Stream

Eric DeFazio

Presentation Videos  - 
 
Richard Sites / Danny Luu great discussion explanation of the Bifurcation of "Servers" v "Desktops". How handling data, and locks lead to tail latency in the Cloud (Google) and how to go about tracking down tail latency (the 1%) ( And why profilers don't work)
3
1
Eric DeFazio's profile photoJames Stansell (Jester of Fabulous)'s profile photo
 
Here's a Link to Dan Luu's article related to the post about Profilers:
https://danluu.com/perf-tracing/
slides:
http://www.pdl.cmu.edu/SDI/2015/slides/DatacenterComputers.pdf
Add a comment...

sorin cristea

General Notices  - 
Now that we have reviewed the core concepts behind GC algorithms, let us move to the specific implementations one can find inside the JVM. An important aspect to recognize first is the fact that, for most JVMs out there, two different GC algorithms are needed – one to clean the Young Generation an...
3
1
Andrey Loskutov's profile photo
Add a comment...

Eric DeFazio

Presentation Videos  - 
 
Eye Opening Presentation about high performance "Databases"(NoSql, et.) Riak, Mongo, Cassandra, ElasticSearch, ...
https://youtu.be/dE3KT7hHkKY
(Is the Engineering really achieving what the marketing claims?)
7
2
Eric DeFazio's profile photoChristoph Engelbert's profile photoJames Stansell (Jester of Fabulous)'s profile photo
 
"Now for the blood" (discuss the Jepsen results of popular databases" https://youtu.be/dE3KT7hHkKY?t=20m42s
Add a comment...

Jean-Philippe BEMPEL

Expert Programming  - 
If you are a regular reader of my blog, you may have noticed that I am (ab)using (of) the PrintAssembly options from the JVM to examine the code generated by the JIT compiler. It helps me a lot to understand how my code is ex...
5
1
Joachim Dorn's profile photo
Add a comment...

Per-Åke Minborg

Expert Programming  - 
 
The JVM can re-capture escaped objects. Read more on Escape Analysis in combination with inlining and why that is good.


7
1
ankit bhavsar's profile photo
Add a comment...

Nitsan Wakart
moderator

Expert Programming  - 
 
JVM Safepoints: meaning, overheads, side effects and PINEAPPLES
I've been giving a couple of talks in the last year about profiling and about the JVM runtime/execution and in both I found myself coming up against the topic of Safepoints. Most people are blissfully ignorant of the existenc...
6
1
Joachim Dorn's profile photo
Add a comment...

William Louth

Presentation Videos  - 
 
Spot Checking Apache Cassandra 3.0 Performance Hotspots
This article attempts to verify the accuracy of a previous Apache Cassandra hotspot analysis captured after many benchmarks iterations in which that bytecode instrumentation is adapted, both online and offline, based on what is learnt from the performance model exported after each stress test.
http://www.autoletics.com/posts/spot-checking-apache-cassandra-3-0-performance-hotspots#java #profiling #satoris #cassandra
This article is a follow-up to a previous article, Advance Adaptive Profiling of Apache Cassandra 3.0, that attempts to verify the accuracy of the hotspot analysis captured after many benchmarks iterations in which that bytecode instrumentation is adapted, both online and offline, based on what ...
4
Add a comment...

William Louth

Presentation Videos  - 
 
Advance Adaptive Profiling of Apache Cassandra 3.0
With Apache Cassandra 3.0 just released it is time to check out whether previous performance hotspot analysis of the codebase, using Satoris, still stands today.
http://www.autoletics.com/posts/adaptive-profiling-benchmarking-of-apache-cassandra-3-0
With Apache Cassandra 3.0 just released it is time to check out whether previous performance hotspot analysis of the codebase, using Satoris, still stands today. But before jumping into the hotspot analysis it is important to first understand how Satoris uniquely approaches profiling and ...
2
Add a comment...

William Louth

Presentation Videos  - 
 
 
Behavioral Variance - The Root Cause of Software Performance Variance
In this screen recording we go back in time, replaying an episodic software memory of Apache Cassandra, to determine whether behavioral variance do really exit in the simplest of method executions using Stenos and Satoris metering extensions.
1
Vachagan Balayan's profile photoWilliam Louth's profile photo
2 comments
 
iTerm2
Add a comment...

Nitsan Wakart
moderator

Expert Programming  - 
 
 
Another great post by Mark Price (from LMAX) about reducing scheduling related jitter on Linux.
In my last post, I demonstrated some techniques for reducing system jitter in low-latency systems. Those techniques brought worst-case inter-thread latency from 11 millliseconds down to 15 microseconds. Next steps I left...
6
2
fernando gonzalez's profile photoRadu Toader's profile photo
Add a comment...

About this community

This community is about learning Java beyond the basics and advanced material. It about thinking about which Java is doing in milli-seconds and at the micro-second level. It is not a general Java programming help line for questions which can be easily answered using google search. You must subscribe under your name, not as a company or group account.

Nitsan Wakart
moderator

Expert Programming  - 
 
Great post from Shipilev demonstrating performance engineering methodology and tools usage(Solaris Studio/JMH perfasm) to look into toArray(new Foo[0]) vs. toArray(new Foo[size])
http://shipilev.net/blog/2016/arrays-wisdom-ancients/
The Java Language and JDK Class Library have two distinct, yet connected, ways to group elements: arrays and Collections. There are pros and cons for using either one, so both are prevalent in real programs. To aid conversion between the two, there are standard methods to make a reference array ...
10
Add a comment...

