That's an interesting techinque, i've never actually seen that processing-instruction command.
I use this approach by...

Thanks Jeff for pointing that out.

Yes the version in my gist is broken. My mistake I've since updated it in my database project but not in this actual test script....

I 100% agree that without a guarantee of ordering, or an explicit item ordinal the function is not complete.

However, I want to reiterate that the function does perform very...

I bet it does, but I'm trying to just measure how fast each function is and well I can say that splitstring is fast. I let the test finish at...

This was run with default settings on the box. So probably 5 for the cost and whatever the default is on dop.

I decided to try to change the number of...

This test script is basically the one in your article, just modified a bit so that it uses dynamic sql to run the cases and requires less effort to add...

Jeff,

I 100% agree with you that his test script wasn't the best. His number table splitter you've already dressed down, and that xml splitter doesn't do any lob caching techniques...

Yeah, I just ran it sorry I lied this box is running sql server 2014 developer edition but the compatibility mode is set to 2012 to match my production environment.

...

Sadly, I don't have access to a box with 2016 at the moment. But Here is a quick summary of the effect of the traceflag on my test script.

without the...

Also sqlperformance article seems to suggest the thing performs fairly well...

http://sqlperformance.com/2016/03/t-sql-queries/string-split

Obviously, his t-sql and xml based splitters are poor choices.

see this post from Paul White :

http://www.sqlservercentral.com/Forums/FindPost1570980.aspx

Paul White (5/14/2014)

---

Is everyone running Jeff's original/updated test rig from the article to get these results?

If so, you might not be testing...

Can you post your test script? I suspect that since delimited8k did so well and string_split did so poorly, this might be related to that optimization that can be disabled...

I'd recommend a filestream over this approach. Filestream is natively supported, and offers asynchronize I/O, which means inserting your blobs will be faster, as will retrieval. Filestream is ideal for...

I modified his test script. And posted it in that same post where I criticized the point. When I didn't modify it my times were much much worse on sql...

The flag just turns off table spool, so that would probably have a very noticeable effect on the function.

When I get some time I might try some of the techniques...