And by weird issue I mean it didn't exactly work.
So I have Digispark ATTiny85 and a couple of raw chips, 10u and 20u. I can program them with Nano programmer I made and as far as I can tell they work fine. In fact they run 1.5 times faster than Digispark on respective CPU frequency. I measured loop runtime and displayed it on I2C diplay which works on all boards with no problems.
The neopixel strip works on Digispark board 4 LEDs are changing color as I wanted but on raw chips it just lighting up 4 LEDS white and nothing. I thought the timing could be off, so I played with timing in every possibe way, you have to believe me, and no change. The only change I noticed when I uploaded selecting Disgispark in menu instead of generic ATTiny45/85...etc option. Then the LEDs would change colour but they all would be lit with different colours and changing to random colours.
I suspect the timing is still off, but I havent used oscillosope yet. Need to make test sketch for that. And while I was troubleshooting I found how to flush Digispark board to remove startup delay, so Im ok now with using that. But still curious if anyone came across this timing weirdness. Maybe these new chips are optimised somehow hardware wise, because I dont see any reason why they would run 1.5 times faster than Digispark.