This sounds like the action of a Congressman trying to discredit the NSA. The NSA obviously is not going to respond to this - if they did, they'd be inundated with requests from every small-town prosecutor wanting some more evidence (ironically, some might even get warrants for it). That would be worse than what will happen instead, which is that an anti-NSA legislator gets a talking point about how the NSA isn't using its data and isn't cooperating with the rest of the government (namely Congress).

Yes, it's just a political point being scored. But it's a point hopefully in our favor - or at the very least, one against our common enemy.

The more I think about it, the more I think this is the best way to get the NSA shut down. The general public has no control over it; trying to get them angry about it is pointless. The only way the general public could shut it down is by a revolution, and we're too well-fed and content to do that. But Congress could shut it down, so let's find every way to get Congressmen upset about the NSA. I wonder what a FOIA request for some congressional metadata would do...

You're arguing with the antecedent. I'm saying "if you care about X, the Titan is good", and you're accusing me of cherry-picking because the Titan is bad at Y and Z, even though I specifically called it out as not being good for anything except X in a performance-per-penny measure.

I am saying that one of the principal reasons to buy a Titan is if you have a heavy double-precision compute load. I then provided a benchmark showing that a Titan beats the 295X2 in such a load. It would be cherry-picking if I picked the one double-precision benchmark that showed the Titan in a good light, but a single-precision benchmark does not invalidate that.

If you are accusing me of cherry-picking, please provide a benchmark that shows a 290X beating a Titan in a double-precision workload. AFAIK the only double-precision benchmark Anandtech uses is the F@H benchmark I linked to originally.

I am not at all arguing that the results in the double-precision benchmark somehow invalidates the single-precision or integer results. If your workload isn't mostly double-precision, the Titan is not for you. But if your workload *is* mostly double-precision, the Titan is a viable card.

Don't worry, you can just download an e-goat. There's a good website for it, but they keep having to change the TLD for some reason and I can't keep track of which one it is.

Workstations, perhaps? There's a lot of scientific computing done using desktop-sized workstations, not supercomputers. And they're spending several grand on Xeon CPUs anyways, so a $3K GPU isn't that much more.

It's not cherry-picking if benchmarks like that are the primary reason to use a Titan. Particularly when I explicitly said so.

The Titan shouldn't be considered a top-end gaming card. It should be treated as a budget Tesla card - even at $3k, it's the cheapest card in Nvidia's lineup with full double-precision floating point performance (which no game uses, but is common for scientific computing, Tesla's market). And on tests using that, the single-gpu Titan and Titan Black outperform the 295X2 by a large amount. AT hasn't gotten to test a Titan Z yet, but you can tell it's going to wipe the floor with the 295X2.

Yes, Nvidia advertised the original Titan as a super-gaming card, and to be fair it was their top-performing gaming card for a while. But once the 780 Ti came out, that was over, and since everyone expects a 790 dual-GPU gaming card to be announced soon, buying any Titan for gaming is a fool's choice.

Nvidia seems to still be advertising it as a top-end gaming card, presumably trying to prove the old adage about fools and their money. It just comes off as a scam to me, but anyone willing to spend over a grand without doing some proper research probably deserves to be ripped off.

You forgot the USB port for keeping your phone charged.

When you've been getting your ass kicked for years, "not losing" can count as a major victory.

I think it's more "if you compile your game without Kinect, you will have access to that processing slice and Kinect won't". Whether the hardware is physically there or not is irrelevant to the reserved processor time.

I'd be keeping OS X on there - I would be switching from dual-boot OSX/W8 to OSX/Linux. I guess I could triple-boot, but that seems a bit excessive when I'd have a dedicated Windows machine as well.

I bought a 6870 as an upgrade to my Mac Pro, mainly because it was highly compatible with OS X (it only fails to show the grey apple screen during boot) and is far cheaper than officially-supported cards. It's also a good mid-tier card on Windows.

And according to this, the 6870 is also basically the best card for use under Linux using open-source drivers, so I guess it's just a very good card in general. When I do a new from-scratch build, I might put Linux on the old Mac so I can play around with Linux gaming more.

