Community
cancel
Showing results for 
Search instead for 
Did you mean: 
TBarr1
Beginner
1,114 Views

New Floating Point Math Error in Subtraction?

To duplicate type the following in to an Excel spreadsheet.

=(6.377-6.376)*1000

I get: 0.99999999999945 (or 0.999999999999446) if I allow the decimal places to be shown.

I get a similar error using the calculator in Windows 10.

I understand rounding error, but look at the significant digits and this occurs in subtraction. There should NOT be any rounding error.

The error exists before multiplying by 1000, that is done, just for convenience.

This error occurs on the latest generation of Intel Processors (i7-8550U as well as my older i7-4770).

I had to track down this error from a 1 > 1 problem (when the number was put into a logical statement it failed).

0 Kudos
3 Replies
idata
Community Manager
70 Views

Hello Get_A_Grip,

 

 

I understand that you are having issues with a floating math point when doing a subtraction.

 

 

Regarding this, what I can recommend you to do is to go to our developer's zone support site, there you should be able to get the information you need. You can refer to the link below:

 

 

https://software.intel.com/en-us/forum https://software.intel.com/en-us/forum

 

 

You can register to this forum and ask the question there, you should receive the best support to get an answer for your inquiry.

 

 

 

Regards,

 

David V
TBarr1
Beginner
70 Views

Your reply assumes the source of error is software. My suggestion is that the source of the error is hardware. Is it your assertion that the https://software.intel.com/en-us/forum https://software.intel.com/en-us/forum is the correct location to post a hardware problem? I took a look through that forum and was not satisfied that it was the correct location. I understand that many people will assume hardware problems exist where they are entirely and most often caused by software issues. I even understand that should be default assumption. I am a programmer and would not suggest a hardware problem by default, bugs are almost always caused by software. Except for the few times when they are not. I already attempted they in the Windows 10 calculator and online where I received imprecise results as well.

If you will recall Intel has had floating point processor errors in the past. I am sure that people would have liked to blame those on the software as well. If I am correct and you failed to investigate and suppressed my post by giving me a one star rating you should be ashamed of your self. I say this because when I searched the posts mine did not appear so I can only assume that steps were taken to suppress the post.

To me it is illogical to assume that the math result came from a software error at the math would have been performed by the processor and only reported by the software. That puts the source of the error at a hardware one not a software one. Since the error is way too large to be rounding error I can't call it simply that.

If there is a better place to post a hardware error I am happy to listen.

idata
Community Manager
70 Views

Hello Get_A_Grip,

 

 

Thank you for your response.

 

 

Math errors, rounding errors, etc. Are supported by the Intel ® Developer Zone, in fact, if you search for contents you will be able to find different posts and articles about rounding errors and math problems. Please address your inquiry to that support site.

 

 

 

Regards,

 

David V
Reply