I couldn’t agree more! Commodore64 for me, but I’m with ya on those “IBM Compatible” machines.
I find this kind of struggle refreshing. I went into the purchase (I went straight with the NVidia branded one) knowing this was a lab/developer device and not a consumer device. I was actually surprised how easy it was to setup compared to what I was expecting.
I did the customary “wipe and start again” when I realized I went down the wrong rabbit hole trying to figure out why my wired network kept failing to work and the respective fixes. So I went through and downloaded the latest release and did the USB restore via the BIOS, and the second setup was even smoother.
Learned that the updates screen and process has a timer-trigger for whether or not the updates failed instead of another signal. So, it will show that the updates failed even though they are still running. This was only a problem on the first batch of updates since there were over 1,000.
But for me, that’s all normal stuff when working with bleeding-edge tech, you’re gonna get cut.
I’m a bit disappointed with the performance and driver support. My laptop’s RTX 4070 seems to get better it/s than this blackwell chip, but I’m trying to get a better benchmark comparison going, so far it hasn’t been quite a 1 to 1 test. And I’m sure there is some fine tuning I still need to do.
Network wise, I have it wired on the 10G port to my router’s 10G and 7G internet and it flexes it to the max speed.
But, I was able to repeat the learning I did on my laptop over a year, in about 3 days on the Spark. Offloading my AI playtime and learning to this device has reignited my interest in learning the innerworkings of the current state of AI.