- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I installed the Intel/Python 2.7 on Mac OS X 10.10.5 and ran a script with a lot of linear algebra (svds) and was surprised to find that it ran half as fast as the Anaconda version on the same machine.
That's unexpected. Is something obviously wrong?
Link Copied
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Jamie,
Thank you for reporting this! Answering your question, no this is not expected and surprises me too. Essentially both Anaconda* and Intel Distribution for Python take benefit from the Intel MKL. In theory the performance should be the same.
Can you provide extra details about versions of Anaconda* and Intel Python that you used?
Would it be possible to provide the test case so we can reproduce exactly what you get?
Also what hardware do you run on?
Sergey
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Jamie,
In earlier releases of Intel Python, I had a similar issue when running some tests on macOS. However, it happened on early beta versions. Are you sure you are working with the latest version?
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page