Parallel Amplifier 2011 from Intel

Optimize Performance and Scalability

Intel Parallel Amplifier 2011 makes it simple to quickly find multicore performance bottlenecks - without needing to know the processor architecture or assembly code. Intel Parallel Amplifier eliminates the guesswork and analyzes performance behavior in Windows* applications, providing quick access to scaling information for faster and improved decision making.

• Helps build applications that scale for multicore and manycore

• Delivers insight into performance hotspots and their calling sequences

• Ensures legacy serial application performance has not declined with threading

• Integrated with Microsoft* Visual Studio*

• Increases application headroom for richer feature sets and next-gen innovation

Hotspot analysis: Where is my application spending time?

Find the functions in your application that consume the most time. This is where you'll want to tune or add threading to make your program faster. Intel Parallel Amplifier also shows the stack, so you know how the function is being called. For functions with multiple calling sequences, this lets you see if one of the call stacks is hotter than the others.

Source view: See the results on your source
Source view shows you the exact location on your source. Just double-click on the function names in any of the analysis views to see the source.

Concurrency analysis: When are cores idle?
Like hotspot analysis, concurrency analysis finds the functions where you are spending the most time. But it also shows you how well you are utilizing multiple cores. Color indicates the core utilization while the function is running. A green bar means all the cores are working. A red bar means cores are underutilized. When red, add threading and get all the cores working for you. This helps you ensure application performance scales as more cores are added.

Locks and waits analysis: Where are the bad waits?
Waiting too long on a lock is a common source of performance problems. It's not bad to wait while all the cores are busy (green), but it is bad to wait when there are unused cores available (red).

Compare results: Quickly see what changed
This gives you a fast way to check progress when tuning and also makes for handy regression analysis.

Overview video highlighting the advantages of using Parallel Amplifier in Microsoft Visual Studio* to tune threaded applications for best performance.

Waiting too long on a lock is a very common parallel performance bottleneck. Amplifier shows you where you are waiting and how many cores are idle. Just what you need to know to go faster.

When tuning for multicore, you need to know where you are spending a lot of time and not using all the cores. This is an opportunity to add parallelism and improve scaling.

Waiting too long on a lock is a very common parallel performance bottleneck. Amplifier shows you where you are waiting and how many cores are idle. Just what you need to know to go faster.

Did your performance fix work? Compare results highlights the changes. This is perfect for performance regressions because it shows which function caused the slow down.

Start, pause and restart the collection of data to focus your tuning effort on a specific performance issue.