The Performance War -- Win it 5% at a time

re: The Performance War -- Win it 5% at a time

BlackTigerX — Tue, 18 Oct 2005 01:51:20 GMT

great post, I couldn't agree more

re: The Performance War -- Win it 5% at a time

Cheong — Tue, 18 Oct 2005 05:51:00 GMT

Sure. But managements simply can't understand.

They want good improvement figures. If you've persuding your boss to give you time for profiling, they'll surly be much happier to see a 50% improvement than a 5% one.

re: The Performance War -- Win it 5% at a time

Andrew D. — Sat, 22 Oct 2005 02:05:57 GMT

That is a great post! Very interesting! I hope I will some day be able to perform both the "Superhero" type And the "real" performance work, as appropriate/necessary. ;)

Thanks.

re: The Performance War -- Win it 5% at a time

oastorga — Tue, 25 Oct 2005 03:42:48 GMT

So to make managers happy, you create a bad desing and then later you fix it. Maybe you'll get even 90% performance improvement. :-) The truth is that educating managers is also part of the process. I think they can understand (it's difficult, I know), maybe we are using the wrong approach. There are some obstinate individuals out there I must admit. How about this, let's try to make them think is their idea.... keep your spirits up...

re: The Performance War -- Win it 5% at a time

Rüdiger Klaehn — Fri, 28 Oct 2005 11:39:40 GMT

I don't know. I would love to have a project where I have to fight for each 5% of performance.

But mostly I just have to explain to people what a StringBuilder is for, and find O(N^2) parts and replace them with O(N*log(N)) algorithms.

The most complex part is to get the people to recognize that their design is broken and to make them change it without insulting them.

re: The Performance War -- Win it 5% at a time

Norman Diamond — Tue, 01 Nov 2005 04:05:36 GMT

Here's something where 5% won't cut it: Windows Vista beta 1 checked build.

I'm installing it on a Pentium III 600 MHz with 320 MB of RAM and 17 GB of available hard disk space, because the machine is available for use as a crash box. mscorsvw.exe is taking more than 90% of the CPU time, so we know that RAM isn't a problem. The total commit charge is around 580 MB but paging isn't a problem. If the machine were doing a lot of paging then the program would spend most of its time waiting and CPU usage would be down around 3%. Also if the machine accessed the hard disk a lot then I'd see the activity icon in the LCD a lot more often than I'm seeing it.

So mscorsvw.exe is really doing stuff during this install, but what? Let's assume there are lots of cache misses. The memory bus is probably 100 MHz and the RAM probably responds in 2 clock cycles. So if the CPU is spending most of its time waiting for RAM to respond, then there have probably been 150 trillion RAM accesses since the install started.

Want to compute that or should I give a spoiler? Yes I do wonder how many months it's going to take for Vista beta 1 checked build to finish installing itself. "Do not restart your PC during this time" for how many months?

And after it finishes, will it be usable? After some number of weeks without being activated, it won't allow logins, right? Without logging in, it won't be possible to configure the network card to use a fixed IP address[*], so I won't be able to activate it over the internet. And guess which company is refusing to activate Vista betas by phone.

[* Of course in the retail build, which installed successfully, it's also hard enough to configure the network card to use a fixed IP address. The settings dialogs go through the motions but the OS keeps trying to get an address from DHCP.]

re: The Performance War -- Win it 5% at a time

Travis Owens — Wed, 02 Nov 2005 23:44:47 GMT

Of course if that's 5% bursts across multiple changes and there's a 3 year pause between releases, 50% is possible.

I mean look at some of the XML bragging in .Net 2.0 where somebody here on MSDN blogging showed improvements as large as 400%.

And just recently I came across a chart showing many boosts in the .Net Compact Framework v 2.0, one improvement was a whopping 800%.
http://www.zintel.net/Blog/NETCFPerfProgress.jpg

re: The Performance War -- Win it 5% at a time

Travis Owens — Wed, 02 Nov 2005 23:48:00 GMT

Here's one site showing the massive XML speed improvements with XML in .Net 2.0
http://msdn.microsoft.com/msdnmag/issues/04/02/XMLFiles/default.aspx

Of course looking back at old code, and having more optimized code, it's easy to see myself calling that old code "a problem".

The Performance War: Using counts to help navigate flat performance reports

Rico Mariani's Performance Tidbits — Wed, 19 Apr 2006 04:07:09 GMT

A while ago I wrote about how you often win the performance war 5% at a time.&nbsp; The theme of that...

Paint.NET and Startup Performance

Rick Brewster's blog — Thu, 20 Apr 2006 21:39:31 GMT

I remember Raymond Chen gave a talk at the last PDC and had this fun quote:

"One of the questions...

Paint.NET and Performance, part 2

Rick Brewster's blog — Tue, 25 Apr 2006 02:46:40 GMT

Last
week I described an optimization that helps Paint.NET's startup performance by
avoiding our...

Performance Quiz #8 -- The problems with parsing -- Part 3

Rico Mariani's Performance Tidbits — Wed, 24 Jan 2007 05:26:51 GMT

The thing about performance work is that it's very easy to be fooled into looking into the wrong areas.