Intel Cannonlake touted to offer >15pc performance improvement

[X4n63r]

CannonLake is 14 nm according to Anand, Guru3D, etc. Not 14 nm after-all (guessing yield problems?). Also, the Intel roadmap makes it clear that CoffeeLake is the high-end desktop part (should be 10 nm, tho'). Desktop is all KabyLake for the time being ('till 2018)? CannonLake for U and Y parts?

I'm trying to say that this 'news' is chaff from Intel. We're going to be getting a lot of this the next few months

[ETR316]

Saracen sounds like he either has another agenda (to bash intel) or you do still drive a 1999 car with 185.000 miles on it and its still going strong. go write a book.

[badass]

I'm interested to know what others, if anyone's excited about this, thinks the real-world actual benefit to their daily PC usage will be? What will it let you do you can't do perfectly well already, or what prodictivity gains will be really noticeable?

I think the main use case where a 15% gain makes a difference is someone like me except with an older system.

It looks like the core i7 7700k averages about 35% faster than the Core i7 3770k and about 80% faster than an (8 year old!) Core i7 880

15% performance increase could be the difference between waiting a generation and buying now.

However I do not believe at all that there will be a real world 15% performance increase There will be 1-2 benchmarks out of loads where there is a 15% performance increase with at best a 5% average.

The sysmark benchmark mentioned by intel will be for a cherry picked mobile processor to its replacement.

[DanceswithUnix]

There will be 1-2 benchmarks out of loads where there is a 15% performance increase with at best a 5% average.

I'm fully expecting Intel to carefully bin the fastest silicon and put out what is basically a factory overclocked part. A lot of people buy a CPU and then overclock to 4.8GHz, perhaps Intel will just be making that legit. That should get some increase across the board.

[peterb]

Saracen sounds like he either has another agenda (to bash intel) or you do still drive a 1999 car with 185.000 miles on it and its still going strong. go write a book.

Welcome to HEXUS....

Saracen is pragmatic and is asking what real world difference will the average user see. As he says, he was looking for applications where the headline 15% figure might make a difference.

And I'd be interested too, especially where CPU performance is unlikely to be the limiting factor in most applications. The exceptions where I might see a difference would be transcoding video files, where processor speed and available RAM would be critical. Applications that don't have demanding video output requirements might use a GPU for complex calculations or batch preocessing, but these are applications that the majority of users would run (although I'd be surprised if there aren't any Hexites doing these things as some are at the forefront of technology.)

However, instead of having a poke at one of the more respected HEXUS contributors, (always a dangerous thing) why not make a positive contribution of your own, rather than knock someone else's? I realise you haven't been around long enough to get a real flavour of contributing to HEXUS, so I suggest a bit of research mupight be in order to ensure you get the best out of the forums.

[Saracen]

Saracen sounds like he either has another agenda (to bash intel) or you do still drive a 1999 car with 185.000 miles on it and its still going strong. go write a book.

As a tech journalist, I've worked closely with Intel, and MANY others, for oh, since a '386 was state of the art.

And, in those days and for quite a while afterwards, every (or nearly every) new generation of processor saw a marked performance increase that either resulted in real-world productivity gains, or made some new processor-bound application feasible.

I've lost count of how many times I've tested new processor releases, usually either golden sample or engineering samples, and many's the time I've bought state-of-the-art processors on, or even before, release, because of the real-world benefits it gave to something I was doing, be it handling very large scanned images, or 3D modelling on a workstation, or Photoshop filters, or whatever.

But those days have gone, and as far as I'm concerned, new generations add little to my actual productivity, and even less to enabling new uses that were too processor-bound to be practical before.

I'm not knocking Intel, partly because I have considerable respect for them, and partly because the same holds true of AMD.

If I go out and buy a new PC, based on newgen CPUs, but otherwise matching the spec of an existing machine fairly closely in terms of memory, disk or SSD space, etc, exactly what will that new machine do that one even several generations can't, in mainstream real-world tasks? Will my letters be written any faster in a WP? Will my email be sent or received any faster? Will my accounts do themselves? If a Photoshop filter runs in 0.5sec rather than 0.7sec, is it going to make my day more productive?

If my existing machines do what I need of them, and work perfectly, what justifies the capital cost of replacing one with another that'll costs hundreds of pounds to do, and make no difference to how well or how fast my work gets done.

Yet, to survive, most PC component manufacturers have to convince us to keep replacing perfectly satisfactory hardware with new, faster versions when that extra performance will make no real difference.

And no, I don't drive a 1999 car, and I've never owned a car with 185,000 miles on it. But if I did, and was happy with it in terms of performance, comfort, running costs and reliability, I'd need a hell of a lot of convincing that it was worth spending several tens of thousands of pounds replacing it unless the benefits I got for my money were clearly quantifiable.

