similar to: how to run linear regression models at once

Hi, We wrote an optimization that eliminates common sub-expressions in a group of similar GEPs for the NVPTX backend. It speeds up some of our benchmarks by up to 20%, which convinces us to try to upstream it. Here's a brief description of why we wrote this optimization, what we did, and how we did it. Loops in CUDA programs are often extensively unrolled by programmers and compilers,

Hi Hal, Thanks for your comments! I'm inlining my responses below. Jingyue On Sat, Apr 19, 2014 at 6:38 AM, Hal Finkel <hfinkel at anl.gov> wrote: > Jingyue, > > I can't speak for the NVPTX backend, but I think this looks useful as an > (optional) target-independent pass. A few thoughts: > > - Running GVN tends to be pretty expensive; have you tried EarlyCSE

Hi I would like to shorten mod1 <- nls(ColName2 ~ ColName1, data = table, ...) mod2 <- nls(ColName3 ~ ColName1, data = table, ...) mod3 <- nls(ColName4 ~ ColName1, data = table, ...) ... is there something like cols = c(ColName2,ColName3,ColName4,...) for i in ... mod[i-1] <- nls(ColName[i] ~ ColName1, data = table, ...) I am looking forward to help Christof

I'm building this with llvm-c, and accessing these intrinsics via calling the intrinsic as if it were a function. class F_SREG<string OpStr, NVPTXRegClass regclassOut, Intrinsic IntOp> : NVPTXInst<(outs regclassOut:$dst), (ins), OpStr, [(set regclassOut:$dst, (IntOp))]>; def INT_PTX_SREG_TID_X : F_SREG<"mov.u32 \t$dst, %tid.x;",

Hello, FYI, this is a bug http://llvm.org/bugs/show_bug.cgi?id=13324 When compiling the following code for sm_20, func params are by some reason given with .align 0, which is invalid. Problem does not occur if compiled for sm_10. > cat test.ll ; ModuleID = '__kernelgen_main_module' target datalayout = "e-p:64:64-i64:64:64-f64:64:64-n1:8:16:32:64" target triple =

I am using a Leveno M58 (model CTO7483 from 2008) for C6/7 experimentation purposes. The thing has a dual core and ~4Gb RAM with a (newly installed) 500GB HDD. I successfully installed CentOS-6.5-minimal from a freshly burned, and verified, DVD from and ISO image whose SHA has been checked and passed. everything about the display was normal throughout that process. When the system boots from

I've examining a number of linear regression models on a large dataset following the basic ideas presented here http://www.r-bloggers.com/r-calculating-all-possible-linear-regression-models-for-a-given-set-of-predictors/ Calculating all possible linear regressions . I run into a problem with ldply when I have a formula that includes no intercept. Here's a simple test to show what happens.

su-2.05a# ping 192.168.1.1 PING 192.168.1.1 (192.168.1.1): 56 data bytes ping: sendto: No route to host ping: sendto: No route to host ^C --- 192.168.1.1 ping statistics --- 2 packets transmitted, 0 packets received, 100% packet rl1: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500 inet 192.168.1.1 netmask 0xffffff00 broadcast 192.168.1.255 inet6

Ok, as I said, the most precise way to figure out what's wrong is to emit LLVM IR first (use clang -emit-llvm ...) and check out how it differs from working examples, for instance, nvptx regression tests. ----- Original message ----- > I'm building this with llvm-c, and accessing these intrinsics via calling > the intrinsic as if it were a function. > > class F_SREG<string

Hi Timothy, I'm not sure what you mean by this working for other intrinsics, but in this case, I think you want the intrinsic name llvm.nvvm.read.ptx.sreg.tid.x. For me, this looks like: %x = call i32 @llvm.nvvm.read.ptx.sreg.tid.x() Pete On Fri, Mar 1, 2013 at 11:51 AM, Timothy Baldridge <tbaldridge at gmail.com> wrote: > I'm building this with llvm-c, and accessing these

