku's and release dates.... I want to build a new rig solely for gaming. No streaming, no content creation, etc. Wondering about the PCI-e lanes

on these mainstream coffee sku's though. I don't think it will make a huge difference, but I would like to continue using my SLI config with 2 1080 Ti cards. I am also one of the old die hard dedicated sound card users, so I want to continue using my X-fi card in the PCI-E 1x slot. Also, I would like to try out a

new NVME drive to roll a fresh windows 10 OS onto, and keep a SSD or large 6TB plus mechanical drive for my gaming needs. I think the upcoming coffee lake 8700k would be the best choice. I really don't care about price, but if I can't justify spending over a grand for a cpu just to game, why spend the

money if won't benefit my needs and I don't have to. thoughts?
 
Island said:
Soo.....I know it's been asked but..... 3770K here for the last 5 plus years....Really been dying to upgrade and always keep holding back due to these continued new sku's and release dates.... I want to build a new rig solely for gaming. No streaming, no content creation, etc. Wondering about the PCI-e lanes

on these mainstream coffee sku's though. I don't think it will make a huge difference, but I would like to continue using my SLI config with 2 1080 Ti cards. I am also one of the old die hard dedicated sound card users, so I want to continue using my X-fi card in the PCI-E 1x slot. Also, I would like to try out a

new NVME drive to roll a fresh windows 10 OS onto, and keep a SSD or large 6TB plus mechanical drive for my gaming needs. I think the upcoming coffee lake 8700k would be the best choice. I really don't care about price, but if I can't justify spending over a grand for a cpu just to game, why spend the

money if won't benefit my needs and I don't have to. thoughts?
Click to expand...

7700k would do just fine but if pricing not a concern go for 8700k. 7700k or 7740x on x299 could definitely allow the highest overclock means u'll get highest gaming performance. 5.3ghz ish is possible with just 4cores.
 
Island said:
Soo.....I know it's been asked but..... 3770K here for the last 5 plus years....Really been dying to upgrade and always keep holding back due to these continued new sku's and release dates.... I want to build a new rig solely for gaming. No streaming, no content creation, etc. Wondering about the PCI-e lanes

on these mainstream coffee sku's though. I don't think it will make a huge difference, but I would like to continue using my SLI config with 2 1080 Ti cards. I am also one of the old die hard dedicated sound card users, so I want to continue using my X-fi card in the PCI-E 1x slot. Also, I would like to try out a

new NVME drive to roll a fresh windows 10 OS onto, and keep a SSD or large 6TB plus mechanical drive for my gaming needs. I think the upcoming coffee lake 8700k would be the best choice. I really don't care about price, but if I can't justify spending over a grand for a cpu just to game, why spend the

money if won't benefit my needs and I don't have to. thoughts?
Click to expand...
I think you can swing it on PCIe 3.0 x8 for each GPU and have the X-Fi and a NVMe SSD run off the DMI 3.0-provided lanes left over without much performance loss.

After all, you did just admit it was a pure gaming build.

However, when I start factoring in things like additional USB 3.0 host controller cards and video capture/framegrabber cards that want PCIe x4 slots each, that's when the lack of lanes really hurts. I may very well abandon Intel for my next build if they don't cut the crap on their HEDT platform with their needless hobbling, since I've already got my 4770K system nearly maxed out on PCIe lane usage.
 
Gulvan said:
I can't wait to see these benched vs Ryzen. Man exciting times!!

Now if only SSDs battled each other to drive down prices...
Click to expand...
Super Super likely an 8700k will outright beat an 1800x in almost all benchmarks (multithreaded included...) - go look at 1800x reviews when it came out, a heap of multithreaded benchmarks, the 7700k was really not that far behind.

That clock speed....
That IPC.

and no, I'm not an intel fanboy, I buy what's best for my wallet.
 
AbRASiON said:
Super Super likely an 8700k will outright beat an 1800x in almost all benchmarks (multithreaded included...) - go look at 1800x reviews when it came out, a heap of multithreaded benchmarks, the 7700k was really not that far behind.

That clock speed....
That IPC.

and no, I'm not an intel fanboy, I buy what's best for my wallet.
Click to expand...

well lets see, 8 cores is 33.3% more cores than 6 cores. 8700k assuming it can hit 5ghz at 6 cores and 1800x can hit 4ghz at 8 cores, intel's frequency is about 25% higher. intel has about 6% IPC advantage over ryzen so overall ryzen should still have an edge in multi threaded scenarios in ideal situation.

of course we know real world ideal situation dont exist, CB15 is very close as it is well optimized and scales perfectly with ryzen cpu but more cores means worse scaling so 6c each core efficiency should be better than each cores in 8, unless CCX design says otherwise.

but yes i agree 6c 5ghz will come VERY CLOSE to 8c at 4ghz ryzen in multi threaded scenarios. in ST it'll just blow ryzen away.
 
Been following this thread and waiting to upgrade the 3570k. I grabbed some 3000mhz ddr4 and an asus z270 strix itx board which I haven't opened. At this point if it doesn't support z270 which I highly doubt since what's the point of z370 followed closely with z390 other than compatibility I'm planning on selling the z270 and waiting for z390 in the spring. Since 4c/8t ~5 real core and 6c/12t ~7.5 real cores the 8600k may be in my future if the 8700k is priced at $400 and 8600k closer to $200. Seems like scaling sweet spot is between 6 and 8 cores
 
If it's backwards compatible, people are going to say Intel changed their mind at the last minute because of all the outrage from the leaks. Mark my words.
 
TaintedSquirrel said:
If it's backwards compatible, people are going to say Intel changed their mind at the last minute because of all the outrage from the leaks. Mark my words.
Click to expand...

THIS.

Though its more than likely that Asrock were wrong if that's the case, I had a friend contact Asrock and they said Broadwell E wasn't on x99
 
oleNBR said:
well lets see, 8 cores is 33.3% more cores than 6 cores. 8700k assuming it can hit 5ghz at 6 cores and 1800x can hit 4ghz at 8 cores, intel's frequency is about 25% higher. intel has about 6% IPC advantage over ryzen so overall ryzen should still have an edge in multi threaded scenarios in ideal situation.

of course we know real world ideal situation dont exist, CB15 is very close as it is well optimized and scales perfectly with ryzen cpu but more cores means worse scaling so 6c each core efficiency should be better than each cores in 8, unless CCX design says otherwise.

but yes i agree 6c 5ghz will come VERY CLOSE to 8c at 4ghz ryzen in multi threaded scenarios. in ST it'll just blow ryzen away.
Click to expand...

I mean you don't even need to do much math or extrapolation. Any Ryzen 1800x review, all the instances where the Ryzen clearly wins, is only about 1/3 or 1/4 of the recorded benchmarks. Let's say it's 1/3
Of that 1/3 of the benchmarks in the review, only 1 or 2 of them in total is the 1800x running a massively scaling piece of software which is significantly faster. Almost every benchmark it wins, it's not by a significant amount, despite double the cores, double the threads.

Logically, the 8700k even at 4.6ghz allcore would probably win 95% of (average) benchmarks and only miss out on a handful of others.

Assuming of course the 8700k does what we expect it to.
PLUS the damn thing comes with a "free GPU" for those of us who've given up gaming. Totally capable of running dual 4k monitors, encoding / decoding x265 and even running 5 year old games and older, generally pretty fine.
 
You must log in or register to reply here.
Back
Top