4 cores vs. 6 cores

>>61341041
Still in the 100s. Wow. Intel truly is shit I cant even. Amd for the win. Much wow.

File: Untitled-1.png (462KB, 1920x1080px) Image search: [Google]

462KB, 1920x1080px

>>61341041
How to ruin performance with cache changes and memory latency increases in one easy step.

Hope Coffee Lake S can put on a better 6 core show than this, or Intel isn't going to sell a single one of these to MUH GAYMERS.

>>61341041
It's because the 7700k has the lowest core to core latency.

Intel has to figure out to get that efficient core to core communication with 6,8,10 core CPUs.

>>61341195
>Hope Coffee Lake S can put on a better 6 core show than this, or Intel isn't going to sell a single one of these to MUH GAYMERS.
That was exactly my thoughts when seeing these results. I highly doubt the new 6c/12t Coffee Lake i7 is going to be much different than the 7800x. That would be a first for Intel to release new mainstream CPUs that perform worse than the generation from before. Looking forward to that.

>>61341204
Yeah it looks like AMD is on the right path with their 2 CCXs design. Hardware Unboxed said in the comments he was working on comparing the 7800k against Ryzen, it'll be interesting to see the results. My bet is on Ryzen delivering better performances per clock.

Kinda disappointing
I was hoping the 7800k was the CPU to get.
But it doesn't preform well
Intel's entire lineup doesn't scale well with added cores.

Wow, Intel shitting the bed on a shitty platform that they had to come up with as soon as they got their shit slapped by AMD, who woulda thunk!

>>61341144
Yeah, because it's half the price and doesn't burn your house down.

>>61341041
not really that interesting
schedulers are spastic, gaymes barely use any cores

>>61341262
New Coffeelake S leak on Sandra

http://ranker.sisoftware.net/show_system.php?q=cea598ab9fa79fab9cbadde0cdfcdaa895a482ebd6e6c0a895a583fbc6f6d0b5d0edddfb88b58d&l=en

3.2 GHz base 3.6 GHz boost

Not looking good for 6 core Intel if you want muh gaymes.

>>61341712
What'll be fascinating is how the 4 core cut will be, or if they'll make an entirely separate die for it.

>>61342158
I see two options here

1. i5 is a 6 core 6 thread CPU and is basically crap
2. i5 is a 4 core 8 thread CPU with lower clockspeeds than the i7-7700K
3. i5 is a 4 core 8 thread CPU that's basically an i7-7700K

Option 1 and 2 would suck for Intel fans. Option 3 would completely BTFO AMD's mid-range CPUs in terms of gaming value.

File: 1454197684503.gif (531KB, 800x450px) Image search: [Google]

531KB, 800x450px

>>61342190
>I see two options here
I'm retarded and refuse to delete my post just to fix this.

>>61341195
>7800K
2.9 ghz base and 3.9 Ghz turbo
>7700K
4.2 ghz base and 4.5 Ghz turbo

>>61342158
>>61342190
>>61342210
It was already confirmed a while ago guys :

i7 = 6c/12t
i5 = 4c/8t
i3 = 4c/4t
Pentium = 2c/4t

>>61342190
ii pray to God that Intel doesn't release a 6C6T CPU, it would easily be the most confusing CPU of all time. Everyone would constantly ask which is better, 6C/6T or 4C/8T.

>>61341195
New arch, worse performance

The worse part is that some cloud customers required and Intel thought it was a kick ass idea. No one is against bigger I-20, as long you don't kill your l3 and make it slow

>>61342233
Confirmed where? It's all still just speculation and rumors as far as I can tell.

>>61342286
https://www.3dcenter.org/news/coffee-lake-portfolio-bietet-durchgehend-mehr-cpu-von-core-i3-bis-core-i7

>>61341262
Engineer samples seen that coffee uses same arch of kabylake so so

>>61342300
I've seen that, still just rumors and speculation. Until it's confirmed on an Intel marketing slide, it's all up in the air.

>>61342190
>>61342233
I'm talking if the 4 core 8 thread will be a harvested 6 core 12 thread die or not. If it is harvested then it'll probably get worse performance than a 7700K at the same clock.

>>61343056
Oh, I see.

Daily reminder that very strong 4 cores and 8 threads will be the best option for gaymes at least until 2025-2027 or whenever the ps6 comes out.

>>61343216
>closing all background tasks and stuttering

Nah, I prefer to not use my PC as a console.

>>61343056
Knowing Intel, it will probably use 3 separate dies for their Coffee Lake chips.
>6 cores/6 or 12 threads 16MB L3$ with GT2 iGPU
>4 cores/4 or 8 threads 8MB L3$ with GT2 iGPU
>2 cores/2 or 4 threads 4MB L3$ with GT2 or GT1 iGPU

This doesn't include the separate dies that they make for the GT3/GT4e mobile CPUs .

>>61343328
4 core 8 thread i5 would pretty much end AMD's mid-range push instantly. It would just be dead. an i7-7700K for $100 cheaper would destroy the 1600/1600X.

>>61343343
Yeah, if only Intel drops prices.

>>61343366
Not dropping prices, just shifting the product stack down. They already did it with X299.

>>61343343
Coffee Lake is coming out before Zen2, which we already know will have 6 cores per CCX on a high frequency-optimized 7nm process. That means that the second gen APUs will not only run circles around the HT'd Core i5, but the Core i7 as well.

>>61343405
Thought Zen 2 was Q4 2018. Coffee Lake will certainly be out before that.

>>61343425
CFL is looking to be a Q4 launch now, it'll have to compete with Pinnacle Ridge

>>61343435
Source? I've heard it's coming out Q1 2018 at the latest.

>>61341041
I was actually seriously considering the 7820 since I mainly game, but do use it for a good deal of encoding/rendering as well so the extra cores/threads would be worthwhile while the IPC would still be (at least theoretically) be better than Ryzen.

This is looking... to not be the case.

>>61343453
That's cannonlake (mobile)

>>61343435
This, AMD will just lower prices for R5/7 is coffelake is indeed dropping prices down(I can't imagine this being good for Intel's margins) and wait it out for 2-3 months before Zen+ releases

>>61343343
Certainly it will be bad bins from 6 cores.
Intel always hold an arch and bin it makes no sense to trash semi usable dies.
If it could hold kaby frequencies it would certainly conflicts with 7800x line.
If frequencies are worse, especially on the bad binned 4 cores, say 4.0 GHz or lower, then ryzen is better.

These Skylake X chips overclock very well, too bad the chip and HIS have bad tolerances that cause housefires.
One dude is running a 7800x 4.4GHz 1.055 V.

>>61343737
Intel managed to make a monolitich die with 10 cores that can make to 5ghz on a reasonable size of bins, consuming less voltage. That is worth of note, however, once you step after 4.5 GHz the current needed becomes insane. This and the whole l2/l3 fucked slx.

File: coffee lake geekbench.png (192KB, 1920x1080px) Image search: [Google]

192KB, 1920x1080px

>>61343716
Yeah a 6 core with 2 dead cores would be pretty crap compared to an actual 4 core.

More Coffee Lake S leaks, this time from Geekbench.

>>61343926
>only 12MB of L3 cache
CACHELET CONFIRMED
Looking at the size of the L2 cache, this is almost certainly not going to have the multi- BINGBUS matrix bullcrap.

>>61344011
Yeah it's Kaby Lake with more cores, basically.

>>61344099
No he's talking about how Skylake-E did the bus, it's got so many ring busses they're pretending it's a mesh now.