Per-Åke Minborg

Performance Libraries  - 
1
Per-Åke Minborg's profile photoMarkus TET Schulz's profile photo
3 comments
 
I know that pattern from genericdao, but not in your perspective... 
Add a comment...
 
Agner Fog (3) great new posts on his blog:

Moore's law hits the roof:
http://www.agner.org/optimize/blog/read.php?i=417

Proposal for ideal extensible instruction set:
http://www.agner.org/optimize/blog/read.php?i=421

Test Results for Broadwell and Skylake
http://www.agner.org/optimize/blog/read.php?i=415


Author: Agner, Date: 2015-12-26 09:02. Through the last 40 years we have seen the speed of computers growing exponentially. Today's computers have a clock frequency a thousand times higher than the first personal computers in the early 1980's. The amount of RAM memory on a computer has increased ...
5
1
William Louth's profile photoEric DeFazio's profile photoChris Ainsley's profile photo
2 comments
 
Yes, Agner is pretty blunt about how our SOFTWARE needs to improve:
"The biggest potential for improved performance is now, as I see it, on the software side..."
Add a comment...

Attila-Mihaly Balazs
moderator

Presentation Videos  - 
 
A good technical discussion about GC with some interesting comparison / contrasting between Go and Java: http://www.infoq.com/interviews/hudson-go-gc
6
2
Thomas Junk's profile photoBrian Oxley's profile photo
Add a comment...

Daniel Shaya

Expert Programming  - 
 
Following on from Per's post I was a bit confised about when escape analysis is actually effected.  I'm left with some unanswered questions so please feel free to comment.
3
James Stansell (Jester of Fabulous)'s profile photo
 
Perplexing observations. I wondered why some numbers in Per's post were the way they were (but not enough to dig into it myself.)

Add a comment...

William Louth

Expert Programming  - 
 
Performance Measurement Budgeting of Apache Cassandra 3.0
In this post, performance measurement budgetary control is applied alongside trace sampling to further reduce the impact of nested call (method) measurements on the accuracy of the measure caller chain.

http://www.autoletics.com/posts/performance-measurement-budgeting-of-apache-cassandra-3-0
This is the third post in a series looking at the performance analysis of Apache Cassandra. In the first post, Advance Adaptive Profiling of Apache Cassandra 3.0, adaptive profiling over many benchmark iterations was introduced. The post offered a comprehensive performance model for an entire ...
1
Add a comment...

Per-Åke Minborg

Expert Programming  - 
 
Do Not Let Your Objects Escape: Escape Analysis and how that can improve performance and coding style. This post explains the fundamentals of EA and also features a concrete example where EA is used in Java 8.


6
1
Emil Forslund's profile photo
Add a comment...

Ethan Millar

General Notices  - 
 
Latest features added to the new JavaFX include accessibility support, custom and standard dialogs, support for text formatting and a spinner control. Let’s learn about these features in more detail to enhance java development experience.

#JavaApplicationDevelopment #Java #JavaSE #JavaDevelopers #JavaFX
Latest features added to the new JavaFX include accessibility support, custom and standard dialogs, support for text formatting and a spinner control. Let’s learn about these features in more detail ...
2
Обучение по English Нива-A1 A2- B1 B2's profile photo
Add a comment...

William Louth

Expert Programming  - 
 
 
Beyond Reactive - QoS & Adaptive Control for Software
Slides from my µCon 2015 (London) talk that explores various approaches to service request level resource management starting with basic back pressure and moving on to advance reservation based quality of service (QoS) and adaptive control execution (ACE) valves.

https://speakerdeck.com/autoletics/beyond-reactive-qos-and-adaptive-control-for-software
Slides from a µCon 2015 (London) talk that explores various approaches to service request level resource management starting with basic back pressure and moving on to advance reservation based quality of service (QoS) and adaptive control execution (ACE) valves.
1
1
William Louth's profile photoPhil Crow's profile photoJames Stansell (Jester of Fabulous)'s profile photo
5 comments
 
this presentation is not almost useless without sound :) nice graphics. thanks mate
Add a comment...

Eric DeFazio

Expert Programming  - 
 
Stupid mistakes people Make when they don't use JMH (Use Case)

Unsafe users; I noticed a strange issue in Unsafe.allocateMemory, (specifically the read performance of memory allocated with Unsafe.allocateMemory(...) is dependent on the primitive TYPE (int or long) (not its value) you pass to the allocateMemory method.  Tried (2) Sun JVMs and (1) OpenJDK VM (Zulu) and got the same results... I realize this isn't a Q&A board, but seems like (if I'm right)  people here could be affected. 

I would like either some corroboration or correction on my assertions  on other platforms I'm Windows 7 (more details and code on gitHub) 
Unsafe.allocateMemory(...) performs 2x BETTER when we pass in an int verses a long (of the same value)
2
Eric DeFazio's profile photo
11 comments
 
Nitsan's (Video) "The Illusion of Execution" describing JVM execution whole video is great, relevant to this (non-counted Loops and safe points) starts at about ~12:15:
https://vimeo.com/138863976
Add a comment...