search for: 60x

Since the contract of ls() is to sort, there is nothing wrong with programmers depending on it. And there are many functions that could be made 60X faster, but is it worth it? But I did notice that as.list.environment has a sorted=FALSE argument already, so I guess identical(names(x), names(as.list(x))) could be made to be TRUE, assuming the order is at least persistent, if undefined, so that is a nice property. I guess I'm OK it with....

...throughout the > > methods package. These functions are currently among the major > > contributors to execution time in package loading. > > > This two-line addition to attrib.c gives a significant speedup for > > listing an environment's names/keys (2-60X depending on the 'sorted' > > argument). It also simplifies the environment API by making it more > > like the other basic types. We already have $ and [[ after all. > > > Rather than sprinkling sorted=FALSE throughout the methods and base > > code...

...;> Michael Lawrence <lawrence.michael at gene.com> >>>>> on Tue, 27 Jan 2015 07:59:59 -0800 writes: > Since the contract of ls() is to sort, there is nothing wrong with > programmers depending on it. And there are many functions that could be > made 60X faster, but is it worth it? But I did notice that > as.list.environment has a sorted=FALSE argument already, so I guess > identical(names(x), names(as.list(x))) could be made to be TRUE, assuming > the order is at least persistent, if undefined, so that is a nice property....

...still get it to go fast. As for precedent, we have several Python scripts already shipping with LLVM, including lit, and Python is listed in the requirements, so I imagine it's fine from that perspective. > > Having said that, the C++ version is quite a lot faster. I'm seeing about a 60x difference in how long it takes to generate HTML. This is with the C++ version doing a bit more work - specifically, it correctly encodes characters that have special meaning to HTML. I've been testing this on a build of the Chromium web browser with reports for the inline pass. On my machine,...

Hi, I''ve been experimenting with Puppet for a few days now, and I really like it. But! I''m having real CPU usage problems. Puppet is still happily eating away 100% CPU for almost an hour at a time, with no apparent things happening. (puppetd -tv --trace --debug, but nothing appearing in the console during this.) I believe the following resource is to blame: file {

...wrence.michael at gene.com> > >>>>> on Tue, 27 Jan 2015 07:59:59 -0800 writes: > > > Since the contract of ls() is to sort, there is nothing wrong with > > programmers depending on it. And there are many functions that could > be > > made 60X faster, but is it worth it? But I did notice that > > as.list.environment has a sorted=FALSE argument already, so I guess > > identical(names(x), names(as.list(x))) could be made to be TRUE, > assuming > > the order is at least persistent, if undefined, so that is a...

...ly in performance-critical code like base/R/namespace.R and throughout the methods package. These functions are currently among the major contributors to execution time in package loading. This two-line addition to attrib.c gives a significant speedup for listing an environment's names/keys (2-60X depending on the 'sorted' argument). It also simplifies the environment API by making it more like the other basic types. We already have $ and [[ after all. Rather than sprinkling sorted=FALSE throughout the methods and base code, let's use names. Would you be open to this change? I...

That's compared to the implementation with the Python parser. So if the libYAML parser is 6x the speed of that, the C++ version would be about 10x the speed of the implementation with libYAML, instead of 60x. On Tue, Nov 15, 2016 at 10:50 AM, Adam Nemet <anemet at apple.com> wrote: > > On Nov 15, 2016, at 10:33 AM, Bob Haarman <inglorion at google.com> wrote: > > Thanks for your comments, everyone! I'll try to answer the questions > people have asked. First, let me say...

..., Bob Haarman <inglorion at google.com<mailto:inglorion at google.com>> wrote: That's compared to the implementation with the Python parser. So if the libYAML parser is 6x the speed of that, the C++ version would be about 10x the speed of the implementation with libYAML, instead of 60x. On Tue, Nov 15, 2016 at 10:50 AM, Adam Nemet <anemet at apple.com<mailto:anemet at apple.com>> wrote: On Nov 15, 2016, at 10:33 AM, Bob Haarman <inglorion at google.com<mailto:inglorion at google.com>> wrote: Thanks for your comments, everyone! I'll try to answer the...

...using a DPU this is around 40us baseline without going to a remote server (4k, QD=1). This patch queues requests when the Virtio queue is full, and when a completed request is taken off, immediately fills it back up with queued requests. This reduces the 99.9th percentile latencies in our tests by 60x and slightly increases the overall throughput, when using a queue depth 2x the size of the Virtio queue size, with a DPU-powered virtio-fs device. Signed-off-by: Peter-Jan Gootzen <peter-jan at gootzen.net> --- fs/fuse/virtio_fs.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletion...