[DanceswithUnix]

Yet, to survive, most PC component manufacturers have to convince us to keep replacing perfectly satisfactory hardware with new, faster versions when that extra performance will make no real difference.

I think most of the improvement recently has been aimed at 15W to 35W laptop footprint with a nod towards lower powers. Such users have probably seen a big change, but to a desktop user like me it has been a bit of a dry spell.

[Saracen]

I think most of the improvement recently has been aimed at 15W to 35W laptop footprint with a nod towards lower powers. Such users have probably seen a big change, but to a desktop user like me it has been a bit of a dry spell.

Well, quite. One of the 'generations' that enabled things not previously possible was when designs started targeting thermal footprint, low power consumption, and even passive cooling. I entirely accept that improvements don't always involve performance, but this generation seems to be aiming at "15% performance" and I just find myself wondering "So what?"

We've seen this so many times .... when some new technology hits the market, early generations make huge leaps, but then the msrhibal rate of improvement slows, and while marketing departments milk it for all it's worth, sooner or later they move on to other features as the product cycle goes from new, to innovative, to stable and then to mature.

Think colour inkjet printers. I remember getting my first HP Deskjet 500C, and it was fabulous bring able to print colour, at home, at all. And while it's ability to digitally print photos at all was novel, it coukdn't compare to a 'proper' photo. The next few years saw the 'resolution' wars, then four, six, seven eight ink colours, and so on.

And now? While those of us for ehom photography is a serious hobby, or business, are still very careful about pointer choice, and I have several for different purposes, most people can get perfectly acceptable enprints from just about any generic colour inkjet. The 'resolution wars' are over, because printers reached the point (years ago) where further increases didn't result in any further perceptible improvement in print quality. So printers started targeting speed, paper handling, and so on.

Like printer resolution, and probably camera resolution too, I wonder if CPU performance has already reached the point where current performance is, excepting special needs, fast enough.

Using the car example thrown at me earlier, if my car does 180mph and 0-60 in 5.0 seconds, am I going to spend £50k upgrading becsuse the new version does 185mph and 0-60 in 4.9 sec? Not unless I'm seriously sad and need to get a life, I'm not, because the difference offers me no real world difference. There might be other reasons to change cars, but performance isn't one of them.

[CAT-THE-FIFTH]

I think most of the improvement recently has been aimed at 15W to 35W laptop footprint with a nod towards lower powers. Such users have probably seen a big change, but to a desktop user like me it has been a bit of a dry spell.

But TBH I kind of agree with what Saracen is saying - a family member was using a 2GHZ P4 laptop until recently and the only reason it was struggling was since the IGP could not decode flash.

They are currently using an Atom based tablet as a replacement and its more than fast enough.

If it were not for the fact that I game,and I am interested in photography,even Core2 level performance in a laptop is enough as long as the IGP is modern enough,and a console will be good enough for gaming. In fact for most photography stuff people do,a Core2 is enough as long as you have solid state storage.

The thing is for most people the most content creation people will do is the odd document,ie,a CV and covering letter and maybe the odd spreadsheet - even a tablet could handle it.

You can kind of see this in the market - more and more people are just their phones or tablets as their main computing device.

This is why companies foster this one-upmanship amongst PCMR now,so it drives sales for smallish improvements.

Edit!!

Its like that A6 3670K system I won here - still fine after many years. The only thing which is noticeable is the lack of SSD and I might add one. In fact an SSD is probably one of the biggest improvements I have seen in years,and I suspect that is why the initial iPads did well - the solid state storage and the decode abilities of the IGP meaning people had a responsive experience.

Well, quite. One of the 'generations' that enabled things not previously possible was when designs started targeting thermal footprint, low power consumption, and even passive cooling. I entirely accept that improvements don't always involve performance, but this generation seems to be aiming at "15% performance" and I just find myself wondering "So what?"

We've seen this so many times .... when some new technology hits the market, early generations make huge leaps, but then the msrhibal rate of improvement slows, and while marketing departments milk it for all it's worth, sooner or later they move on to other features as the product cycle goes from new, to innovative, to stable and then to mature.

Think colour inkjet printers. I remember getting my first HP Deskjet 500C, and it was fabulous bring able to print colour, at home, at all. And while it's ability to digitally print photos at all was novel, it coukdn't compare to a 'proper' photo. The next few years saw the 'resolution' wars, then four, six, seven eight ink colours, and so on.

And now? While those of us for ehom photography is a serious hobby, or business, are still very careful about pointer choice, and I have several for different purposes, most people can get perfectly acceptable enprints from just about any generic colour inkjet. The 'resolution wars' are over, because printers reached the point (years ago) where further increases didn't result in any further perceptible improvement in print quality. So printers started targeting speed, paper handling, and so on.

Like printer resolution, and probably camera resolution too, I wonder if CPU performance has already reached the point where current performance is, excepting special needs, fast enough.

