They sent the letter on the 22nd March, which may not have been received for a couple of days. The license was revoked on the 1st April. That is no way "enough time to respond" unless you're assuming that nobody ever goes on holiday or is away from home for more than a few days at a time on business.
Give people a month to respond and then these costly (to the taxpayer) mistakes will happen less often, and there's little difference a couple of weeks will make in the case that the license is actually fraudulent.
I guess that games have ends, so Minecraft can be more than a game. Like Lego, what are the gameplay elements of that? Some people just like creating, destroying, exploring, building, and even engineering.
Or he can get a lawyer and sue them for defamation and/or libel if what they are claiming isn't true. If it isn't, I am sure that he is already talking to a lawyer if there is no basis in the claim.
Some people take a lower paycheck in return for future returns - in addition this is an incentive for them to work harder to grow the business.
Of course these people should have their contracts well covered so that they can't be stiffed out of the future rewards part of the equation, so I don't really feel sorry for them. I'm sure they got glorious severance packages and retained stock - if they had covered their backs.
Slashdot strips it, it's a known bug with Slashdot since 1998, and still not fixed. What's that? 13 years? I think the HTML works... £
It also appears that if this library was used for any other hashing, that this bug would arise. And if that included binary files then it almost certainly would have been triggered.
The main issue is that it consistently generates the wrong hash, so that it actually appears to work fine.
However the bug fix means it now generates the correct hash. Which is going to be different. So you may have websites with significant numbers of the user base now unable to log in, for example (I am certain that accented characters will be popular in passwords in Europe, for example, and let's not even consider other areas of the world). I also wonder how many Mac users use the (S with a circle in the middle) symbol in a password. Which of course raises support queries, which costs money/time to handle...
I had a lovely big comment but hit reload instead of new-tab when going to check something, so you'll get a much rougher version now. All FLOPS are single precision.
Theoretical Cell: 25.6 GFLOPS (PPE) + 179.2GFLOPS (Cell SPU) + 400 GFLOPS (RSX, not general purpose).
So what's the unknown? I am going to assume a 3.2GHz dual-core variant of Power 7 (the architecture can go significantly faster, I'm presuming a lower clock speed for power consumption reasons; full Power 7 has eight cores). That would get 51.2 DP GFLOPS (http://forum.beyond3d.com/showpost.php?p=1363413&postcount=2946), hence 102.4 SP GFLOPS. It can also run 8 threads, compared to 2 on the PPE.
So four times the CPU FLOPS and 2.5x the GPU/Computation FLOPS (although a modern GPU will probably be far more efficient).
I am guessing that it has the same amount of shader cores as the 4870, and hence someone said that's the most similar chip.
What is interesting is the fact that it's meant to be 32nm - which means GlobalFoundries making it, rather than TSMC (who have made all ATI/AMD GPUs for a long time, Llano excepted and that's an APU). I am sure that we don't know a tenth of the details right now, hell the demo boxes might be plain old 4870s now because the actual GPU design is still being baked by AMD. In that case we actually know nothing, it could be 4870, 6xxx, or even 7xxx series, with an overall GPU performance around that of a 4870 (allegedly 1.3 TFLOPS for the Wii U).
How does the current HP site handle all that traffic then? (at least when it was being used)
Firstly, you're losing all the HP related traffic. Secondly you're mainly moving Apple traffic from their current leased offices to the new site. Overall there is probably going to be a traffic reduction until the other offices get re-leased and more people come to the city.
I don't think this algorithm is primarily meant for run-time upscaling, but to provide automated vector extraction from existing sprites for re-implementations.
This means that post extraction, the images can be manually tweaked, but you have most of the required information there already rather than having to build the vector sprite from scratch. I guess this can then be animated by hand.
The fact that this method is vectorising is one of the major benefits. The filters in ScummVM improve the image, but the results are still quite blocky or incorrect, and it's only scaling up 2x, 3x, etc, with simple heuristics. This also has a good guess at smooth shading, which could really improve some situations.
