In More on Performance Improvements in Excel 2010, Excel Program Manager Joseph Chirilov writes about improved performance in Excel 2010 over 2007.
According to Chirilov, Excel 2010 has made improvements in the following areas:
- Large Data Set Performance
- Calculation Performance
- Multi-core functionality
Here's the chart he posted showing how Excel 2010 stacks up against its predecessor:
For the record, I don't have a copy of Excel 2010. My comments are based solely on the blog post.
While I'm glad that MS is working on improving performance, the charts presented in the blog post seem to be nothing more than a marketing gimmick carefully crafted to make it appear as if Excel 2010 is better than (one of) its predecessor. What seems to be ignored is that overall, performance has degraded since Excel 2003. I think it's dishonest to ignore this fact when claiming speed improvements.
Am I the only one amused by the strawman-like comparisons between Excel 2007 and 2010? The charts (especially the bar chart) are really amusing. I ran my own informal test using the same conditions: Two workbooks, one with 6 million cells filled using RAND() (3m on sheet1, 3m on sheet2), and another with 12 million cells (6m on sheet1, 6m on sheet2). I don't know how MS measured their test, but mine were measured by clicking a mouse button and then looking at a clock.
In Excel 2003 (32-bit), running on Windows XP Pro with 2 GB RAM, opening the first workbook (6m cells) took 5 seconds, while the second workbook (12m cells) took roughly 10 seconds to open. This is anywhere from three to four times faster, depending on which new version of Excel you're comparing against.
Another way of looking at it would be:
Personally I'm put off by the false speed gains, and I feel bad for anyone who looks at that blog post and comes away with the idea that Excel 2010 is fast. Or maybe I'm just not getting it, in that case, enlighten me.
About JP
5 Comments:
Why not compare to 97, XP, 2000?
This post can be held to the same bias as the Excel team.
The fact that 2010 is faster than 2007 means a lot to my business as we currently use 2007.
Thanks for sharing, Jeremy. That was my point. You can't just pick and choose how you want to compare Excel 2010 and ignore the fact that a lot of users aren't happy with the new UI.
I ran your 6m test and got inconsistent results using my "accurate" finger tapping timing method. First pass with 2003 I got 18 seconds, and then levelled out at around 6 seconds. With 2007, I was at around 10 seconds each pass.
I also saved the file as an xlsb format (file size reduced from 190k to 82k). It didn't seem to help 2007.
Yet I've been using that format at work and found it has really cut down on my file opening and saving times, especially for files stored on the network server. I'm assuming that is because this is a more compressed format resulting in much less network traffic. For my larger spreadsheets, I've found the file size reduced as much as 90% using the xlsb format vs xls (from 2Mb to 200k). That reduction didn't show up in this test, so that may be why I didn't see the improvement.
Bottom line, in my environment, using the binary file format, I'm seeing definite benefits in file opening speed with XL2007.
The Microsoft blog posts claiming speed improvements are the biggest joke I have ever seen! Excel 2007 degraded performance from Excel 2003 by a factor of about 20 in my experience, for anything involving calculating, plotting, etc. (both the spreadsheet itself and Visual Basic). I have seen many other posts by others consistent with that observation. Now they brag about a 40% improvement from Excel 2007??? A 40% improvement from that standard is a miserable failure, they need a 2000% improvement just to get back to Excel 2003 levels. It's easy to understand why they "chose" not to compare to 2003. Excel 2007 is just horrible, both in its vastly more cumbersome user interface and its horribly degraded performance. I haven't tried 2010 yet, but if the "improvements" are what MS claims they are (and I can't believe they are better than that), it won't be worth it. Any "development" team that destroys a product the way they have done should be fired on the spot. Only at Microsoft!!