RSstudio 1.1 perfomance issues

I just updated to RStudio 1.1 and it's unacceptably slow.

The View(data) pane is incredibly slow. When scrolling horizontally the variable names don't follow and the view often freezes with or without a "Loading" message. 309 cases and 148 variables, which my older version could handle just fine.

Also, the prompt is slow and unresponsive. I takes a few milliseconds before the letters that I write appear.

EDIT: The prompt is actually seems to be slow only when the View(data) pane is open.

Please fix these very annoying perfomance issues. Running on Mac OS Sierra.

4 Likes

Sorry to hear you're having trouble! Can you provide some more information as to what you're seeing?

  • Is R, RStudio, or your installed R packages installed on a network drive by any chance? (We've received some reports on Windows that seem to indicate this occurs most often when RStudio is installed on a network drive)

  • Is the UI otherwise responsive? (Can you type in the R console while this hang is occurring, or does the entire IDE appear to lock up?) I understand that the R session is likely locked up so you can't actually submit any commands, but just want to confirm whether you're able to perform other actions.

Anything else you can share about your environment / the environment in which RStudio is installed will be helpful, in case there's anything unique.

@brjann I can confirm that Data Viewer is also slow for me using data of that size.

x <- matrix(rnorm(309 * 148), nrow = 309)
View(as.data.frame(x))

@kevinushey I am running RStudio 1.1.383 on Ubuntu 17.04. Similar to @brjann, I can type into the R console while the Viewer is loading, but the text does not appear in the R console until after the Viewer loads.

Are you using R / RStudio on a networked drive or anything similar here? What is the output of Sys.getenv("HOME")?

@kevinushey I am running R / RStudio on a vanilla installation of desktop Ubuntu.

> Sys.getenv("HOME")
[1] "/home/jdb-work"

I do have some external directories on my university's HPC cluster mounted on my local computer, but unmounting them (i.e. sudo umount -a) does not affect the observed behavior.

No networked drives attached to my computer. And my previously installed version of RStudio did not show this behavior.

Much the same here. I am running Win7 64bit, R v3.4.2 64bit and RStudio 1.1.383 on 2 separate PCS.

View() is hanging on 1.2MB of gene/cell counts.
Memory usage looks OK - no leaks evident. Eventually I have to abort the R session.

I have rolled back to RStudio 1.0.153 and the problem has gone away.
Both PCS have the same issue and both resolved the same.

So I am suggesting not to upgrade to RStudio 1.1.383
Stay with 1.0.153 for now.

Urgent fix on this is needed.
Thanks.

2 Likes

I am experiencing the same issue, also immediately after the 1.1.383 update (2017 iMac 27", 40GB RAM, 4.2Ghz CPU, 1TB SSD). I have also noticed that the typing lags on the 27" iMac, but not on my laptop... what's up with this?

Agreed, I deleted the 1.1.383 and reinstalled the older version.

1 Like

I'm having the same problem, Windows 10. Seems to be only with wider data. Long data (even over 1M rows) loads fine but much smaller but wide data takes a very long time to load into a view() tab and occasionally crashes Rstudio all together.

I am having this problem as well using 1.1.383 on Linux Mint. Has there been any progress on a solution?

I too am having similar problems, although in my case it does not require the view data pane to be open (i.e., I can observe significant lags between entering an R expression such as "1+2" at the console and seeing the result, even if only editor tabs are open).

I am running everything off a local (SSD) drive. RStudio is Version 1.1.383 on Windows 10 with plenty of memory.

As others have commented, the performance issue is severe enough that I think I will downgrade to an earlier version.

EDIT: One more detail, I am using the Fira Code font with ligatures enabled. I just disabled ligatures and performance seems to have improved somewhat.

Also seeing a ridiculously slow load time for viewing and scrolling in large data frames. RStudio 1.1.447. Older versions did perform vastly better....why hasn't this been fixed yet???

Any update on when this will be fixed? I just downloaded 1.1.453 and it's still a pain to use View on large datasets.

Is the situation improved at all in the daily builds? https://dailies.rstudio.com

1 Like

no it isn't improved.
To demonstrate the clunky feel of view(), I recorded two clips for viewing the same matrix in 1.0.153 and 1.2.687:
code:

test.mat <- matrix(data=runif(2000000,0,1),nrow=20000,ncol=100)

1.0.153
https://imgur.com/a/FbEnKle

1.2.687
https://imgur.com/a/ZMlsXdC

*click image to view the clip

I'm in ubuntu 16.04

Just an update on this:

As of the end of October, I'm still running into this issue with 1.1.463. It makes the new version of RStudio unusable for work, which is frustrating - and have to use 1.0.153, where wide data can still be viewed fine.

System spec:
R 3.5.1
Ubuntu 18.04 LTS
500 GB SSD (no networked storage)
128 GB RAM

4 posts were split to a new topic: Performance Issues with View() for large data frames RStudio Preview Version 1.2