Hi, table1D() below can be up to 60x faster than base::table() for the 1D case. Here are the detailed speedups compared to base::table(). o With a logical vector of length 5M: 11x faster (or more if 'useNA="always"') o With factor/integer/numeric/character of length 1M...

I think that the "sorted" and "all.names" arguments are really only appropriate for pretty printing to the screen. I think it is a bit unfortunate that environments have a names accessor that is 60X slower than all the other types. This is likely due to the history of environments, which were originally just for behind-the-scenes tasks. Now that users can use environments as hashes, we really need something like a "keys" function. We don't want programmers depending on the sorte...

...using a DPU this is around 40us baseline without going to a remote server (4k, QD=1). This patch queues requests when the Virtio queue is full, and when a completed request is taken off, immediately fills it back up with queued requests. This reduces the 99.9th percentile latencies in our tests by 60x and slightly increases the overall throughput, when using a queue depth 2x the size of the Virtio queue size, with a DPU-powered virtio-fs device. Signed-off-by: Peter-Jan Gootzen <peter-jan at gootzen.net> --- V4: Removed return value on error changes to simplify patch, that should be chang...

...ch queues requests when the Virtio queue is full, > > >> and when a completed request is taken off, immediately fills > > >> it back up with queued requests. > > >> > > >> This reduces the 99.9th percentile latencies in our tests by > > >> 60x and slightly increases the overall throughput, when using a > > >> queue depth 2x the size of the Virtio queue size, with a > > >> DPU-powered virtio-fs device. > > >> > > >> Signed-off-by: Peter-Jan Gootzen <peter-jan at gootzen.net> > >...

...Haarman < <mailto:inglorion at google.com> inglorion at google.com> wrote: That's compared to the implementation with the Python parser. So if the libYAML parser is 6x the speed of that, the C++ version would be about 10x the speed of the implementation with libYAML, instead of 60x. On Tue, Nov 15, 2016 at 10:50 AM, Adam Nemet < <mailto:anemet at apple.com> anemet at apple.com> wrote: On Nov 15, 2016, at 10:33 AM, Bob Haarman < <mailto:inglorion at google.com> inglorion at google.com> wrote: Thanks for your comments, everyone! I'll try...

...). > >> This patch queues requests when the Virtio queue is full, > >> and when a completed request is taken off, immediately fills > >> it back up with queued requests. > >> > >> This reduces the 99.9th percentile latencies in our tests by > >> 60x and slightly increases the overall throughput, when using a > >> queue depth 2x the size of the Virtio queue size, with a > >> DPU-powered virtio-fs device. > >> > >> Signed-off-by: Peter-Jan Gootzen <peter-jan at gootzen.net> > >> --- > >&gt...

...gt;Within -O3 the overhead is still far away from the 60 slowdown we have >seen on the other test case. A 2x slowdown for a kernel that we fully >analyse and code generate is not optimal, but probably not an issue. The >question is if the increasing depth of the loop nest can yield this 60x >slowdown. This is because the depth of the loop nest is only two. If we increase the depth of the loop nest, then the compile-time would significantly increase. I have posted results for the original code which contains 6 nested loops on http://llvm.org/bugs/show_bug.cgi?id=16843#c5 In fac...

...using a DPU this is around 40us baseline without going to a remote server (4k, QD=1). This patch queues requests when the Virtio queue is full, and when a completed request is taken off, immediately fills it back up with queued requests. This reduces the 99.9th percentile latencies in our tests by 60x and slightly increases the overall throughput, when using a queue depth 2x the size of the Virtio queue size, with a DPU-powered virtio-fs device. Signed-off-by: Peter-Jan Gootzen <peter-jan at gootzen.net> --- V1 -> V2: Not scheduling dispatch work anymore when not needed and changed del...

...requests when the Virtio queue is full, >>>>> and when a completed request is taken off, immediately fills >>>>> it back up with queued requests. >>>>> >>>>> This reduces the 99.9th percentile latencies in our tests by >>>>> 60x and slightly increases the overall throughput, when using a >>>>> queue depth 2x the size of the Virtio queue size, with a >>>>> DPU-powered virtio-fs device. >>>>> >>>>> Signed-off-by: Peter-Jan Gootzen <peter-jan at gootzen.net> &g...

...sys 0m0.004s Within -O3 the overhead is still far away from the 60 slowdown we have seen on the other test case. A 2x slowdown for a kernel that we fully analyse and code generate is not optimal, but probably not an issue. The question is if the increasing depth of the loop nest can yield this 60x slowdown. [..] > These store instructions significantly complicate the "polly-dependence" pass, and thus leading to high compile-time overhead. What do you mean by "complicated the 'polly-dependence' pass"? Without -polly-prepare the scop that is detected is a lo...