Intel Coffee Lake Leaked News

[Taf the Ghost]

44 minutes ago, tom_w141 said:

Costing less than a 4 core/4 thread i5 it is already the budget option and the best option for a mid range builder or gamer/streamer. Also the 7700k only helps you if you have a 144Hz refresh rate monitor anyway. You can't even see a difference between 120 and 140fps (and that is with the difference exaggerated. irl its closer). FPS has diminishing returns above 100. 30-60 is a stark difference, 60-100 is also a large difference, 100-144 not so much. 

The one subtlety we don't know going forward is the performance needed to saturate GPUs. To really get the most out of the 1080-class requires one of a few CPUs at 1080p Gaming. The Coffee Lake CPUs are going to be interesting because it's the high clocks on the 7700k that really do most of the work. Coffee Lake is going to be clocked lower and there probably won't be a 4c/8t that's unlocked. 

 

Without the Turbo numbers, we can't say completely what the results will be like yet. However, if Max Turbo is lower than the 7700k, we're technically going to see Regression in a lot of cases.  Under a high OC 6c/12t CPU, it'll be the best gaming CPU for possibly a very long time, if Intel switches to Mesh on their Desktop CPUs.

[EPPHO]

On 7/22/2017 at 3:47 AM, That_PC_Kid said:

Anyone with a Skylake or Kaby Lake CPU will be able to upgrade to the 300 series chipset motherboards

Why would someone do that?

[Princess Luna]

2 minutes ago, frozeNNN said:

Why would someone do that?

Also beats to me to understand the purpose of it.

 

As I said previously I think I'll replace my i7 7700 only on Ice Lake~

[That_PC_Kid]

4 hours ago, frozeNNN said:

Why would someone do that?

I meant to also add in that you would be able to use Z170/Z270 with Coffee Lake unless they remapped the pins. But maybe there is something you want on a 300 series board or maybe yours dies and you end up looking for a new board. Just upgrading for no reason really wouldn't make any sense though.

[Mr.Meerkat]

On 2017-7-22 at 10:12 AM, Jito463 said:

 

Except that one of the rumors floating around is that Intel may have remapped the pins on the 1151 socket, making it electrically incompatible with existing 1151 socket CPUs.  Similar to how AMD's TR uses effectively the same physical socket as Epyc, but they're not cross-compatible.

Another (better known) example is X79 and X99, both are LGA 2011 sockets but are not cross-compatible as well  (with one being v1 LGA2011 with the other being v3 LGA2011). 

[Jito463]

Just now, Mr.Meerkat said:

Another (better known) example is X79 and X99, both are LGA 2011 sockets but are not cross-compatible as well  (with one being v1 LGA2011 with the other being v3 LGA2011). 

True, I suppose I could have mentioned the 2011 sockets (which would have been even more apropos, given that it's all Intel), but I had recently been reading some info on TR/Epyc, and that's what came to mind first.

[Drak3]

@MageTank Regarding the issue of higher core parts not achieving as high a clock as lower cores, such as deca/octo/hexacore vs quadcore, you're right up to a point. In older chips, efficiency stood in the way of clocking higher when you throw in more cores, but with Haswell, the 4790K and 5930K were hitting clocks that were in the small double digits of each other on water. Same with the 4770K and 5820K, effectively the same achievable clocks on consumer/prosumer cooling, provided one cooled them properly.

 

As it stands, Skylake X doesn't have a decent sample size, but the 7800X has enough submissions to make some early observations. It's hitting 4790MHz on water, 4442MHz on air, and 5878MHz on LN2, according to HWBot. The 6700K hits 4765MHz on water, 4642MHz on air, and 6048MHz LN2. We're at a point where we can expect the same clocks between with water, and the disparity being extreme cooling or cooling inadequate for pushing high clocks. Moving to octocores and up, and clocks drop to add cores.

 

Also, the 7800X follows the same trend as the 5930K and 6850K, air and water averages are reversed.

[MageTank]

Just now, Drak3 said:

@MageTank Regarding the issue of higher core parts not achieving as high a clock as lower cores, such as deca/octo/hexacore vs quadcore, you're right up to a point. In older chips, efficiency stood in the way of clocking higher when you throw in more cores, but with Haswell, the 4790K and 5930K were hitting clocks that were in the small double digits of each other on water. Same with the 4770K and 5820K, effectively the same achievable clocks on consumer/prosumer cooling, provided one cooled them properly.

