Just read any non-mainstream book. It's the best thing you can do to get out of the way society has formatted you. In particular, you might try reading books from a variety of countries and cultures instead of just reading English "classics", that are only classics because people read them in the past and they ask the next generation to read them as well..
Cough isn't even a word in German... Amen is a Latin word, and not an English word of Latin origin. While pronouncing it the English way is accepted by laymen, it's not the proper pronunciation. Amen is a word which is used in many different languages too.
Trivial linear algebra kernels, even sparse ones, are not real applications. Sparse linear algebra should be interesting to do on the GPU, since you don't suffer from the lack of gather instruction and you have no cache. Apart from that, FLOPS is not a good unit really, since for non-trivial algorithms the number and types of the operations being done are non trivial, you might have dependencies stalling the pipeline, you might need to spill and thus depend more on local memory... What's more interesting is finding how many cycles are required for each kernel. Real applications are things like physics simulation, computer vision, quantitative finance, or audio/video transcoding. They're not as simple to optimize as a simple BLAS routine.
Extracting the peak performance of a GPU is often very difficult, while it's reasonably straightforward to get close on the CPU. In practice, I personally often see CPUs starting to beat GPUs at around 32 cores for real applications, but that's with a GPU 3 to 4 times less powerful than above.
Are you sure you know what China is? While they care less for human rights than other countries, it is one of the biggest economies in the world, and nearly all other economies are dependent on them.
Slashdot Mirror
It fails as soon as I move.
More like warmongering definition.
You realize Japan is a sovereign nation, right?
Surely you're speaking of integral logarithm and division, not floating-point?
monotone is written in C++, and we all know what Torvalds thinks of C++
Just read any non-mainstream book. It's the best thing you can do to get out of the way society has formatted you.
In particular, you might try reading books from a variety of countries and cultures instead of just reading English "classics", that are only classics because people read them in the past and they ask the next generation to read them as well..
Is this some sort of joke?
A Unix-style shell is the most useful command line interface. It is itself its own scripting language.
There are various ones. The most popular is bash, but there are also fairly different ones like fish.
The patents are in the OpenGL implementation, not in the hardware specifications.
Cough isn't even a word in German...
Amen is a Latin word, and not an English word of Latin origin. While pronouncing it the English way is accepted by laymen, it's not the proper pronunciation. Amen is a word which is used in many different languages too.
It's Latin. You don't get to use English pronunciation with it.
Nice way of displaying you're a hardcore weeaboo.
Call it a "script" or "writing system" if you want to be anal. It's the same thing than an alphabet to normal people.
There are three japanese alphabets.
The Japanese are the ones whose alphabet uses the same letters for r and l sounds.
Trivial linear algebra kernels, even sparse ones, are not real applications. Sparse linear algebra should be interesting to do on the GPU, since you don't suffer from the lack of gather instruction and you have no cache. Apart from that, FLOPS is not a good unit really, since for non-trivial algorithms the number and types of the operations being done are non trivial, you might have dependencies stalling the pipeline, you might need to spill and thus depend more on local memory... What's more interesting is finding how many cycles are required for each kernel.
Real applications are things like physics simulation, computer vision, quantitative finance, or audio/video transcoding. They're not as simple to optimize as a simple BLAS routine.
don't confuse japan with china, you insensitive clod!
10k is the normal price of a server with decent computing power. If you're surprised by this price tag, clearly you have never bought one.
Most Xeon processors are vastly different from "equivalent" consumer-grade hardware. If yiu can't tell the difference, they're just not for you.
More processors also means more bandwidth. Processor clock speed isn't all that matters.
How is Qt for exampke not comparable to Cocoa?
Extracting the peak performance of a GPU is often very difficult, while it's reasonably straightforward to get close on the CPU.
In practice, I personally often see CPUs starting to beat GPUs at around 32 cores for real applications, but that's with a GPU 3 to 4 times less powerful than above.
I was talking about porting to another OS on the same type of device.
You should be using an abstraction layer that can be easily made to target any platform.
Unlike you American, I read several books every week.
Are you sure you know what China is? While they care less for human rights than other countries, it is one of the biggest economies in the world, and nearly all other economies are dependent on them.