well, I have been living in the US for a few years and my job (researcher) requires me to read and write english frequently. I do understand where the confusion can come from. Certainly punctuation would help clarifying. Though, it is rare that context does not allow you to understand the intended meaning.
You are right, let's distrust a press release about car problem and trust a slashdot article instead. Slashdot is so much expert on the POTUS's car and in investigative journalism.
While I understand your point, my understanding of punctuation is not fine enough for me to correct it by myself. Note that even in French I do not understand the punctuation rules properly. But anyway, I'll read the rule again. Maybe I'll end up understanding it one day.:)
So my parents used to be car dealers in the 80's. At that time, there were only crap automatic cars with so much "latency" in gear shifting that it was difficult to drive in heavy traffic. (My parents live in Paris.) When they visited me in the US, they realized that automatic gear shifting is decent now. Probably not as precise as a experienced driver, but good enough in most situations.
I found it interesting to see that because of bad previous experience, Americans have a huge biais against diesel which is common in Europe. Meanwhile, because of bad previous experience, European have a huge biais against automatic gear shifting which is common in the US.
Well, If you do not need massive bandwidth, you do not. Personnally, I do not use GPUs to do graphic computation for sparse computations (multiplying sparse matrices or traversing graphs). On these computation the main bottleneck is memory bandwidth. So if the memory bandwidth increase by a factor of 3, I will see an immediate improvement of performance by at least 50%, potentially by a factor of three once the kernels are optimized for that new architecture.
1TB/s of memory bandwidth is indeed impressive. I am doing quite a bit of memory intensive kernels (graph algorithms) on accelerators (GPU, Xeon Phi). And bandwidth is a significant bottleneck. Kepler did not bring a significant bandwidth improvement over Fermi. Xeon Phi is in the same areas. But 1TB/s seems tremendous. I am impatient putting my hands (or my ssh) on one of these.
My bad, I realize later that AVX was the new instruction set for sandy bridge and not for xeon phi. AVX (version whatever) and IMCI instructions are quite similar (gather/scatter, Fused Multiply Add, swizzling/permute). Their main different is the SIMD width.
My overall point remains valid. Doing floating point arithmetic by packs of 256 bits is overall useful.
* Expansion of most integer AVX instructions to 256 bits
* 3-operand general-purpose bit manipulation and multiply
* Gather support, enabling vector elements to be loaded from non-contiguous memory locations
* DWORD- and QWORD-granularity any-to-any permutes
* Vector shifts
* 3-operand fused multiply-accumulate support
Intel Xeon Phi relies on avx (version 1 I believe) and using avx gets you good improvement compared to not using avx for both sequential and parallel codes. Of course, course sequential code on Xeon Phi is typically slower than a regular sandy bridge processor.
Many applications can use 16 float operations simultaneously. Certainly many video codecs and physics engine.
GPUs can be good for many computations but tehre are many case where they are not so good. Most pointer chasing type of application tend not to be so GPU-friendly. If you need to go back and forth between CPU and GPU, then you pay some latency. GPUs suffer from programming abstraction problems (no CUDA on AMD, OpenCL is suboptimal on NVIDIA, openacc is only good for simple tasks).
Larger SIMD lanes on the CPU side will certainly be a good thing for performance.
What is the use for them? for "personal" use, floats are all you will ever need. Many physics computation stays in single precision to avoid doubling the memory usage. I guess fluid mecanic computation use double, but is there really a use for quad. Who needs that kind of precisions?
But would the legal system in the US allow you to sue (and potentially win) UMG for making you lose your job? Not saying you should do it; it is only a theoretical question.
I love git and latex and use them internally at my university. But experience proved that that combination does not work well with all collaborators...
What I am doing and all the communications with my colaborator which are spread in various institute all over the world is fairly asynchronous. What do you suggest as a technology which should be perfectly portable, trivial to use and deployed everywhere that I can convince my collaborator to switch to?
I am pretty much sticking to my HTC G2 for that reason. I want my phone to have a reasonnable keyboard. The G2 has a 4 row qwerty keyboard, and that's really useful. I do not really have a need for a fifth row though. I am not going to write perl on my phone:) I am mainly using my phone to answer short text-based communications (emails, IM or text messages.) Acording to my phone logs, I make less than 3 phone calls a day. (1 of which I could make from my office phone and 1 of them I could make from my home phone if it was not ridiculously priced.)
It is frequently mentionned on slashdot. Many emblematic games when they are ported to new architectures are ported by porting SDL on the new architecture.
Android document is pretty bad as well. Everything is explained assuming you read everything from cover to back and understood. Except it is not a book so there is no start and no end. So it is good for looking something up, but for learning or understanding a problem the documentation is not sufficient.
StackOverflow on the other hand plays the role of a huge FAQ.
Slashdot Mirror
well, I have been living in the US for a few years and my job (researcher) requires me to read and write english frequently. I do understand where the confusion can come from. Certainly punctuation would help clarifying. Though, it is rare that context does not allow you to understand the intended meaning.
Why at random interval? My mail client fetches email automatically at a fixed interval. Random interval would actually look stranger during an audit.
You are right, let's distrust a press release about car problem and trust a slashdot article instead. Slashdot is so much expert on the POTUS's car and in investigative journalism.