Dunno how well it would work on tile graphics though, as the individual tiles would all need to be vectorised, and then might not work well together. So I guess it's really only useful for vectorisation of sprites, not the backgrounds (unless you want a vector still of a screenshot).
One of the aims of HOTOL was human payload, for ~2 hour low-orbit flights to Australia. I presume that this aim would still be there for Skylon. Not to mention that it presumably can also use existing airport infrastructure, which is a major advantage too. Of course, at $1000 a kilo we're looking at $100,000 per person per flight, but possibly the low-orbit flights don't need as much fuel as satellite launching flights as they don't need to achieve such a high orbit.
Anyway, they're targeting half the cost of Falcon 9, albeit much further down the line. This will drive competition and lower prices across the industry over the next twenty years.
Yes. Don't take on anything more than your core skills, or the other stuff will eat into your time and stop you doing what you aim to develop. When starting up you might even find free cloud hosting whilst you develop.
You will want to move from Microsoft products as a one-man band, as this will make your cloud choices cheaper and more varied. Look into PostgreSQL or MySQL, and PHP, Java, Python or any one of the myriad other Linux web server languages.
But do put some thought into how you are going to scale your platform so that it will run on the dirt cheap hosting platform initially, yet expand and scale across multiple cloud hosts down the line.
i.e., you load your FPGA core at startup, and then you're stuck with it until you re-boot the FPGA. And yes, I know that there are some FPGAs out there that can be re-programmed as it is running.
Their system, being far closer to a CPU, lets you just load the new program into its RAM which it executes. The photo app runs face detection until you take the photo, then RAW image processing, then image enhancement, then it runs JPEG encode. The game runs physics and head tracking for 3D effects. The music apps runs DSP effects.
Of course if we were talking about a 64-way SMP machine what you write would be correct.
However in this case we are talking about a 1 or 2 way SMP machine with an array of 64 (and so far only 16 implemented) C-programmable cores that can be used.
The OS may provide a concurrent API to manage sending tasks to these cores, and getting the results out at the end. Or the applications themselves will handle this aspect. But the OS will not be scheduling standard applications themselves onto these cores.
I don't even know if this array of cores can access shared memory outside its domain - each core has 32KB of local memory.
The advantage over a GPU here is that it is very low power, and can get high GFLOPS/W compared to what's currently on the market.
To be fair with Grand Central and Apple's 'blocks' they could probably make some use of such a chip. And it's 'only' a couple of hundred million transistors for a 64-core implementation.
What I see these being is actually more along the line of a Cell SPU, but optimised for low power and inter-core communications. Like Transputer shagged an SPU. As programmers have raised issues with the SPU's 128KB of local RAM, I wonder if the 32KB of local RAM on this design will prove to be an issue in the future.
Might sound silly, but if they could provide an OpenCL front-end to their chip they will make it more desirable, but maybe that would not utilise the features that make their design worthwhile over a low-power GPU.
Well the eetimes article says they have achieved 25GFLOPS/W on their 65nm 16-core testbed running at 1GHz. This is using regular C apparently.
To compare, an AMD Zacate processor at 1.6GHz with 80 shader cores can get similar results, but in 18W on 40nm. And you have to use OpenCL, not regular C, although the tools will be a lot better at auto-parallelising.
Their aim is 50GFLOPS/W, and they're planning a shrink to 28nm where they expect to use a quarter of the power, or 100GFLOPS/W. I presume the 28nm design will be a 64-core design in around 160m transistors.
However they need to license the end design to ARM SoC manufacturers if they want to get into mobile devices. Or maybe get bought by Apple for some unique feature on their future ARM SoCs.
Even though I like Macs and Mac OS X and even iOS devices, I strongly suspect that Jobs is a bit of a demanding, selfish cunt. And that's probably why he is good at what he does, and why Apple's devices are so good at whatever he wants them to be able to do.
Of course his personal wealth is still a fraction of Gates' and Allen's. Also he is very very quiet about anything he does with his personal life, so maybe he donates anonymously.
