Home » The PARIS Forums » PARIS: Main » OT: Waster CPU Cycles when tracking @ 44.1/48k?
| OT: Waster CPU Cycles when tracking @ 44.1/48k? [message #93095] |
Mon, 26 November 2007 21:50  |
|
|
Guys, I just read (Sound on Sound DEC 07 ) that those of us who track @ 44.1
& 48k are actually using "More" CPU cycles than those who track & Mix at
88.2/96/192..WOW!!
Meaning (If I'm getting this right) The computer works Harder to track and
mix sessions at those low rates (44.1/48k)..
If this is correcxt, then starting fresh in 08(after this last mixing session)
, all of my session will be @ 88.2.
Since hard driver are going for very little theses days (500 gig seagate
ATA 3.0 $99.00 bucks everywhere, this makes going to the higher rates a must..
Please (all) chime in..DJ, Neil, Thad, John,Chris, Bill, Kim, Don, Sarah,
Chuck, Doug W etc..
|
|
|
|
| Re: OT: Wasted CPU Cycles when tracking @ 44.1/48k? [message #93096 is a reply to message #93095] |
Mon, 26 November 2007 21:52  |
|
|
Sorry
>
>Guys, I just read (Sound on Sound DEC 07 ) that those of us who track @
44.1
>& 48k are actually using "More" CPU cycles than those who track & Mix at
>88.2/96/192..WOW!!
>
>Meaning (If I'm getting this right) The computer works Harder to track and
>mix sessions at those low rates (44.1/48k)..
>
>If this is correcxt, then starting fresh in 08(after this last mixing session)
>, all of my session will be @ 88.2.
>
>Since hard driver are going for very little theses days (500 gig seagate
>ATA 3.0 $99.00 bucks everywhere, this makes going to the higher rates a
must..
>
>Please (all) chime in..DJ, Neil, Thad, John,Chris, Bill, Kim, Don, Sarah,
>Chuck, Doug W etc..
|
|
|
|
| Re: Waster CPU Cycles when tracking @ 44.1/48k? [message #93098 is a reply to message #93095] |
Mon, 26 November 2007 21:01 |
Aaron Allen
Messages: 1988
Registered: May 2008
|
Senior Member |
|
|
How do they verify this information as correct?
AA
"LaMont" <jjdpro@gmail.com> wrote in message news:474ba231$1@linux...
>
> Guys, I just read (Sound on Sound DEC 07 ) that those of us who track @
> 44.1
> & 48k are actually using "More" CPU cycles than those who track & Mix at
> 88.2/96/192..WOW!!
>
> Meaning (If I'm getting this right) The computer works Harder to track and
> mix sessions at those low rates (44.1/48k)..
>
> If this is correcxt, then starting fresh in 08(after this last mixing
> session)
> , all of my session will be @ 88.2.
>
> Since hard driver are going for very little theses days (500 gig seagate
> ATA 3.0 $99.00 bucks everywhere, this makes going to the higher rates a
> must..
>
> Please (all) chime in..DJ, Neil, Thad, John,Chris, Bill, Kim, Don, Sarah,
> Chuck, Doug W etc..
|
|
|
|
| Re: Waster CPU Cycles when tracking @ 44.1/48k? [message #93099 is a reply to message #93095] |
Mon, 26 November 2007 21:09 |
DJ
Messages: 1124
Registered: July 2005
|
Senior Member |
|
|
I doubt this. The HD is having to stream twice the data. How could this be
less resource intensive?
"LaMont" <jjdpro@gmail.com> wrote in message news:474ba231$1@linux...
>
> Guys, I just read (Sound on Sound DEC 07 ) that those of us who track @
> 44.1
> & 48k are actually using "More" CPU cycles than those who track & Mix at
> 88.2/96/192..WOW!!
>
> Meaning (If I'm getting this right) The computer works Harder to track and
> mix sessions at those low rates (44.1/48k)..
>
> If this is correcxt, then starting fresh in 08(after this last mixing
> session)
> , all of my session will be @ 88.2.
>
> Since hard driver are going for very little theses days (500 gig seagate
> ATA 3.0 $99.00 bucks everywhere, this makes going to the higher rates a
> must..
