Yeah, yeah. Discrete is dead, and IGP is good enough now... as long as you don't want hardware-accelerated realtime raytracing:

(drool)

  http://www.siliconarts.co.kr/gpu-ip

(/drool)

When I can have Windows or Linux with full eye candy and zero performance hit (vs Win2k or something XFCE-like) at 2560x1920@240fps, I'll accept current GPUs as "good enough".

When I can open a pdf document on my phone or tablet and effortlessly fling through it without any perceptible lag waiting for the fonts to render, current GPUs will be "good enough". Newsflash: 80% of the reason why ebooks suck so miserably is the fact that current hardware CAN'T effortlessly render them in realtime

OK, the last one is a bit unfair, because the blame for current shit pdf-rendering performance lies mostly at the feet of the videocard industry, for throwing away everything it learned and developed relating to 2D acceleration in the mad rush to cheap 3D. Basically, they took 3D GPUs developed for strap-on videocards, grafted on enough extra silicon to let them stand alone as their own self-hosted minimalist frame buffer, and called it a day.

^^^ Argh. Proofreading typo-blindness. "~25 years ago, I got a Vic-20 on Christmas Eve. By dinner on Christmas Day, ..."

It's both. Comparing OpenGL (ES) to sprite-based and tile-based 2D is kind of like comparing J2EE in all its distributed splendor to PHP.

OpenGL ES 2 was a pain, but dear Jesus God, I spent the better part of a day just TYPING IN the HelloWorld code for an OpenGL ES 3 Android app, and ended up with something like 8 or 10 classes that compiled into a .apk file several hundred kilobytes in size just to draw a yellow triangle on a black screen. Now, admittedly, the increased HelloWorld complexity eventually pays off by making it more straightforward to do COMPLICATED things, but GETTING to that point has absolutely become more painful over the years.

~25 years ago, I got a Vic-20 on Christmas Eve. By dinner on Christmas Eve, I was writing programs with custom characters, animation, and music. Today, you'd spend 2-3 days with a new computer just waiting for Windows Update to finish installing one or two service packs and several hundred individual updates. Some people might grouse about "cryptic code", but I dare anyone to compare the amount of code you need to open a native Window and make its background black under .net or Java to:

POKE 36879, 8 (*)

Butart is a big, huge problem too. Back when graphics were made from 8x8 characters that could either be 2 colors, or 4 colors at half-resolution, there were only *so* many ways to meaningfully make something, and most of them were dictated by a need to have at least 1 pixel separating major features (like eyes) so they'd be recognizable as such. The lack of resolution and subtle colors basically solved the problem for you.

If anything, having an artist involved 25 years ago usually made the gameplay worse, because they'd force the programmer to make horrible performance compromises to implement their artistic vision. Instead of being able to play tricks with barrel-shifting and video timing, the programmer would be stuck shoveling raw bits around the slowest way possible because the artist designed an image whose width wasn't a whole multiple of 8 pixels.

(*) of course, understanding WHY the value was 8, and not 0, as opposed to just blindly copying the value out of a book, required a few semesters in college ;-)

Actually, in at least half the cases, yes... we should be able to. The problem with this whole goddamn industry is that everyone in it is determined to pretend that drives don't fail, rather than designing them to fail gracefully and in ways from which data can be recovered by means besides black magic.

In the case of rotating drives, something as straightforward as storing the drive's firmware on real flash, instead of on one of the platters, would probably eliminate the root cause of at least 20-30% of unrecoverable (without thousands of dollars) drive failures. Modern hard drives have literally dozens of failure points, any one of which is enough to completely torpedo it beyond consumer recovery. Expecting a mountable filesystem is a bit much, but expecting to be able to rip raw bits from the drive for offline recovery is, IMHO, 100% reasonable.

Regardless, SSDs need it more badly than rotating disks, because rotating disks don't totally crap and lose everything out if they lose power at the wrong moment. And SSDs have very, very few things that can mechanically fail in ways that someone with ham radio or arduino experience couldn't fix, given adequate documenetation and community-developed software to do it with. In Sandforce's case, it's like they decided that data-recovery companies have a civil right to extract extortionate amounts of money for data recovery in perpetuity, even if they could -- with minimal effort -- make recovery by motivated and technically-adept end users borderline trivial.

> Isn't this why god created UPS?

