I was recommended to post some real hardware benchmarks I have been performing over the last few weeks, as no one else has.
As I've stated before, although perhaps not on Stardot, all emulation currently of the 32016 co pro, and especially 32081, is vastly inaccurate, speed wise.
Clock Speed 5 - BAS32F (6mhz 32081)(More accurate, later version):
Real REPEAT Loop: 4.457MHz
Variant REPEAT Loop: 4.455MHz
Integer REPEAT Loop: 3.408MHz
Real FOR loop: 7.334MHz
Variant FOR Loop: 7.343MHz
Integer FOR loop: 3.636MHz
Trig/Log test: 82.57MHz
String manipulation: 3.503MHz
Procedure call: 3.572MHz
GOSUB call: 3.019MHz
Average: 12.33MHz
Clock Speed 3 - BAS32F (6MHz 32081) (the one grannyg was using)
Really real REPEAT Loop: 4.18MHz
Integer REPEAT Loop: 3.22MHz
Really Real FOR loop: 7.52MHz
Integer FOR loop: 3.70MHz
Trig/Log test: 62.54MHz
String manipulation: 6.43MHz
Procedure call: 3.40MHz
GOSUB call: 3.21MHz
Combined Average: 12.08MHz
WHETS - Panos 1.4 w/ 32081:
Loop content Result MFLOPS MOPS Seconds
N1 floating point -1.12475000000000000 3.500 .384
N2 floating point -1.12274700000000000 3.675 2.560
N3 if then else 1.00000000000000000 1.617 4.480
N4 fixed point 12.00000000000000000 3.132 7.040
N5 sin,cos etc. .49914560000000000 .228 25.600
N6 floating point .99999980000000000 1.439 26.240
N7 assignments 3.00000000000000000 .879 14.720
N8 exp,sqrt etc. .75113200000000000 .136 19.200
MWIPS 6.984 100.224
WHETD - Panos 1.4 W/ 32081 @6mhz (double precision):
Loop content Result MFLOPS MOPS Seconds
N1 floating point -1.13145740591122500 .041 .467
N2 floating point -1.13105217644797000 .042 3.210
N3 if then else 1.00000000000000000 .047 2.190
N4 fixed point 12.00000000000000000 .054 5.800
N5 sin,cos etc. .49981457027447220 .003 26.160
N6 floating point 1.00027757247772000 .023 23.320
N7 assignments 3.00000000000000000 .028 9.200
N8 exp,sqrt etc. .75020059172019020 .002 21.420
MWIPS .109 91.767
WHETS - Panos 1.4 W/32081 @6mhz:
Will be added later today.
As I've stated before, although perhaps not on Stardot, all emulation currently of the 32016 co pro, and especially 32081, is vastly inaccurate, speed wise.
Clock Speed 5 - BAS32F (6mhz 32081)(More accurate, later version):
Real REPEAT Loop: 4.457MHz
Variant REPEAT Loop: 4.455MHz
Integer REPEAT Loop: 3.408MHz
Real FOR loop: 7.334MHz
Variant FOR Loop: 7.343MHz
Integer FOR loop: 3.636MHz
Trig/Log test: 82.57MHz
String manipulation: 3.503MHz
Procedure call: 3.572MHz
GOSUB call: 3.019MHz
Average: 12.33MHz
Clock Speed 3 - BAS32F (6MHz 32081) (the one grannyg was using)
Really real REPEAT Loop: 4.18MHz
Integer REPEAT Loop: 3.22MHz
Really Real FOR loop: 7.52MHz
Integer FOR loop: 3.70MHz
Trig/Log test: 62.54MHz
String manipulation: 6.43MHz
Procedure call: 3.40MHz
GOSUB call: 3.21MHz
Combined Average: 12.08MHz
WHETS - Panos 1.4 w/ 32081:
Loop content Result MFLOPS MOPS Seconds
N1 floating point -1.12475000000000000 3.500 .384
N2 floating point -1.12274700000000000 3.675 2.560
N3 if then else 1.00000000000000000 1.617 4.480
N4 fixed point 12.00000000000000000 3.132 7.040
N5 sin,cos etc. .49914560000000000 .228 25.600
N6 floating point .99999980000000000 1.439 26.240
N7 assignments 3.00000000000000000 .879 14.720
N8 exp,sqrt etc. .75113200000000000 .136 19.200
MWIPS 6.984 100.224
WHETD - Panos 1.4 W/ 32081 @6mhz (double precision):
Loop content Result MFLOPS MOPS Seconds
N1 floating point -1.13145740591122500 .041 .467
N2 floating point -1.13105217644797000 .042 3.210
N3 if then else 1.00000000000000000 .047 2.190
N4 fixed point 12.00000000000000000 .054 5.800
N5 sin,cos etc. .49981457027447220 .003 26.160
N6 floating point 1.00027757247772000 .023 23.320
N7 assignments 3.00000000000000000 .028 9.200
N8 exp,sqrt etc. .75020059172019020 .002 21.420
MWIPS .109 91.767
WHETS - Panos 1.4 W/32081 @6mhz:
Will be added later today.
Statistics: Posted by Cruxinc — Tue Jan 27, 2026 1:01 pm