Intel Skylake / Kaby Lake

Page 172 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Mar 10, 2006
11,715
2,012
126
Reasons I went with a i5 6600K instead of a 6700K

1. I had the itch to upgrade for a year and didn't want to wait for the 6700k. If it was released day 1, I probably would have got it.

2. I plan on moving to Skylake-E as soon as it becomes available. I would have moved to Haswell-E if I didn't read how awesome the Purley platform will be. (Damn I hate Intel for making HEDT wait 1.5 years.)

3. A 6600K OC was showing impressive results with good RAM, so the money saved on the CPU bought better RAM.

Skylake-E is going to be pretty darn awesome. I think we will see 10 cores for the highest end part.
 

tential

Diamond Member
May 13, 2008
7,355
642
121
I should mention that my current system is the most stable I have ever had. I'm very hesitant to do a complete system upgrade to just gain a couple of fps.

The GTA5 benchmark here shows roughly an 8% increase: http://hexus.net/tech/reviews/cpu/85193-intel-core-i7-6700k-14nm-skylake/?page=7
But I could not find anywhere in this test which frequency they ran the CPUs on, so I have to assume they were run at stock, which means the 2700 was run at 3.5Ghz. But I run mine at 4.4, so the gap would obviously be smaller than 8%. But yeah, the Skylake can of course also be overclocked, as you say.

I suppose it depends on the specific game. The difference in Shadow of Mordor is negligable. Have you found any benches which show a significant fps increase?
My point is that you still don't need to upgrade right? So a person should get an i7 at the high end given how long a processor will last them.
Your gpu lasts you half as long but you spend twice as much on it (not you people in general). Makes 0 sense. My gpu will be upgraded twice probably before I need to upgrade my cpu....
 
Aug 11, 2008
10,451
642
126
You haven't seen much tests it seems. Skylake doubles performance in some cases.

Well if you want to get snippy about it this test shows only about 10 to 20 percent gain in three popular demanding titles. hexus.net igp gaming .

So like I said, results are all over the place, probably depending on resolution and image quality among other things. Intel igps are notorious for pooping out at higher resolutions/image quality.
 

dark zero

Platinum Member
Jun 2, 2015
2,655
138
106
Intel IGPU is an interesting way to evolve. However when HBM hits the lowest tier, it's game over for HMC and by default, Intel iGPU
 

Pandamonia

Senior member
Jun 13, 2013
433
49
91
Useless. OK. Do you have something useful? I like learning, I'd take a look, I'm open-minded. Is hyperthreading good at helping minimum frames?
Yes it can. Whilst games don't use the cores fully they often have threads which are used by HT now. Now sometimes that is needed when the cores are bogged down and you get a negative fps spike. Smoother game play is always a winner
 

Sweepr

Diamond Member
May 12, 2006
5,148
1,142
131
Say hello to the Dell’s Inspiron 13 7000 Series convertible with Intel Skylake

7359_01.jpg


Intel will probably launch its first Skylake chips aimed at laptops and 2-in-1 tablets in September, and PC makers will probably introduce a whole bunch of new and updated portable computers at around the same time (perhaps at the IFA trade show in Berlin).

...Sam’s Club has posted a product page for a new version of the Dell Inspiron 13 7000 Series convertible notebook with a screen that folds back 360 degrees to let you use the PC like a tablet.

The new version has the model number I7359-5984SLV, and it looks a lot like the Broadwell-powered models that are currently available, but the new version features an Intel Core i7-6500U processor and ships with Windows 10 software.

The model listed by Sam’s Club is priced at $899, and features 8GB of RAM, a 500 GB hard drive, a 1920 x 1080 pixel display, 802.11ac WiFi, Bluetooth 4.0, a backlit keyboard, two USB 3.0 ports, one USB 2.0 port, an HDMI port, a headset jack, and an SD card reader. It has a 43 Wh battery and the system also comes with a passive stylus.

http://liliputing.com/2015/08/say-h...00-series-convertible-with-intel-skylake.html

10 days to go till IFA 2015.
 

mikk

Diamond Member
May 15, 2012
4,111
2,105
136
Well if you want to get snippy about it this test shows only about 10 to 20 percent gain in three popular demanding titles. hexus.net igp gaming .

So like I said, results are all over the place, probably depending on resolution and image quality among other things. Intel igps are notorious for pooping out at higher resolutions/image quality.


RAM advantage for Haswell and no driver infos. Three games are not enough also. This is a poor test.
 

Deders

Platinum Member
Oct 14, 2012
2,401
1
91
It may have to do with Prime95 FMA3 loads that punish a CPU at high voltages. Asus recommend 1.42 max voltage for extreme testing, and 1.45 for gaming.
 
