New I7- 8086k CPU Anniversary Edition

The name being an homage is kinda ironic in that this is a 6 core "meh" and should've been 8 by now.
 
The name being an homage is kinda ironic in that this is a 6 core "meh" and should've been 8 by now.
We are kind of at a crossroads here though, there are a lot of cases where the two cores outmatch but then the ability to run 5GHz or more outmatch the additional core-age (made up word :D). It all depends on what software is being run. Overclocking tons of cores just isn't as fun with all the heat and lower clocks..
 
I wonder if these are really better binned chips or they just use soder on the ihs?
 
The name being an homage is kinda ironic in that this is a 6 core "meh" and should've been 8 by now.

The former 6-core beated the 8-core Zen in a well-number of multithreaded scenarios including Blender or Handbrake. The new Zen+ is about 10% faster than the 1800X. So if this new 8086k is ~8% faster than the former 8-core it could beat the 2700k in some multithreaded scenarios.
 
Binning has really gone to extreme levels these last few years, first with Ryzen and now these chips from Intel. Like there's not much space for overclockers left (other than extreme overclockers). Take my piss poor i5-8600K that is just barely stable at 4.8GHz @ 1.34v and that's obviously without any multithreading support too. To think an retail sample does factory 200MHz higher clock at significantly lower voltage than mine.... well it's a bit crazy how much things have changed in that regard. They really push everything out of the chips these days with tight binning process.

Even the piss poor samples would always OC a couple hundred MHz more than retail samples not too long ago. Is this perhaps the first time overclockers may lose out to retail samples? I'm not the only one who got some Coffee Lake stopping at 4.8GHz on highend air or medium~highend water at "reasonable" voltage levels/temps, not to mention 4.9GHz which is also far more common, seems to be a split between 4.9 / 5.0 GHz at 1.32~1.34v or thereabouts for a pretty common chip (best ones usually do 5.2GHz at say 1.32v or so, a very few perhaps 5.3GHz at 1.33~1.35v or so) which is roughly where reasonable voltage/temps upper limit goes for these chips.
 
Last edited:
In a non boost state.

Rather than arguing with you, sigh......

Care to explain just how in the holy hell these chips can even come close to 220watts of thermal dissipation? I really really really want to know. And you may not fully understand what TDP is but I do
Binning has really gone to extreme levels these last few years, first with Ryzen and now these chips from Intel. Like there's not much space for overclockers left (other than extreme overclockers). Take my piss poor i5-8600K that is just barely stable at 4.8GHz @ 1.34v and that's obviously without any multithreading support too. To think an retail sample does factory 200MHz higher clock at significantly lower voltage than mine.... well it's a bit crazy how much things have changed in that regard. They really push everything out of the chips these days with tight binning process.

Even the piss poor samples would always OC a couple hundred MHz more than retail samples not too long ago. Is this perhaps the first time overclockers may lose out to retail samples? I'm not the only one who got some Coffee Lake stopping at 4.8GHz on highend air or medium~highend water at "reasonable" voltage levels/temps.

As we get more and more cores and software that is built for high thread counts.... the days of overclocking are going to end as clock speed isn't going to matter as much as parallel operation.
 
As we get more and more cores and software that is built for high thread counts.... the days of overclocking are going to end as clock speed isn't going to matter as much as parallel operation.

The problem is how we will get the software utilizing the high thread counts effectively (other than huge server load environments). For a non-server environment it's challenging. Typically of the max potential performance benefit an extra core can add, parallellisation only adds a little bit of the potential typically and the higher the core count the more lost potential benefit. More complex code also costs more development time.

It's going to be quite a few years where a desktop computer will see great added benefit from more than roughly 6~8 cores. It's a combination of added complexity (cost of development), RAM pricing to go down (the more cores the more RAM we need) and various standards of built-in APIs of parrallell operation to become more common so that more companies can cheaply invest in support for it. As I see it currently 6 - 8 cores will be relevant for at least another 2 - 3 or so years.
 
Last edited:
Just smoke and mirrors folks. It's only 5GHz on ONE core.

Yes but for the price and the quality of the slab of Si, I would definitely get one over the 8700k if I wanted 6 cores but I want moar cores so I am happy with my smoking fast 7820x
 
Binning has really gone to extreme levels these last few years, first with Ryzen and now these chips from Intel. Like there's not much space for overclockers left (other than extreme overclockers). Take my piss poor i5-8600K that is just barely stable at 4.8GHz @ 1.34v and that's obviously without any multithreading support too. To think an retail sample does factory 200MHz higher clock at significantly lower voltage than mine.... well it's a bit crazy how much things have changed in that regard. They really push everything out of the chips these days with tight binning process.