Using the car example thrown at me earlier, if my car does 180mph and 0-60 in 5.0 seconds, am I going to spend £50k upgrading becsuse the new version does 185mph and 0-60 in 4.9 sec? Not unless I'm seriously sad and need to get a life, I'm not, because the difference offers me no real world difference. There might be other reasons to change cars, but performance isn't one of them.

Welcome to DIY PC building circa 2017.

[Saracen]

But TBH I kind of agree with what Saracen is saying - a family member was using a 2GHZ P4 laptop until recently and the only reason it was struggling was since the IGP could not decode flash.

They are currently using an Atom based tablet as a replacement and its more than fast enough.

If it were not for the fact that I game,and I am interested in photography,even Core2 level performance in a laptop is enough as long as the IGP is modern enough,and a console will be good enough for gaming. In fact for most photography stuff people do,a Core2 is enough as long as you have solid state storage.

The thing is .....

.... snip ....

Exactly.

I guess I'm saying, as with the car or inkjet printer analogy, how fast is fast enough, and how many pixels (yes, for purists, I know they're not pixels, but that's how most people see it) is enough, and at what point does more pixels make no visible difference.

After all, the ultimate limit (for photo quality) is the eyes of the person viewing the photos, and once printer technology reaches the point we can't see the difference, is there really a difference?

And once processor performance has reached the point where the user can't tell whether they're using the older, slower or the newer, faster chip, then what's the difference? Unless, of course, all that extra CPU power enables some new usage which we don't currently have due to inadequate CPU grunt. Real-time voice recognition and dictation suffered that for a while, where software led hardware, but hardware caught up to that years ago.

As for DIY building, if I need an extra machine, or if an existing machine croaks and needs replacing, then I'd look long and hard at available hardware options, and pick based on my needs, and price differentials. Unlike early days when CPU performance enhancements meant noticeable real-world benefits and I tended to go very high end, these days my get is that for the vast majority of users, mid-range or even budget CPUs would do the job just fine, as long as your objective is real-world capability (excepting those rare specialities and, perhaps gaming) and not willy-waving your benchmark results.

Put it another way.

Most people don't have unlimited budgets. If I have a defined budget that's reasonable but not huge, do I go state-of-art next-gen CPU, at leading edge prices, or a midrange current-gen CPU, and maybe more RAM, a much bigger SSD and/or an extra chunk of cash on a gaming graphics card?

Or, just go for the modest CPU and use the cash I save to take the Missus away for a nice weekend. Or buy a new TV. Or .... etc.

Given the nominal (IMHO) real-world benefit of a supposed 15% performance gain, does they represent a good use of money over modest current-range processors? IMHO, no. We're at risk of chasing marketing fairy dust.

[DanceswithUnix]

But TBH I kind of agree with what Saracen is saying - a family member was using a 2GHZ P4 laptop until recently and the only reason it was struggling was since the IGP could not decode flash.

Wow, I'm impressed with their frugal use. I found the P4 pretty unusable when they were new and the slowest I had to put up with was a 2.4GHz one

OTOH, my Shield tablet has been a joy to use so I do get what you mean about good enough computing.

[CAT-THE-FIFTH]

Wow, I'm impressed with their frugal use. I found the P4 pretty unusable when they were new and the slowest I had to put up with was a 2.4GHz one

OTOH, my Shield tablet has been a joy to use so I do get what you mean about good enough computing.

All they were doing is basic web-browsing and a bit of YT,but flash got heavier and heavier,so it was eventually retired. Also they needed to get off XP,since it is not very secure now.

Remember this thread I made about the latest Atom CPUs:

http://forums.hexus.net/pc-hardware-...o-l-chips.html

It will be interesting to see how things progress. I suspect in a few years you could use your phone to power a desktop experience,ie,a larger screen,web browsing,office work,watching videos,etc.

[j.o.s.h.1408]

tick or tock?

[CAT-THE-FIFTH]

tick or tock?

Not sure but I suspect its another overclocked Skylake unless the next generation has been pull forward but apparently that has 6 cores,so would be considered more than 15% overall I suspect.

[DanceswithUnix]

tick or tock?

There never really was a tick-tock it is all marketing nonsense, and Intel have completely given up on it now. Officially in Intel's new Process-Architecture-Optimisation replacement for tick-tock, Cannonlake is supposed to be a "Process" where the cpu is mainly a die shrink, yet it incorporates things like some new AVX512 instructions so it is more than that.

Most interestingly, on Wikipedia it says that Cannonlake tops out at a 15W part with a 5W part also available, so as some of us expected this is just about "ultrabook" sustained performance though better thermals.

https://en.wikipedia.org/wiki/Cannonlake_(CPU)

However, Coffee Lake should be released in 14nm for desktop use with 6 cores. A bit more interesting, but I wonder how that compares to the old Haswell era 6 core Xeon I use at work.

[kalniel]

tick or tock?

Rat-a-tat-tat