On Fri, 23 Dec 2022 22:28:34 GMT, Markus KARG wrote:
> I/O had always been much slower than CPU and memory access, and thanks to
> physical constraints, always will be.
> While CPUs can get shrinked more and more, and can hold more and more memory
> cache on or nearby a CPU core, the distance
On Fri, 23 Dec 2022 22:28:34 GMT, Markus KARG wrote:
> I/O had always been much slower than CPU and memory access, and thanks to
> physical constraints, always will be.
> While CPUs can get shrinked more and more, and can hold more and more memory
> cache on or nearby a CPU core, the distance
On Mon, 2 Jan 2023 10:03:02 GMT, Peter Levart wrote:
> Here, the benefit of increasing buffer from 8k to 16k gets from about 10%
> (doing IO) up to 20% (reading from cache) increase in performance.
I think 10% to 20% is good enough as an argument to go with 16k instead of 8k.
-
On Fri, 23 Dec 2022 22:28:34 GMT, Markus KARG wrote:
> I/O had always been much slower than CPU and memory access, and thanks to
> physical constraints, always will be.
> While CPUs can get shrinked more and more, and can hold more and more memory
> cache on or nearby a CPU core, the distance
On Fri, 23 Dec 2022 22:28:34 GMT, Markus KARG wrote:
> I/O had always been much slower than CPU and memory access, and thanks to
> physical constraints, always will be.
> While CPUs can get shrinked more and more, and can hold more and more memory
> cache on or nearby a CPU core, the distance
On Tue, 27 Dec 2022 14:55:31 GMT, Peter Levart wrote:
> Hello Markus! Could you show the JMH code that produced the benchmark results?
The following lines make use of a custom method I have added to `InputStream`
in a custom build of JDK 21, so JMH can control the size of the buffer. The
test
On Fri, 23 Dec 2022 22:28:34 GMT, Markus KARG wrote:
> I/O had always been much slower than CPU and memory access, and thanks to
> physical constraints, always will be.
> While CPUs can get shrinked more and more, and can hold more and more memory
> cache on or nearby a CPU core, the distance
I'm also looking forward to io_uring support on Linux kernels which support
it and the Windows equivalent :-) The first one should come with one of the
project Loom updates sometime.
Kind regards
Johannes
Markus KARG schrieb am Sa., 24. Dez. 2022, 10:11:
> I/O had always been much slower than
I/O had always been much slower than CPU and memory access, and thanks to
physical constraints, always will be.
While CPUs can get shrinked more and more, and can hold more and more memory
cache on or nearby a CPU core, the distance between CPU core and I/O device
cannot get reduced much: It