Even the piss poor samples would always OC a couple hundred MHz more than retail samples not too long ago. Is this perhaps the first time overclockers may lose out to retail samples? I'm not the only one who got some Coffee Lake stopping at 4.8GHz on highend air or medium~highend water at "reasonable" voltage levels/temps, not to mention 4.9GHz which is also far more common, seems to be a split between 4.9 / 5.0 GHz at 1.32~1.34v or thereabouts for a pretty common chip (best ones usually do 5.2GHz at say 1.32v or so, a very few perhaps 5.3GHz at 1.33~1.35v or so) which is roughly where reasonable voltage/temps upper limit goes for these chips.

I am not really sure why you are complaining, rofl. 4.8 to 5ghz+ with 2+ more cores looks okay to me.
 
Basic question from someone who is still on a 2500K: could I get a H370 Mobo that has muticore enhancement, turn that on, and then it would all-core boost to 5GHz?

Assuming sufficient VRM, VCore, and cooling of course. I might not even have to wait for a Z390 board then. 5GHz be plenty.

EDIT: On further research, it looks like MCE doesn’t exist on non-Z boards. Bummer.
 
Last edited:
The 8% base clock increase doesn't seem "just smoke and mirrors".

Sure, that base clock increase is great, no qualms there. It's the people who are excited about running at 5Ghz that I'm posting that for because realistically, the difference between running 4.7 vs 5.0 on a single core will be negligible. How many applications do you know of that would benefit from that extra frequency and are single-threaded?

Gaming - Nope
Encoding - Nope
Office Worker - Nope

¯\_(ツ)_/¯
 
Sure, that base clock increase is great, no qualms there. It's the people who are excited about running at 5Ghz that I'm posting that for because realistically, the difference between running 4.7 vs 5.0 on a single core will be negligible. How many applications do you know of that would benefit from that extra frequency and are single-threaded?

Gaming - Nope
Encoding - Nope
Office Worker - Nope

¯\_(ツ)_/¯

It would my Civ VI turns faster I would hope.
 
Basic question from someone who is still on a 2500K: could I get a H370 Mobo that has muticore enhancement, turn that on, and then it would all-core boost to 5GHz?

Assuming sufficient VRM, VCore, and cooling of course. I might not even have to wait for a Z390 board then. 5GHz be plenty.

EDIT: On further research, it looks like MCE doesn’t exist on non-Z boards. Bummer.

You can get a pretty cheap Z370 board for ~$100 new. Since you're not pushing it to the limits, it should work fine for MCE.
 
The name being an homage is kinda ironic in that this is a 6 core "meh" and should've been 8 by now.

Well, if Intel would quit fucking around with shitty paste under the IHS, then they could have definitely done an 8 core for this one at those speeds...would have been about a 130-140W part, but 8 cores at 5GHz boost would be bad ass.
 
The name being an homage is kinda ironic in that this is a 6 core "meh" and should've been 8 by now.

We need faster single cores more than we do more cores these days as so few apps can efficiently use more cores.
 
So this is confirmed real. There's an Intel Sweepstakes that opens 2 days, 8 hours from now, and will stay open for entry for 24 hours. 8086 of these will be given away. 50,000 are being made in total.

My question is: where is Z390? Not sure what I'll do if I wait out Computex and there are no announcements. Seems like a matter of when, not if, I'll get burned.
 
All core boost is 4.4, compared to 8700k's 4.3. Single core boost to 5.0 as opposed to 4.7. It's not a big difference. Might be worth it for some for the better binned silicon, though, especially if it OCs better.

Much more excited about the upcoming 8 core CFL-S CPU.
 
Well, if Intel would quit fucking around with shitty paste under the IHS, then they could have definitely done an 8 core for this one at those speeds...would have been about a 130-140W part, but 8 cores at 5GHz boost would be bad ass.
How about 28 cores at 5Ghz?
Intel showed one the other day.

image591.jpg
 
I wonder if these are really better binned chips or they just use soder on the ihs?

If they was soldered we'd see a stronger base clock than this... Plus Intel has to be giving itself some sort of head room for the i7 extreme line up to go, and all soldered parts would be one (of a few) possible directions. I'm of the thinking that Intel has to love the fact that AMD seems to be pushing in the direction of more cores vs. walking down the path to higher MHz as I believe that while Intel has the big bucks and R/D, they wouldn't ramp up/win the clock speeds (match them at best) in a MHz race with AMD.
 
