Test for floating point result consistency

Overview

This page presents results of an experiment done by IT. We

• ... instructed Matlab to do a small computation
• ... on multiple computers
• ... using multiple Matlab releases

You should draw your own conclusions about the results. IT's minimum suggestion is: If you have to use Matlab, do not change releases mid-computation and stick to a specific matlab release .

CPU/Matlab-Release matrix

Each number represents one possible result of the calculation. The respective result is show in the tooltip help.

CPU	Host	7.1	7.3	7.4	7.5	7.6	7.7	7.8	7.9	7.10	7.11	7.12	7.13	7.14	8.0	8.1	8.2	8.3	8.5	8.6	9.0	9.1	9.2	9.3	9.4	9.5	9.6	9.7	9.8	9.9	9.10	9.11
Intel(R) Xeon(R) CPU E5-4650L 0 @ 2.60GHz	springsteen	1	2	3	3	2	2	2	2	2	2	4	4	4	4	4	4	4	4	4	4	4	2	2	2	5	5	5	5	5	5	5
AMD EPYC 7702 64-Core Processor	malteser	1	1	1	1	1	1	1	1	6	6	6	1	1	1	1	1	1	1	1	1	1	1	1	1	1	2	2	7	7	8	8
AMD Ryzen 5 1500X Quad-Core Processor	supremes	-	-	-	-	-	-	-	-	-	-	-	1	1	1	1	1	1	1	1	1	1	1	1	1	1	2	2	7	7	8	8
AMD Ryzen 5 2600 Six-Core Processor	fohlen	1	1	1	1	1	1	1	1	6	6	6	1	1	1	1	1	1	1	1	1	1	1	1	1	1	2	-	7	7	8	8
AMD Ryzen 5 3400G with Radeon Vega Graphics	okapi	1	1	1	1	1	1	1	1	6	6	6	1	1	1	1	1	1	1	1	1	1	1	1	1	1	2	-	7	7	8	8
AMD Ryzen 5 3600 6-Core Processor	puma	1	1	1	1	1	1	1	1	6	6	6	1	1	1	1	1	1	1	1	1	1	1	1	1	1	2	2	7	7	8	8
AMD Ryzen 7 2700 Eight-Core Processor	beagle	1	1	1	1	1	1	1	1	6	6	6	1	1	1	1	1	1	1	1	1	1	1	1	1	1	2	2	7	7	8	8
AMD Ryzen 7 5800X 8-Core Processor	reh	1	1	1	1	1	1	1	1	6	6	6	1	1	1	1	1	1	1	1	1	1	1	1	1	1	2	2	7	7	8	8
AMD Ryzen 9 5950X 16-Core Processor	pavian	1	1	1	1	1	1	1	1	6	6	6	1	1	1	1	1	1	1	1	1	1	1	1	1	1	2	2	7	7	8	8
Intel(R) Core(TM) i5 CPU	algerien	1	2	3	3	2	2	2	2	2	2	4	4	4	4	4	4	4	4	4	4	4	2	2	2	5	5	5	5	5	5	5
Intel(R) Core(TM) i7 CPU	wehner	-	-	-	-	-	-	-	-	-	-	-	4	4	4	4	4	4	4	4	4	4	2	2	2	5	-	5	5	5	5	5
Intel(R) Core(TM) i7-2600K CPU	anna	-	-	-	-	-	-	-	-	-	-	-	4	4	4	4	4	4	4	4	4	4	2	2	2	5	5	5	5	5	5	5
Intel(R) Core(TM) i7-2600K CPU @ 3.40GHz	nevada	1	2	3	3	2	2	2	2	2	2	4	4	4	4	4	4	4	4	4	4	4	2	2	2	5	5	5	5	5	5	5
Intel(R) Core(TM) i7-2700K CPU @ 3.50GHz	enz	1	2	3	3	2	2	2	2	2	2	4	4	4	4	4	4	4	4	4	4	4	2	2	2	5	5	5	5	5	5	5
Intel(R) Core(TM) i7-3770K CPU @ 3.50GHz	aisch	1	2	3	3	2	2	2	2	2	2	4	4	4	4	4	4	4	4	4	4	4	2	2	2	5	5	5	5	5	5	5
Intel(R) Core(TM) i7-4770K CPU @ 3.50GHz	alphaville	1	2	3	3	2	2	2	2	2	2	4	4	4	4	4	3	3	3	3	9	9	10	10	10	7	7	7	7	7	8	8
Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz	bap	1	2	3	3	2	2	2	2	2	2	4	4	4	4	4	3	3	3	3	9	9	10	10	10	7	7	7	7	7	8	8
Intel(R) Core(TM) i9-10900X CPU @ 3.70GHz	momo	1	2	3	3	2	2	2	2	2	2	4	4	4	4	4	3	3	3	3	9	9	10	10	1	11	12	12	12	12	8	8
Intel(R) Core(TM)2 CPU 660	guido	1	2	3	3	2	2	2	2	2	2	4	4	4	4	4	4	4	1	1	1	1	1	1	1	13	5	5	5	5	5	5
Intel(R) Core(TM)2 CPU 6600	boron	-	-	-	-	-	-	-	-	-	-	-	4	4	4	4	4	4	1	1	1	1	1	1	1	13	5	5	5	5	5	5
Intel(R) Core(TM)2 Duo CPU	adorno	1	2	3	3	2	2	2	2	2	2	4	4	4	4	4	4	4	1	1	1	1	1	1	1	13	5	5	5	5	5	5
Intel(R) Core(TM)2 Quad CPU	actinium	1	2	3	3	2	2	2	2	2	2	4	4	4	4	4	4	4	1	1	1	1	1	1	1	13	5	5	5	5	5	5
Intel(R) Xeon(R) CPU E5-1660 v3 @ 3.00GHz	luke	-	-	-	-	-	-	-	-	-	-	-	4	4	4	4	3	3	3	3	9	9	10	10	10	7	7	7	7	7	8	8
Intel(R) Xeon(R) CPU E5-2670 0 @ 2.60GHz	ufo	1	2	3	3	2	2	2	2	2	2	4	4	4	4	4	4	4	4	4	4	4	2	2	2	5	5	5	5	5	5	5
Intel(R) Xeon(R) CPU E5-2680 v3 @ 2.50GHz	comps06h01	1	2	3	3	2	2	2	2	2	2	4	4	4	4	4	3	3	3	3	9	9	10	10	10	7	7	7	7	7	8	8
Intel(R) Xeon(R) Gold 6244 CPU @ 3.60GHz	maki	1	2	3	3	2	2	2	2	2	2	4	4	4	4	4	3	3	3	3	9	9	10	10	1	11	12	12	12	12	8	8
Intel(R) Xeon(R) Silver 4108 CPU @ 1.80GHz	silbermond	1	2	3	3	2	2	2	2	2	2	4	4	4	4	4	3	3	3	3	9	9	10	10	1	11	12	12	12	12	8	8

Script

This is the script being used to compare results of floating point calculations:

format long

x = [0 1 2 3 4]'
y = [1 2 5 10 17]'

A = [sin(x), x.^2, ones(size(x))]

ATA = A'*A
rhs = A'*y

c = ATA\rhs

F = @(x) c(1)*sin(x) + c(2)*x.^2 + c(3) * ones(size(x));

S = sum(abs(y - F(x)))