Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?
Check out the new SourceForge HTML5 internet speed test! No Flash necessary and runs on all devices. ×

Comment Re:Can it beat the doctors (Score 1) 138

Yes, and we already have that. There are people who die every day waiting for a transplant organ. There's a limited amount available so they must be rationed and someone (or a panel of people most likely) has to determine where the limited supply will do the most good.

That's what other western countries do. The US goes by who has the best health insurance, like how big of a bill can we justify sending...

Comment Re:Down with Putin - Down with Trump (Score 1) 191

Yes. Timing is the key to understanding it. Sanders would have defeated Trump easily. The timing of the releases were carefully placed so as to build suspicion with independants while not hurting her primary bid. Then once she clenched that, proof that it was a rigged primary sent a lot of independants away from the DNC to either Green, Libertarian, and even a number to Trump. If they had released it all in the beginning, we would be swearing in Sanders tomorrow.

Let's not forget that the DNC wanted Trump to win the Republican nomination. So first Trump let Hillary drag him center stage as the enemy, then he let her eliminate Sanders before landing a final blow nobody saw coming until late election night? If all of that was planned Machiavelli could take lessons from him. If could simply be that they know the media has the attention span of a humming bird on speed, let's just pace this out so we get a good buzz and can keep it going until the election and that the rest was plain lucky. It's either that or we're in a Bond movie.

Comment Easy to save on RAM, but RAM is cheap (Score 1) 89

It's easy to save on RAM, but RAM is cheap. With the zram module in Linux, you can create a zram block device 2x the size of RAM with mem_limit set to 50% of RAM and experience approximately no performance hit--faulting out of zram is approximately twice as heavy as a worst-case cache miss. I've had a 1GB server run 700MB into zram swap trying to run Gitlab, with 40MB of available RAM (including disk cache), and not show any visible sign of performance degradation; note that that's about 230MB of RAM acting as a compressed cache area for that 700MB, and 770MB of flat RAM available.

This works because CPU isn't pegged to 100% on average across 1 second, and decompression requires something like 23-26 instructions per byte. That means decompressing one page per second on a 1.2GHz core consumes about 0.00887% CPU at 1 cycle per instruction, or 0.0266% at an average 3 cycles per instruction. RAM prefetching is actually huge--a cache miss can cost 48 cycles for 64 bytes (on x86-64) or 0.000256% for a 4096-byte page, at a minimum, with 8-cycle CAS across a CPU, or a whopping 1,200 cycles or 0.0064% for 4096 bytes, although that's never going to happen (it's physically impossible: sequential reads don't need the expensive row precharge before RAS after the first read).

Basically, if your code uses memory infrequently, it has no reason to swap; and if it uses it frequently, then the cost of swapping can be absorbed by prefetch algorithms similar to the ones used by the CPU itself to avoid the above cache miss costs. Standard LRU swap algorithms will prevent swapping out of frequently-used memory; and the delay waiting for a swap-in consumes the bits of unused CPU time in a 99.7%-pegged processor.

The performance hit explodes exponentially at a certain point. If you have 1GB RAM and use 900MB as a compressed swap such that you have 2.8GB available, you're going to have a bad time. If you have 1GB and use 500MB for swap such that you have 2GB available, you'll be fine even under high load.

The problem is the whole phone is made of a SOC which isn't that much cheaper on 1GB versus 2GB; expensive NAND storage; radio chipsets; a battery; an expensive display; and so forth. The SOC isn't even the biggest part, with a cost of like $35 or sometimes in the $20 range for something current-generation for a $400 phone, up to $70-ish for state-of-the-art SOCs. Slap a $100 screen, $80 of TLC NAND, and $40 of boards and components and case around a $30 chip and you have a $250 phone.

Comment Re:That doesn't change anything (Score 1) 107

Making money here means net profit. For a growing business, this should be negative. If you can borrow money at a 5% interest rate and use it to grow the business at a rate of 10%, then you report a loss but your company value increases. Amazon's strategy has been along these lines for most of its existence: all gross profits are reinvested in the business, so they always show a loss and their increase in value gives them the access to more capital through loans or by issuing more stock.

Comment Re:Can it beat the doctors (Score 4, Informative) 138

This is a common problem with most AI announcements. Is 80% accurate better than a simple statistical model? Often not. Does it scale up from a small sample size? Remember the recent face recognition thing that managed with only a hundred or so pixels? Sounded impressive, until you realise that the training set and the testing set were the same and that they only included around 1,000 faces, so simple information theory tells you that you only need 10 bits of information to identify each one and 800 bits doesn't sound quite so impressive.