If they was soldered we'd see a stronger base clock than this... Plus Intel has to be giving itself some sort of head room for the i7 extreme line up to go, and all soldered parts would be one (of a few) possible directions. I'm of the thinking that Intel has to love the fact that AMD seems to be pushing in the direction of more cores vs. walking down the path to higher MHz as I believe that while Intel has the big bucks and R/D, they wouldn't ramp up/win the clock speeds (match them at best) in a MHz race with AMD.

I suppose I'm probably confused by how you stated this.

Intel already has a mainstream desktop CPU (i7-8086K) that will boost to 5 GHz out of the box. AMD does not (yet)...and 4.2 GHz seems to be the Ryzen 2xxx series OC'ing lid.
 
If they was soldered we'd see a stronger base clock than this... Plus Intel has to be giving itself some sort of head room for the i7 extreme line up to go, and all soldered parts would be one (of a few) possible directions. I'm of the thinking that Intel has to love the fact that AMD seems to be pushing in the direction of more cores vs. walking down the path to higher MHz as I believe that while Intel has the big bucks and R/D, they wouldn't ramp up/win the clock speeds (match them at best) in a MHz race with AMD.

I suppose I'm probably confused by how you stated this.

Intel already has a mainstream desktop CPU (i7-8086K) that will boost to 5 GHz out of the box. AMD does not (yet)...and 4.2 GHz seems to be the Ryzen 2xxx series OC'ing lid.

Currently we are not any where near a MHz race between AMD and Intel... AMD is clearly moving into a (number of) cores race to which Intel is gladly stepping up to the plate for - AMD has already been to 4.7 GHz base clock and 5 GHz turbo 8 core 9590 (not in any way whatsoever implying AMD had the performance crown though, and yes, its not Ryzen).

For me (and I realize I'm not the majority), I'd love to see a MHz race, see new materials along with invitation improvements in cooling solutions which could drive MHz up over 7 GHz, and then circle back around to more cores races with base clocks on all those cores 7 GHz - Of course I'm thinking out a good amount, however if we move to who has the biggest amount of cores race, there will be less interest in MHz race (including not as much interest in per clock cycle improvements).

Also interesting would be seeing true evolution (vs moving to tons of cores so rapidly) to perhaps 128/256/384 bit evolving beyond the whole x86/x64 model.
 
Fucking badass. I love seeing movement from Intel these days, it's great.

This chip will cost an arm, a leg, and probably another arm. But daaaaaayum.

Here's to hoping Intel can base clock 28 cores at 5 GHz and actually mass produce it. This would make a great i9 upgrade even at $1.5k or so.... Maybe we'll see an Intel Video Card around the same time ;)
 
For me (and I realize I'm not the majority), I'd love to see a MHz race, see new materials along with invitation improvements in cooling solutions which could drive MHz up over 7 GHz, and then circle back around to more cores races with base clocks on all those cores 7 GHz - Of course I'm thinking out a good amount, however if we move to who has the biggest amount of cores race, there will be less interest in MHz race (including not as much interest in per clock cycle improvements).

Everyone would love to see a MHz race again. Not happening in this decade. Moreover, the main problem those days is memory speed. You can clock cores to 7GHz but the bottleneck is on memory. It is named the "memory wall"

140364245678419-640x315.jpg


Also interesting would be seeing true evolution (vs moving to tons of cores so rapidly) to perhaps 128/256/384 bit evolving beyond the whole x86/x64 model.

64bit will be with us for many many years. The only ISA has defined a 128bit version for future is RISC-V.
 
Impressive. Will be interesting to see what its actual TDP is when running 5Gz on all 28 cores and how they are cooling this sucker... I seriously doubt this is on air. And as to price, I bet it's well above even what most hardcore PC enthusiasts would be willing to pay.
 
Last edited:
Everyone would love to see a MHz race again. Not happening in this decade. Moreover, the main problem those days is memory speed. You can clock cores to 7GHz but the bottleneck is on memory. It is named the "memory wall"
.

If that is the case wouldn't increasing memory speed have a profound effect on performance? Yet in practice upping memory clocks results in negligible performance gains, often in the realm of measurement errors.
 
Everyone would love to see a MHz race again. Not happening in this decade. Moreover, the main problem those days is memory speed. You can clock cores to 7GHz but the bottleneck is on memory. It is named the "memory wall"

View attachment 78817


64bit will be with us for many many years. The only ISA has defined a 128bit version for future is RISC-V.


Brings back the old days (sort of) - Rambus, this was one of their agreements if memory serves me right in that the way to beat the wall is with Rambus memory technology with the understanding Rambus (in theory) could ramp speeds well beyond other technologies (on the market at that time) - NOT saying in any way I'm a Rambus fan, lol...

We need(ed) more dreamers to push the envelope past the 64 bit hurdle and frequency challenges.
 
Back
Top