Also announced were an i5 and a Pentium-branded Devil's Canyon processors. All three have the same TIM upgrade and overclocking focus. The i5-4690K is similar to the i7-4690K, dropping hyperthreading, a bit of cache and some stock clock, but for $100 cheaper ($242 instead of $339, if reports are accurate).

The really interesting one is the Pentium G3258. Two cores, no hyperthreading, but with an unlocked multiplier, for $72. If you care more about single-threaded performance than multi-threaded, this might be a very cool thing. Buy one, and a good aftermarket cooler, and overclock it into the 4GHz range. If your load is mainly single-threaded, like far too many games are, that can give you the same performance but be much, much cheaper.

I have a laptop with both a Crucial M4 and a regular hard drive. It's been going strong for two years now (well, the display is dying but that's not really relevant). Going off history, I expect the hard drive to die first, but I admit that's a completely unscientific prediction.

You were right to be cautious, though - back in the early days of SSDs, there were many that were absolute crap (OCZ drives had horrible failure rates, and JMicron controllers were rubbish performance-wise). Intel was really the only one worth buying from. Nowadays there are plenty of good companies to buy from. Intel and Samsung are probably the best, but Crucial is up there in the lists.

The Space Shuttle is a really weird mix of qualities. The boosters are actually very good at their job - they're extremely powerful, and surprisingly reusable. The main engines are also good - they're some of the most efficient engines to be flown, period, and they're the most efficient that ever flew regularly. Using an external tank also is a good move - it's much cheaper, and it means the only thing getting thrown away is an empty tank. On paper, the Shuttle should have been an amazing craft.

The biggest problems with the Space Shuttle are deeper.

The first problem is the choice of fuel. Liquid hydrogen is amazingly efficient, but it's both bulky (look at the external tank) and expensive. I suspect NASA thought that, by flying dozens of shuttle missions per year, they could build up a large LH industry in the US, the same way UDMH and other fuels went from chemical curiosities to made-by-the-ton commodities. That didn't happen, possibly because the Shuttle never flew as often as it was designed to. But a more conventional fuel would have been both cheaper to use, and would have allowed for a smaller vehicle.

The second problem is the airframe. The basic idea of the Shuttle is a good one ONLY if you regularly need to recapture satellites and deorbit them intact. This basically never happened. Without that, the Shuttle is a massive, heavy airframe with no purpose. This is getting fixed with SLS/Orion, which is basically a Space Shuttle with a capsule instead of pseudo-spaceplane. Well, assuming NASA actually makes it. Considering how simple the design is, I don't know why it isn't flying already, except for politics.

The third problem is the politics. To get Congressional support, parts for the Shuttle were made all over the country. That's inefficiency for the sake of inefficiency. Then, once Challenger happened, bureaucrats went through everything and OSHA-fied it. Things that were designed to be reused a few times were made disposable, or were rebuilt after every flight. Training times went through the roof. That made the program as a whole slower and less effective - so Congress started slashing funding, because who wants to fund such an ineffective program?

That third problem is honestly the biggest one. If they had been flying them according to the original plan, and using all the capabilities of the Shuttle, it would have been a great spacecraft. And you could easily use the parts of the Shuttle program to build a great spacecraft still. But you won't be seeing that from NASA, at least without some major changes in other parts of the government.

Still, I hope someone can buy up the SSME design. One of those would make a good upper stage for a heavy lift rocket.

The cars shouldn't have a manual override for emergencies - it should have manual controls for when the computers can't handle the regular driving.

Imagine this: you're driving down a country road. It goes from a 2-lane paved road to a 2-lane dirt road to a 1-lane dirt road. At some point during that progression, the AI no longer has enough information to be able to safely operate. It come to a full stop, plays a prerecorded "Manual assistance required" message, and waits for the human to start driving. Once it's back to a point where it can resume automatic mode, it waits for the driver to hand over control.

Basically, we need a manual mode because, especially in these first generations, automatic mode won't be able to handle everything. But it should only ever force a switch from automatic to manual control when the vehicle is stopped, and the driver can safely take however long he wants to start driving.