Serving the Quantitative Finance Community

 
User avatar
Cuchulainn
Posts: 20254
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

sample cuda problems in finance

April 25th, 2015, 3:44 pm

N/A
Last edited by Cuchulainn on April 24th, 2015, 10:00 pm, edited 1 time in total.
 
User avatar
Cuchulainn
Posts: 20254
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

sample cuda problems in finance

April 25th, 2015, 7:02 pm

QuoteOriginally posted by: outrunparallel binomial tree works fine.Do you have the paper? Have you read it?BM prices at a _single_ S with a speedup of 285 is not spectacular. A real FDM can price a whole matrix of values in one go.
 
User avatar
Cuchulainn
Posts: 20254
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

sample cuda problems in finance

April 26th, 2015, 10:58 am

QuoteOriginally posted by: outrunQuoteOriginally posted by: CuchulainnQuoteOriginally posted by: outrunparallel binomial tree works fine.Do you have the paper? Have you read it?BM prices at a _single_ S with a speedup of 285 is not spectacular. A real FDM can price a whole matrix of values in one go...a tree ..a grid: that's not the point... The point is that explicit is more efficient parallelizable. You can have cores work on various parts of the tree (or grid)Well, I take it then that you have _not_ read the article.There authors are talking about a speedup factor of 2! here They have conclusions why the speedup is so low.Even more worrying is they use floats!(??) The book by Sandors and Kandrot also use float. Is there no support for double.QuoteThe point is that explicit is more efficient parallelizable.Not necessarily. Example: FTCS(aka explicit Euler) scheme for heat PDE.
Last edited by Cuchulainn on April 25th, 2015, 10:00 pm, edited 1 time in total.
 
User avatar
Cuchulainn
Posts: 20254
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

sample cuda problems in finance

April 26th, 2015, 11:01 am

QuoteWe learned from the experiments that the lack of explicit mechanism in CUDA for synchronising CPU and GPU executions is a major obstacle for the hybrid processing to achieve high performance.N. Zhang et al.Really?
 
User avatar
Cuchulainn
Posts: 20254
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

sample cuda problems in finance

April 26th, 2015, 11:04 am

QuoteOriginally posted by: outrunparallel FFT is yet another option ..I think this is good case.QuoteThe Cooley-Tukey algorithm that FFTW uses to perform one-dimensional transforms works by breaking a problem of size N=N1N2 into N2 problems of size N1 and N1 problems of size N2. The way in which we parallelize this transform, then, is simply to divide these sub-problems equally among different threads. Ray tracing of course .. that's why GPU was incarnated.
Last edited by Cuchulainn on April 25th, 2015, 10:00 pm, edited 1 time in total.
 
User avatar
Cuchulainn
Posts: 20254
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

sample cuda problems in finance

April 27th, 2015, 7:45 am

Here is another approach Quote* Only very old GPUs don't have doubles, that was the first generation.So I can throw that CUDA book you recommended in the bin; it uses float.
Last edited by Cuchulainn on April 26th, 2015, 10:00 pm, edited 1 time in total.
 
User avatar
Cuchulainn
Posts: 20254
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

sample cuda problems in finance

April 27th, 2015, 8:19 am

QuoteOriginally posted by: outrunQuoteOriginally posted by: CuchulainnHere is another approach Quote* Only very old GPUs don't have doubles, that was the first generation.So I can throw that CUDA book you recommended in the bin; it uses float.Yes, you can put it on a blanked outside your home if you can't come up with the obvious solution yourself. I've said it before: I'm an expert in spoon feeding a toddler Brussels sprouts -one of the most difficult tasks on earth-, and you are acting like one.Indeed; communication is not your strong point.
 
User avatar
Polter
Posts: 1
Joined: April 29th, 2008, 4:55 pm

sample cuda problems in finance

April 28th, 2015, 12:56 pm

Some food for thought: Stepanov's litmus test in a scary, heterogeneous world: does C++ still pass?Slides: http://accu.org/content/conf2015/AlexVo ... s_Test.pdf