optimal stagger size for 1gb ram gpu plotting and mining?



  • im setting up my first plot file. it will be 2.75 tb plot using the gpu plotter what stagger size should i use for 1 gb ram gpu?



  • @masterhamby stagger size is for your physical RAM 😉



  • ok i figured that out looking around but im still getting an error

    does this look ok for device? 0 0 512 64 8192

    and for the plot? xxx_0_11520000_40000

    keep getting a buffer error



  • @masterhamby i use

    0 0 1792 128 4096
    

    but my card has 2GB of video ram so not sure if it would help you.



  • this is the error im pulling

    [ERROR][-61][CL_INVALID_BUFFER_SIZE] Unable to create the OpenCL GPU buffer



  • @masterhamby ahh i had that one 🙂 i think what i did was decrease 8192 to 4096 but cant be sure 100%, but i am sure that i decreased one of those numbers in devices.txt



  • thanks ill give it a try

    still pulling an error anyone have any idea why?



  • Hi, there. If your system only has 1GB ram, your OS probably needs at lest 512MB to run. So this means you have about ~500MB of ram to plot on. I would really recommend to upgrade to 2GB or more. Try 2,000. 🙂

    @gpedro said in What is AVX2:

    @Tate-A Well i will try to answer those questions xD

    1 - I'm not the correct person to answer this but as far as i know AVX2 is an advanced codepath that deal with the way CPU threads manage processes...
    2 - While mining makes it read the plots faster 😉
    3 - Right
    4 - The stagger is the number of nonces used in each scoop... When plotting the plotter fills the RAM (the RAM used in the plotting is the stagger of a given plot) and then writes what is in the RAM into the storage (plot file) one scoop at a time.
    5 - Bigger stagger, bigger scoop and less scoops in the same file... Then a scoop is the amount of nonces readed at a time while mining! When you are mining the miner reads 1/4096 nonces of each scoop and then goes to the next scoop and so on until it finishes read the plot file. Then goes to the next plot and do the same and so on... So basically the biggest the scoop is the less processes will be started and stopped while reading plots and that correspond to miliseconds gained during the reading process, multiplied for a lot of scoops and plots it will gives you some seconds of diference while reading the plots during mining...
    6 - With the previous answers i already responded to this but basically is because the number of nonces inside each scoop have to be exactly the same, so if it is not the same, the last scoop will always be ignored and the plot will be easily corrupt over time ;P

    I think all my answers are correct but if some of the guys with more correct knowledge than me can confirm all i said i'd appreciate 😉



  • no my system has 8gb ram. the graphics card is 1gb. but that was still very helpful information. thank you. i still havnt gotten the gpu plotter to work for me idk what the problem is at this point. i started to plot it with the aio wallet tool but its still less than 50% in almost 24hrs and it started on 600,000,000 nonces and i wanted to start on 0. does anyone know if its worth continueing the plot or if i should stop it and use the new tool blago released or if anyone can help me get the gpu plotter working some how



  • @masterhamby depends on the size your plotting 8TB for me took 24H+ i think 🙂



  • its only a 2.75tb plot



  • @masterhamby a bit slow, how many nonces per minute you get? (mine was doing 19k) if its low it is slow 🙂



  • 4400 on adverage. way slow. thats why i want to get the gpu working ort ry blagos new plotting tool. but i dont know if its worth stoping this plot to restart it with one of the other tools.



  • @masterhamby whats your cpu? it might be faster than your video card in this case.



  • im using the cpu now i still cant get the gpu working. its definately not faster.

    cpu=intel xeon w3565 3.2 gh 8 threads
    gpu= ati radeon hd 4600 series