Also, a lot of his wealth is in shares currently - selling them all to give to charity is (a) not a sensible thing to do when you're head of a company, and (b) if you believe the shares will be higher in the future, then selling them now is silly even if you do intend to donate the proceeds to charity, as you could donate even more down the line. Warren Buffet has the same belief.
Read character number from display RAM. Read row X of that character definition from character ROM. Display that bit pattern.
Do that 40 times in succession and you've got a single pixel row of your display. Then you need a pause for HSYNC. Eight more times, and you've got a single row of text characters on your display! Then all that some 24 times. Then a pause for VSYNC.
All done with very limited hardware (mainly counters and bit comparators).
It would be interesting to see the patent in question. Does it also cover keeping the text image stable on the display device (i.e., synchronisation) for example?
It's probably more than just: Character ROM A, Display RAM B, Character Generator C comprising of a counter through display RAM (0..39), a char row counter (0..7) to decide which row of the character you are generating, a display row counter (0..24), and logic that reads display RAM based upon display counters, then character ROM based upon the character you've just read, and the char row counter, then then outputs the bit pattern correctly. Display synchronization requires more counters (see the VHDL for the 6845 for example, but at least this allows you to customise the display width and height).
Somehow I suspect that to Steve Wozniak it was obvious, but maybe he isn't your average person.
Awesome, each of my cats can have its own directly accessible IP webcam finally. And my gerbils. And the goldfish. And the mice in the walls. And the spiders. And the dust mites.
Slashdot Mirror
They sent the letter on the 22nd March, which may not have been received for a couple of days. The license was revoked on the 1st April. That is no way "enough time to respond" unless you're assuming that nobody ever goes on holiday or is away from home for more than a few days at a time on business.
Give people a month to respond and then these costly (to the taxpayer) mistakes will happen less often, and there's little difference a couple of weeks will make in the case that the license is actually fraudulent.
Survival. Adventure (coming in 1.8).
I guess that games have ends, so Minecraft can be more than a game. Like Lego, what are the gameplay elements of that? Some people just like creating, destroying, exploring, building, and even engineering.
Minecraft is also horrendously addictive.
It is works it's British. If it fails it's Scottish.
Much like Andy Murray.
You have to accept that is just the way it is (tm).
Or he can get a lawyer and sue them for defamation and/or libel if what they are claiming isn't true. If it isn't, I am sure that he is already talking to a lawyer if there is no basis in the claim.
Some people take a lower paycheck in return for future returns - in addition this is an incentive for them to work harder to grow the business.
Of course these people should have their contracts well covered so that they can't be stiffed out of the future rewards part of the equation, so I don't really feel sorry for them. I'm sure they got glorious severance packages and retained stock - if they had covered their backs.
Slashdot strips it, it's a known bug with Slashdot since 1998, and still not fixed. What's that? 13 years? I think the HTML works ... £
It also appears that if this library was used for any other hashing, that this bug would arise. And if that included binary files then it almost certainly would have been triggered.
The main issue is that it consistently generates the wrong hash, so that it actually appears to work fine.
However the bug fix means it now generates the correct hash. Which is going to be different. So you may have websites with significant numbers of the user base now unable to log in, for example (I am certain that accented characters will be popular in passwords in Europe, for example, and let's not even consider other areas of the world). I also wonder how many Mac users use the (S with a circle in the middle) symbol in a password. Which of course raises support queries, which costs money/time to handle ...
I had a lovely big comment but hit reload instead of new-tab when going to check something, so you'll get a much rougher version now. All FLOPS are single precision.
Theoretical Cell: 25.6 GFLOPS (PPE) + 179.2GFLOPS (Cell SPU) + 400 GFLOPS (RSX, not general purpose).
