similar to: Using objects within functions in formulas

All, If we can speculatively execute a load instruction, why isn’t it safe to hoist it out by promoting it to a scalar in LICM pass? There is a comment in LICM pass that if a load/store is conditional then it is not safe because it would break the LLVM concurrency model (See commit 73bfa4a). It has an IR test for checking this in test/Transforms/LICM/scalar-promote-memmodel.ll However, I have

I think gcc is right. It inserted a branch for n == 0 (the cbz at the top), so that's not a problem. In all other regards, this is safe: if you examine the sequence of loads and stores, it eliminated all but the first load and all but the last store. How's that unsafe? If I had to guess, the bug here is that LLVM doesn't want to hoist the load over the condition (which it is right

Thanks for the background on the concurrent memory model. So, is it sufficient that the loop entry is guarded by condition (cbz at top) for preventing the race? The loop entry will be guarded by condition if loop has been rotated by loop rotate pass. Since LICM runs after loop rotate, we can use ScalarEvolution::isLoopEntryGuardedByCond to check if we can speculatively execute load without

Hi All, I found some cases where a spill of a live range in a block is reloaded only in one of its successors, and there is no reload in other paths through other successors. Since the spill is reloaded only in a certain path, it must be okay to sink such spill close to its reloads. In the AArch64 code below, there is a spill(x2) in the entry, but this value is reloaded only in %bb.1, not in

On 2018-02-22 11:14, gberry at codeaurora.org wrote: > FROM: llvm-dev [mailto:llvm-dev-bounces at lists.llvm.org] ON BEHALF OF > Jun Lim via llvm-dev > SENT: Thursday, February 22, 2018 11:05 AM > > Hi All, > > I found some cases where a spill of a live range in a block is > reloaded only in one of its successors, and there is no reload in > other paths through other

From: llvm-dev [mailto:llvm-dev-bounces at lists.llvm.org] On Behalf Of Jun Lim via llvm-dev Sent: Thursday, February 22, 2018 11:05 AM Hi All, I found some cases where a spill of a live range in a block is reloaded only in one of its successors, and there is no reload in other paths through other successors. Since the spill is reloaded only in a certain path, it must be okay to sink such

> From: junbuml at codeaurora.org [mailto:junbuml at codeaurora.org] > Sent: Thursday, February 22, 2018 11:39 AM > > On 2018-02-22 11:14, gberry at codeaurora.org wrote: > > FROM: llvm-dev [mailto:llvm-dev-bounces at lists.llvm.org] ON BEHALF OF > > Jun Lim via llvm-dev > > SENT: Thursday, February 22, 2018 11:05 AM > > > > Hi All, > > > > I

I keep getting the 'memory.size' error message when I run a program I have been writing. It always it cannot allocate a vector of a certain size. I believe the error comes in the code fragement below where I have multiple arrays that could be taking up space. Does anyone know a good way around this? w1 <- outer(xk$xk1, data[,x1], function(y,z) abs(z-y)) w2 <- outer(xk$xk2,

Hi Quentin, > It sounds like you only need the earlyclobber description for the N, N > variant. > In other words, as long as you use different opcodes for widen-op NN and > widen-op WN, you model exactly what you want. > > What am I missing? > we are using different opcodes for widen-op NN and widen-op WN. My understanding is that not setting earlyclobber to the W, N

Dear R-Help List, I'm trying to simulate data from a conditional distribution, and haven't been able to modify my existing code to do so. I searched the archives, but didn't find any previous post that matched my question. n=10000 pop = data.frame(W1 = rbinom(n, 1, .2), W2 = runif(n, min = 3, max = 8), W3 = rnorm(n, mean=0, sd=2)) pop = transform(pop, A = rbinom(n, 1,

Dear R users, I used to "OPTIM" to minimize the obj. function below. Even though I used the true parameter values as initial values, the results are not very good. How could I improve my results? Any suggestion will be greatly appreciated. Regards, Kathryn Lord #------------------------------------------------------------------------------------------ x = c(0.35938587,

Dear R users, I used to "OPTIM" to minimize the obj. function below. Even though I used the true parameter values as initial values, the results are not very good. How could I improve my results? Any suggestion will be greatly appreciated. Regards, Kathryn Lord #------------------------------------------------------------------------------------------ x = c(0.35938587,

Dear R Help, I read through the archives pretty extensively before sending this email, as it seemed there were several threads on using predict with GLM. However, while my issue is similar to previous posts (cannot get it to predict using new data), none of the suggested fixes are working. The important bits of my code: set.seed(644) n0=200 #number of observations

Dear R-helpers, I am struggling with an optimization problem at the moment and decided to write the list looking for some help. I will use a very small example to explain what I would like to. Thanks in advance for your help. We would like to distribute resources from 4 warehouses to 3 destinations. The costs associated are as follows: Destination >From 1 2 3 Total

Dear R helpers I have following data stocks <- c("ABC", "DEF", "GHI", "JKL") prices_df <- data.frame(ABC = c(17,24,15,22,16,22,17,22,15,19),                                          DEF = c(22,28,20,20,28,26,29,18,24,21),                                           GHI = c(32,27,32,36,37,37,34,23,25,32),                                          

Hi, Sorry for the confusion. I would like to estimate a model wherein the marginals of z with respect to w1 and w2 are smooth functions of x and y. I have data on z, x, y, w1 and w2. so E[dz/dw1] = f(x,y) and E[dz/dw2] = g(x,y) and I would like to estimate f(x,y) and g(x,y) I suppose I could try to fit something more general using projection pursuit, but the nature of the problem suggests

I have a bit of an issue with rsync. I am using to keep directories in sync via another server for backup. Here is the server config [w1] path = /w1 comment = w1 web dir [w2] path = /w2 comment = w2 web dir Now on the client i run this command rsync -avv --delete --force domain.com::w1/ /w1/ It will NOT delete anything that is no on the server anymore.. for example on the server/client there

Hello, I'm trying to get percentiles (PERCENTRANK for excel users) by factor in the following data.frame: myExample <- data.frame(Ret=seq(-2, 2.5, by=0.5),PE=seq(10,19),Sectors=rep(c("Financial","Industrial"),5)) myExample <- na.omit(myExample) Thanks to Patrick I I managed to put together the following lines which does it for the "Ret" column: myecdf

Hi all developers, I'm changing compiler from gcc to llvm on a RISCV target now. but I found in some case the assembly code generated by llvm is much more than gcc. It cause my program's performance about 40% decrease. The flowing is a simple test code. It shows the problem. We can see than gcc prefer to use pointer to iterate the array, but llvm perfere to use index to iterate

Hi, I'm looking to use the grep function (or something else) to return the matched pattern as opposed to the whole element. For example: x <- c("pjhj24jhjhd") grep("[[:digit:]]{2}", x, value=T) returns "people", whereas I simply want "24". Any help would be appreciated, Thanks, Aidan [[alternative HTML version deleted]]