Intel claims 8th-gen chips will be 30 percent faster than 7th-gen

Intel claims 8th-gen chips will be 30 percent faster than 7th-gen

Intel's upcoming 8th generation processors could be more interesting than we initially thought. While the company previously said they would be a mere 15 percent improvement over 7th-gen chips -- a performance gain it's been achieving steadily over the years -- Intel revealed today at Computex that the new processors actually perform around 30 percent better. Sure, that figure is only based on a Sysmark benchmark, but it's an encouraging sign for overall performance improvements.

Intel didn't have much else to share about the chips during its Computex keynote, unfortunately. They're still expected to arrive by the end of the year, and they're still based on a 14 nanometer architecture, like before. The company briefly showed off a Quanta machine powered by one of the new chips, but aside from looking respectably thin, there wasn't much we could gather from it.

 

Same way 7th gen was "15% faster" than 6th gen and 6th gen was, supposedly, "15% faster" than 5th gen? Yeah, I bet this is iGPU talk again. Maybe a 10% CPU improvement as well.

EDIT: Also don't forget Intel licensed tech from AMD for their newer iGPUs - I wouldn't be surprised if that's where the performance improvement comes from. As we already know, R7 iGPUs are substantially faster when compared to the HD series from Intel.

 

True, but it seems Intel is comparing the i7-7500U 15W to the new 4 core, 8 threads 15W chip....so a performance gain from a dual core to a quad core is expected...although it is impressive that they are managing to offer a quad core at 15W for ultrabooks. Let´s see...

 

Larger L2 caches by a factor of 4 = more work done.

I want to see gen 9 in a couple of years with even larger caches (L1, L2 and L3) along with other evolutionary improvements...

An igpu doesn't make the processor more powerful, btw. Rather; the other way around.

QC @ 15W TDP is insane progress - regardless what some may think.

When a two year old, 4.5W TDP, $300 2in1 Tablet/Notebook (Asus T100TA) has as much snap as a $2K desktop of not too long ago... the progress is undeniable.

The 'insignificant' upgrades that Intel has put into play in the previous generations to get to here take time to iron out as needed and fully optimize - soon though; we'll be able to reap the rewards.

Comparing CPU generations (aka IPC...) on their own is not doing yourself a good service - the platform they are expected to run on is just as important too... At the clock rates they were built for... When platforms are compared properly to otherwise identically setup systems (Same O/S, RAM, storage drives...) and workflows/workloads - those Intel predicted improvements are obviously there for all to see.

Not each and every platform change, of course. But the progress from one to the other is clearly seen with as simple a 'test' as browsing the 'net. Responsiveness is a high priority for my systems, along with raw horse power to push productivity to the max. Intel actively seeks to improve both while keeping power, heat and noise at their lowest possible values across all the different platforms they are used in.

 

Intel refuted those rumors so we will have to see

 

The day after they "refuted" the circumstantial evidence in favour of those claims, new ones emerged and Intel hasn't responded yet. the last Hard[OCP] article about this story is 1-2 days older than the intel statement and it still holds firm that it is, indeed, happening. I am also going to say it likely as Intel already announced their intention to support Free-sync a long while back.

 

But I wouldn't be surprised if any full CPU load longer than a few seconds will result in the CPU speed being significantly throttled. I generated this graph for a dual core Core m5-6Y54 in a Dell Latitude 7370 using data logged by HWiNFO.



Put two of those on a chip and we have a fully loaded quad core chip running at 1.7GHz and using less than 10W and likely to be no more than 2.5GHz at 15W. In fact, one can get a current quad core CPU down to 15W by lowering the maximum CPU state. I created this graph for my Precision 5510:



A maximum CPU state of 80% resulted in a power drain of 14.6W when running 8 threads. So the question is how much faster the newer CPUs can run within the 15W power rating?

John

 

@John Ratsey could you please try loading iGPU as well, to see how low the frequencies will go?

 

John, those are very interesting graphs, to be sure. Thank you for sharing.

I still think that there are other advances that greatly help with the peak performance (even if it's less than a second... i.e. 'snappiness') and the sustained performance in low TDP and in non actively cooled, computing platforms. Things like hardware accelerators for video, sound and soon; Bluetooth, WiFi and networking performance too within very low (and always on) power levels.

Not to mention totally silent, very low weight and easy to stuff into any big pocket or binder you happen to have with your with barely noticing you have it.

It is great to be able to use my workstation for 8+ hours on a single battery charge by using great utilities like TS or just Windows Power scheme options. But it would still be a big honking notebook to carry around and when I used it for real work - the local power company is still my friend.

When Intel releases their latest tech crowns at both the low and high end; I have very good reason that their performance will be better (and hopefully 'much better') than what we have now.

Particularly when the 'overall' experience is compared to old tech (which to me becomes ancient tech overnight).

Like I hinted at above; when a 4.5W TDP platform is capable of almost doing real work (much, much more than any so called smartphone I've used can), I can easily see the day when my workflow can change to having one or two monster multicore setups onsite with a few 'lessor' 15W QC notebooks to assist vs. the multiple workstation class notebooks my workloads require today.

I remember reading a few years ago that processors are not meant/capable to operate over a wide range of TDP's (at least; efficiently...).

~4W, ~15W and ~45W (and higher) TDP ranges for mobile setups still seems optimum to me.

Along with any added raw performance capabilities, the addition of 4K, 8K and 12K video capabilities for the higher end parts, is worth it for the longevity usefulness I expect from any system being considered today.

 

passively cooled?

