Hello All,

I was just going over Peldi's algorithm for optimizing camera/mic
settings, and I noticed that 1024 was used in some of the bw
calculations. (e.g. bw = bwUp*1024/8;) Maybe I haven't read the
algorithm close enough because it's such an amazing tool--fitting
into the 'if it's not broke, don't fix it' category.

However, my understanding of bw settings is that their kilo- bytes
and bits are actually 1000-based rather than the normal 1024-based k
used with real bits and bytes. (Asa once laid this gem on us I
believe, noting that the BW bits and byes were really a marketing
ploy and the extra 24 bits used with processor-related kilos were
shaved off.) So if we allocate bits and bytes by the K we need to
multiply by 1000 instead of 1024, no? Or is the 1024 referring to
something else?

BTW, if you're not familiar with Peldi's work on this, see:

http://www.peldi.com/blog/archives/...ing_opti_1.html

TIA,
Bill


bill sanders | www.sandlight.com | bloomfield, ct | 860-242-2260



=-----------------------------------------------------------
Supported by Fig Leaf Software - http://www.figleaf.com
=-----------------------------------------------------------

To change your subscription options or search the archive:
http://chattyfig.figleaf.com/mailman/listinfo/flashcomm

[ Post a follow-up to this message ]

I ran into the exact same issue recently. Whats worse is that as I researche
d it on the internet, I found that most of the bandwidth calculators were us
ing
1024 instead of 1000 also. I was in the process of calculating 'potential' b
andwidth usage per customer on my webcam chat app. So then my obsessive
compulsive disorder kicked into full throttle and I fretted about it for sev
eral days until my wife pointed out that I was wasting more money worrying a
bout it
than I would be by dropping a few bits on the way to the bit-bucket now and 
then.  So, I  said the hell with it. What's a few bits one way or the other.
... I
used '1000' in my calculations ....now nothing matches up.....Bottom line, i
t ain't so easy knowing what you're paying for or selling when it comes to
bandwidth.

Let's see now I buy 'per Gigabyte'....I measure in kilobits...but they may b
e calculated wrong in FCS...so if I correct by a percentage......

James Voss

On Mon, 19 Dec 2005 06:07:31 -0500, Bill Sanders wrote:

>Hello All,

>I was just going over Peldi's algorithm for optimizing camera/mic
>settings, and I noticed that 1024 was used in some of the bw
>calculations. (e.g. bw = bwUp*1024/8;) Maybe I haven't read the
>algorithm close enough because it's such an amazing tool--fitting
>into the 'if it's not broke, don't fix it' category.

>However, my understanding of bw settings is that their kilo- bytes
>and bits are actually 1000-based rather than the normal 1024-based k
>used with real bits and bytes. (Asa once laid this gem on us I
>believe, noting that the BW bits and byes were really a marketing
>ploy and the extra 24 bits used with processor-related kilos were
>shaved off.) So if we allocate bits and bytes by the K we need to
>multiply by 1000 instead of 1024, no? Or is the 1024 referring to
>something else?

>BTW, if you're not familiar with Peldi's work on this, see:

>http://www.peldi.com/blog/archives/...ing_opti_1.html

>TIA,
>Bill


>bill sanders | www.sandlight.com | bloomfield, ct | 860-242-2260



>=-----------------------------------------------------------
>Supported by Fig Leaf Software - http://www.figleaf.com
>=-----------------------------------------------------------

>To change your subscription options or search the archive:
>http://chattyfig.figleaf.com/mailman/listinfo/flashcomm



=-----------------------------------------------------------
Supported by Fig Leaf Software - http://www.figleaf.com
=-----------------------------------------------------------

To change your subscription options or search the archive:
http://chattyfig.figleaf.com/mailman/listinfo/flashcomm

[ Post a follow-up to this message ]

One thing's for sure: kilo means 1000, it's the metric system.
1000 grams = 1 kilogram.

However when dealing with bits and bytes the term kilobyte is used for
simplification. 1 kilobyte is actually 1024 bytes and not 1000 bytes. So
strictly speaking the term kilo is incorrect here, they have used it because
1 kilobyte is 'roughly' 1000 bytes - but not exactly.

Bottom line: Peldi's calculations are correct.

