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 .
this makes me curious, looks like core architecture cant get optimized anymore so no more "core" ipc increase. when will we expect to see a new build soon from intel? AMD's got ryzen, hope intel can bring some nice things to the table, something like actual 10% ipc increase in a single generation due to architecture change would be nice.

The 8700k is going to be a definitive cpu I suspect. First real leap in several generations.

When (if) they make an 8 core 10nm chip, that leap will also be significant but likely the last in ages.

One positive thing about this difficulty in shrinking technology is that eventually, one day AMD and Intel will be very close in process technology, this means, eventually AMD just need to figure out how to make the design better and better to catch up properly.
Then competition will be even hotter, benefiting us
 
One positive thing about this difficulty in shrinking technology is that eventually, one day AMD and Intel will be very close in process technology, this means, eventually AMD just need to figure out how to make the design better and better to catch up properly.
Then competition will be even hotter, benefiting us

They certainly wont. If you didn't know it, not a lot of companies can afford shrinks because without billion $ IC designs you only increase transistor cost. There is a reason why companies are dropping of left and right from the nodes and legacy nodes are such a huge business. AMD is already struggling cost wise with 14nm, despite getting higher prices for the CPUs. Their margins are actually decreasing! On the GPU front 14LP and 12LP looks simply to be a loss for them.

Unless AMD regains a massive marketshare I cant see them making any profit on "7nm".

A real 5nm or 7nm single IC design will cost multi billion $ and have a development time for that will pretty much last ages on that alone.

I doubt more than Intel and Apple can afford it since it requires ultra high revenue/volume/profit. You can also take a hit on the transistor utilization rate and end up with a higher cost transistor to transistor. Meaning if you increase transistor count you increase cost. Besides the other costs related.

14nm-Ramp-IBS.jpg
 
Last edited:
Skylake, Kaby and Coffee lake are all the same skylake core on refined nodes and more cores, wait till Icelake (new core) to make the judgement of no IPC increases....
& Shintai

honestly dont need to wait till icelake. as much as intel love to milk consumer, its becoming more transparent they are running out of stuff to improve other than chipset features or rebinned silicon. honestly i think they are hitting a limit. looking at haswell to broadwell we got a 3% ish ipc boost? then from broadwell to skylake is like 2% ish and since skylake, its been nothing. boosting cache/imc is one thing but when cpu itself is very similar to previous gen thats when we see some actual improvement rather than rely on making memory/caching faster if u know what i mean.

if icelake/tigerlake do get a boost, we probably gonna see around 1-2% this time.. to the point of irrelevant and better just to stick to 14nm++ 6 cores, or 10nm+ 8 cores.
 
& Shintai

honestly dont need to wait till icelake. as much as intel love to milk consumer, its becoming more transparent they are running out of stuff to improve other than chipset features or rebinned silicon. honestly i think they are hitting a limit. looking at haswell to broadwell we got a 3% ish ipc boost? then from broadwell to skylake is like 2% ish and since skylake, its been nothing. boosting cache/imc is one thing but when cpu itself is very similar to previous gen thats when we see some actual improvement rather than rely on making memory/caching faster if u know what i mean.

if icelake/tigerlake do get a boost, we probably gonna see around 1-2% this time.. to the point of irrelevant and better just to stick to 14nm++ 6 cores, or 10nm+ 8 cores.

Your numbers are way off. Also increasing IPC=lowering clocks.

Big IPC increases stopped with the Pentium Pro.
You can only make IPC increase look artificially good since if you use P4 or FX speed racer designs to compare.
intel-xeon-ipc-over-time.jpg


You can always create another bottleneck however and claim there is no IPC changes.
 
Last edited:
As expected, stock 8700K leading @ the majority of actual MT apps tested, at stock (vs more expensive 1800X). Both 8600K/8700K were OCed to 5.1 GHz with relative ease. We got a winner product here.

- Core i7-8700K is faster in: Handbrake Photoshopt, 3D Studio Max, Blender, WinRAR, 7-ZIP, wPrime and all gaming tests, with a huge advantage in minimum FPS in almost all titles (except Titanfall and Metro)
- Ryzen R7 1800X is faster in: Cinebench R15, POV-Ray

I might go for 8700K, but let's not overlook the fact that 8600K is looking like sweet gaming chip for $100 less.

g3-2.jpg


g6-1.jpg
 
Last edited:
Your numbers are way off. Also increasing IPC=lowering clocks.

Big IPC increases stopped with the Pentium Pro.
You can only make IPC increase look artificially good since if you use P4 or FX speed racer designs to compare.
intel-xeon-ipc-over-time.jpg


You can always create another bottleneck however and claim there is no IPC changes.