>
> Please (all) chime in..DJ, Neil, Thad, John,Chris, Bill, Kim, Don, Sarah,
> Chuck, Doug W etc..
|
|
|
|
| Re: OT: Waster CPU Cycles when tracking @ 44.1/48k? [message #93102 is a reply to message #93095] |
Mon, 26 November 2007 23:59 |
Kim
Messages: 1246
Registered: October 2005
|
Senior Member |
|
|
The only logical support I can see for this idea would be something to do
with data flow and bus size depending on the app (but it's a stretch). Kinda
like 64 bit computers are more or less just wasting the extra 32 bits much
of the time.
I can't see immediately how this would work out though. One potential is
that perhaps at 88.2 they ship two 24 bit packets in one hit, making a single
48 bit packet, precisely 3 x 16. Make it stereo and you're 6 x 16 or 3 x
32 which fits neatly down the bus. Of course you could do the same with 44.1
but you'd need to pack the data appropriately. Perhaps some apps don't.
Even on the wild stretch that I'm correct in this theory or something similar,
I can't see how you'd end up using LESS. You would use less than double,
sure, but not less overall.
They may be right, but I'm not a believer yet.
Cheers,
Kim.
"LaMont" <jjdpro@gmail.com> wrote:
>
>Guys, I just read (Sound on Sound DEC 07 ) that those of us who track @
44.1
>& 48k are actually using "More" CPU cycles than those who track & Mix at
>88.2/96/192..WOW!!
>
>Meaning (If I'm getting this right) The computer works Harder to track and
>mix sessions at those low rates (44.1/48k)..
>
>If this is correcxt, then starting fresh in 08(after this last mixing session)
>, all of my session will be @ 88.2.
>
>Since hard driver are going for very little theses days (500 gig seagate
>ATA 3.0 $99.00 bucks everywhere, this makes going to the higher rates a
must..
>
>Please (all) chime in..DJ, Neil, Thad, John,Chris, Bill, Kim, Don, Sarah,
>Chuck, Doug W etc..
|
|
|
|
| Re: OT: Waster CPU Cycles when tracking @ 44.1/48k? [message #93105 is a reply to message #93102] |
Tue, 27 November 2007 01:21 |
DJ
Messages: 1124
Registered: July 2005
|
Senior Member |
|
|
"Kim" <hiddensounds@hotmail.com> wrote in message news:474bc063$1@linux...
>
>
> The only logical support I can see for this idea would be something to do
> with data flow and bus size depending on the app (but it's a stretch).
> Kinda
> like 64 bit computers are more or less just wasting the extra 32 bits much
> of the time.
>
> I can't see immediately how this would work out though. One potential is
> that perhaps at 88.2 they ship two 24 bit packets in one hit, making a
> single
> 48 bit packet, precisely 3 x 16. Make it stereo and you're 6 x 16 or 3 x
> 32 which fits neatly down the bus. Of course you could do the same with
> 44.1
> but you'd need to pack the data appropriately. Perhaps some apps don't.
>
> Even on the wild stretch that I'm correct in this theory or something
> similar,
> I can't see how you'd end up using LESS. You would use less than double,
> sure, but not less overall.
>
> They may be right, but I'm not a believer yet.
>
> Cheers,
> Kim.
Perhaps so, but we're talking about overall system resource use at 88.2 as
well. You certainly don't get additional capabilities to run plugins, for
instance, and when using dedicated DSP cards like the UAD-1, system
resources are half what's available at 44.1. I'm going to have to get this
issue and read the article. SOS is usually pretty on top of things when it
comes to accuracy in their articles. This sounds to me like a big FU as far
as taking the overall big picture into account.
|
|
|
|
| Re: OT: Waster CPU Cycles when tracking @ 44.1/48k? [message #93106 is a reply to message #93105] |
Tue, 27 November 2007 03:06 |
Kim
Messages: 1246
Registered: October 2005
|
Senior Member |
|
|
Well yes you're dead right. Even in the scenario I described we're talking
about a "less than double" scenario, not a "less than status quo" situation.
Like you say, hardware based stuff will pretty much always use precisely
double the resources. Indeed I recall some sketch ideas one of the experts
(chuck maybe?) put forward about the highly unlikely possibility of running
a Paris system at 88.2 and that, with EDS at least, you could probably program
it to simply process double the stuff, but you'd end up with half the tracks...
ignoring fundamentals like changing the converters etc.