When my UPS battery starts going bad, the first sign is that it just cuts the power without warning. If you have a SSD, that could be the deathblow that sends your data bye-bye.

The bigger question, though, is WHY THE FUCK can't we either disable whole-drive encryption, or at least set it to a key WE control, with some means to read the bits from even a drive that's totally nonfunctional SATA-wise (JTAG, SPI, whatever) and reconstruct it offline? That's why I despise Sandforce so much. As if it's not bad ENOUGH that Sandforce-based drives can just die from a single corrupted write, they have to go a step further and make it impossible for end users to do any kind of meaningful data recovery. There's NO REASON why a corrupted SSD should require thousands of dollars of commercial data recovery. If they'd just give us some fucking way to rip the raw bits from the drive, document the data structures, and give us control over the encryption, a fucked up SSD would just be an annoyance.

I'm still trying to process the fact that there are new SSDs that DON'T have ultracaps. You'd think that what happened to OCZ might have taught the industry a lesson or something. Well, besides, "There's a sucker born every minute!"

No, but it would dearly love to. If Apple could use "I desperately want _(some-non-negotiable-feature)_" as a way to force Android users to grudgingly surrender themselves into the soul-crushing (if tastefully-appointed) captivity of Apple's walled garden, it would do it in a heartbeat. Apple and Microsoft would dearly love to close "the Android Hole" that empowers users to run whatever they fucking feel like running, instead of limiting their software to apps that are neatly aligned with the priorities and agendas of the manufacturer and mobile carrier.

> LEDs are cheaper

Unless you're talking about outdoor light fixtures in Florida... where the atmosphere is maybe two or three steps less-corrosive than the atmosphere of Venus insofar as light bulbs with active electronics are concerned. LED and CFL bulbs in a porch light have a lifespan measured in MONTHS here.

It's not even the endless rain per se... it's the dew that condenses inside the bulb just about every night/morning. People who've never lived in South Florida just don't "get" how quickly and completely stuff here gets destroyed when it's left directly exposed to outdoor air, even in a sheltered location that doesn't get directly exposed to actual rain.

Going off slightly onto a diagonal tangent, but relevant due to Christmas shopping and the annual agony of trying to pick a new router before giving up in disgust... is there actually such a thing as a high-end router that DOESN'T seem to have page after page of 4- and 5-star reviews, sprinkled with 5-10% of 1-star reviews, and a pattern something like...

***** Awesome! Kicks ass! The greatest router I've ever had! Problem free, works flawlessly, and perfect in every way. {technical details}.

* Total garbage. Pure shit. 5GHz connections dropped after a day, and the router had to be rebooted to fix it.

***** (another overwhelmingly-positive review)

* Worked like a champ for {3-9 months}, then crapped out and left me in misery until I finally gave up and bought a new one.

***** (another positive review claiming it's a gift from ${deity})

* Junk. 2.5GHz works for 3 hours, then the router forgets how to route traffic between wireless and wired. 5GHz doesn't work reliably with Apple devices, and works reliably with Android devices only if you remove the network, reboot the phone, then add it as a new network.

* Terrible range and speed. I connected to this AP with my laptop from 5 feet away, and got barely 1mbps on an unused 5GHz channel. I disconnected it, reconnected my old $49 access point, and benchmarked 36mbps. WTF?!? ... and so on. Case in point: just about every dual-band 3-antenna router from ASUS, Netgear, Linksys, and Buffalo router that costs more than $150.

As far as I can tell, it basically comes down to:

* None of them have adequate heat-removal, especially if they're in a closet or cabinet of any kind. The electrolytic capacitor plague continues unabated 15 years later.

* Poor antenna impedance-matching, so RF gets reflected back into the radio module and progressively damages it.

* RF modules have real limits that nobody ever talks about, and certain permutations of features that just can't work, but because nobody from the manufacturer will ever come out and identify what those precise constraints are, end users are left to randomly flail about and wonder why certain things just don't work.

* Crap component quality pushed to the absolute limit of its design capabilities, then pushed 5% further, and guaranteed to fail eventually.

* Zero quality control besides "could we power it up enough to flash it"?

The problem with using a FPGA is that THEN you're buying a chip that costs more than Intel's second- or third-most expensive i7, and getting a CPU with the approximate performance of a 500MHz Pentium III.

