How big is a 64 bit register?

[btbass]

I have a chip with a 64 bit register and an overflow flag.
I clear the register to zero and increment it with a fast clock of 1GHz.
Thats a 1 followed by 9 zeros.
1,000,000,000 counts per second!

Now 64 bits is a big number.
How long will it take before the overflow flag is set?

 

[alexan_e]

64 bit is 2^64 = 18446744073709551616
if the counter works at 1GHz then you need 18446744073.709551616 seconds

Alex

 

[arthur0]

Is this a trick question?
If not, I think it's something like 585 years, so you'd better buy a faster chip

 

[alexan_e]

converted to days it is 213503,98 days

Alex

 

[btbass]

No, it's not a trick question, it's just that 64 bits is a big number.
I have seen an audio dac that says it's 64 bits. Now audiophiles have some strange notions, but unless they are doing something fancy with the bits, the lsb of a 64 bit dac must be less than the charge on an electron!

 

[arthur0]

Hehe, the reason I was a bit unsure is (given your capable previous posts), I was conviced you could do the calculations and find the answer yourself (which I still am sure you did)...

 

[DXNewcastle]

. . . audiophiles have some strange notions, but unless they are doing something fancy with the bits . . .

isn't that exactly the reason? That there is processing to be done to the samples?
Assuming that we accept that 16 bits are not quite enough to represent audio adequately, and we have to do some basic processing such as cross-fading (mixing) between two sources and just a modest amout of equalisation (gently sloping first-order high and low pass filters), then it becomes a lot easier for a designer just to use 64 bit words for the processing, as any truncation errors can quickly become very audible as unpleasant artifacts. I'm sure I've seen 128bit processing too, though I'll offer no evidence in support of consumer media holding audio in any more than 24bit samples!