👩💻 Join our community of thousands of amazing developers!
NumPy and CuPy converted my computations to float64 despite being told to work with float32, and that prompted some readers to discard my previous article as faulty. My article doesn't make CuPy and NumPy more or less perfect, though, and that forceful conversion stings everyone who is using Nvidia's GTX gaming GPUs. I don't have grants that would cover $10,000 for uncrippled float64 support on Tesla GPU. So, if you're poor like me, that article probably offered valuable information. I've rec...