No wonder laptop sales are down

Hello,

Sorry for being so candid, but I have to get something off my chest here.

I bought a laptop in about 2011, its great. I love it. I still use it. It's fast, SSD, half way decent video card, 32GB RAM. Solid.

I want to upgrade, its been almost 5 years. I buy a new laptop and here's the comparison between processors:

http://cpuboss.com/cpus/Intel-Core-i7-4710HQ-vs-Intel-Core-i7-2820QM

You're telling me, almost 5 years they can only get a marginal increase in performance. I'm tempted to sell this laptop, the video card is much better... but everything else is about the same (and it only supports 16GB). Even if I wait till skylake (I think that's the name of the next one), it'll probably be another marginal jump again.

Come on Intel!!

Jeff.

 

It only gets worse bro, we are going backwards not forward......

If you were waiting for the SkyFlake, please don't........

New leaked benchmarks.....













not even close to 3-5% in procesor performance increase

http://wccftech.com/intel-skylake-core-i7-6700k-review-gaming-performance-5820k/

 

The ONLY reason I would consider waiting, is the 5 series chips supposedly supports 16GB RAM chips on a single slot.... would allow me to get back to 32GB RAM at least in a 2 slot laptop. I would have to confirm this though.

That's insane stats! Do they only care about power consumption at this point, have we reached the fastest processors we need?

Thanks
Jeff.

 

There is nothing out there pushing CPU limits other than gaming and photo/video. For eral power in any of those you really need to go desktop there is no need to push the laptop envelope. This especially since the lack of a competitive option.

 

Efficiency has taken over performance as the main driving metric for Intel for quite some time now.

 

cpuboss? Really?

Sure, not much better as far as 'scores' go, but passmark has been pretty good comparing different platform capabilities for me.

See:
http://www.cpubenchmark.net/compare.php?cmp[]=885&cmp[]=2243

~23% better single thread improvement and ~20% better overall improvement. Not too bad in my books for a 3 year difference in platforms and all this with a mere 100MHz higher clock rate (3% higher) in the same effective TDP.

If you pick a Broadwell platform which is today's/right now current gen, it is almost 10W less TDP required to just push past the 4 year old platform you had. Again, not bad for a mobile solution.

If you really wanted a real jump in performance you wouldn't be still on a mobile platform. But if mobile is a requirement a desktop processor is still the way to go (even if you have to cram it into a mobile chassis).

We haven't reached the pinnacle of processor performance by far, but the other subsystems need to catch up a little first (see the 3D Xpoint Memory thread for more of that info).

Thinking that performance has stagnated is not a sign of where things are - it is an indication that we need to look further than fluff websites like cpuboss.


As for what Matrix Leader is posting all over the forums, that is a little laughable too.

Not really comparing apples to apples, are we?

The platform as a whole is greater than the sum of it's parts. Picking a low end cpu on one end of the spectrum and comparing to platforms much higher, even if they're now older and even against cpu's with 50% more cores and L2/L3 cache, etc. is pitiful. Especially when that old tech is over 100W TDP higher (37W vs. 140W TDP's).


But enough with 'scores' - what is your workflow for your new/old machines? Do you think your new GPU can do what it is capable of in the old chassis (even if it was possible to install it there) - here is a hint; it can't.

Yeah, the RAM limit is a tough break - but it is not the new processors fault; that is strictly on you for choosing the wrong platform (there are systems out there with 4 SoDimm slots available with the newer platform you have, after all).

Don't think I'm picking on you or that I'm defending Intel.

Just showing you another way to look at things. A way that is more helpful than simply repeating the mantra that Intel or whoever is 'holding back on the performance we think we should have by now'.

Viewing the new platforms in that light is illuminating.

A Hex Core/12 Thread platform with over 100W more TDP and with the support and benefits of a desktop chassis and chipset and without being held back by an on board igpu and with double the cache and 100MHz higher clocks is barely able to stay ahead of the latest Skylake platform.

Even worse, the i7-4790K compares even worse against the i7-5820K Hex Core, performance-wise - especially clocks, but at least it has only an 88W TDP vs. the one quarter newer gen Hex Core's 140W requirements.

Hmmm... very interesting, eh?

So, now what do you think? Are your workflows faster, overall? Do you simply have buyers regret for not waiting for Skylake to go into full swing? After all, you say you're still using your old setup and it still performs great.

But that is more to do with the 32GB of RAM and the SSD it has than anything else.

