The Spectrum had major limitations with the graphics, which I'll get to, but the 'sound' was nothing more than a single io pin on the Z80 wired to a piezo sound

The Spectrum had major limitations with the graphics, which I'll get to, but the 'sound' was nothing more than a single io pin on the Z80 wired to a piezo sounder (or the cassette 'line-out'). To make a tone you had to run an assembly program to toggle the io port pin on and off at the correct frequency. When you made a tone in basic your program paused until the tone was done. The graphics were a 256x192 memory mapped 1bit buffer (6144 bytes). There was an additional 768 byte buffer that determined what the on and off color was for each 8x8 square of screen pixels. There were 8 possible colors, the byte used 3 bits to select the on color, 3 bits to select the off color, 1 bit to determine if the colors were 'bright' and 1 bit to determine if the colors would flash (alternate every second). Many games would have a 'play area' that was monochrome to avoid the issues of so called color clash where differently colored sprites overlapped and fought over the color selection. The further problem with the screen (and the major headache as soon as you wanted to do anything performant) was that the buffer was not laid out linearly. The top line of the screen (y=0) was at ram address 0 (byte 16384 in memory, rom space went first) and spanned 32 bytes (256 pixels). The next 32 bytes were y=8 on screen. Then y=16 etc. Once you got to byte 256 you were back at y=1 on the screen. Next 32 bytes was y=9 etc. Line y=2 started at byte 512... If you're paying attention (and if you are just imagine what the code to blit sprites looked like) we've only covered the top third of the screen. At byte 2048 you get into the second third of the screen (y=64) and the pattern repeats. I loved the Spectrum (had the 48k and +2, with MGT +D and Swift Disk disk interfaces) but damn that screen layout!