 

That's a condition I hadn't looked at when making my graphs some months ago as it's not an operating condition I encounter in my normal workflow.

Yes, the Latitude 7370 uses passive cooling which is why the CPU is allowed only a short burst at full speed. My point is that with active cooling a pair of those chips could run at around 2GHz within a 15W thermal envelope. Move forward a couple of generations and there's the prospect of 4 cores / 8 threads at around 2.5GHz (provided the iGPU isn't pulling significant power).

However, as much of my work involves single-threaded software I will be more interested in the speed and power drain running fewer threads. I also hope that notebook manufacturers don't take the availability of lower TDP to shrink the fans when there's the opportunity to give users an almost inaudible cooling under full CPU load.

John

 

Considering the previous iNtel claims they were going to be 10nm and then suddenly going back to 14nm that tells me iNtel isn't ready for prime time. As for the 30% I'll believe it when I see it saying it doesn't mean it. And by that time AMD would have their own equal CPU to iNtel so if they really want to keep AMD at bay they will have to do more then 30% but also drop their prices if they want to compete now with AMD.

 

I would like to see intel make an i7 with no igpu. For kaby lake the difference between the k and x variants is mainly the binning and the fact that they cut the power flow to the igpu. I personally have no desire to ever use the igpu even on a laptop. I know that many do but to me it does nothing.

 

Yeah, his question isn't clickbait at all. sigh...

So, in conclusion - he's dropped everything Ryzen up to now and if and when both Intel and AMD get their HCCP* out, he'll test then.

Yeah; just what I've been saying all along. Posturing and marketing bs can always be safely ignored. Including any 'questions' from other parties interested in their own agenda...

If/When they actually bring actual products to the consumer... just let them do the talking (the products...) then.

Another 11 minutes of wasted time.

*High Core Count Processor

 

I think the point more so that the loyalists in the media that usually just swallow Intel's hype unquestioned aren't this time around. This is a good thing but it is true we need to let the hardware speak for what it is once here. Now if they are smart they will be sure to have boards with UDIMM support for x399 as I do not think you will see that (ECC) for x299.

 

Yup, Intel has really helped out AMD with these latest and continuing confusions around their new releases, configurations, and limitations.

 

I haven't met any loyalists (Intel's or other) that purchase on anyone's hype.

But I do show my clients what is real and what isn't. Just trying to do the same here.

If my personal computer was stolen or destroyed today and I needed to replace it, pronto; it would be with an identical build (most likely). Why? Because I know that those parts work in my workloads, period.

When I'm contemplating of upgrading... Yeah; I read reviews. I verify facts on manufacturers websites and I even more carefully read other's user experiences more closely too.

But in the end; I buy and test a complete platform vs. what I have and use it as I've used the previous one before.

It either surpasses what I have by enough margins and benefits - or, I return it and continue using my already superior systems and setups.

If it is deemed unworthy of keeping; it doesn't matter what the rest of the world is saying (in videos, text and 'scores'). I know the truth (for me). And back it will go (yeah; full refund except for my time).

If there are some others out there that exist by swallowing hype and have made it to 2017, I haven't seen them around here.

Sure; we've all done that at least once when we were young and the 'tech' was 'toys' - but we've all learned some solid basics by now, no?

 

90% of consumers have brand preferences and over 50% of them are brand loyalists. If that weren't the case, Razer would have been out of business years ago.

 

On the surface that may seem to be the case with me too. But the reason to be loyal or have a preference isn't in what's promised or hyped.

It's in what is actually delivered.

 

Nah, in most cases people prefer a brand because of past positive experiences with that brand and reject a brand because it failed on them in the past. It's not really about promises or what is offered, it's a lot simpler. The majority of people don't buy based on logic or reason, they buy based on what "looks" good on paper in a vacuum.

 

Or it's based on the brands associations. People buy MacBooks to join the club and be seen with the latest model, not necessarily because of the performance. It's like what people think a car or clothes "say about you".


Sent from my iPhone using Tapatalk

 

So 8th gen CPU's aren't Coffeelake, they are revamped Kabylake's?

Intel Coffee Lake CPUs Delayed to 2018, 8th Gen Gets Kaby Lake Refresh This Year
http://wccftech.com/intel-coffee-lake-delayed-2018-8th-gen-kaby-lake-refresh/

So, have the 30% improvements moved out with Coffeelake to the "9th" gen now??

 

The 30% iGPU improvements are being pushed back, yes . We are refreshing Skylake refresh atm

 

Hmmm... this is interesting. This may be a real delay or, lets play the popular rumors game here.

What does Intel know about it's competition that it doesn't seem to care about delaying Coffee Lake by a quarter?

Is TR still missing in action?

 

If they're planning on taking on 16-core TR4 CPUs with a 6-core 1151 CPU, I feel sorry for Intel, genuinely so. 1151 is being delayed due to 10nm not being ready likely. They'll come around the same time AM4+ and Zen+ come out. Aka 2018.

 

So the rumored 30% is for next year or is it for the refresh and what value would there be in refreshing Kabylake if it's going to eclipsed by Coffeelake?

 

I guess we really don't know...

The value in keeping a shipping date is that the vendors that buy Intel's parts synchronize their new model rollout plans with Intel's parts promises.

If Intel can't make a release date, they can substitute a backup plan to give something for vendors to make a big whoop about in their model "refresh" release.

It's not as good as a real new CPU, but Intel's been doing this for a while anyway.

The Kabylake refresh could unlock some more IPC performance, more clock speed, and/or a new iGPU to pair with it all.

 

Guess we'll know as we get closer to the date.