 

As stated previously, unless you do something extremely CPU intensive, there is no need to upgrade from Sandy Bridge. Even gaming doesn't really suffer much at all on a sandy bridge quad core coupled with a decent video card. Web browsing, media usage, and H.264 encoding all work fine and quite well on Sandy Bridge and above. I wouldn't suggest Intel is regressing, but they are changing their packaging methods and cutting down on user updatability. My daily typer is a 2011 Thinkpad X220 tablet with a i5 2520M and a modern/fast SSD. I use it more than any of my other computers, and it does the job just fine. I honestly think we will not need to upgrade/change computers until 4K becomes the standard, which would probably hamper/kill most systems with iGPU from 2011 till....who knows whenever the iGPU becomes able to handle 4K?

 

Except that it's getting worse. *looks at Haswell/Broadwell vs Ivy Bridge for power efficiency under heavy loads*

 

@webjeff as TOTE says, what sort of load are you using with your CPU and what laptop is it? Try posting a couple of CPU benches for it relevant to your personal usage so we can see your current performance level.

Broadwell has been shown to support 16GB modules but still need to check it has BIOS support for the system you are interested in. Seems Micron's 16GB DDR3 SO-DIMM's, are finally in production also.

 

It's a combination of middling performance increases, particularly on the CPU side, and what TANWare said in post #4 - you can indeed still do most tasks on computers that are quite old. For the average user, a Wolfdale E8400 CPU from mid-2008 is likely good enough. Throw in a RAM upgrade and maybe an SSD, and they're good. I still use my 2007 laptop occasionally - for browsing Hacker News or posting on NBR, it's more than sufficient. Try using a 1997 laptop in 2005, and you probably would not have nearly as good of an experience.

That said, Intel has done a poor job of giving those who could benefit from higher performance a good reason to upgrade. I have a 2500K; why would I want to dole out for a Haswell or Broadwell system and a new motherboard for perhaps 10% better performance? It only makes sense to upgrade if you're made of money, or are making your income from the computer and are highly CPU-bottlenecked. Intel has switched towards focusing on efficiency, and perhaps even more so iGPU performance.

In a way, it's impressive that AMD and nVIDIA have managed to improve GPU performance so much since 2011 despite being stuck at 28nm, while Intel has done hardly anything to (non-IGPU) CPU performance in that time, despite moving down from 32nm to 14nm.

For CPU performance, I think AMD's Zen in 2016 will be the next interesting jump. Despite having had higher year-over-year CPU performance increases the past few years, AMD's still well behind Intel in performance since they were so far behind 4-5 years ago. But if Zen lives up to projections (40% increase in IPC), AMD will be respectably close to Intel in single-threaded performance (and potentially ahead in multithreaded if they keep reasonably priced desktop octo-cores; Intel's octo-cores are still very expensive). And that would mean both that AMD could potentially move the high-end needle forward in Zen's 2017 successor, and that Intel would finally have a reason to focus more on performance again.

Until that time, though, I'm sticking with Sandy Bridge.

 

The difference between central processor workloads and graphics workloads is why graphics continue to improve dramatically but processors are improving incrementally. There are a great number of programs that (even now) require single-thread performance more and are not fully optimized to divide up the workload according to the number of threads the processor can handle. Proof of this can be seen in the fact that 6 years after mainstream quad-core mobile processors were introduced, more than half of notebook models sold still use dual-core processors. Graphics, on the other hand, use massively parallel workloads, so adding more "cores" or "shader units" always results in higher performance.

Intel hit a wall with increasing clock speed in 2006, and they have only been able to move it slowly. The solution at that time was to add multiple cores. That worked for a while, but then performance plateued and they came up with Turbo Boost. Now they're at a point where performance gains plateued again. But the consumer landscape is different now, and for the majority of people, the current processing power is sufficient. As AMD is still light years behind on performance, Intel has been focusing on power efficiency and increasing battery life. Batteries haven't changed significantly for 20 years, so any improvements in battery life are the result of the components using less power to do the same amount of work. And since that also allows thinner systems with less cooling requirements, both the manufacturers and most customers are behind it. So that's the direction Intel is going, both because they can't make their processors significantly more powerful and also because few would care if they did.

 

See:
http://forum.notebookreview.com/threads/no-wonder-laptop-sales-are-down.779406/#post-10055634


The post above was directed to you. Still waiting for the questions I asked to get answered.

Business and tech direction of a company only coincide when more $$$$$$$$$ is to be made than otherwise. Complaining about this direct connection of the #1 reason for a company to exist (to make $$$$ for it's shareholders) is simply wishing for tech to wither and die and never move forward, starting today. Investors would jump off that ship (a company giving everything it has all at one go) in droves and the company would be just a footnote in history (if that).

The talk about AMD's next offerings are humorous too. See the post in the link above and then revisit the idea that AMD in one massive jump will not only catch up, but even surpass Intel in compute performance.