And effects calculations for double the frequency are not like, say, increasing
from 24 bit to 48 bit audio. With 88.2 you literally simply have twice as
many numbers rather than numbers twice as big. Sometimes numbers twice as
big can be pushed through more efficiently, especially if you're on 64 bit
hardware, but if you've got twice as many numbers it pretty much just takes
twice as many cycles.
It's fairly much identical in every respect to trying to play a 44.1 project
twice as quickly on your DAW. It seems pretty clear that it will just take
twice as much grunt.
I'll be interested to hear more details. It does seem that either there is
a FU of some sort, or that there is some missing info here. It wouldn't surprise
me if there are hidden efficiencies somewhere, but I'd be amazingly surprised
if you used less grunt overall. Bamboozled in fact. ;o)
Cheers,
Kim.
>Perhaps so, but we're talking about overall system resource use at 88.2
as
>well. You certainly don't get additional capabilities to run plugins, for
>instance, and when using dedicated DSP cards like the UAD-1, system
>resources are half what's available at 44.1. I'm going to have to get this
>issue and read the article. SOS is usually pretty on top of things when
it
>comes to accuracy in their articles. This sounds to me like a big FU as
far
>as taking the overall big picture into account.
>
>
>
|
|
|
|
| Re: OT: Waster CPU Cycles when tracking @ 44.1/48k? [message #93108 is a reply to message #93106] |
Tue, 27 November 2007 03:48 |
Kim
Messages: 1246
Registered: October 2005
|
Senior Member |
|
|
Having said all that...
I do believe noting somewhere that they included a function with 64 bit processors
whereby they can actually act as two 32 bit processors working in parallel.
I think it required that the exact same operation was being performed in
each half, but this would appear to apply in audio. If you were able to use
that particular feature of the processor within a 64 bit OS (which I'm not
sure about) you just MIGHT almost break even... ignoring UAD scenarios
etc, but on the host only side, you should theoretically come close. The
host app would obviously need to be written specifically to do this on 64
bit hardware, but it may be a possibility.
You still wouldn't use less of course, but you might only use, say, 5-10%
more grunt than otherwise given that memory bandwidth is also 64 bit. Your
only real bottleneck would be HDD access and, well, any other parts of the
DAW software which couldn't use this feature but still had to deal with the
extra data. I could envisage this might be minimal though.
So I could see that maybe, on 64 bit hardware with specifically written host
software to suit, you may come close to the same performance as 44.1...
....maybe...
....did I say maybe?
Best I can come up with. Anybody else?
Cheers,
Kim.
"Kim" <hiddensounds@hotmail.com> wrote:
>
>
>Well yes you're dead right. Even in the scenario I described we're talking
>about a "less than double" scenario, not a "less than status quo" situation.
>Like you say, hardware based stuff will pretty much always use precisely
>double the resources. Indeed I recall some sketch ideas one of the experts
>(chuck maybe?) put forward about the highly unlikely possibility of running
>a Paris system at 88.2 and that, with EDS at least, you could probably program
>it to simply process double the stuff, but you'd end up with half the tracks...
>ignoring fundamentals like changing the converters etc.
>
>And effects calculations for double the frequency are not like, say, increasing
>from 24 bit to 48 bit audio. With 88.2 you literally simply have twice as
>many numbers rather than numbers twice as big. Sometimes numbers twice as
>big can be pushed through more efficiently, especially if you're on 64 bit
>hardware, but if you've got twice as many numbers it pretty much just takes
>twice as many cycles.
>
>It's fairly much identical in every respect to trying to play a 44.1 project
>twice as quickly on your DAW. It seems pretty clear that it will just take
>twice as much grunt.
>
>I'll be interested to hear more details. It does seem that either there
is
>a FU of some sort, or that there is some missing info here. It wouldn't
surprise
>me if there are hidden efficiencies somewhere, but I'd be amazingly surprised
>if you used less grunt overall. Bamboozled in fact. ;o)
>
>Cheers,
>Kim.
>
>>Perhaps so, but we're talking about overall system resource use at 88.2
>as
>>well. You certainly don't get additional capabilities to run plugins, for
>
>>instance, and when using dedicated DSP cards like the UAD-1, system
>>resources are half what's available at 44.1. I'm going to have to get this
>
>>issue and read the article. SOS is usually pretty on top of things when
>it
>>comes to accuracy in their articles. This sounds to me like a big FU as
>far
>>as taking the overall big picture into account.
