• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Skylake Clockspeeds and benchmarks!

/jams his tongue firmly in his cheek.

Oh stop moaning. Just buy a new motherboard you tight git.

/removes tongue from cheek.

Oh goody. Another new socket, another new board, and 5-10% gains over the last.

*yawn*

The benchmarks show at least a 15% improvement over Haswell in cinebench, assuming they are legitimate benchmarks.

The usual numbers thrown around here are 5% per generation, such as Sandy to Ivy, or Ivy to Haswell etc. So Haswell > Skylake being 15% shows Intel have done more work for us this time.

Intel have the market share already, their CPU's have no competition from AMD at this stage, so a 15% improvement over Haswell is quite decent imo. It further extends the huge lead Intel have over AMD.
 
Broadwell provides around 5.5% IPC increase over Haswell. So how much IPC increase can we see going from Broadwell to Skylake ?

http://www.anandtech.com/show/9070/intel-xeon-d-launched-14nm-broadwell-soc-for-enterprise

Intel%20Broadwell_zps1cf1bzv2.jpg


I am actually wondering if Intel could have squeezed more IPC from Skylake on the same 14nm process.

There are no top tier Broadwell CPU's being released for desktop. There will only be 2 desktop Broadwell CPU's, both of with have only 6MB l3 cache (compared to 8MB on the 4790k, 6700k etc). They are aimed for PC's without dedicated GPU's, and as such have Iris Pro, Intel's most expensive IGPU.

Best to forget about Broadwell for desktop, and just consider Skylake as the next step after Haswell.
 
I take it there won't be a Broadwell-E after all then? Your post read that way to me anyways that there is only going to be two
Broadwell CPUs being released for desktop and both of them have integrated GPUs and they are not high end CPUs.

Do you have source that they have officially stated there won`t be a Broadwell-E release for the x99 platform?

I meant only the mainstream Broadwell CPU's, LGA1150.

Broadwell-E should still be coming, either very late this year or very early next year, though Intel themselves describe a 5.5% IPC increase from Haswell > Broadwell, so it's completely pointless for existing X99 owners. The only saving grace would be if the 5930K equivalent is also an 8 core product, leaving the 5820k equivalent as the only 6 core part, though this is unlikely.
 
Last edited:
Got a niche usage case here and a few questions. I run simultanious video encodes, x264 on the CPU and QuickSync on the IGPU. Would I be better off with Broadwell or Skylake? If I went x99 would a 5960X be fast enough to make up for the loss of the IGPU?

Skylake has native X265 encoding support on the IGPU, so I'd wait for that.
 
I've read some more about Skylake, it does seem IPC has got a 10%+ jump over Haswell.

Got my eyes on the i7 6700T, might be ideal for my Mini-ITX, if it can match a 4770K/4790K for gaming etc, that's the chip I would go for. 4/8 core/thread with 8MB Cache and only 35W, not bad at all.

As I posted many days ago - the Cinebench performance alone shows a 15% IPC increase over Haswell (4790k), even when the 4790k has a 200Mhz clock speed advantage over the 6700k.

Glad you are finally seeing the light! :D

Now we just need to see average overclocking results, and we'l know just how good Skylake will be.
 
What price range seems likely for a 6700k, £250-300? Just wondering if we can expect it to be around the price of a 4790k due to the similarities in performance. Hopefully cheaper (obviously) :p

Should be the same price as a 4790K. Maybe £10-30 higher the first week or two, as usual for new releases.
 
Given how underwhelming Skyake's turned out to be for desktop users, would a 5820 be a reasonable upgrade path from 4790K?

Skylake's not released yet. The benchmarks here are leaked ones - they may be all fake for all we know.

We also don't know Skylake's overclocking potential. It may overclock extremely well.

Don't consider anything fact in this thread :)
 
OCing is irrelevant, unless it goes above 5GHz which it doesn't judging by stock clocks.

Overclocking is very relevant - we're posting on 'Overclockers UK' forum after all, of course people care about this stuff.

Stock clocks don't represent the overclocking potential of a CPU, or CPU process. Never have, never will.

For all we know Skylake could overclock well. It could also not overclock at all, we just don't know. No-one has tried overclocking using Intel's new 14nm process yet.
 
Skeptical that are they gonna launch two Es in the same year.

Cannonlake has been pushed back to 2017 so maybe they could use it to massage the disappointing sales numbers in 2016.

Intel's sales records show record profits, up year after year.

I think you'll find their 2015/2016 sales numbers will be even better than this years - since we're getting a brand new architecture, Skylake.

Remember that Haswell has been out since June 2013 on desktop - so a new architecture release will serve to deliver greater sales records than the year before.

