HNHacker News
TopNewBestAskShowJobs

miloshh

420 karma · joined October 25, 2008

submissionscomments
miloshh··on The Dangerous Trend Of Thinking That Ideas Can Be Owned, Sold Or Stolen
For example in academia, being "scooped" is a very often discussed and a very real threat. In the research community, hard-working people that are excellent at execution are a dime a dozen, and ideas that have not been done before are a really scarce resource. In the industry it might be less obvious, but it's still quite possible for a small company with good ideas AND good execution to be killed by a large competitor that decides to reimplement their product, but with 200 engineers.
miloshh··on Real World Haskell
Especially program transformations, compilation, correctness proofs, model checking... Other ML-style languages might be good too, but Haskell feels cleaner and more modern, and seems to have a more vibrant community these days.
miloshh··on Does Gödel Matter?
I think physics would be very happy to come up with a theory of "almost everything", like set theory is for mathematics. That would be a huge step forward, and would be completely unaffected by Godel's theorem.

Also, the halting problem's insolvability is straightforward to see without knowing any of Godel's work. :)

miloshh··on The Manifesto of the Futurist Programmers
What exactly are they trying to say? What they don't like and want to change?

The Communist Manifesto was similarly arrogant and angry, but at least somewhat understandable. :)

miloshh··on NVIDIA Announces Tesla Personal Supercomputer
The Larrabee SIMD width will be 16 and NVIDIA's is currently 32, so that's almost the same.

Not just the rasterizer, but any application that wants to take full advantage of Larabee will have to use the SIMD vector units to the max.

Larrabee might turn out to have great performance (which I hope), but if it does, the reason will not be black magic or breaking laws of physics. The reason will be SIMD.

miloshh··on NVIDIA Announces Tesla Personal Supercomputer
OK, this will take me some time to think about, but I'll get back to you...
miloshh··on NVIDIA Announces Tesla Personal Supercomputer
No, Larrabee will have exactly the same issues. Except threads will be called fibers, and blocks will be called threads. Read the Larrabee paper from Siggraph 08.

The way to get maximum performance out of a given chip area is to use SIMD, and that's here to stay, with all the associated issues.

miloshh··on NVIDIA Announces Tesla Personal Supercomputer
Hmmm.. OK. You mind posting a link to the pseudo-code of the algorithm? The problem of different number of iterations per thread is quite common, but can often be fixed.

If all threads in a block have the same number if iterations, you're fine; different blocks can take different number of iterations. As long as the number of blocks is much higher than the number of SMs, the machine will dynamically schedule the blocks to available SMs.

If each thread takes a different number of iterations, it's more difficult, but you can still do dynamic allocation yourself, by running as many blocks as you have SMs, and having each thread pick up work as it needs. You can also randomize the assignment of work to threads, so the expected amount of work is roughly the same. This all depends on the particular problem and data, though...

As for the 8-bit nature of the problem - this is true, you're not utilizing the floating point units that are the biggest advantage of the GPU. How large are the vectors you need to do PSADBW over?

miloshh··on NVIDIA Announces Tesla Personal Supercomputer
What are some of these complex tasks you had in mind? I think that almost any computation-bound task, however complex, could benefit from GPU acceleration. But I would like to hear about exceptions.
miloshh··on "LoseThos" 64-bit Open Source OS: For Programming as Entertainment
However, the easy access to graphics hardware is less appealing when you realize it only supports 640x480 and 16 colors. Yes, you could write directly to ports to change the screen mode, but that would break the "non-moving target" argument.
miloshh··on Python+Scipy+Matplotlib vs Matlab?
I don't agree Matlab is a horror - and I am a CS person, not an "engineer" as you say. It is an excellent domain-specific language. It might not be a good general-purpose language, but it was never intended to be that. True, some of the "bolt-on" parts of Matlab feel inconsistent, but the matrix/vector/tensor core is very elegant and powerful.
miloshh··on Python+Scipy+Matplotlib vs Matlab?
If you're affiliated with a university, it's very likely that a Matlab license is already available to you. I do research in graphics, and I find it extremely productive to test ideas in Matlab. you can hardly find a better time-saving procedure than testing an idea in Matlab and knowing in 10 minutes that it doesn't work, instead of spending a week implementing it in C++. Once you find what works, re-implementation in C++ is a breeze.
miloshh··on [dead]
The NLP concepts seem so broad and vague that whatever a socially skilled person is saying might be cast in terms of them. For example, can you imagine a single good public speaker that's not doing any "pacing and leading" or "emotion transfer"?
miloshh··on Start-Up Teaches Math to Americans, Indian-Style
Yes, part of the problem might be that mathematical achievement is not valued highly. But it cannot be the full explanation, since math is unpopular across the world, not just in the US. For example, I'm originally from Eastern Europe, one of the places where mathematical talent is supposedly valued highly, but I can tell you it absolutely isn't. I even went to a special math/programming-oriented high-school, and even there the cool kids were the ones that didn't seem to care about math at all, but instead had the looks/fashion/sports abilities.

Also, the comparison between American kids and Indian kids in the US is not quite fair, since these Indians are a highly biased sample of the true Indian population.

So the difference must be in something else, too. My personal guess is that the math in the US schools must be just too "easy". So, if kids everywhere on average learn about 60% of the math they are taught, they will be better in countries where the math taught is more advanced.

miloshh··on Ask YC: How to survive grad school?
I think the way you idealize research is not naive by any means, and you should stick to your view.

By all means, one year is not too late - you should change your research project to something that's your very own. If your advisor doesn't let you, change your advisor too.

I think you're exactly the type that needs to lead a project, and be the first author, instead of contributing bits here and there, implementing somebody else's "vision". Leave that for those that think academic success is putting their names on the maximum number of papers.

miloshh··on A Critique of Krugman's Economic (not political) Commentary
The article has some good points, but Paul Krugman does too. The modern "keynesian" view that the Great Depression was caused by very bad policy of the Fed (keeping very high interest rates and desperately clinging to the gold standard) seems very close to the truth. But, on the other hand, the people running the Fed now, trying to avoid the mistakes of that era, might be going to the other extreme of keeping interest rates too low and printing too much money.While that might make the current recession relatively short and mild, it will prepare the ground for further bubbles - people, worried about their cash losing value, will invest like sheep into the next big thing, until it bursts again. So, maybe the right balance is somewhere between "keynesian" and "austrian"?
← PreviousPage 6 of 6