WTF has gone wrong with the PC World?

The electrons don't move faster or slower. There are just more or less of them moving for a given time slice.

 

Yes, of course Notice they've turned up again with electric cars?

But when we're between cars and computers anyway, the question is what you do with each clock-cycle. Let's.. take a practical and real example. So you have some.. array of some sort.

And you want to find the biggest split between two consecutive numbers. Something... I don't know.. to find out if input is varying over a set threshold, or to find the entry point for the most significant amplitude or something like that.

So you prepare two registers with data from the array, and have an index pointer primed with the byte-length so you can use INC or add on the register. And then run the array, subtract and put the sum in a new register if it's higher. Easy, and you only have one cmp, where you can get away with instructions that are guaranteed to collapse into fewer operations.

Except that you don't know how long the array is from high-level language, even though it definitely has a fixed length. And you actually have to fetch data from an indirect pointer before every "atomic" operation, which the assembler will try to collapse. With the result typically being that for each unknown next pointer that hasn't been loaded, the code-snippet has to wait before each arithmetic operation for IO. And you don't even know if it's going to be done regularly. And what could have been done in as many clock-cycles as the length of the array actually isn't defined for certain. All you really know is that the code-snippet is going to perhaps complete once in a while, and essentially therefore be useless.

If this is a mouse pointer where you're trying to smooth the input, you're very likely making things worse by making a pretty expensive routine to just filter out the odd read on the input buffer. Which in turn would lead you to choose to avoid routines like that altogether and just filter the visual output instead.

The same goes for literally anything that is supposed to do fairly simple maths on very short memory areas - you don't do it.

Because when you don't have fixed values and actually need to compute something and fetch pointers, you introduce uncertainty on the run. The compiler cannot predict the future, basically. This is also the problem with microcode optimisation - you can go quite far with it, but it cannot predict the future.

So what could you do instead? Say you had the opportunity for some device to read in the array of data every clock-cycle (it'd just dump the data there, for example ticking two-three new data into the array over time, but placing an entire data-array in a specified buffer). And then had a processor array where each processor could individually do complex math on the entire register? You want to have the biggest difference in the numbers, so you create a set of instructions that this processor is supposed to perform. Which then compiles, is collapsed, you can test it on the individual processing elements and ensure that it takes such and such time to complete - and now it depends on an existing array; It perhaps computes in one clock-cycle, and the result is even guaranteed to be up to date. And suddenly the requirement isn't to have a million clock-cycles just in case, but to just ensure that the processor completes a simple operation in one cycle, or by the time we need IO to happen. Suddenly you turn the entire thing upside down, and just want to figure out ways to transfer very large amounts of data to different memory areas simultaneously, so you can map that damned dna molecule before lunch, instead of until Christmas. Etc.

But to do this effectively we need a processor array of processors that can execute longer instruction words asynchronously on working memory. Rather than a central processing unit that can only serve immediately local memory in parallel.

And... in the meantime we keep believing that clock-speed is everything that matters. While it's not. Like engines with fifteen tiny cylinders that run at incredible speeds - it's not going to move the car any faster than a four-stroke, even though a bunch of stuff is happening in the engine really, really fast. That's the whole point here. The programs we create are directly affected by platform design in this way, where we exploit it's strengths and ignore literally everything else -- and it effectively prevents any actual improvements over time when processor speed pans out, as it already has.

 

if my Air 2 had a pointer, I'd never look back.
and please, don't try to sell me your pig of a Surface.

why all the chatter about batteries? the main thing wrong with the Wintel battery life is Windows.
I have been fighting for a week to find a way to TURN OFF a Yoga 2. and It is one of the BETTER current laptops on the market. horrible, cheap touchpad buttons virtually unusable [but Dell's are worse and they charge more! inverse equation], pretty good basic architecture but leaks. in the Wacky Windows 10 world, Power-off doesn't mean much - just another way to hibernate, slumber, leak.

Apple could decimate the PC market even more than it has, at will, but it is a lousy bottom-feeder business to be in. Wintel hath no happy partners - I doubt even Lenovo can turn a profit in the business.

There is a reason why schools are being taken over by Chromebook

 

It's called capitalism. Why would a company spend money producing a quality product? If they own the market there's little incentive.

 

@nipsen how long has AVX been available and how much is it used today.

• Each core has it's own memory which it can operate on, cache.
  
• Pipelines for simultaneous instructions.
• Out of order execution.
• Branch prediction.
  
• Interrupts for IO.

Clock speed isn't everything but certainly a part of it. Many softwares could be optimized to complete quicker but costs more in development so as long as the system it runs on is quick enough not many care.