Those are truly some of the worst graphs I've ever seen

i have to agree with abrasion here, that looked like chery picked results tbh. those numbers are wayyyy too high. anandtech clearly shows average gain from haswel to broadwell is 3% and broadwell to skylake is like 2.5%. even pcper showing broadwell to skylake is like less than 2%.
 
i have to agree with abrasion here, that looked like chery picked results tbh. those numbers are wayyyy too high. anandtech clearly shows average gain from haswel to broadwell is 3% and broadwell to skylake is like 2.5%. even pcper showing broadwell to skylake is like less than 2%.


He wasn't talking about the IPC graphs, he was talking about the 8700k and 8600k graphs...
 
i have to agree with abrasion here, that looked like chery picked results tbh. those numbers are wayyyy too high. anandtech clearly shows average gain from haswel to broadwell is 3% and broadwell to skylake is like 2.5%. even pcper showing broadwell to skylake is like less than 2%.

If you keep the bottlenecks elsewhere, sure. or maybe you are one of those that believe Broadwell got an IPC around 20% over Skylake? Or was it just an effect of the EDRAM?

There is a reason why SKL/KBL/CFL scales to 4000Mhz+ memory and beyond.
 
If you keep the bottlenecks elsewhere, sure. or maybe you are one of those that believe Broadwell got an IPC around 20% over Skylake? Or was it just an effect of the EDRAM?

There is a reason why SKL/KBL/CFL scales to 4000Mhz+ memory and beyond.

nope, i simply go by what anandtech's graph show, unless you can somehow explain how they are incorrect like for example, where would be the bottleneck you're talking about could be causing the 2-3% improvement since haswell. for the memory thing, it maybe related but i think thats a bit less relevant, i mean i have seen some 3600 or 4000mhz memory few yrs ago not just recently with SKL/KBL.
 
I see "5G" and have a bit of a freakout. I'm trying to make that work for your next phones RIGHT NOW.

The board doesn't look completely awful, assuming you can turn off the lights. It's massive overkill for me, I may pick up a prime which is not terribly obnoxiously styled. I still have a thing for ASUS engineering in general. They have some great talent. Much less experience with other vendors, so am timid.
 
The first 4 reviews of a 1700 showed 4.0, 3.9, 3.9, and 4.0 ghz. The 1800x MIGHT have a 100 mhz advantage.

So when someone mentions OC clocks obtained by reviews, the standard reply is that they aren't significant, because reviews got golden samples, but clocks obtained on reviews of AMD chips are always significant. Right?
 
& Shintai

honestly dont need to wait till icelake. as much as intel love to milk consumer, its becoming more transparent they are running out of stuff to improve other than chipset features or rebinned silicon. honestly i think they are hitting a limit. looking at haswell to broadwell we got a 3% ish ipc boost? then from broadwell to skylake is like 2% ish and since skylake, its been nothing. boosting cache/imc is one thing but when cpu itself is very similar to previous gen thats when we see some actual improvement rather than rely on making memory/caching faster if u know what i mean.

if icelake/tigerlake do get a boost, we probably gonna see around 1-2% this time.. to the point of irrelevant and better just to stick to 14nm++ 6 cores, or 10nm+ 8 cores.

You forgot to mention how AMD loves to milk consumer and the tiny or even negative IPC changes AMD did bring during last years.
 
You forgot to mention how AMD loves to milk consumer and the tiny or even negative IPC changes AMD did bring during last years.

naw, they have had less than 1/10th of money intel had and had to split that between cpu and gpu. with intel's underhanded tactics over the last few decades its actually rather amazing they able to still pull ryzen out. if AMD goes down, i'd guess we'll see 8 core mainstream last 20 yrs, with no IPC gain, milked by intel to the max, yep.
 
You forgot to mention how AMD loves to milk consumer and the tiny or even negative IPC changes AMD did bring during last years.
Key word is did.

I don't see much of a difference in gaming between either maker. Intel has the epeen of more fps but so far I have not seen anybody have a bad experience with either at 1080. Go to higher resolutions and it's way less of a gap.
Now that we have a core war, everybody has cores. The big difference here is going to be price/performance. Unless wasting money is a good thing. I have 0 brand loyalty and could care less who did what and when.
Personally my 8/16 chips is more than I even use except when I do some DC stuff.
 
Last edited:
As expected, stock 8700K leading @ the majority of actual MT apps tested, at stock (vs more expensive 1800X). Both 8600K/8700K were OCed to 5.1 GHz with relative ease. We got a winner product here.

