A general discussion of population count:
https://stackoverflow.com/questions/8590432/when-to-use-parallel-counting-mit-hakmem-for-bitcount-when-memory-is-an-issue
At 08:49 PM 5/04/2019 +, you wrote:
>Hi Kyle,
>
>hat's a really interesting problem, and the government (NSA) wanted this badly
>and done FAST.
>
>they asked Seymour Cray to create a specific instruction for this and they
>called it 'population count'
>
>Anybody know the why and how it is
> On Apr 5, 2019, at 4:49 PM, Randy Dawson via cctalk
> wrote:
>
> Hi Kyle,
>
> hat's a really interesting problem, and the government (NSA) wanted this
> badly and done FAST.
>
> they asked Seymour Cray to create a specific instruction for this and they
> called it 'population count'
>
and Off-Topic Posts
Subject: Re: PDP-8: count number of set bits in a word
From: Kyle Owen via cctalk: Friday, April 05, 2019 8:59 AM
> Just wondering if anyone has come up with a fast way to count the number of
> 1s in a word on a PDP-8. The obvious way is looping 12 times, rotating the
From: Kyle Owen via cctalk: Friday, April 05, 2019 8:59 AM
Just wondering if anyone has come up with a fast way to count the number of
1s in a word on a PDP-8. The obvious way is looping 12 times, rotating the
word through the link or sign bit, incrementing a count based on the value
of the link
I suppose you could test each nybble for zero, then equate a 16-element LUT
on nybbles not zero?
--
Anders Nelson
+1 (517) 775-6129
www.erogear.com
On Fri, Apr 5, 2019 at 11:59 AM Kyle Owen via cctalk
wrote:
> Just wondering if anyone has come up with a fast way to count the number of
> 1s
Just wondering if anyone has come up with a fast way to count the number of
1s in a word on a PDP-8. The obvious way is looping 12 times, rotating the
word through the link or sign bit, incrementing a count based on the value
of the link or sign.
With a small lookup table, you can reduce the