Theoretical Wii U: 1300 GFLOPS (GPU) + unknown GFLOPS (CPU)
So what's the unknown? I am going to assume a 3.2GHz dual-core variant of Power 7 (the architecture can go significantly faster, I'm presuming a lower clock speed for power consumption reasons; full Power 7 has eight cores). That would get 51.2 DP GFLOPS (http://forum.beyond3d.com/showpost.php?p=1363413&postcount=2946), hence 102.4 SP GFLOPS. It can also run 8 threads, compared to 2 on the PPE.
So four times the CPU FLOPS and 2.5x the GPU/Computation FLOPS (although a modern GPU will probably be far more efficient).
I am guessing that it has the same amount of shader cores as the 4870, and hence someone said that's the most similar chip.
What is interesting is the fact that it's meant to be 32nm - which means GlobalFoundries making it, rather than TSMC (who have made all ATI/AMD GPUs for a long time, Llano excepted and that's an APU). I am sure that we don't know a tenth of the details right now, hell the demo boxes might be plain old 4870s now because the actual GPU design is still being baked by AMD. In that case we actually know nothing, it could be 4870, 6xxx, or even 7xxx series, with an overall GPU performance around that of a 4870 (allegedly 1.3 TFLOPS for the Wii U).
Ah yes, I see, a Hitachi CPU halfway between an 8080 and a Z80.
The author really needs to sort that space bar issue out.
It's not the only JavaScript classic system emulator.
For example, here are TWO different JavaScript Amstrad CPC emulators: http://www.cpcbox.com/ http://roland.antoniovillena.es/
A Spectrum emulator! http://jbacteria.antoniovillena.es/
TRS-80: http://jtandy.antoniovillena.es/
Of course the first two systems here are 4MHz Z80s, not 8MHz Z80s like the Gameboy Color.
How does the current HP site handle all that traffic then? (at least when it was being used)
Firstly, you're losing all the HP related traffic. Secondly you're mainly moving Apple traffic from their current leased offices to the new site. Overall there is probably going to be a traffic reduction until the other offices get re-leased and more people come to the city.
I don't think this algorithm is primarily meant for run-time upscaling, but to provide automated vector extraction from existing sprites for re-implementations.
This means that post extraction, the images can be manually tweaked, but you have most of the required information there already rather than having to build the vector sprite from scratch. I guess this can then be animated by hand.
The fact that this method is vectorising is one of the major benefits. The filters in ScummVM improve the image, but the results are still quite blocky or incorrect, and it's only scaling up 2x, 3x, etc, with simple heuristics. This also has a good guess at smooth shading, which could really improve some situations.
Dunno how well it would work on tile graphics though, as the individual tiles would all need to be vectorised, and then might not work well together. So I guess it's really only useful for vectorisation of sprites, not the backgrounds (unless you want a vector still of a screenshot).
I like their space invaders, I think they look more evil and alien. Kinda like the things from Pitch Black maybe.
One of the aims of HOTOL was human payload, for ~2 hour low-orbit flights to Australia. I presume that this aim would still be there for Skylon. Not to mention that it presumably can also use existing airport infrastructure, which is a major advantage too. Of course, at $1000 a kilo we're looking at $100,000 per person per flight, but possibly the low-orbit flights don't need as much fuel as satellite launching flights as they don't need to achieve such a high orbit.
Anyway, they're targeting half the cost of Falcon 9, albeit much further down the line. This will drive competition and lower prices across the industry over the next twenty years.
Yes. Don't take on anything more than your core skills, or the other stuff will eat into your time and stop you doing what you aim to develop. When starting up you might even find free cloud hosting whilst you develop.
You will want to move from Microsoft products as a one-man band, as this will make your cloud choices cheaper and more varied. Look into PostgreSQL or MySQL, and PHP, Java, Python or any one of the myriad other Linux web server languages.
But do put some thought into how you are going to scale your platform so that it will run on the dirt cheap hosting platform initially, yet expand and scale across multiple cloud hosts down the line.
"not restricted to running specific tasks"
i.e., you load your FPGA core at startup, and then you're stuck with it until you re-boot the FPGA. And yes, I know that there are some FPGAs out there that can be re-programmed as it is running.
