We may all live in a great big global community, but in my Blog, it's my world.
Excel 2007 fails at simple math
Published on September 29, 2007 By terpfan1980 In Personal Computing

I can't help but snicker at this news: Microsoft Excel {2007} Fails Math Test.

So Microsoft is pulling an Intel (remember the Pentium math processing issues?) and putting out software that can't do simple math?

A display error. Hmmm, nice to know that Microsoft may be the cause for overbillings to the tune of approx. $35,000 because of a display error related to multiplaction in Excel. I hope they get it fixed soon and in the meantime am glad I don't rely on Excel to run a business (but then again, I could be depending upon someone that is running Excel when I purchase a product or pay for a product that I've purchased.


Comments
on Sep 29, 2007
Fun (NOT!).
on Sep 29, 2007
  
on Sep 29, 2007

Here's some of the details:

From IWon.com (linked in original article):

In a blog post, Microsoft employee David Gainer said that when computer users tried to get Excel 2007 to multiply some pairs of numbers and the result was 65,535, Excel would incorrectly display 100,000 as the answer.

Gainer said Excel makes mistakes multiplying 77.1 by 850, 10.2 by 6,425 and 20.4 by 3,212.5, but the program appears to be able to handle 16,383.75 times 4.

"Further testing showed a similar phenomenon with 65,536 as well," Gainer wrote Tuesday.

He said Excel was actually performing the calculations correctly, but when it comes time to show the answer on the screen, it messes up.

Gainer said the bug is limited to six numbers from 65,534.99999999995 to 65,535, and six numbers from 65,535.99999999995 to 65,536, and that Microsoft is working hard to fix the problem.

on Sep 29, 2007
Like anyone's ever been able to add 1 + 1 around here and get 2!!!  

Give me a break, I don't even have children and I can recognize New Math when I see it!      

Haven't you ever heard the phrase "Check you math" ?   

                
on Sep 29, 2007

"Check you math"

Because of the decimal to binary conversion, if you want accuracy to decimal places, you always have to on a computer.  Strange but true.

on Sep 29, 2007
Because of the decimal to binary conversion, if you want accuracy to decimal places, you always have to on a computer. Strange but true.


Say what!  
on Sep 29, 2007
Say what!


Yep. COmputers do not do decimal math accurately. If you convert to whole numbers and do the math, you are fine. But if you want to divide $1.44 by .12, prepare to get change(d).

but 144 divided by 12 will get you 12 each time.
on Sep 29, 2007
Because of the decimal to binary conversion, if you want accuracy to decimal places, you always have to on a computer. Strange but true.


I have a hunch that the error is somehow related to binary, I just don't get why it would be 100,000. 65,536 is, of course, 16 bits. But that conversion would be 1 quadrillion, not 100,000.

So, of course, I'm still clueless.
on Sep 29, 2007
So, of course, I'm still clueless.


This one is just a bone head blunder on MS' part, but yea, probably due to some binary conversion. As you say, the number is just a "leeeetle tooo convenient".
on Sep 29, 2007
It's likely related to how float type is stored. Float type definitely is not stored in memory the same way as integrals.
on Sep 30, 2007
Float type definitely is not stored in memory the same way as integrals.


And it gets the accountants most of all!
on Oct 01, 2007


New excuse of bulking companies funds: "Uhhh all the floating numbers got confused up in my mind, and I ended up taking company money."
Meta
Views
» 2073
Comments
» 12
Sponsored Links