Cracking in Google Cloud Instance Questions.

Im gonna start cracking in a google cloud instance, just for test, i have some questions.
What gpu type/number choose. For maximize the 300$ Free credit.
*all instances has 500GB HDD(ssd would be better, but i have 100GB LIMIT) + 8 vcpu + 7.2GB RAM  price included for be able to upload the wordlist, first i will create instance witouth the GPUs and once wordlist is already uploaded , i unrar, put GPUs inside and start cracking.

Assuming i have just 300$ so what im asking is what is more worth? to have 8x Nvidia Tesla K80 running for 6 days or 8xNvidia Testla v100 for 20 Hours?. How i can check/calculate these?(depending on hash obviusly)

- 8x Nvidia Tesla K80 = 2.004,52 $  / month | 2,7 per hour

- 4x Nvidia Tesla P100 =  3.149,16 $ /month  | 4,3 $ per hour

-4x NVIDIA TESLA P100 Virtual workstation = 3.733,16 $ / month | 5,1$ / hour

- 8x Nvidia Tesla  V100 = 10.303,16$ /month  | 14,1$ per hour

-4x Nvidia Tesla P4 = 1500$ / month | 2,069$ per hour

-4x Nvidia Tesla P4 Virtual workstation = 2000$ / month | 2,869$ per hour



Another question is, i will change between servers alot, does hashcat creates the output file every it founds a hash? or when it finish all the procces? if its everytime it founds a hash i could rsync to my raspberry for have a updated file of output/folders.



And last question, is there any guide for Cracking in google cloud/creating instance/tips etc¿? thanks



PS: I also read thath google cloud request you to put some cash inside for see its a legit request(i read from 40-90$) these can be take from free credit? if not then im fucked.