Stefan



> -----Original Message-----
> From: flashcomm-bounces-1Ss2GqJETD3yZ38Mhd3e/9ZfFG6BLHNm@public.gmane.org
> [mailto:flashcomm-bounces-1Ss2GqJETD3yZ38Mhd3e/9ZfFG6BLHNm@public.gman
e.org] On Behalf Of
> James Voss
> Sent: 20 December 2005 03:37
> To: FlashComm Mailing List
> Subject: Re: [FlashComm] BW algorithms & 1000 vs 1024
>
> I ran into the exact same issue recently. Whats worse is that
> as I researched it on the internet, I found that most of the
> bandwidth calculators were using
> 1024 instead of 1000 also. I was in the process of
> calculating 'potential' bandwidth usage per customer on my
> webcam chat app. So then my obsessive compulsive disorder
> kicked into full throttle and I fretted about it for several
> days until my wife pointed out that I was wasting more money
> worrying about it than I would be by dropping a few bits on
> the way to the bit-bucket now and then.  So, I  said the hell
> with it. What's a few bits one way or the other..... I used
> '1000' in my calculations ....now nothing matches
> up.....Bottom line, it ain't so easy knowing what you're
> paying for or selling when it comes to bandwidth.
>
> Let's see now I buy 'per Gigabyte'....I measure in
> kilobits...but they may be calculated wrong in FCS...so if I
> correct by a percentage......
>
> James Voss
>
> On Mon, 19 Dec 2005 06:07:31 -0500, Bill Sanders wrote:
> 
> 
> tool--fitting into 
> 
> bytes and 
> 1024-based k used 
> ploy and the 
> off.) So if 
> 1000 instead 
> 
> 
> 
>
> 
>
>


=-----------------------------------------------------------
Supported by Fig Leaf Software - http://www.figleaf.com
=-----------------------------------------------------------

To change your subscription options or search the archive:
http://chattyfig.figleaf.com/mailman/listinfo/flashcomm

[ Post a follow-up to this message ]

Stefan,

Actually, I don't think they are--unless Peldi was doing calculations
based on something that the processor does. That was Asa's point some
time ago. In dealing with bits/bytes 1024 is based on the way a
processor works and the bits stored in registers. (I believe).
However, you don't get 1024 bits in a BW kilobit. So a kb is 1000b
and not 1024b. So in a 10mb FCS license, you get 10,000,000 bits and
not 10,240,000 -- so basically the difference is all of the BW that
was in the FCS developer's version.

I don't mean to quibble over inconsequential values.  My hope is to
better understand the algorithm used. What I'm wondering, actually,
is whether Peldi's calculations were based on something other than BW
bits and bytes--such as something the processor does. I'll try it
with the different values and see if it makes a difference.

Thanks,
Bill

On Dec 20, 2005, at 4:22 AM, Stefan Richter wrote:

> One thing's for sure: kilo means 1000, it's the metric system.
> 1000 grams = 1 kilogram.
>
> However when dealing with bits and bytes the term kilobyte is used for
> simplification. 1 kilobyte is actually 1024 bytes and not 1000
> bytes. So
> strictly speaking the term kilo is incorrect here, they have used
> it because
> 1 kilobyte is 'roughly' 1000 bytes - but not exactly.
>
> Bottom line: Peldi's calculations are correct.
>
> Stefan
>
>
> 
>
>
> =-----------------------------------------------------------
> Supported by Fig Leaf Software - http://www.figleaf.com
> =-----------------------------------------------------------
>
> To change your subscription options or search the archive:
> http://chattyfig.figleaf.com/mailman/listinfo/flashcomm

bill sanders | www.sandlight.com | bloomfield, ct | 860-242-2260



=-----------------------------------------------------------
Supported by Fig Leaf Software - http://www.figleaf.com
=-----------------------------------------------------------

To change your subscription options or search the archive:
http://chattyfig.figleaf.com/mailman/listinfo/flashcomm

[ Post a follow-up to this message ]

Ah I see.

So you mean that when dealing with kbps (data transmission speed) it's often
the 1000 multiplier that's used whereas when dealing with conversion from
bits to kilobits in general it's 1024 (and rightly so).

Very interesting... And confusing.
I agree, this is very important to get right when dealing with a product
that's licensed and priced based on bandwidth usage.

You're right, when dealing with data transmission speeds (bitrates) it's the
multiplier of 1000 that's used. I was first under the impression that you
wanted to establish how many bits are in a kilobit - and in non-data
transmission circumstances it would be 1024.