2015/2016 are looking fantastic for Intel.
 
One thing I found interesting was the potential for dual M.2 slots on the Skylake chipset, allowing RAID - I think if you had some fast PCI-e drives in there, in RAID, you'd see some pretty fast speeds.

Z170 is a very nice platform upgrade for sure. DMI 3.0, enabling the CPU to talk to the PCH (chipset) much faster, with 20 lanes of PCI-E v3.0 bandwidth from CPU to PCH, in addition to the 16 lanes directly from CPU.

It antiquates Z97 as soon as it launches, pretty much.
 

? Do you not understand the months of the year? :confused:

What I wrote:

"Skylake-k July, August and September (Q3)" - as the roadmaps shows.

You proceed to link to an article claiming Skylake will release in August - which is within Q3 and explicitly written in my post.

Again I have the feeling you just want to troll, reply to everyone one of my posts with a contradictory statement, then posting a link to something that confirms what I originally posted.
 
It was supposed to launch in Q2, which the article says?

Not sure how it being delayed = "on track"? :confused:

The roadmap was updated to 2H 2015 for Skylake months and months ago.

The Q2 was mentioned in some very old roadmaps, dating back to June/July 2014 - those roadmaps are completely out of date now.

Skylake is 'on track' for 2H 2015, as has been for many months now, then the new roadmap leaked yesterday has refined that date to Q3.
 
Another benchmark, again confirming a 15% IPC increase over Haswell!

Today someone ran geekbench3 on a I5 6600K. Engineering sample, or maybe an Intel employee having fun, who knows.

http://browser.primatelabs.com/geekbench3/2491359

I5 6600K = 3967 Single Core Score

Comparing to an identically clocked Haswell I5 4690k (3.5Ghz, 3.9Ghz turbo)

http://browser.primatelabs.com/geekbench3/2495318

I5 4690K = 3426 Single Core Score

--------------------------------------------

Skylake I5 6600K (3.5Ghz, 3.9Ghz turbo, 6MB L3 Cache)= 3967
Haswell I5 4690K (3.5Ghz, 3.9Ghz turbo, 6MB L3 Cache)= 3426

Showing how I calculated the percentage, since Boomstick has a hard time understanding them! (j/k):

3967-3426 = 541
541/3426 = 0.1579100992410975
0.1579100992410975x100 = Rounded up to a nice 15% IPC increase :)

Hopefully we'll get some overclocking leaks soon! If this thing clocks well, it will be the best architecture since Sandybridge.

Also wanted to note that the motherboard the 6600k was tested on was the 'ASRock Z170 Gaming K6', it would appear Skylake is most likely fully ready to launch. Motherboards are ready, perhaps only waiting on mass production to produce enough to satisfy the most likely insane demand on release :)
 
Last edited:
Both done on different version of Geekbench though. The 4690k was done on v 3.3.2 yet the 6600k was done on the older v 3.1.2. Why would you run a older version apart from making a cpu look better than it actually is? The comparisons of Cinebench at the begining of the thread are the same. You can't compare scores between R11.5 and R15 as they are different. Benchmarks need to be the same version or they are useless.

Hopefully more benchmarks will get leaked soon, with identical versions of the benchmark. I didn't notice the versions of geekbench were different.
 
The benches were apparently actually guesses based on available information originally posted on a Turkish site.

The geekbench wasn't a guess, it was validated on their website.

Best to link sources to statements, or these forums would be full of statements like that.

For example "A Egyptian website found that Skylake can overclock reliably to 23Ghz"
 
Lol, so we have no real idea how these perform yet then. Best to wait for reviews.

How many here will be willing to switch from 4770K / 4790K is it's over 10% quicker?

Is it even worth it?

Orangey failed to give any source for his accusation that both benches were fake, even when I asked him in an earlier post.

You and I, and others, have already commented several times in this thread that it would not be a wise decision to upgrade from Haswell to Skylake. There's simply not going to be enough difference to warrant it. Even if Skylake had 20% improved IPC and similar clocks - the difference in games would be only a couple of FPS etc.

Instead it's the Nehalem and Sandy Bridge users (possibly Ivy Bridge, due to the age of the chipset) that would see the most benefit. It would be like night and day upgrading from a I7 900 series to the top Skylake model, for example.

Then you have to factor in the huge improvements in the Z170 chipset - namely the DMI upgrade from v2 to v3 - which upgrades the link between the PCH and CPU from PCI-E V2 (as found on Z97, Z87 etc) to PCI-E v3 - giving loads of bandwidth to support PCI-E NVME SSD's, without any bottleneck.
 
Back
Top Bottom