Read up on analog-to-digital conversion. Here's my (possibly flawed) simplistic take on it:
Digital effects processors don't work on an analog signal, such as what comes out of your electric guitar or microphone. The first step is to convert the incoming analog signal to a digital representation. When converting an analog signal to digital, the analog signal's voltage is 'sampled' and converted to the closest digital representation. You need two parameters to do this: sampling resolution and sampling rate.
* The sampling resolution is how fine the digital representation of the signal is, and this is measured in 'bits'. Common values are 16, 24 and 48 bits. A higher value here means a more accurate representation, but at the cost of more storage (memory) required.
* The sampling rate is how often the analogue signal is sampled, and is measured in Hertz (typically kilohertz). A higher value here also means a more accurate representation of the original analog signal, and likewise at the cost of more storage.
For example, assuming an 8-bit 16kHZ sampling resolution and rate, this requires 8 * 16000 = 128000 bits = 125 Kbits of storage space for a 1 second sample.
So, now we know that a digital representation of analog audio needs a certain amount of bits, and it's these bits that are stored in memory (RAM). Getting back to the digital delay pedal: it's continually sampling the incoming audio and will replay it back after a delay, which is a variable period set by the 'delay' control. The pedal needs to store the last 'x' seconds of audio before it replays, so it needs a certain amount of memory to store that digital audio. The amount of memory required not only depends on the sampling rate and resolution, but also on how long the delay is:
* Sample resolution
* Sample rate
* Delay time
Total Storage Required (bits) = Resolution (bits) * Rate (Hz) * Time (seconds). The amount of memory you have available determines the constraints of the above three parameters.
The first two parameters are usually constrained by the ADC (analog to digital converter) chip that you're using. For the PT2399 this is built into the chip, and without referring to the data sheets I'm not sure what resolution and rate it uses. But for argument's sake, say it has a sample resolution of 8 bits, and a rate of 8KHz. Given your 45,056 bits of memory, the only unknown in the above equation is the maximum delay time:
Time = 45056 / (8 * 8000) = 0.704 seconds
So what does it all mean? I guess it boils down to this: The more memory you have available, the longer delay times you can acheive and/or you can have increased quality of audio. The less memory you have, either the delay time suffers or the quality of the delayed audio does.