I've written a compiler that outputs PTX code, the result seems fairly reasonable, but I'm not sure the intrinsics are getting compiled correctly. In addition, when I try load the module using CUDA, I get an error: CUDA_ERROR_NO_BINARY_FOR_GPU. I'm running this on a 2012 MBP with a 640M GPU. PTX Code (for a mandelbrot calculation): // // Generated by LLVM NVPTX Back-End //

The identifier INT_PTX_SREG_TID_X is the name of an instruction as the back-end sees it, and has very little to do with the name you should use in your IR. Your best bet is to look at the include/llvm/IR/IntrinsicsNVVM.td file and see the definitions for each intrinsic. Then, the name mapping is just: int_foo_bar -> llvm.foo.bar() int_ prefix becomes llvm., and all underscores turn into

Hi all, I have the next problem: I have a matrix with size 8,000,000x18. My personal computer...blocks...so I have cut my original file into 100 different file. I have written a function that should be run on each of this file. So imagine I need to read data from q1 to q100 file data<-read.table("q1.txt",sep="") and each time I read 1 file execute my personal function

Hello R User, I was trying to display r.squared and p value in table from regression, but I could not display these parameters in the table (matrix) for example individual <- c(1,1,6,8,8,9,9,9,12,12) day <- c(4,17,12,12,17,3,9,22,13,20) condition <- c(0.72, 0.72, 0.67, 0.73, 0.76, 0.65, 0.68, 0.78, 0.73, 0.71) test <- data.frame(individual, day, condition) ind.id <-

Dear all, I?m having trouble getting a list of regression variables back into a dataframe. mydf <- data.frame(x1=rnorm(100), x2=rnorm(100), x3=rnorm(100)) mydf$fac<-factor(sample((0:2),replace=T,100)) mydf$y<- mydf$x1+0.01+mydf$x2*3-mydf$x3*19+rnorm(100) dlply(mydf,.(fac),function(df) lm(y~x1+x2+x3,data=df))->dl here I?d like to use ldply(dl,coef(summary)) or something

Timothy, Those calls to compute grid intrinsics are definitely wrong. In ptx code they should end up into reading special registers, rather than function calls. Try to take some working example and figure out the LLVM IR differences between it and the result of your compiler. - D. ----- Original message ----- > I've written a compiler that outputs PTX code, the result seems fairly >

for example : I have files with the name "ma01.dat","ma02.dat","ma03.dat","ma04.dat",I want to read the data in these files into one data.frame flnm<-paste("obs",101:114,"_err.dat",sep="") newdata<-read.table(flnm,skip=2) data<-(flnm,skip=2) but the data only contains data from the flnm[1] I also tried as below : for

Hi all I have a list of multiple datframes with the same column headers. The last column in each datframe contains a vector of "Interval" class after I have produced this column using "lubridate" package. I needed to convert my list of dataframes to be in a single dataframe for further analysis. I did this using the following syntax : SingleDataframe <- ldply

Hi there, I ran the following code: vols=read.csv(file="C:/Documents and Settings/Hugh/My Documents/PhD/Swaption vols.csv" , header=TRUE, sep=",") X<-ts(vols[,2]) #X dcOU<-function(x,t,x0,theta,log=FALSE){ Ex<-theta[1]/theta[2]+(x0-theta[1]/theta[2])*exp(-theta[2]*t) Vx<-theta[3]^2*(1-exp(-2*theta[2]*t))/(2*theta[2]) dnorm(x,mean=Ex,sd=sqrt(Vx),log=log) }

I need to run binomial tests (binom.test) on a large set of data, stored in a table - 600 tests in total. The values of x are stored in a column, as are the values of n. The data for each test are on a separate row. For example: X N 11 19 9 26 13 21 13 27 18 30 It is a two-tailed test, and P in all cases is 0.5. My question is: Is there a quicker way of running these tests without having to