 

But.. that's the thing, isn't it? If a random user had the choice between consistent and low response time, predictable battery and immediate IO - at the cost of maximum processor speed. Or higher synthetic performance in benchmarks, but with irregular response time everywhere and a fan that revs up every time you open a browser. Then you already know what a user with no computer knowledge will pick.

And sure, AVX and more SSE-like instruction set support will turn up on Intel now. Even if I expect that very few people will find a way to actually use it, just like with SSE2. Because the problem with using it is just that you only have access to the immediate processor cache if you want to do any semi-complex math. And all of the optimisations on the instruction level are based around common occurrences of similar operations, not with collapsing mathematical functions for you (or allowing you to program them without hilarious cost in run-time). I mean, that's the entire trick with x86 and intel. Total focus on one computing core surrounded by an instruction-level aware cache, so this one core can execute single and basic instructions very fast -- and then outside of that there's the cheapest ram and the cheapest engineering money can buy. And you can't get around that by making the instruction-level cache accessible by different computing engines "virtually" at the same time, etc. It's like having a super-computer with, say, 8 cores and 256Kb of RAM for your program. Useless. And only solution to actually improve on that then is to incorporate more cores inside the cpu-pipeline and increase the cache, to get more units possibly accessing the bus with that immediate access. But it's akin to, you know, making stockings that you put on over your head rather than from your toes and up..

Point is - x86 is obviously a very efficient way of making a general computing engine run relatively fast for fairly little money. It's been massively superior in every way that counts, though specially cost, since Pentium. But it's is at the end of the lane when it comes to getting anywhere from now on. It was a sideshow. A trick to get around production cost. A trick that lasted for 20 years, true, but it's not going to last another 20.

 

If apple could as you say "decimate" the pc market, it would have done so by now. I would rather have a surface anyday over a fisher price toy like the ipad.

 

The reason is simply cost and convenience. $250 for a chromebook vs $1000 for an Apple or comparable Wintel product. And everything managed in the cloud so there's no or limited server overhead costs.

 

I wouldn't say that is 100% true. My fiance's daughter's school has all Dell computers (laptops for the kids to use, Dell slim towers for teachers). Schools will buy what works for them. Chromebooks are REALLY restrictive and really only surf the net and can't do anything beyond that. Sure in some schools I'm certain they bought a ton of Chromebooks, and if that works for them great.

 

Hear Hear!!! + rep ... Never buying Apple garbage here...

Agreed, at least here in the UK, the chromebook isn't spreading at all in schools... It's just too restrictive for them and they're mostly using Windows laptops... Very crappy ones though..

 

the new acer cloudbook is the way to go for schools. cheap, reliable and w10.

 

Acer not rugged enough for school use. And cheap costs more money in the long run. Lenovo ThinkPad X131e is a better machine for that kind of application. More rugged.

 

Acer is PLENTY rugged enough for schools. My son's 5252 has fallen down a full flight of hard wood stairs, been dropped of his bed, desk etc numerous times, slept on, and god knows what else. The only thing that happened to it was two keys off the keypad popped off, and I lost the retaining clips. Therefore I could not get them back on. And, no....cheap does not cost more money in the long run when it comes to supplying schools etc with equipement.

 

Kojack,

Think lifecycle cost. Think support when a machine goes down. Not much unlike a business fleet. There is a saying that some bargains are not necessarily. School funding cycles in the USA don't always jibe with broken machines. Acer for your son appears good. But when it is more than one user and the "What the heck it is not mine" mentality. You see where I am coming from.

Acer is best known for consumer machines in the USA. I see many of them in for repairs (Low end Acer Aspires, Gateway, HP consumer, and Dell Inspiron laptops are the most frequently seen machines for repairs in my shop or any other local repair shop. Mostly I think due to being so common. Usually between out of warranty and 3 years old). Better Acer consumer models I see few of, and I have seen only two Acer TravelMates in as many years. (both for upgrades) Acer makes a lot of machines varying from very cheap and flimsy to sturdy TravelMates and all points in between. And their new gaming line is nice.

ChromeBooks are not in the USA suitable for the needs of most schools yet. Which is my biggest thing I'd see as their barrier to entry. And the ChromeOS is the big factor. A lot of educational software is either Macintosh or Windows only. And like businesses, school computers need extremely good management and support options.

P.S. Some of the older Acer Aspire models are favorites of mine for sturdy cheap machines. My main weaknesses I see supporting Acers is harder to get spare parts and accessories past two years old, and the lack of thorough hardware documentation. Those factors do need consideration. eBay does help with this issue often on the parts side.

 

I also used to service Acers as well. I'd say half of their product lines weren't too shabby, not great. But the uber cheap stuff (500 or less) was extremely flimsy, I could flex the top cover by pressing a finger down on it. Any cheap notebook is going to use flimsy plastic for the base/top cover.