While I understand your point, my understanding of punctuation is not fine enough for me to correct it by myself. Note that even in French I do not understand the punctuation rules properly. But anyway, I'll read the rule again. Maybe I'll end up understanding it one day. :)
So my parents used to be car dealers in the 80's. At that time, there were only crap automatic cars with so much "latency" in gear shifting that it was difficult to drive in heavy traffic. (My parents live in Paris.) When they visited me in the US, they realized that automatic gear shifting is decent now. Probably not as precise as a experienced driver, but good enough in most situations.
I found it interesting to see that because of bad previous experience, Americans have a huge biais against diesel which is common in Europe. Meanwhile, because of bad previous experience, European have a huge biais against automatic gear shifting which is common in the US.
I guess everybody is just as biased :)
Well, If you do not need massive bandwidth, you do not. Personnally, I do not use GPUs to do graphic computation for sparse computations (multiplying sparse matrices or traversing graphs). On these computation the main bottleneck is memory bandwidth. So if the memory bandwidth increase by a factor of 3, I will see an immediate improvement of performance by at least 50%, potentially by a factor of three once the kernels are optimized for that new architecture.
There are many comments on Al Gore on slashdot. I did not know what it was refering too. Apparently it comes from a quote in 1999.
http://www.snopes.com/quotes/internet.asp
1TB/s of memory bandwidth is indeed impressive. I am doing quite a bit of memory intensive kernels (graph algorithms) on accelerators (GPU, Xeon Phi). And bandwidth is a significant bottleneck. Kepler did not bring a significant bandwidth improvement over Fermi. Xeon Phi is in the same areas. But 1TB/s seems tremendous. I am impatient putting my hands (or my ssh) on one of these.
My bad, I realize later that AVX was the new instruction set for sandy bridge and not for xeon phi. AVX (version whatever) and IMCI instructions are quite similar (gather/scatter, Fused Multiply Add, swizzling/permute). Their main different is the SIMD width.
My overall point remains valid. Doing floating point arithmetic by packs of 256 bits is overall useful.
replying to self. Xeon Phi uses larger lanes than AVX. It is 512 bits in Xeon Phi and 256 in AVX, I got the names mixed up.
No, there is more to it:
* Expansion of most integer AVX instructions to 256 bits
* 3-operand general-purpose bit manipulation and multiply
* Gather support, enabling vector elements to be loaded from non-contiguous memory locations
* DWORD- and QWORD-granularity any-to-any permutes
* Vector shifts
* 3-operand fused multiply-accumulate support
source: wikipedia http://en.wikipedia.org/wiki/Advanced_Vector_Extensions#Advanced_Vector_Extensions_2
Intel Xeon Phi relies on avx (version 1 I believe) and using avx gets you good improvement compared to not using avx for both sequential and parallel codes. Of course, course sequential code on Xeon Phi is typically slower than a regular sandy bridge processor.
Many applications can use 16 float operations simultaneously. Certainly many video codecs and physics engine.
GPUs can be good for many computations but tehre are many case where they are not so good. Most pointer chasing type of application tend not to be so GPU-friendly. If you need to go back and forth between CPU and GPU, then you pay some latency. GPUs suffer from programming abstraction problems (no CUDA on AMD, OpenCL is suboptimal on NVIDIA, openacc is only good for simple tasks).
Larger SIMD lanes on the CPU side will certainly be a good thing for performance.
What is the use for them? for "personal" use, floats are all you will ever need. Many physics computation stays in single precision to avoid doubling the memory usage. I guess fluid mecanic computation use double, but is there really a use for quad. Who needs that kind of precisions?
But would the legal system in the US allow you to sue (and potentially win) UMG for making you lose your job? Not saying you should do it; it is only a theoretical question.
I love git and latex and use them internally at my university. But experience proved that that combination does not work well with all collaborators...
the government is not the only one interested in data... tax records might not be the only thing you store...
services like dropbox can be handy. privacy concerned people can use owncloud. (I do, it works fine.)
how do I send a message to somebody through dropbox?
What I am doing and all the communications with my colaborator which are spread in various institute all over the world is fairly asynchronous. What do you suggest as a technology which should be perfectly portable, trivial to use and deployed everywhere that I can convince my collaborator to switch to?
I am pretty much sticking to my HTC G2 for that reason. I want my phone to have a reasonnable keyboard. The G2 has a 4 row qwerty keyboard, and that's really useful. I do not really have a need for a fifth row though. I am not going to write perl on my phone :) I am mainly using my phone to answer short text-based communications (emails, IM or text messages.) Acording to my phone logs, I make less than 3 phone calls a day. (1 of which I could make from my office phone and 1 of them I could make from my home phone if it was not ridiculously priced.)
It is frequently mentionned on slashdot. Many emblematic games when they are ported to new architectures are ported by porting SDL on the new architecture.
Wow, there are really people on slashdot that do not know what SDL is?
And they wont accept numerical solution: http://xkcd.com/613/
Android document is pretty bad as well. Everything is explained assuming you read everything from cover to back and understood. Except it is not a book so there is no start and no end.
So it is good for looking something up, but for learning or understanding a problem the documentation is not sufficient.
StackOverflow on the other hand plays the role of a huge FAQ.