>>
>>
>>
>
|
|
|
|
|
|
| Re: OT: Waster CPU Cycles when tracking @ 44.1/48k? [message #93115 is a reply to message #93113] |
Tue, 27 November 2007 09:40 |
|
|
Guys & Gals, I'm sorry the article was from VI (Virtual Instruments) Magazine.
I think it's the Oct/Nov 07 issure.
Dedric Terry <dterry@keyofd.net> wrote:
>On 11/27/07 2:21 AM, in article 474be2af@linux, "DJ" <animix _ at _ animas
_
>dot _ net> wrote:
>
>>
>> Perhaps so, but we're talking about overall system resource use at 88.2
as
>> well. You certainly don't get additional capabilities to run plugins,
for
>> instance, and when using dedicated DSP cards like the UAD-1, system
>> resources are half what's available at 44.1. I'm going to have to get
this
>> issue and read the article. SOS is usually pretty on top of things when
it
>> comes to accuracy in their articles. This sounds to me like a big FU as
far
>> as taking the overall big picture into account.
>>
>
>I don't know DJ, I've been seeing more and more opinionated comments and
>using forums as resource material in SOS reviews.
>
>The cpu use/sample rate issue is of course easy to test and if SOS didn't
>even bother, then they are really getting sloppy.
>
>Dedric
>
|
|
|
|
| Re: OT: Waster CPU Cycles when tracking @ 44.1/48k? [message #93116 is a reply to message #93113] |
Tue, 27 November 2007 09:42 |
Neil
Messages: 1645
Registered: April 2006
|
Senior Member |
|
|
Based on my experience (not measurements of any kind, other
than looking at the CPU meter), this is not the case. I work
primarily in 88.2k, but have done perhaps 3 or 4 44.1k
projects in SX, and it definitely takes more CPU horsepower to
run the higher samplerates. Streaming the files is nothing for
the audio drives anymore, especially if you're using SATA...
I can stream around 40 tracks at 88.2k & the drive meter in SX
barely registers - that's with a Western Digiblah 7,200rpm with
the 8meg buffer - but the CPU absolutely takes a hit when
you're using the higher samplerates. You can see it on the
meter, AND you can feel it in the performance of the machine.
Neil
Dedric Terry <dterry@keyofd.net> wrote:
>On 11/27/07 2:21 AM, in article 474be2af@linux, "DJ" <animix _ at _ animas
_
>dot _ net> wrote:
>
>>
>> Perhaps so, but we're talking about overall system resource use at 88.2
as
>> well. You certainly don't get additional capabilities to run plugins,
for
>> instance, and when using dedicated DSP cards like the UAD-1, system
>> resources are half what's available at 44.1. I'm going to have to get
this
>> issue and read the article. SOS is usually pretty on top of things when
it
>> comes to accuracy in their articles. This sounds to me like a big FU as
far
>> as taking the overall big picture into account.
>>
>
>I don't know DJ, I've been seeing more and more opinionated comments and
>using forums as resource material in SOS reviews.
>
>The cpu use/sample rate issue is of course easy to test and if SOS didn't
>even bother, then they are really getting sloppy.
>
>Dedric
>
|
|
|
|
| Re: OT: Waster CPU Cycles when tracking @ 44.1/48k? [message #93119 is a reply to message #93095] |
Tue, 27 November 2007 08:52 |
EK Sound
Messages: 939
Registered: June 2005
|
Senior Member |
|
|
BUT... the larger files are more taxing on the drive itself to stream,
so you will end up with half the potential track count per drive.
David.
LaMont wrote:
> Guys, I just read (Sound on Sound DEC 07 ) that those of us who track @ 44.1
> & 48k are actually using "More" CPU cycles than those who track & Mix at
> 88.2/96/192..WOW!!
>
> Meaning (If I'm getting this right) The computer works Harder to track and
> mix sessions at those low rates (44.1/48k)..
>
> If this is correcxt, then starting fresh in 08(after this last mixing session)
> , all of my session will be @ 88.2.