Yes, Intel has had the luxury of keeping huge, raw performance jumps on the backburner for a while now, not that advances haven't been made though. But they are not sitting idly by waiting for whoever to catch up to them. Performance per Watt has been their goal and their mantra for a few generations now; and they are almost untouchable in that discipline. What most don't seem to understand is that if AMD or anyone else comes up behind them in a surprise move, all that Intel needs to do is issue 45W/55W versions once again of their current 25W/37W platforms (at that time) and easily regain the performance crown for another few years.

There was a time where I thought like many on the forums here and for mobile platforms, it is mostly true if we look at mere % increases from the last few years to the % increases from a decade ago. But whenever I updated a desktop platform the improvements to my productivity were not so subtle. Why? Because those platforms are not as restricted by noise, power or cooling requirements as much as mobile platforms are.

What Intel has been achieving with each new generation of their tech (whether by happy accident or design, I'm not sure...) is a system that as a whole is better balanced as time goes on. They are slowly erasing the bottlenecks that stand out like a sore thumb and for me, that is the true progress that we see vs. old tech that on the surface seems to be able to stand with and among todays platforms on a somewhat equal footing (or so it seems for some aspects).

I remember when RAM was so expensive that a computer was worth more in RAM than the rest of the chips inside. But the problem back then was that the cpu and the O/S architecture could not leverage more RAM effectively. The 'balance' was off between the two.

I remember when RAID0 arrays (w/VRaptors) increased my productivity. But overall, almost all the benefits could be erased by having one or more of my workstations out of commission when that same array died (be it the RAID card or because of drive failure). Again; that 'balance' that is so important was off (this time; the balance between productivity and reliability).

I remember too when I could easily buy so much RAM for all my systems and ended up returning most of it. Why? The cpu's/platforms still could not scale so much (Vistax64 was the turning point).

No one can forget when we could take our mobile powerhouses out to jobsites and without AC power, they died in 45 minutes or less. Again; balance.

When SSD's finally became viable to me (starting with the Intel 520 series 240GB, mostly because of capacity and the huge effects OP'ing has on sustained performance on any SSD I have used...), another piece of the puzzle was fitted and in my real world use, proven to further the balance between the rest of the components. And to do so with a real and true overall increase (and not just in one or two corner cases; like all the SSD offerings before).

Today on the latest platforms, the O/S, the cpu, the RAM and the storage subsystems we have available are more balanced than they have ever been before. On the latest Broadwell platforms I have played with, this is even more true (some, with the same peripherals/hardware I'm using now...).

As long as Intel keeps pushing for this balance and the erasing of bottlenecks between components with each gen, I will be at least willing to test against what I have at that time and see if the productivity offered in my real world put food on my table and a roof over my head workloads warrants an upgrade to the new platform at that time.


The 5% and 10% + increases of each generation encompass whole cpu/igpu architectures from once upon a time.

Those that deem this insufficient are the ones that have a platform with no real use for that level of power, imo.

Will we never see huge 2x and orders of magnitude jumps again in tech? Of course we will (see 3D XPoint for more). But expecting it as the norm with each new generation? That is not a realistic world today.

 

I still use the laptop in my signature (which at this point is 7 years old).
Switching over to Haswell for instance would be a pretty high boost for me, but I'd be willing to wait until Skylake hits the shelves since the market is in the state of transitioning to it anyway in a couple of months.

I was eyeing a laptop with an i7-4790S inside, however, since it is yet undetermined whether or not I'll be accepted onto a college (I will know in 2 weeks time), I was thinking on waiting for either Broadwell or Skylake equivalent (if it won't be 'pressing' for me to get a new system that is).

 

I'd agree with you, except that I've heard this song before: 4-5 years ago, AMD was supposed to finally catch up to Intel with a brand-new and radically different architecture called Bulldozer... and we all know how that turned out. I'll believe a huge improvement from Zen when I see it.

 

The software side of things has gotten a *lot* better as well. I just installed Windows 10 on my 8-year-old laptop. It has never felt faster, nor more functional, and that's largely on account of the software being more aligned to the hardware (ie: able to take advantage of multiple cores). Also, since mobile has been the big driving force behind software sales, the developer community has put a lot of effort into reducing the impact of code bloat in order to make software run acceptably on mobile devices -- stuff that has leaked back onto the desktop.

I'll buy new laptops when 4K laptop screens and dual 4K outputs are available on normal business-class laptops (ie: not the super-duper mobile "desktop replacement" machines). At this point, would also like to see PCI-E SSDs standard, and even WiFi, Bluetooth, and 4G on the same chipset. Dual (u)SIM's would be nice as well, as would 10gig-E.

 

Come back in... 2020.