And yes, Acer/Gateway/eMachines are infamous for being out of stock on parts after just 1-2 years. Ever since Acer bought eMachines and Gateway, they've shared parts over the same platform, but Acer support tends to lack after 2 years in getting support/parts. That is a major factor why schools don't use consumer brands, lack of parts and actual PC support. Hence most schools use the big 3 of business, Dell, HP, Lenovo and sometimes Apple.

 

...curious factoid, though - the bendy motherboard sandwich, the contacts socketed with tape and glue, the shoddily soldered slots, the screws mashed directly into the plastic frame for armoring, the lid with plastic clamps that wear out and creak a lot, the edge around the screen that peels off, the paint/coating on the wrist-rest that flakes away, the keys that short out the first time you spill something on it, the corroding psu connector and the disintegrating battery, the misaligned external ports, etc. These aren't exactly unique to Acer.

I think maybe the only thing Acer has been original about at some point was using a form-molded bendy plastic chassis, so you basically guarantee that if there's one bad contact somewhere, it's going to fall off sometime soon. And that it's just a question of time before the edges start stabbing you in the hand when you pick it up, or they get stuck in the bag, etc.

On the other hand, the cooling array on Acer has been comical on all the acer models I've seen up until the Haswell-reboot and the newish slim designs.

So yeah - agree that it's so cheap and popular has something to do with the reputation, probably more than the actual quality of the kits..

 

Well, like I said, I am going by my personal experience with using products from apple (god forbid), hp, asus and acer. Acer is by far the most reliable out of them all. My 2 acers right now are a 2007 and 2009 model. Both are looking like the day I bought them have been on numerous trips around north America and have never failed me.

 

Mm. And probably gaming folks use the thing differently. I'm not sure, but my guess is that if you're an enthusiast, instead of having a laptop as a work-computer or for surfing newspapers and writing on, or something like that, you're more likely to be active with feedback to support, and stuff on the intertron, and so on.

But some of the Acer chassis have been terrible. No denying that

 

Maybe, none that I own however. I mostly do photo editing. My son used his for video editing his xbox kills. ha ha. Never had a hiccup. works great. Gaming would probably not that great on my acers. I know the video is lacking for that. But that has nothing to do with how reliable the hardware is. I think the new acer gaming laptop will be a great choice for a reliable gaming system.

 

Interesting thing about an Acer.
Mine costed £600 when I got it on November 1st, 2008.
It's still running, and almost 7 years old (granted, I upgraded it from its baseline specs since then), and had to replace the screen (due to the backlight going out) and the fan (bearings going), but good thing for me was that the laptop has a backpanel which grants access to all internal components (HDD, RAM, CPU and dGPU), so cleaning it from dust and re-applying thermal paste was easy.

I detest laptops that need to be completely disassembled in order to get to the internal components (plus, most of those have inverted layout so you cannot just open it by removing the keyboard and all, you actually need to remove 2 or 3 layers of components before getting to the motherboard that's also upside down and doesn't allow easy access to the fan).

Cleaning such laptops is a nightmare, and using compressed air can be useful in such situations, but it is also extremely problematic because it often pushes the dust further into the laptop.

Laptops should have easy access to all components and maximized modularity in addition to similar layouts in case the motherboard goes out (why are laptop motherboards prohibitively expensive though?).

 

Want a laptop to last you years, be reliable and faithful, not to mention durable and easily serviced/upgraded skip the consumer grade models. Get business class. Major brands in this market are often capable of lasting 6-8 years of use easily. And their parts and service support is the best available. I'd rather have a 4-5 year old Dell Latitude E 5000 or higher, HP EliteBook or Lenovo ThinkPad X, T, or W line than any new Acer or Asus. I had two interesting laptops in, one an Acer Aspire from 2008, it had a bad screen, the other a IBM ThinkPad T 61. Which needed mainly more RAM, and a cracked bezel replaced (his daughter dropped it). I am still hunting for a screen for the Acer, and I have already found my bezel for the ThinkPad in spite of it being older by a year or two.

 

Yep, just like my two. Both were upgraded with 8gb ram, ssd, and I am doing the processor and new fan in my 7720 soon. I need a new fan so might as well drop in the faster processor as well. I have the Bluetooth module for my 7720 too but I have to remove the keyboard, so I am putting a black one back in. make it more "up to date" ah ah!.

 

I would have upgraded my GPU too, from 9600m GT to 240M which would give me roughly 50% boost in graphics capabilities.
The laptop accepted the 240M, but unfortunately, it kept freezing my system when it was sufficiently stressed (black screen and music would just halt with an endless repeating pattern).
It would appear that the 240M I got has faulty VRAM - or at least, it would appear that some other people experienced similar difficulties and explained it as such.
Had to put back the 9600M inside, so things are running fine now.

Sigh... oh well.

 

Both mine are video on board, so that upgrade is out. Too bad too, there is a 7720 build on here and it actually works awesome with the upgraded video.

 

Yeah I'd have to say Acer gets a bad rap. Most Acers I've used or owned have not been bad at all. They are definitely economical but not that bad when it comes to options for the price and are generally reasonably well made and designed.

 

http://economictimes.indiatimes.com...-job-losses-in-split/articleshow/48979011.cms

the root issue as to why notebooks are stuck permanently in archaic, lagging hardware, way-cheap-to-use components, and channel clutter with dead inventory: they can't turn a profit.
the brain drain fled to mobile/tablets/ way back, and won't return to the pc patch. they get 2nd string EEs
HP's "new" printer/PC unit is DOA. left for dead

 

That's been an issue for YEARS, there's zero money in hardware anymore. When everyone screams the cheapest crap, that's exactly what you get, crap. I think I read somewhere that most OEMs on computers 400 or less, they only turn like a 10-20 dollar profit margin per computer. So if you make the cheapest crap, you need to sell the most of it to make any kind of money.

 

Yep. And that's exactly how most PC manufacturers cut off the branch that they were sitting on...

 

Are you guys blaming capitalism?

Sent from my SM-T700 using Tapatalk

 

Nah, I think they're blaming short sighted 'leaders'.

 

It's called "consumerism"....

 

Couldn't agree more, I am really sad too as I love technology.

I enjoy to read tech articles and buy piece of the latest hardware to push my laptop to the limits and have fun with it.
Gaming, movies doesn't matter.

2 years has passed now for my modded P170SM, and I still cannot find anything interested in the market to push my self to buy a new machine.

Besides this, I am really disappointed with windows 10, considering to move to Linux Mint but scared for gaming performance issues...



Sent from my HTC One_M8 using Tapatalk

 

I don't know if it's really either of those Since they'd suggest that the buyer has a certain amount of choice and control over the goods they acquire, and so guide the market - or where the ones producing products do so at the request of a demand. Consumerism is a criticism in a sense of personal buying habits, where we buy things we don't need but can afford, at an unsustainable rate in the long run (though not in the short run, as it makes a company money).

So it's more that we're talking about a market where we entertain the idea that consumers need to buy whatever product is offered at any time, or else there would be no products to buy at all. Some sort of inverse monopsony I guess. It's kind of curious, was talking with some Microsoft rep. about xbox a few years ago. And I just said it as a joke, you know, that it's a curious marketing ploy, where you actually manage to get people to believe that if the company makes ridiculous amounts of money on a sub-par product, that this will automatically result in a better product being made later on (because some gaming enthusiasts actually believe that - that if they buy Star Wars Dance revolution and the Kinect, that we'd eventually get total immersion VR. Or even that getting total immersion VR is even dependent on suffering through Dance Dance Revolution Star Wars.. That the only way to get a better product is to throw money at Microsoft for products you don't really want in the short run).

And the guy looks at me with these really suspicious eyes, you know, like I had said something unforgivable and insulting. That I'd spilled out a well-kept masonic secret by accident in public. But that's really how the laptop market tends to function as well. We are essentially fed the idea that there's no money in hardware, and there's no money in creating specified solutions. And that what we should be paying for is the effort and talent it takes to shape the same hardware with unique software solutions. In the sense that even if the product isn't very interesting right now, we should support a company's temporary solutions in the hopes that they'd be successful and then later on produce something incredible.

Which in practice leads us to the point where the laptop makers are pressured to cut costs on production as much as possible on the same hardware platform, and where the baseline cost really is pressured as far down as it can be from their point of view. And where the budget on specifying and adapting software on top is essentially removed. Because what we're having is a ready-made intel solution in the back, and a Microsoft platform on top. With a very small window where laptop-makers can add new screens or fix the keyboard a bit, and add some custom buttons. And the specification on top is also very limited, and geared into - for example - ready-made platform-dependent solutions.

It's not that I don't understand the problems laptop makers get into, though. But I would sort of hope that I would hear some laptop maker guys start to talk more about for example utilizing efi bios layers for hardware specification that isn't platform or OS dependent. Things like that - having all the acpi functionality, clocking for hardware/ram/cpu being compatible with an OS layer, but not dependent on a specific OS. And it could be something as simple as moving the keyboard layouts with the extended keys (along with lighting control, device management) into efi-abstraction rather than having it being dependent on wmi, for example. In the sense that something like this would allow users to switch OS more readily, along with the laptop maker actually being able to offer different hardware solutions with the same platform specification.

You know -- it's simple and small steps like that that would make a difference, even in the small sphere the market is moving around in right now.