- Core i7-8700K is faster in: Handbrake Photoshopt, 3D Studio Max, Blender, WinRAR, 7-ZIP, wPrime and all gaming tests, with a huge advantage in minimum FPS in almost all titles (except Titanfall and Metro)
- Ryzen R7 1800X is faster in: Cinebench R15, POV-Ray

I might go for 8700K, but let's not overlook the fact that 8600K is looking like sweet gaming chip for $100 less.

g3-2.jpg


g6-1.jpg

honestly 8700k is probably a much better choice as it is not that much over 8600k. given how much more cache we'd get i think its definitely worth it! also knowing on average its probably binned better than the i5s too.

8700k in a laptop pls clevo thanks.
 
So when someone mentions OC clocks obtained by reviews, the standard reply is that they aren't significant, because reviews got golden samples, but clocks obtained on reviews of AMD chips are always significant. Right?

I think the clocks achieved by reviewers are close to what users can use, AMD or Intel. I never said otherwise. Just take 5 min reading user reviews of a retailer to check what they say.
 
naw, they have had less than 1/10th of money intel had and had to split that between cpu and gpu. with intel's underhanded tactics over the last few decades its actually rather amazing they able to still pull ryzen out. if AMD goes down, i'd guess we'll see 8 core mainstream last 20 yrs, with no IPC gain, milked by intel to the max, yep.

AMD also has 1/10th the number of projects and products, not to mention that you would consider AMD plus Glofo since Intel is a foundry as well. Intel has iGPUs and also has the Phi line for GPU-compute tasks.

Even former CEOs and engineers from AMD admit that those "Intel's underhanded tactics" aren't the source of AMD financial problems. The source of AMd problems are the dozens of strategic, financial, and engineering mistakes made by AMD.

Paraphrasing you, if Intel goes down, i'd guess we'll see 4 core mainstream last 20 yrs, with no IPC gain, milked by AMD to the max, yep. Luckily for us Intel is giving 6-core mainstream this year and 8-core mainstream next year. Hopefully we will get a 6-core APU from AMD somewhat in 2020, after a decade of AMD quad-core APUs!!!
 
Last edited:
that review man. kaby lake clocks on 6 cores :D my delidder is ready.... hurry up and launch this already Intel lol.
It looks nice, but comparing with 7700k results - nothing new when it comes to gaming. The best case was 10 fps difference at 1080p in GTA V, when fps is already above 120. Definitely can keep calm and continue using my 6700k for now. The funny thing is that 7700k beats 8600k in gaming tasks when we look at minimums, just like I also expected.
 
It looks nice, but comparing with 7700k results - nothing new when it comes to gaming. The best case was 10 fps difference at 1080p in GTA V, when fps is already above 120. Definitely can keep calm and continue using my 6700k for now. The funny thing is that 7700k beats 8600k in gaming tasks when we look at minimums, just like I also expected.
So basically even if the programmers make better multi-threaded games, we won't see much effect since it will run smooth no matter what. Kind of a bummer! Bring on more eye candy options!
 
Weak USB setup on the Hero, might bump to the Code if it offers more. The Formula looks great.
Compared to the IX Series, I am disappointed. Looks like they are removing a lot of I/O and adding LEDs. lol
 
Weak USB setup on the Hero, might bump to the Code if it offers more. The Formula looks great.
Compared to the IX Series, I am disappointed. Looks like they are removing a lot of I/O and adding LEDs. lol

The Aorus Gaming 7 doesn't have many USBs either. I was really liking that board until I saw the i/o. I may actually end up looking at the Asrock offerings as well.
 
If you keep the bottlenecks elsewhere, sure. or maybe you are one of those that believe Broadwell got an IPC around 20% over Skylake? Or was it just an effect of the EDRAM?

There is a reason why SKL/KBL/CFL scales to 4000Mhz+ memory and beyond.
any intellectually honest person uses an average IPC over multiple tests and includes independent tests but dishonest people will do dishonest things.
 
It looks nice, but comparing with 7700k results - nothing new when it comes to gaming. The best case was 10 fps difference at 1080p in GTA V, when fps is already above 120. Definitely can keep calm and continue using my 6700k for now. The funny thing is that 7700k beats 8600k in gaming tasks when we look at minimums, just like I also expected.

Yeah there's no really compelling reason for most gamers on kaby/skylake to upgrade but certainly the headroom of two more cores and no cost in clocks or IPC for that will make many upgrade i'll be willing to bet. I sold off my 7700k ahead of this launch because the prices are going to crash on quad cores after this release and my peticular 7700k was a really bad overclocker even after delidding so i wasn't ever very happy with it.
 
Yeah there's no really compelling reason for most gamers on kaby/skylake to upgrade but certainly the headroom of two more cores and no cost in clocks or IPC for that will make many upgrade i'll be willing to bet. I sold off my 7700k ahead of this launch because the prices are going to crash on quad cores after this release and my peticular 7700k was a really bad overclocker even after delidding so i wasn't ever very happy with it.
Frys has it for $259 allready...how low can it go
 
Back
Top