Their system, being far closer to a CPU, lets you just load the new program into its RAM which it executes. The photo app runs face detection until you take the photo, then RAW image processing, then image enhancement, then it runs JPEG encode. The game runs physics and head tracking for 3D effects. The music apps runs DSP effects.
Of course if we were talking about a 64-way SMP machine what you write would be correct.
However in this case we are talking about a 1 or 2 way SMP machine with an array of 64 (and so far only 16 implemented) C-programmable cores that can be used.
The OS may provide a concurrent API to manage sending tasks to these cores, and getting the results out at the end. Or the applications themselves will handle this aspect. But the OS will not be scheduling standard applications themselves onto these cores.
I don't even know if this array of cores can access shared memory outside its domain - each core has 32KB of local memory.
The advantage over a GPU here is that it is very low power, and can get high GFLOPS/W compared to what's currently on the market.
To be fair with Grand Central and Apple's 'blocks' they could probably make some use of such a chip. And it's 'only' a couple of hundred million transistors for a 64-core implementation.
What I see these being is actually more along the line of a Cell SPU, but optimised for low power and inter-core communications. Like Transputer shagged an SPU. As programmers have raised issues with the SPU's 128KB of local RAM, I wonder if the 32KB of local RAM on this design will prove to be an issue in the future.
Might sound silly, but if they could provide an OpenCL front-end to their chip they will make it more desirable, but maybe that would not utilise the features that make their design worthwhile over a low-power GPU.
Well the eetimes article says they have achieved 25GFLOPS/W on their 65nm 16-core testbed running at 1GHz. This is using regular C apparently.
To compare, an AMD Zacate processor at 1.6GHz with 80 shader cores can get similar results, but in 18W on 40nm. And you have to use OpenCL, not regular C, although the tools will be a lot better at auto-parallelising.
Their aim is 50GFLOPS/W, and they're planning a shrink to 28nm where they expect to use a quarter of the power, or 100GFLOPS/W. I presume the 28nm design will be a 64-core design in around 160m transistors.
However they need to license the end design to ARM SoC manufacturers if they want to get into mobile devices. Or maybe get bought by Apple for some unique feature on their future ARM SoCs.
Even though I like Macs and Mac OS X and even iOS devices, I strongly suspect that Jobs is a bit of a demanding, selfish cunt. And that's probably why he is good at what he does, and why Apple's devices are so good at whatever he wants them to be able to do.
Of course his personal wealth is still a fraction of Gates' and Allen's. Also he is very very quiet about anything he does with his personal life, so maybe he donates anonymously.
Also, a lot of his wealth is in shares currently - selling them all to give to charity is (a) not a sensible thing to do when you're head of a company, and (b) if you believe the shares will be higher in the future, then selling them now is silly even if you do intend to donate the proceeds to charity, as you could donate even more down the line. Warren Buffet has the same belief.
Read character number from display RAM.
Read row X of that character definition from character ROM.
Display that bit pattern.
Do that 40 times in succession and you've got a single pixel row of your display. Then you need a pause for HSYNC.
Eight more times, and you've got a single row of text characters on your display!
Then all that some 24 times.
Then a pause for VSYNC.
All done with very limited hardware (mainly counters and bit comparators).
It would be interesting to see the patent in question. Does it also cover keeping the text image stable on the display device (i.e., synchronisation) for example?
It's probably more than just: Character ROM A, Display RAM B, Character Generator C comprising of a counter through display RAM (0..39), a char row counter (0..7) to decide which row of the character you are generating, a display row counter (0..24), and logic that reads display RAM based upon display counters, then character ROM based upon the character you've just read, and the char row counter, then then outputs the bit pattern correctly. Display synchronization requires more counters (see the VHDL for the 6845 for example, but at least this allows you to customise the display width and height).
Somehow I suspect that to Steve Wozniak it was obvious, but maybe he isn't your average person.
Awesome, each of my cats can have its own directly accessible IP webcam finally. And my gerbils. And the goldfish. And the mice in the walls. And the spiders. And the dust mites.