I am keen to hear more on this subject.

Stefan



> -----Original Message-----
> From: flashcomm-bounces-1Ss2GqJETD3yZ38Mhd3e/9ZfFG6BLHNm@public.gmane.org
> [mailto:flashcomm-bounces-1Ss2GqJETD3yZ38Mhd3e/9ZfFG6BLHNm@public.gman
e.org] On Behalf Of
> Bill Sanders
> Sent: 20 December 2005 09:42
> To: FlashComm Mailing List
> Subject: Re: [FlashComm] BW algorithms & 1000 vs 1024
>
> Stefan,
>
> Actually, I don't think they are--unless Peldi was doing
> calculations based on something that the processor does. That
> was Asa's point some time ago. In dealing with bits/bytes
> 1024 is based on the way a processor works and the bits
> stored in registers. (I believe).
> However, you don't get 1024 bits in a BW kilobit. So a kb is
> 1000b and not 1024b. So in a 10mb FCS license, you get
> 10,000,000 bits and not 10,240,000 -- so basically the
> difference is all of the BW that was in the FCS developer's version.
>
> I don't mean to quibble over inconsequential values.  My hope
> is to better understand the algorithm used. What I'm
> wondering, actually, is whether Peldi's calculations were
> based on something other than BW bits and bytes--such as
> something the processor does. I'll try it with the different
> values and see if it makes a difference.
>
> Thanks,
> Bill
>
> On Dec 20, 2005, at 4:22 AM, Stefan Richter wrote:
> 
> is used for 
>
> bill sanders | www.sandlight.com | bloomfield, ct | 860-242-2260
>
>
>
> =-----------------------------------------------------------
> Supported by Fig Leaf Software - http://www.figleaf.com
> =-----------------------------------------------------------
>
> To change your subscription options or search the archive:
> http://chattyfig.figleaf.com/mailman/listinfo/flashcomm
>


=-----------------------------------------------------------
Supported by Fig Leaf Software - http://www.figleaf.com
=-----------------------------------------------------------

To change your subscription options or search the archive:
http://chattyfig.figleaf.com/mailman/listinfo/flashcomm

[ Post a follow-up to this message ]

Stefan,

That's it exactly. I believe Asa said it was a "marketing" usage
rather than an actual kb that we all use based on 1024. When I was
going over the algorithm I saw the 1024 and bw in the same scope and
bells went off. In point of fact, the BW measurement uses literal
kilos rather than computer kilos. Just like we still have that
confusing setup in FMS2 where all values are in 'bytes' but a 10mb
license is in bits. So dividing by 8 is still the same as it would be
with an 8-bit processor, but the kb is 1000. Now maybe 24 bits per kb
doesn't amount to a hill of beans, but a mb here and a mb there--it
adds up.

However, if the 1024 that Peldi was using in his algorithm referred
to something else, then that's interesting!

Cheers,
Bill

On Dec 20, 2005, at 4:56 AM, Stefan Richter wrote:

> Ah I see.
>
> So you mean that when dealing with kbps (data transmission speed)
> it's often
> the 1000 multiplier that's used whereas when dealing with
> conversion from
> bits to kilobits in general it's 1024 (and rightly so).
>
> Very interesting... And confusing.
> I agree, this is very important to get right when dealing with a
> product
> that's licensed and priced based on bandwidth usage.
>
> You're right, when dealing with data transmission speeds (bitrates)
> it's the
> multiplier of 1000 that's used. I was first under the impression
> that you
> wanted to establish how many bits are in a kilobit - and in non-data
> transmission circumstances it would be 1024.
>
> I am keen to hear more on this subject.
>
> Stefan
>
>
> 
>
>
> =-----------------------------------------------------------
> Supported by Fig Leaf Software - http://www.figleaf.com
> =-----------------------------------------------------------
>
> To change your subscription options or search the archive:
> http://chattyfig.figleaf.com/mailman/listinfo/flashcomm

bill sanders | www.sandlight.com | bloomfield, ct | 860-242-2260



=-----------------------------------------------------------
Supported by Fig Leaf Software - http://www.figleaf.com
=-----------------------------------------------------------

To change your subscription options or search the archive:
http://chattyfig.figleaf.com/mailman/listinfo/flashcomm

[ Post a follow-up to this message ]