More importantly, even if you DO build your own CPU using a FPGA, at least 95% of your VHDL is going to come from somebody else if you want to have it meaningfully working, with Ethernet and USB, before you die someday. If somebody is so paranoid about security that he doesn't think he can trust a COTS CPU from someone like Intel, what makes him think that ${government-espionage-agency} doesn't have the resources to plant exploits in the VHDL components he'd download and add?

And before someone brings up China... frankly, if my hardware is going to be pwn3d by ANY government espionage agency, I'd PREFER to have it be pwn3d by China's instead of the NSA (or some other American agency, or the agency of some obedient American vassal state). At least China doesn't have the legal authority to deprive me of my life and liberty based upon data mining for technical violations of some obscure law.

Well... in the case of Cyanogenmod, it's more like, "Google kind of turns a blind eye to end users installing them, because it knows that 99.9% of the people downloading them are installing them on phones that shipped with GApps to begin with." It's kind of like how if you ask a Microsoft Licensing Specialist about using an OEM copy of Windows to install a virgin copy of the same version of Windows on a computer that shipped with a crapware-laden copy from the manufacturer, they'll tell you it's officially forbidden... but if you get your hands on an install disc somehow, call Microsoft to phone activate it, and give them the number printed on the COA on the bottom of the computer, they'll activate it anyway.

Google doesn't give a shit if someone with a Samsung phone installs GApps after reflashing it to Cyanogen. They care ENORMOUSLY if a carrier somewhere in the world sells phones from Shenzhen with unlicensed copies of GApps. As long as the barrier to doing it is high enough to require a fair amount of technical skill to install GApps on never-licensed hardware (as opposed to originally-licensed hardware that was just reflashed), they don't really care. And more importantly, they know that if they tried TOO hard to stop people with the skills to reflash from doing it, it would cost them several orders of magnitude more sales and lost goodwill, because we're the ones that two dozen people ask for advice when it's time to buy a new phone.

If you don't believe me, find a friend or family member who owned a Motorola Android phone that got its bootloader permalocked (Photon, Electrify, Atrix2, others) and ask them what they think about Motorola's current phone.... Make sure you're wearing asbestos clothing, because you'll need it...

I hate to break it to you, but frequency bands are the LEAST of our problems. Google "LTE Lock-in" and read the tales of misery about how even two nominally-GSM networks (AT&T and T-Mobile) have managed to make themselves almost as proprietary & hardware-locked as Sprint & Verizon, unless you're willing to live without LTE. As of this moment, there's ONE (maybe two... not sure about the Nexus 5) phone(s) known to be capable of doing LTE on both AT&T and T-Mobile (the HTC One... IF you buy the special Google Edition; the carrier-branded variants are LTE-locked at the radio modem level EVEN IF you remove the SIM lock).

Leave it to American mobile phone networks to find a way to pwn even GSM.God forbid, if Sprint ends up being allowed to buy T-Mobile, they'll probably program their new tower hardware so it refuses to talk to phones with SprinT-Mobile SIM card unless the ESN is in their holy database of carrier-branded phones.

Apparently, other carriers learned nothing from the total well-deserved hate Sprint took over the Photon Q's "sealed SIM" that locked you into their rip-off roaming rates when outside the US on GSM networks.

In the end, the Photon Q was one of the worst-selling phones in Sprint history. The sealed SIM was such a staggeringly huge anti-feature, even people who didn't CARE about international travel wouldn't even give it a second glance once they found out about it.

> This led me to wonder what a feminist programming language would look like,
> one that might allow you to create entanglements

Ah... in other words... a language based upon dependency-injection for non-deterministic multithreaded runtime environments with planned monthly maintenance cycles. It's mostly interrupt based and requires extensive exception-handling. :-D

If you want to see some real fun, find a way to post a sequence of example pics that supposedly show how people with anomalous trichromatic color vision see the world, then pull out the bowl of popcorn when actual deuteranomalous and protanomalous individuals scream, "WTF, these examples are just WRONG... but THIS is an example that works and is, to me, indistinguishable from the control picture" and the editors defend keeping the wrong pics as examples because the edits and new example pics made by actual individuals with anomalous color vision don't represent a "Neutral Point of View" (as if there can possibly BE a neutral point of view over something that is by definition about subjective sensory perception).