 

Yeah, the key word is "if" in "if Zen lives up to projections". No one really knows yet.

 

They have, just not in single thread performance.

From SB to HW, the EP/EX chips have seen ~100% increase in general throughput. From 32nm to 22nm the managed to scale from 8 cores up to 18 cores with similar, if not better single core performance. They are improving as much as the dGPU guys.


On the mobile platform, however, they see no point going beyond 4 cores anyway.

 

OK, true, but I was thinking of the more generally affordable chips. In mobile, it's been pretty much dual-core i5 and quad-core i7 (with a few dual i7s) since Westmere, whereas desktop has been mostly dual-core i3 and quad-core i5/i7 since Nehalem, with a few hexacores in the top end, and with Haswell finally an octocore option for a cool grand. Servers have increased core counts a lot more, but I doubt very many people at NBR will be using those for their primary desktop.

 

Check out the anandtech review on Skylake ( http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation): the gain is quite minor, even smaller than as in the past. If you're potentially moving away somewhere and need to buy, then you can do so largely without regret. Check the chipset features list first and see if any excite you.

Even 2020 might be a bit optimistic. Especially if you're mixing in 10gigE on the list too.

 

No, forget anandtech. They've sold out...

See:
http://www.tweaktown.com/reviews/72...0k-cpu-z170-chipset-gt530-review/index14.html


For a much better real world impact of the Skylake architecture.

 

Sold out to who...?

 

Hint; it's who Anand himself is working for.

See:
http://www.anandtech.com/show/8456/the-road-ahead

See:
http://appleinsider.com/articles/14/08/31/apple-hires-noted-tech-journalist-anand-lal-shimpi

 

Anand left the company a long time ago

 

Less than a year ago, officially. But he was bought long before that.

The quality of the work coming out of anandtech was below par for a long time before he left.

 

For example?

 

For example, apple was always applauded while MS or almost everyone else was shown the error of their ways. Even the Intel news was more positive after apple used their cpu's in their systems. And that was at a time when Intel was indisputably at the top of their game.

 

That's so vague, I have no idea what you're talking about. If you dislike AT so much, stop linking to them for SSD stuff. I read AT because they dive deep into CPU/GPU architectures while managing to keep the esoteric stuff to a minimum so as to be intelligible for average readers, which no one else seems to be able to do.

 

Sorry, it is vague. But I have no time to go back 4 or 5 years and show where the slippery slope started (or at least, where I noticed it...).

As for the SSD 'stuff' - since people here don't believe my empirical evidence, sites like AT shows what they think is objective evidence of the ideas I try to convey.

And I do agree, when AT goes into a deep dive on a topic, they are untouchable in being able to make it understandable, even for mere mortals like me. But even those articles are few and far between lately from them.

 

Sorry, couldn't resist. The grammar nerd is strong in me.

 

Intel doesn't push top-end CPU performance because it doesn't need to. They don't have any competition from AMD on the high-end, so why bother throwing gobs of R&D money competing against yourself?

On the other hand, Intel is trying to enter the low-power space, so it can compete with things like embedded computing, smartphones, and tablets. So a processor like Broadwell focused on lowering power and heat, rather than performance boosts. It DOES make sense for Intel to put their R&D focus behind low-power computing, because that's a part of the market where they can grow as a company.

On the bright side, high-end CPU performance has become largely irrelevant. There's actually very little that you do today that depends on your CPU performance. Once you get a CPU that is "fast enough", then you're actually much better off focusing on other components to get tangible performance benefits (like SSDs, battery life, screen quality, and better GPU performance).

 

They should give us a mobile phone with true desktop experience and decent battery life at last. Those previous efforts sucked battery-wise, performance not being stellar either.

 

Thanks for pointing this out, I hadn't saw that he was now working for apple. Lately I've found a dip in articles that really keep me reading for a long time, but I had chalked that up to being natural with a change in management. Interesting that there might be something else behind it.

Their coverage of apple content was more positive than other PC-related sites I visit, but I didn't find it anywhere near as biased or outrageous as mac-related sites that were out there. It is true though that everybody says that Apple will outright remove your access to all future devices if they're unhappy with anything that is written. In other words, you'd have to demonstrate some bias in order to keep receiving the goods, so to speak.

 

Last summer. Not really what I would consider to be a long time ago, like, say, Thomas Pabst leaving Tom's Hardware. It is unfortunate for the site that he left, but if there's something consistently better, I haven't found it yet (and please do link it). They've still got a lot of good content and writers.

 

Anand's interests changed every few years while he was still writing for AT. In the years leading up to his departure, he wrote almost exclusively about mobile (smartphones and tablets), so I don't really miss him.