 

As it stands, Skylake X doesn't have a decent sample size, but the 7800X has enough submissions to make some early observations. It's hitting 4790MHz on water, 4442MHz on air, and 5878MHz on LN2, according to HWBot. The 6700K hits 4765MHz on water, 4642MHz on air, and 6048MHz LN2. We're at a point where we can expect the same clocks between with water, and the disparity being extreme cooling or cooling inadequate for pushing high clocks. Moving to octocores and up, and clocks drop to add cores.

 

Also, the 7800X follows the same trend as the 5930K and 6850K, air and water averages are reversed.

It's hard to compare smaller vs larger dies (and IHS heat transfer for that matter). They are likely going to be using the same size die (or at the very least, same size IHS to transfer the heat) with Coffeelake as they are on Kaby/Skylake. Comparing the consumer SKU's against the enthusiast ones are simply difficult due to the very distinct differences between the two.

 

If you would like to see a pattern comparing the same generation of hardare, compare the 6950X against the 6900k/6850k/6800k. I personally can't find that kind of data (sadly, no OCN owners club for Broadwell-E apparently) but that would give us a realistic way to gauge same-platform core clock/count scaling. 

[Drak3]

Just now, MageTank said:

If you would like to see a pattern comparing the same generation of hardare, compare the 6950X against the 6900k/6850k/6800k. I personally can't find that kind of data (sadly, no OCN owners club for Broadwell-E apparently) but that would give us a realistic way to gauge same-platform core clock/count scaling. 

Broadwell-E overclocks like shit, this I know. For HWBot tells me so.

1 minute ago, MageTank said:

It's hard to compare smaller vs larger dies (and IHS heat transfer for that matter). They are likely going to be using the same size die (or at the very least, same size IHS to transfer the heat) with Coffeelake as they are on Kaby/Skylake. Comparing the consumer SKU's against the enthusiast ones are simply difficult due to the very distinct differences between the two.

Should they be the same die size, the hexacores and quadcores should hit damn near the same clocks with a delid and watercooling. The only issue I see coming is the TIM being a bottleneck, unless Intel bins the chips for overclockability, in which case, I see the bad ones being the quad cores.

[VagabondWraith]

More Coffee Lake leaks.

 

Quote

Exclusive: Intel® Core™ i7-8700K Specifications (Coffee Lake)

6C/12T
12MB L3
3.7 GHz Base
4.3 GHz 6-core Turbo
4.4 GHz 4-core Turbo
4.6 GHz 2-core Turbo
4.7 GHz 1-core Turbo
95W TDP

Disclaimer: The source this info comes from has provided accurate Intel leaks in the past. For now, treat it like a rumor, but rest assured I will let you know about any changes. No pricing (MSRP) information yet. Next in line, Core i5-8600K specifications.

 

 

https://forums.anandtech.com/threads/intel-skylake-kaby-lake-coffee-lake-thread-skylake-x-reviews-out-page-501.2428363/page-547#post-39000209

 

[Taf the Ghost]

Sweeper is legit, so it means the announcement date is somewhere close.

 

So it would appear that the Coffee Lake's "new feature" is going to be more multi-core Turbo ability, something beyond even the 2-core from SKL-X. It really is going to be the 7700k + 2 cores.  Though those turbo numbers point to using the Power Draw as the turbo limiter. I'm a tad surprised we haven't seen that type of turbo option on the SKL-X.

 

But, hey, at least it'll be an awesome gaming CPU.

[Citadelen]

On 22/07/2017 at 8:05 PM, That_PC_Kid said:

It was a horrible architecture. How do you manage to put 8 cores on a CPU and have the performance of a low-end quad core and a dual core?

It only had 8 integer units, not 8 FPUs. Two cores shared an FPU.

[You_are_a_cunt]

On 7/22/2017 at 8:06 AM, SCHISCHKA said:

I think its cofefe

You mean covfefe, right?

[SCHISCHKA]

1 minute ago, revsilverspine said:

You mean covfefe, right?

no i meant cofefe

[You_are_a_cunt]

Just now, SCHISCHKA said:

no i meant cofefe

Despite the constant negative press?

[SCHISCHKA]

36 minutes ago, revsilverspine said:

Despite the constant negative press?

MSM is fake news. The crooked Hillary wants you to think I am collaborating with the Russians, cofefe?