>
> Since hard driver are going for very little theses days (500 gig seagate
> ATA 3.0 $99.00 bucks everywhere, this makes going to the higher rates a must..
>
> Please (all) chime in..DJ, Neil, Thad, John,Chris, Bill, Kim, Don, Sarah,
> Chuck, Doug W etc..
|
|
|
|
| Re: OT: Wasted CPU Cycles when tracking @ 44.1/48k? [message #93120 is a reply to message #93096] |
Tue, 27 November 2007 08:53 |
EK Sound
Messages: 939
Registered: June 2005
|
Senior Member |
|
|
You forgot to correct the speeling in the body as well... ;-)
David.
LaMont wrote:
> Sorry
>
>
>>Guys, I just read (Sound on Sound DEC 07 ) that those of us who track @
>
> 44.1
>
>>& 48k are actually using "More" CPU cycles than those who track & Mix at
>>88.2/96/192..WOW!!
>>
>>Meaning (If I'm getting this right) The computer works Harder to track and
>>mix sessions at those low rates (44.1/48k)..
>>
>>If this is correcxt, then starting fresh in 08(after this last mixing session)
>>, all of my session will be @ 88.2.
>>
>>Since hard driver are going for very little theses days (500 gig seagate
>>ATA 3.0 $99.00 bucks everywhere, this makes going to the higher rates a
>
> must..
>
>>Please (all) chime in..DJ, Neil, Thad, John,Chris, Bill, Kim, Don, Sarah,
>>Chuck, Doug W etc..
>
>
|
|
|
|
| Re: OT: Wasted CPU Cycles when tracking @ 44.1/48k? [message #93121 is a reply to message #93095] |
Tue, 27 November 2007 10:24 |
LaMont
Messages: 828
Registered: October 2005
|
Senior Member |
|
|
Opps, upun further review.. I mis-read the print. And it stated that If you
don't want to get half or quarter of the performance out of your DAW when
streaming samples, then work @ 44.1/48 k..
Sorry about that..Gota get a new eye glass persciption..
"LaMont" <jjdpro@gmail.com> wrote:
>
>Guys, I just read (Sound on Sound DEC 07 ) that those of us who track @
44.1
>& 48k are actually using "More" CPU cycles than those who track & Mix at
>88.2/96/192..WOW!!
>
>Meaning (If I'm getting this right) The computer works Harder to track and
>mix sessions at those low rates (44.1/48k)..
>
>If this is correcxt, then starting fresh in 08(after this last mixing session)
>, all of my session will be @ 88.2.
>
>Since hard driver are going for very little theses days (500 gig seagate
>ATA 3.0 $99.00 bucks everywhere, this makes going to the higher rates a
must..
>
>Please (all) chime in..DJ, Neil, Thad, John,Chris, Bill, Kim, Don, Sarah,
>Chuck, Doug W etc..
|
|
|
|
|
|
| Re: OT: Waster CPU Cycles when tracking @ 44.1/48k? [message #93125 is a reply to message #93116] |
Tue, 27 November 2007 12:17 |
rick
Messages: 1976
Registered: February 2006
|
Senior Member |
|
|
i agree, there's definitely a hit on all aspects.
On 28 Nov 2007 03:42:56 +1000, "Neil" <OIU@OIU.com> wrote:
>
>Based on my experience (not measurements of any kind, other
>than looking at the CPU meter), this is not the case. I work
>primarily in 88.2k, but have done perhaps 3 or 4 44.1k
>projects in SX, and it definitely takes more CPU horsepower to
>run the higher samplerates. Streaming the files is nothing for
>the audio drives anymore, especially if you're using SATA...
>I can stream around 40 tracks at 88.2k & the drive meter in SX
>barely registers - that's with a Western Digiblah 7,200rpm with
>the 8meg buffer - but the CPU absolutely takes a hit when
>you're using the higher samplerates. You can see it on the
>meter, AND you can feel it in the performance of the machine.
>
>Neil
>
>
>Dedric Terry <dterry@keyofd.net> wrote:
>>On 11/27/07 2:21 AM, in article 474be2af@linux, "DJ" <animix _ at _ animas
>_
>>dot _ net> wrote:
>>
>>>
>>> Perhaps so, but we're talking about overall system resource use at 88.2
>as
>>> well. You certainly don't get additional capabilities to run plugins,
>for
>>> instance, and when using dedicated DSP cards like the UAD-1, system
>>> resources are half what's available at 44.1. I'm going to have to get
>this
>>> issue and read the article. SOS is usually pretty on top of things when
>it
>>> comes to accuracy in their articles. This sounds to me like a big FU as
>far
>>> as taking the overall big picture into account.