Jun 23, 2006
104
0
71
Ok I have a question....
I have around 500 bucks that I have been saving for the eventual successor to the FX-8350 I have now..
Also I should note that I am not very familiar with Intel so here goes..

Im interested in a i7 6700K and am confused about what motherboard I would need?
And would I be able to reuse my ram or will I need DDR4?
 

Absolute0

Senior member
Nov 9, 2005
714
21
81
Skylake is down, repeat Skylake is down.

http://www.overclock.net/t/1571038/my-6700k-dead

Have read previously that Prime95 stress test is insane on these CPUs. Makes them draw more juice and get hotter than any real-world applications, stresses the package side power (the caps on bottom?)?

Voltage should be suspected as the culprit but ASUS/Intel engineers have had a long time to test this and have stated 1.45v. Now that sounds crazy but they probably weren't at 4.7 Ghz and running Prime 95 either.
 

Deders

Platinum Member
Oct 14, 2012
2,401
1
91
If you want the K version for overclocking then you need a z170 motherboard. Most of those need DDR4 but some will accept low voltage DDR3L sticks.
 

VirtualLarry

No Lifer
Aug 25, 2001
56,226
9,990
126
Have read previously that Prime95 stress test is insane on these CPUs. Makes them draw more juice and get hotter than any real-world applications, stresses the package side power (the caps on bottom?)?

Voltage should be suspected as the culprit but ASUS/Intel engineers have had a long time to test this and have stated 1.45v. Now that sounds crazy but they probably weren't at 4.7 Ghz and running Prime 95 either.

The thing is, Prime95 is NOT a "power-virus". It's a distributed-computing app. Lots of people use it for the intended purpose (searching for prime numbers) every day.

If Intel were sloppy in terms of engineering SKL for "max loads", then perhaps SKL isn't a good CPU to use for DC rigs.

Edit: Maybe leaving off FIVR was a mistake?
 

Deders

Platinum Member
Oct 14, 2012
2,401
1
91
The thing is, Prime95 is NOT a "power-virus". It's a distributed-computing app. Lots of people use it for the intended purpose (searching for prime numbers) every day.

If Intel were sloppy in terms of engineering SKL for "max loads", then perhaps SKL isn't a good CPU to use for DC rigs.

Edit: Maybe leaving off FIVR was a mistake?

It seem if you are using accelerate instruction sets that hammer the cpu with unnatural loads at voltages beyond the spec then it can cause damage. Intel would have tested and tuned for longevity within spec. so best not use these programs when overclcked, especially at high voltages.

Asus mention that for extreme lads, 1.42v is the max you want to put through it, for normal loads 1.45 is ok apparently.

The error cods 32 and 55 point to something going wrong with communication to the memory, so maybe the IMC went?
 

.vodka

Golden Member
Dec 5, 2014
1,203
1,537
136
I have to say the same as all of you, once I had started seeing people pushing 1.3-1.4v on their shiny new 14nm CPUs, it was a matter of time before one would release the magic smoke. These kind of voltages are what are usually considered upper limits for long term safety on decent cooling (70-80°C as maximum for stress testing, 60-70°C for normal workloads) for 22nm-32nm CPUs, not 14nm ones...


Yet the guy says his board defaults to 1.3v at stock settings. That seems a little excessive. On top of that, he also mentions it reaching 87°C and being fed 1.35-1.4v. That CPU was being abused, to put it lightly.


How do these CPUs clock when tested at sane voltages, like 1.2v? Is there even a need for such stupid high voltages to reach, say 4.5 GHz? I don't remember seeing such voltages on day one reviews and overclocking repots.
 

CakeMonster

Golden Member
Nov 22, 2012
1,384
482
136
My 6700K goes up to 1.32 (possibly more) at times when putting it under stress. It seems to me that the motherboards are running in "Haswell mode" still. It does dynamically adjust though, and downvolt a lot more on idle.
 

CakeMonster

Golden Member
Nov 22, 2012
1,384
482
136
It may be easy to "fix", I haven't had time to play with it at all yet.

My experience with Haswell in this regard was a headache. I had to lock the voltage in place IIRC to avoid it boosting up way beyond what was needed. I ended up undervolting significantly on a permanent basis just to get a mild OC stable.
 

Walter E Kurtz

Junior Member
Jun 5, 2015
18
0
0
Everyone saying he was pushing too much volts though that chip, remember that the vccin is not the vcore.

Actual vcore might be 1.2 or so, afaik it is not clear yet how big the difference is.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
Considering he is using Asus with their own custom LGA1151 socket, it may simply be the socket. Its seen before with Asus.

But again, he is getting memory errors. He would get no error codes like this with a completely dead CPU.

From the Asus tech:
Code 32 and 55 are memory related codes. If the CPU were completely dead, it would not be showing these codes
 
Last edited: