Slashdot Mirror
Life's Too Short For Slow Computers (theverge.com)
Nilay Patel, the Editor-in-Chief of The Verge looks back the Apple Watch, the company's first wearable device which went on sale roughly a year ago. In the article, Patel notes that Apple Watch, a computing product, is just too slow at doing some of the most basic things such as running apps. He writes: Here's the problem with the Apple Watch: it's slow. It was slow when it was first announced, it was slow when it came out, and it stayed slow when Watch OS 2.0 arrived. When I reviewed it last year, the slowness was so immediately annoying that I got on the phone with Apple to double check their performance expectations before making "it's kind of slow" the opening of the review. [...] The grand ambition of the Apple Watch is to be a full-fledged computer on your wrist, and right now it's a very slow computer. If Apple believes the watch is indeed destined to become that computer, it needs to radically increase the raw power of the Watch's processor, while maintaining its just-almost-acceptable battery life. And it needs to do that while all of the other computers around us keep getting faster themselves.
is not designed to be a full fledged computer, treating it as such is stupid.
Though I will freely admit that many third party apps suck (and are often trying to solve problems that are not suited to the watch).
"In America, first you get the sugar, then you get the power, then you get the women..." -H. Simpson
As long as the smartwatch needs to be tethered to your phone, it shouldn't have to be a computing powerhouse. Apple's Watch has been cart-before-horse from the start, and the battery tech just doesn't exist yet to make it a product that doesn't suck. It'll be years before the Watch, as Apple has it envisioned, is decent enough to not be a pain in the ass.
Pebble got this one right, and Apple should have taken their cues from them. E-ink displays are where it's at now if you need to maximize battery life, and their latest color e-ink displays are actually quite pretty. They're not chocked head-to-toe with features like the Watch, but back in the day Apple used to be about user experience first, features second. Amazing how Cook managed to derail all that in such a short time.
Its just a non-essential toy anyway. It does absolutely nothing that isn't done on a phone far better. The only reason to buy it is for the oneupmanship that fanboys love to play.
The whole point of having a watch is to tell the time, and if the watch is slow, then it isn't fit for the purpose for which it is made, is it? :)
File under 'M' for 'Manic ranting'
life, for one.
I'll settle for a very basic computer (clock, simple games, light source, etc.) plus a simple interface to selected iPhone apps, such as getting notifications, sending canned responses to texts, etc. I don't expect much in the way of zorch on the actual watch.
Knowledge is how to play a game, intelligence is how to win, wisdom is knowing what game to play.
Sorry, but the apple watch has no need to run slowly. What's slow is shitty code written by shitty programmers who are tied to their shitty IDEs and have no idea what a processor is or does. Spend 2 hours and a case of beer reading the processor spec sheet at a drunken level of detail and your code can magically become ten times as fast, just by not doing stupid shit. (I assume, of course, that you understand how a compiler works at some high level.)
No, we are surrounded by powerful, capable computers, and we run them into the ground interpreting (or JIT-compiling) Javashit frameworks to interpret/JIT-compile Javashit code, and then use that to manipulate the DOM on the fly, all to produce a little fade-in/out effect to make up for the delay while other Javashit contacts an ad auction amongst another few dozen offsite machines bidding for the right to serve yet more Javashit, when all we wanted to do was read static text on a http://motherfuckingwebsite.com/
We do it because programmer time is more expensive than user time, but the net effect is that we now employ sledgehammers to swat flies because flyswatters aren't clever enough.
let me note that plenty of much older and slower hardware managed to run just fine, even snappy. Thus, hardware as such cannot be the problem. It's software that demands more from the hardware than the hardware can deliver. And that, as they say, is fixable in software.
I think life is way too short to waste it immersed in the Internet of Things. There is so much more to do in life than spend it following an obsession with technology for technology's sake. I don't want computers dangling from my body, following me to the bathroom (that's what my dogs are for), monitoring my every breath, and of course, reporting every detail of my life to a bunch of marketers.
I want to go outside, ride my bike, hike with my dogs, and enjoy time with my family, all without being constantly bothered, interrupted, and monitored by some device.
Now, get off my lawn, you meddling kids.
Maybe your data and personal area network can run from your wrist (or arguably something a bit bigger, with more battery), being a watch gives it a function (yeah a crappy function, but a function). But putting the computing power in a tiny wearable just isn't the future. .. in the future your data/PAN will most likely be an implanted device that runs off ambient power ...
It doesn't need to be hardware accelerated. Just programmed directly.
I should use this sig to advertise my book ISBN-13 : 978-1501515132.
Processors today are orders of magnitude faster and more capable than just a few years ago. There shouldn't be a question that our apps run faster on them.
The problem is we are loading them down with extraneous cruft. Remove the bloat and you remove the problem. Throwing hardware at it may solve some of the problem - but that is just a bandaid, and definitely won't allow you to lead the market if your competitor is producing leaner, faster code.
Lodragan Draoidh
The more you explain it, the more I don't understand it. - Mark Twain
It's not so much that the Apple Watch is just slow, it's that it's slower than a mechanical watch that has a centuries-old design.
If you've got to wait for your wristwatch to tell you the time, you've got an irritating product.
You are welcome on my lawn.
I came here to say this. Windows 3.11 can now run entirely in the cache of my laptop's mid-range processor. Applications that could be envisioned for that level of power include just about everything you would want to do with a watch. You're not short on computronium, you're too cheap to program for the environment.
If video games influenced behavior the Pac Man generation would be eating pills and running away from their problems.
No, really. Do you type your essays on a phone? You could, but the device is not made for the task and you are going to have a frustrating experience. Watches are for scrolling through a few notifications, not interactive apps. What is needed is a fully water/dust proof device with WiFi access, payment support and reliable weekend battery life. Maybe also ability to unlock my house door. Apple watch is not it yet, and neither are other smartwatches. But if such a thing existed, it would finally make it practical to leave cellphone, wallet and keys at home for a short while.
Sorry, I tried to make this the first post but my raspberry pi too so long to load the page I had to switch over to my $39 kindle and it took me a while to tap in these words.
Some drink at the fountain of knowledge. Others just gargle.
Releasing the watch when it was still too unpolished, or fast enough hardware just not available, to avoid feeling 'slow' seems like a particularly strange and foolish move for Apple given how much positive press and user satisfaction they enjoyed from the fact that iOS enjoyed the perception of being much snappier and more responsive than Android(less true now, thanks to a combination of Google's 'project butter' and other improvements, plus sheer brute force on the hardware side: but definitely true in the bad old days and on bottom-feeder handsets). Apple, of all consumer electronics outfits, seem like they should most understand that "if it doesn't feel fast, it's too slow; if it does feel fast, spec-sheet preening is pointless". This is how they've always sold their mobile devices; and largely how they've approached specs for all but workstation computer products.
I hate to break it to everyone: but computers aren't getting faster in general like they used to. The train is reaching the end of the track. You will see small bumps in performance year-over-year but don't expect the computers next year to be much faster than the ones this year. Digital processor technology is reaching its limits.
While that's kinda important you also need to have active communication to be useful, which puts a rather significant drain on your power budget. And every time the user raises his arm so you think he might be looking at the watch and not just reaching for something on a shelf, the screen has to turn on. On the kind of battery you can fit in a watch, that's a pretty big deal. Ever put your cell phone in flight mode and not use it to game or listen to music? It'll barely sip power because it's not doing anything. On the other hand, it's not very useful either.
Live today, because you never know what tomorrow brings
Processors today are orders of magnitude faster and more capable than just a few years ago. Remove the bloat and you remove the problem.
Not quite. Processors today are mostly more power efficient, performance gains that we are got used to during 80s and 90s largely stalled quite a few years back. For example, 6th generation I5 (Skylake, 2015) is only marginally (~20%) faster than 2nd generation I5 (Sandy Bridge, 2011). At the same time, historical bloat growth rate remained constant and overtook meager computational gains.
I'm torn here.
There's no need for desktop-performance on a watch.
But equally, there's no need for a watch to "feel" sluggish at all.
Apple are even skilled at such tricks. Load up a modern Apple Mac OS X image in a VMWare machine. Dial down the specifications of the machine to the bare minimum. Make sure you DON'T have graphical acceleration etc. on.
Now slide over the bottom row of icons that you get on Mac OS. It will look and slide as smooth as silk. How do they do that? They pre-rasterise the icons in a variety of sizes and keep them loaded in RAM so you're basically seeing a flickbook of all the different sizes depending on which icon your mouse is over at that moment. No, they ARE NOT using "scalable vector icons" - it's pre-rendered from an those vectors into multiple bitmaps instead ahead of time.
It's simple, beautiful, fast. But it's also a con. It's not ACTUALLY resizing those icons or blitting them to the screen via an OpenGL filter or similar in real-time. It's just been optimised to its precise usage... to look "slippy, slidey, silky, bulgy" for the first time you log onto the machine and the thing you'll use to start all your programs. But you have to say that it "feels" nice even on a machine incapable of running anything else at a comfortable speed.
Apple could do similar. There's no reason the watch can't be high-res and super-responsive and just a fraction of a second latent to the message coming in or whatever. But they haven't done that. They haven't spent time optimising it to its intended usage. They've rushed it to market. There's no need for it to be able to render 4K video at 120fps or whatever. It just needs to show a simple interface fast, something a Z80 could do in the same position if you were to really want it to, just tie it to a bluetooth chip to talk to the phone and make the phone do the heavy lifting and the watch just display what it needs to.
That Apple, master of such tricks, hasn't done this means they aren't really interested in spending time on it, I think. It also means that competitors can have easy-wins. Nobody cares that the Apple Watch is 100 MHz and the Competitor Watch is only 99MHz if the competitors just feels so much sleeker to use.
Now, personally, I hate Apple and have never owned a single product of theirs in my life. But some things, especially where appearance matters over substance, they utilise clever tricks to good effect. With the Apple Watch, it just ... seems sloppy. Like they don't expect it to be successful, or like they expect it to be successful no matter what it actually does (the famous "It's expensive, it must be good" factor).
But just because the chip may not be the latest and greatest? That's no excuse for a bad user experience. The CPU and RAM specifications in my client machines haven't changed in years but going from 7 to 8 to 10 actually makes the same machine "feel" faster, even if statistically it may not be.
That Apple can't avoid this tells me they had no idea what to do with it when it was being designed, or just don't care because they don't plan on any more of them.
Processors today are orders of magnitude faster and more capable than just a few years ago. There shouldn't be a question that our apps run faster on them.
The problem is we are loading them down with extraneous cruft. Remove the bloat and you remove the problem. Throwing hardware at it may solve some of the problem - but that is just a bandaid, and definitely won't allow you to lead the market if your competitor is producing leaner, faster code.
What you gain by the layers of complex software is mostly flexibility. Older computers had vastly slower processors that could run application at speeds similar (or even slower) to those that we use today, but:
- The software came distributed through magnetic media, or had to be run on a centralized mainframe accessed through a dedicated network.
- You could only run one program at a time on a workstation.
- Interpreted languages were awfully slow - you wouldn't run full-fledged applications on top of them.
- Forget about multi-platform; most software was tied to a single environment, and porting applications to different platforms required wizardry-like knowledge of several operating systems and architectures.
- Of course, such a thing as a "hypervisor" allowing several OSs to run on the same machine at the same time was unthinkable.
Sure, there could be a lot of streamlining and cutting corners to gain speed - on problems that you understand thoroughly; for those problems, such streamlining it is still possible and frequently done, as long as you don't need the flexibility because the problem won't change.
What that "bloatware" gains you is the possibility to work efficiently and explore new workflows on problems that you barely understand, or not at all, without having first to assemble dedicated of engineers to build the software to solve the problem.
Singularity: a belief in the "God" idea with the "demiurge" relation inverted.
Ya, I want a toaster strapped to my wrist..
It's because smart devices have to look swish too, so they need to piss away several megabytes on some fancy graphics and throw in a ridiculously powerful GPU so that it moves at 60 FPS. Of course that means it takes 5 seconds to actually load the alarm setting app, but once it's done man does it move satisfying.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
A stat I like to tell people: in the 90s, in the same length of time - 4.5 years - that it took for today's 20% improvement, we went from P54C to Athlon and Coppermine - roughly an 800% improvement.
Yes, we do. And the fact is that we like it that way. We could all be using command-line tools on CP/M to launch our single-execution sovereign application; but most of us don't, because we enjoy the benefits of a flexible system - where you can distribute software through the Internet, run many applications at the same time in a windowing environment, or port software between different architectures.
Singularity: a belief in the "God" idea with the "demiurge" relation inverted.
That's an amazing observation, because I always use my smart phone not to game! And rarely use it for music. I usually use it as a phone and communications device. It gets OK battery life. I have to recharge it every night, which is inconvenient. I miss the decades old flip phone that would last for a week, and had communication stuff built in as well. No 3rd party apps!
Embedded systems have worked well with far more restrictive environments. Looking at its specs, it does not seem impossible task to have responsive system with what they have.
And how long did it take to build a new application?
How fast did your dynamic programming language with generics run?
And could your users buy commercial software downloaded through an online store to a portable device?
And how well did your grandma use it?
Are you sure you had the same?
Singularity: a belief in the "God" idea with the "demiurge" relation inverted.
If you turn off mobile data you'll likely see a week or so of battery life. If you like to rely on always on Internet connectivity you're not making a fair comparison with the flip phones of yesteryear.
We're limited by DRAM bandwidth and have been for _decades_.
Funny how the performance increase is in line with DRAM performance changes.
Car analogy: it doesn't matter what kind of engine you have, if you're sucking through an asthmatic intake.
John McAfee 'It was like that time I hired that Bangkok prostitute; to do my taxes, while I fucked my accountant'
Life is too short for our currently blazing-fast computers to be bogged down by shitty bloated code.
Fix that problem, and many of your other problems will magically vanish.
Still waiting on Serviscope_minor to wake up to fucking reality and realize that Jessica Price isn't going to fuck him.
Apple watch? Bah. Why PS3 is so slow? It's slow to start, it's slow to load its dashboard, it's slow to load apps. Has always been, and it's getting more annoying each time I'm waiting for it. I mean, why does it take a good minute or two to load Netflix? What is it doing while booting up? Probing the hardware? It should already know everything about the hardware since the console was produced (it's the same for every console, and there is a handful of SKUs). Even if something was replaced, it should know the configuration from the last time it started. Why does it take so long to load the Netflix app?! What is it doing, exactly? When I click netflix.com on my notebook, everything comes up in a fraction of a second. And I'm pretty sure PS3 is way more powerful than my notebook on the hardware side. So, in short, it's shitty software. Including Apple watch. If they can't fix iTunes for so many years, there is really no hope.
But there's a huge gap between the amount of bloat and the amount of actual increase in functionality. Look at your example of window management, the only real improvements over the past 20 years have been composition, better multi-monitor support, and virtual desktops (does Windows even include that yet?) Many websites I use do the same exact thing they did 5+ years ago, but are a hell of a lot slower now.
Isn't it kind of stupid for you to opine on something you've never used?
The Apple Watch is in fact, fast in all of the ways you laid out. Animations are very fast and fluid. Moving between various screens is quite fast.
The singular thing that is very slow, is loading MOST (but not all) apps, because whatever the initial screen in requires watch to phone communication, and that takes just a bit too long right now.
What that means in practice is that for most of the built in features and apps the Watch works quite well and is not slow.
As a developer I also think it's possible for apps to not be AS SLOW as they are, but I don't know many companies have really taken the time to optimize as well as they could have.
"There is more worth loving than we have strength to love." - Brian Jay Stanley
We do it because programmer time is more expensive than user time, but the net effect is that we now employ sledgehammers to swat flies because flyswatters aren't clever enough.
Sometimes sledgehammers are the thing you need to tackle complex problems. I suspect the problem is that we don't actually use big-enough sledgehammers yet (for example, compilers and language environments are still too much stupid to solve complex automated programming problems on their own).
No, we are surrounded by powerful, capable computers, and we run them into the ground interpreting (or JIT-compiling) Javashit frameworks to interpret/JIT-compile Javashit code, and then use that to manipulate the DOM on the fly,
Nothing wrong with the concept per se, VPRI would approve. It's just that the implementation is horrible.
Ezekiel 23:20