Comment Re:Don't realize who the robber barons are, do you (Score 2) 107

And yet, in the UK where it is illegal to make union membership compulsory, unions seem to work better. If people see the unions working on their behalf, then they're generally happy to pay the dues (sure, you get a few freeloaders, but not enough to break the system). If a union is not representing the interests of the majority of its members, then it will quickly see its funding dry up. Importantly, voluntarily paying union dues is a big signal to the employer that the union actually does have negotiating power: it implies that the the union is trusted by the majority of the employees to bargain on their behalf. In many places, you have two or more competing unions (though the law says that any deal reached by one union must be offered to all employees, irrespective of whether they are members of that union) and so not only does a union have to represent its members' interests to retain its income, it has to represent those interests better than the competition.

Comment "frankly unlikely"? (Score 3, Insightful) 195

ZDNet adds: ... And while the likelihood that the company is doing anything nefarious with users' information is frankly unlikely ...

This quote is a case of somebody writing something to just fit a grammatical template, rather than thinking about what they're writing. Substantiate that wild speculation, ZDNet, or turn in your beard-stroking license asap.

Comment Re:battery life a braindead argument (Score 1) 289

Only if you never use suspend to RAM. 32GB of DDR4 will use 12W, constantly, for as long as the machine is storing data in memory, including in sleep mode. Currently, the sleep mode uses around 1W, so you're cutting the sleep time to 1/12th before you even start using the machine. In fact, with the current FAA rules on battery size allowed on flights, you'd only get about 8 hours of standby time in the model you're describing - not even enough to leave it overnight without needing to suspend to disk. In idle use (CPU and GPU not doing much, but screen on), you'd double the power consumption. In heavy use, you'd increase it by about a quarter. Unless you're spending basically all of your time with the CPU and GPU saturated and swapping heavily, you'd see far less battery life with 32GB of DDR4 than with 16GB of LPDDR3 (the choices that current Intel chips provide).

Comment Re:a little late, no? (Score 1) 289

The batteries in the MBP are as big as the FAA allows on planes. Even if you're not using it in the cabin, you're not allowed lithium ion batteries in the hold at all, so they'd have created a laptop that no one could take on a flight. That makes it useless for a lot of Apple's current customers and having two lines, one for people who might want to fly and one for people who definitely won't would be a pain.

Comment Re: They said they want us to die... (Score 1) 289

A C++ compiler will happily use 2-300MB of RAM. A MBP has 4 cores plus hyperthreading, so to make sure that you're using the CPU you're doing 8-way parallel builds. That will easily fit in 4GB, until you get to the small handful of template-heavy files that use 1-2GB each, and suddenly you're at 16GB and swapping, which kills performance for the whole build. The linker will take 4GB or so if you're not doing LTO, if you are then it will happily chew through 16GB.

Comment Re:Expected /. response (Score 1) 493

Firstly, you apparently didn't read my comment that I wasn't discussing how apt works, only yum.

When Yum downloads something, it fetches a bunch of repo information (like apt-get update), then it downloads files (like apt-get install). To do this, it does... all the shit I described apt doing.

Secondly, the critical issue that you are missing is that if I install a package from an alternate repository (eg EPEL), my systems don't tell the main CENTOS mirrors about those EPEL packages.

No, of course not. You tell Georgia Tech, the NSA CentOS Mirror, or Microsoft's Redmond CentOS mirror, at random, who you are and what you're downloading.

Multiple distributions and mirror maintainers coordinate in secret to keep security exploit details quiet until a patch is ready from everyone. There's an entire network of quiet discussion that happens, intentionally hidden from everyone, to make sure everyone hits the ground running. If you report a remote exploit in Firefox directly to Mozilla, Debian, RedHat, Slackware, or Gentoo, marked as a security bug, they will keep the details private until everyone has patches ready; then they all release at once.

So you believe Microsoft is doing secret things dealing your data to secret partners in secret; but that Linux distributions might not be secretly collecting your data, or that various Linux mirrors who aren't controlled by those distributions aren't under the influence of others. That is: although AT&T was sucking up your phone data and piping it to the NSA, they apparently won't collect what scraps of OS update telemetry data hits their servers in the same way.

You're basically saying there's no network of bad actors out there, so instead of trusting "Debian", you trust everyone.

Finally, there is no fingerprinting involved in the yum transactions. If I have multiple machines behind a single IP address, the server doesn't have sufficient information to distinguish them. As well has having insufficient information to fingerprint individual systems, no user information is transmitted.

We've been able to identify individuals based on their Internet usage and TV usage, even from the same account, device, and browser. We can tell if your 16 year old daughter or her 17 year old sister is currently using the PC or watching TV.

I might have two x86-64 PCs running the same version of Ubuntu, and a Raspberry pi; you can fingerprint at least three systems out of my usage habits, and identify one distinctly at least.

Through all of that...

In summary, yes I am leaking some information, but it is benign.

The leaking of what Microsoft software you've installed to Microsoft's servers is benign as well. Who fucking cares that Microsoft knows you have Office 2013 installed?

Slashdot Top Deals

We all like praise, but a hike in our pay is the best kind of ways.