Re: What will they do with the data?
No he doesn't realise.
Also doesn't realise that you don't need to be working on upstream to write drivers. It open source! Everyone can write a driver, and many people do. And then get them upstreamed.
2645 publicly visible posts • joined 12 Jun 2009
Yes, they are dumping them because is cheaper. The cost to recover is the barge to land on, the time to get it back, the cost of craning it off the barge, the cost of storing it until you spend even more money and time to dismantle it. And actually, the environmental cost of that lot is probably higher than just dumping it.
But tbh, dropping a few in the water is still less impact that a couple of containers falling off a ship, and you don't seem to be complaining too much about that.
Ahh, the old government subsidy attack. It was wrong before and it's wrong now.
Tesla WERE helped in the initial stages by government subsidies. The same subsidies that were available to other businesses. Any business in the same area was able able to take advantage, if they didn't that was not Musk's fault. Subsidies are used by governments all round the world to kick start business and new tech.
Tesla paid back their government loads (not subsidies).
SpaceX were paid to provide a service to NASA (not a subsidy). They have been providing that service cheaper than anyone else, so in fact, NASA have saved money by using SPaceX.
SpaceX performance history is pretty good, checkered is entirely wrong, they have over 50 flights of the F9, with two failures early on, the causes of which have been fixed.
There has been a accident with a ModelX - whoopy doo,. There were 37,461 road deaths in the US in 2016. In the same year 38,658 death by gun. Do you really think that ONE death in a X is statistically significant? Even now I suspect that computer driven cars are safer than human driven ones!
As with any company investment, on the whole the investors know what they are doing. And at the moment they are happy with Tesla. When they start dumping shares, then maybe a cause for concern, but until then Tesla are pretty safe.
SpaceX itself was a massive gamble from the very start.
To says that 'all' they have done is evolutionary is probably true. And yet they are the only people to have EVER done it (ie reusability of S1). So whilst what that have done is evolutionary, what they have done to the launch market is revolutionary.
SLS is the same - its an evolution. As are ALL other rocket projects.
BFR will be the real revolution. And will be REAL heavy lift. Heavier than SLS but at a fraction of the cost. If it works, and I see no reason to believe it won't.
Some answers to questions above:
New SoC: I reckon about three years work. We need to ensure its robust, performant, certified with a solid SW base that works on day of release.
H265. Not purely in HW, that would require new HW blocks, but we are reusing some H264 blocks, plus NEON to get HEVC 1080p30 working for the majority of the use cases.
We've had problems with the supplier of the BT chip updating their firmware at any sort of reasonable rate. This is a newer chip with newer firmware, although same supplier, so we hope there are some improvements.
BT is just a PITA in general I find.
EDIT to add: BT/Wifi coexistence is the culprit I suspect. A good idea in principle, a PITA in practice.
@DougS The problem is that Android is a sprawling mess of millions of lines of code. Stripping out 'unnecessary' stuff is a horrendous job, and there is still no guarantee it will be any more reliable than something written from scratch (albeit based over the Linux kernel which handles all the complex stuff).
It's not that the development process is shit, it is indeed that CPU development is very hard indeed.
Combine that with some very clever people who can figure out utterly obscure exploits such as these, and this sort of thing can happen.
As someone above said, it's taken about 10 years to figure out this exploit. If Intel/ARM/AMD had spent an extra 10 years figuring it out before releasing the chips, well, nobody would be in business.
@Jake. Feel free to continue polluting the world until you die coughing, or my children do. Electric is the future, you might not like it, but to get this planet back on the road to long term climate stability, something needs to change. Going electric is one way of doing it, it's going to be a long process, and people like you won't like it. But I guess once the electric vehicles start out performing petrol and diesel across the board, even the most hardened petrol head (and I am one - used to race cars, children race karts) will finally have to admit defeat.
Indeed, Which is why they are fixing the production issues.
This is the bit I don't get when this is continually brought up. They have loads of people working endless hours sorting out the productions issues. Do people actually think they WONT sort them out, that they will encounter some intractable problem? Seems horribly unlikely to me. Production line are a solved problem, they take time to get up to speed, but they (almost) always get where they need to be.
In Ely a few months ago, in a queue of cars waiting at a T junction. Car in front drives away, I start moving up to the stop line. Some guy in full on lycra running gear runs straight across the junction in front of me. Because there are houses on each side of the junction and he was at speed there was no way I could see him before he ran out. He rolls along the front the car. Then gave me an earful*. He hadn't even slowed down or looked as he ran across the road. He did slow down as he slid over the car. I was doing about 3mph at the most, which is not a huge speed, he was running faster than I was moving, yet he was still unable to avoid the incident.
If you are reading this you wankpuffin, you will, in all likelihood, die through behaviour like this. Cars are heavier than people, and somewhat more robust.
