Page 2 of 2

Re: IBM mainframe 360/30 vs. iPad 2019

Posted: Thu May 16, 2019 2:29 am
by GeorgeMcGinn
Duh, I forgot to replace the Function.

Here it is, I changed the font size so it fits in one screen: SET OUTPUT FONT SIZE 12

This is what I get.
IMG_1200.PNG
IMG_1200.PNG (1.25 MiB) Viewed 316 times

Re: IBM mainframe 360/30 vs. iPad 2019

Posted: Thu May 16, 2019 2:34 am
by GeorgeMcGinn
I had to run the program 32 times to get any errors, and this is my first screen with errors:
IMG_1201.PNG
IMG_1201.PNG (1.31 MiB) Viewed 316 times

Re: IBM mainframe 360/30 vs. iPad 2019

Posted: Thu May 16, 2019 2:54 am
by rbytes
You are running on an iPad Pro, I believe. That accounts for being able to increase the font size. Your speeds are quite a bit faster than mine and Henk's. I'm not sure how to account for the improved accuracy. Is there better math processing on the Pro?

Re: IBM mainframe 360/30 vs. iPad 2019

Posted: Thu May 16, 2019 5:31 am
by Henko
rbytes wrote:
Thu May 16, 2019 2:54 am
You are running on an iPad Pro, I believe. That accounts for being able to increase the font size. Your speeds are quite a bit faster than mine and Henk's. I'm not sure how to account for the improved accuracy. Is there better math processing on the Pro?
No, i lowered the bias that you added to the mat_mul function untll i got no more "erroneous" matrices in 10 runs, but stated that that bias was not a guarantee on a "clean" unit matrix.
George just did more runs and proved that statement, using 32 runs.