Hi ![]()
I've created an UTFT Library benchmark, test it, and tell me your results !!!
I have 1344 points on my due.
Hope you'll like the demo ![]()
ArduMark.ino (6.28 KB)
Hi ![]()
I've created an UTFT Library benchmark, test it, and tell me your results !!!
I have 1344 points on my due.
Hope you'll like the demo ![]()
ArduMark.ino (6.28 KB)
It is a little unfair to expect Benchmarks for such a big display.
Anyway, I ran your sketch on a 800x480 with a CTE DUE shield : 960
I then ran it with UTFTGLUE class on top of MCUFRIEND_kbv library : 537
It would be far more interesting to compare the ILI9481 type of 480x320 controllers.
David.
p.s. I removed the UTouch calls. They do not do anything useful.
the MCUFRIEND_kbv library does not support angled text
Have the same LCD ![]()
Read the sketch, its scales the points according to the number of pixels ![]()
Well, you've got better results than me, because the lower the better
David wins with his glue library!!
I hand (butchered) optimised UTFT to 16 bit only, and I got 614... Kudos David ![]()
Regards,
Graham
Actualy, theres a bug in my code, I will upload the good version tomorow (The points are not change correctly with the resolution, it was a silly error from me xD)
Now that I overclocked my due (114MHz), i'm getting 1305 points (with the bug still in)
I found the over-clocking ideas that are out there for the DUE, actually reduce performance of UTFT....
I moved away from those long ago...
But, will be interested to see your new version!
Regards,
Graham
PS 800x480 CTE50
Here is the new version (hasn't been tested)
I cant test it now, but if you can, tell me your score !
Now, the higher, the better (This is the correction I was talking about) ![]()
ArduMark.ino (6.28 KB)
Hand butchered UTFT score = 576
Standard UTFT score = 960
Not sure higher is better!!!!!!!!
Regards,
Graham
WHOO HOO!!! my 1500th post!!! ![]()
Yes, sorry ^^
But normaly, il you test on a lcd with a different size, it should be the same score (If you dont change of library)
Sorry for bad english ![]()
david_prentice:
It would be far more interesting to compare the ILI9481 type of 480x320 controllers.
TFT_HX8357_Due does not support angled text but here are the results for a 480 x 320 screen.
ArduMark test with HX8357_Due library
TFT width = 480
TFT height = 320
TestA time = 1697
TestB time = 219
TestC time = 1453
TestD time = 20
TestE time = 698
TestF time = 80
Points = 61
Results on a Mega:
ArduMark test with HX8357 library
TFT width = 480
TFT height = 320
TestA time = 4648
TestB time = 219
TestC time = 1347
TestD time = 66
TestE time = 593
TestF time = 114
Points = 92
I suspect the points calculation does not correctly compensate for the resolution so I have printed the execution times in milliseconds. Only TestA seems to have a fixed size and thus would give a result that is equivalent to a larger display. On the Mega millisecond timing variables were set to "unsigned long" to avoid overflow.
Results with Due running the basic UTFT library are:
ArduMark test with UTFT library
TFT width = 480
TFT height = 320
TestA time = 5462
TestB time = 650
TestC time = 8900
TestD time = 78
TestE time = 2027
TestF time = 2860
Points = 291
Results on a Mega with UTFT:
ArduMark test with UTFT library
TFT width = 480
TFT height = 320
TestA time = 24429
TestB time = 1727
TestC time = 23079
TestD time = 422
TestE time = 5235
TestF time = 9492
Points = 983
In these UFTF cases TestF is drawing rotated text so would be expected to take longer.
Serial.println("ArduMark test with HX8357_Due library");
Serial.print("TFT width = ");Serial.println(width);
Serial.print("TFT height = ");Serial.println(height);
Serial.println();
Serial.print("TestA time = ");Serial.println(a);
Serial.print("TestB time = ");Serial.println(b);
Serial.print("TestC time = ");Serial.println(c);
Serial.print("TestD time = ");Serial.println(d);
Serial.print("TestE time = ");Serial.println(e);
Serial.print("TestF time = ");Serial.println(f);
Serial.print("Points = ");Serial.println(points);