* I did return the earful.
I once went to a talk by Doug Trumbull (yes, the one mentioned in the article - Siggraph 94 IIRC), and he was asking for 16kx16k resolution for IMax (specifically, the Back to the Future ride at Universal, again IIRC). So resolution is important, it just needs to be applied in the right place.
1080p is not necessary on a smartphone.
It's actually a lot easier nowadays to build you own silicon. Good tools, libraries and cheap FPGA's make the process cheaper. Still expensive making the actual silicon, even on a MPW.
I'm not surprised in this move TBH, getting a custom designed chip exactly targeted at what you want, rather than buying something not quite right from a third party for more money seems like a good move.
I don't think this is overstretching. Its not like Musk is sitting there doing the design himself, he has employed a lot of smart people to do that for him.
Important to remember - Investors, in general, are not stupid people. If they think DONT think Tesla are a worthwhile investment, they don't need to put any money in. And yet, here they are, putting money in. Ergo, they think Tesla are a good investment. And I think they are right, certainly better than the money sinks that are the other car companies in the states.
@dainB
If you are going to give examples of failure, at least give examples of failure. I think everything you mentioned is actually still in progress. Failures happen when they fail. And none of the examples you have given have failed, in fact, most of them are successes. And having read you post, were you being sarcastic? Because after I wrote the below I realised that everything you wrote was actually the exact opposite of whats actually happening.
Gigafactory - production ramping up.
Hyperloop - people starting to build prototypes.
Powerwell - installations going on around the world. Related - Pueta Rico and the Australian thing that just went online.
Solar Shingles - early days, but still going and a great idea.
Constellation/Global Broadband (same thing) - Two prototype satellites going up early next year.
Boring Company - currently building a test tunnel.
Model 3 production ramping as we speak.
SpaceX - undercutting its competitors by a considerable margin and now launches about half the worlds satellites. And is bringing back boosters for reuse. Whether they use them is up to SpaceX, some already used, the F9H is two reused cores. Block 5 about to come on line which is the final variant and should be good for 100 flights each.
Of course, any of the above might not pan out, but to claim any of them are failures at this point is clearly absolutely incorrect.
Yes, the Ethernet and 4 port USB hub all go via a single USB port into the SoC. Yes, it is a bottleneck, no it hasn't really stopped a lot of sales, because in general people are not too worried about it. Those who really need the throughput get alternative devices and live with the less than useful support.
As for whether this is a toy, I'd suggest that if Los Alomos National Lab though it was a good idea to make one, then it's not really a toy. After all, it's not like the Pi is the new kid on the block with an unexpected design that will catch them unawares.
This is a project for testing code in an environment similar to the real HPC's, so you don't take up valuable compute time ironing out the kinks. It gives you 750 * 4 core A53 devices running at 1.2GHz, with slow interconnects and only 1GB RAM per node, buts costs less than $35k or so (my figures, not sure of final cost, assuming $35/Pi ex VAT plus costs for the Bitscope racks)
The latest Pi version is a quad core arm A53 running at 1.2gig so considerably faster than the original Pi model. They also support neon in the latest models so that if you can take advantage of it give you a massive speed improvement.
There's a thread on the Raspberry Pi forums that has a lot of linpack style testing done on it which might be worth a look if you're interested.
Raspberry Pi make what is it about 500k a month or something ridiculous so buying in quantities is not a problem.
As for using something like a banana pie or an orange Pi yes they do have faster ethernet but the os support is absolutely rubbish. Since this is a educational project rather than a serious high-performance computing project it seems that the Ethernet is probably not that important anyway.
There is a single consistent GUI/desktop. The one you chose to use. Just ignore any others.
As for compiling applications, just chose one where the distro has already done it for you. I haven't had to compile an app for years, I just use whatever Canonical has in their repo.
And using one distro does not means a monopoly, since the kernel development is done out of distro land.
Note: I have Windows10 on the laptop, but only for web and some Windows only apps (not many at all). We handle a huge amount of stuff on the cloud (email, source control, bug tracking, office management, task management, which can actually be done under Linux as well). All of my software work in done on a Linux VM, or VNC to a Linux server, apart from a particular debugger which runs under Windows, but access the source tree on a Linux server.
Generally, the Windows fails exceed the Linux fails by quite some margin.
I've got a couple of Dysons and a Henry.
The Henry is shit at actually sucking stuff up compared with the Dysons. The Dysons (one handheld, one upright) are fairly robust. The upright has a broken bit of plastic, but is over 5yrs old and still works fine. One day I'll get round to fixing it. The handheld is great, cost about £80 IIRC, so maybe twice the cost of a really cheap nasty, but well made and very effective.
I have no issues with Dyson's.