>>>
>>
>>I don't know DJ, I've been seeing more and more opinionated comments and
>>using forums as resource material in SOS reviews.
>>
>>The cpu use/sample rate issue is of course easy to test and if SOS didn't
>>even bother, then they are really getting sloppy.
>>
>>Dedric
>>
|
|
|
|
| Re: OT: Wasted CPU Cycles when tracking @ 44.1/48k? [message #93128 is a reply to message #93121] |
Tue, 27 November 2007 13:11 |
Chris Ludwig
Messages: 868
Registered: May 2006
|
Senior Member |
|
|
Hi Lamont
I think you gonna need to clean out the bong AND get new glasses. :)
That said it sounds like some that would be passed around on the
Gearsluts forum. Hey did you know that audio sounds different playing
from a firewire than it does from a internal drive !!! NO really I can
hear the difference. But thats because of the pure silver oxygen free
firewire cables and SATA cables. :)
Chris
LaMont wrote:
> Opps, upun further review.. I mis-read the print. And it stated that If you
> don't want to get half or quarter of the performance out of your DAW when
> streaming samples, then work @ 44.1/48 k..
> Sorry about that..Gota get a new eye glass persciption..
>
>
> "LaMont" <jjdpro@gmail.com> wrote:
>> Guys, I just read (Sound on Sound DEC 07 ) that those of us who track @
> 44.1
>> & 48k are actually using "More" CPU cycles than those who track & Mix at
>> 88.2/96/192..WOW!!
>>
>> Meaning (If I'm getting this right) The computer works Harder to track and
>> mix sessions at those low rates (44.1/48k)..
>>
>> If this is correcxt, then starting fresh in 08(after this last mixing session)
>> , all of my session will be @ 88.2.
>>
>> Since hard driver are going for very little theses days (500 gig seagate
>> ATA 3.0 $99.00 bucks everywhere, this makes going to the higher rates a
> must..
>> Please (all) chime in..DJ, Neil, Thad, John,Chris, Bill, Kim, Don, Sarah,
>> Chuck, Doug W etc..
>
--
Chris Ludwig
ADK Pro Audio
(859) 635-5762
www.adkproaudio.com
chrisl@adkproaudio.com
|
|
|
|
|
|
| Re: OT: Waster CPU Cycles when tracking @ 44.1/48k? [message #93134 is a reply to message #93108] |
Tue, 27 November 2007 16:30 |
TCB
Messages: 1261
Registered: July 2007
|
Senior Member |
|
|
INTEGER! 64 bit INTEGER PROCESSORS!!!!!! PLEASE FOR THE LOVE OF GOD DON'T
MAKE ME GO OVER THIS AGAIN. FPU PROCESSING IS ALREADY 128 BIT WIDE ON MODERN
PROCESSORS, AND ALMOST ALWAYS PARALLEL. SO THE FPU (IF THE ALGORITHM IS PROPERLY
CODED) COULD DO TWO 64 BIT WORDS, FOUR 32 BIT WORDS, EIGHT 16 BIT WORDS,
ETC. AS LONG AS IT'S THE SAME PROCESS.
INTEGER! INTEGER! INTEGER!
Sorry,
TCB
"Kim" <hiddensounds@hotmail.com> wrote:
>
>
>Having said all that...
>
>I do believe noting somewhere that they included a function with 64 bit
processors
>whereby they can actually act as two 32 bit processors working in parallel.
>I think it required that the exact same operation was being performed in
>each half, but this would appear to apply in audio. If you were able to
use
>that particular feature of the processor within a 64 bit OS (which I'm not
>sure about) you just MIGHT almost break even... ignoring UAD scenarios
>etc, but on the host only side, you should theoretically come close. The
>host app would obviously need to be written specifically to do this on 64
>bit hardware, but it may be a possibility.
>
>You still wouldn't use less of course, but you might only use, say, 5-10%
>more grunt than otherwise given that memory bandwidth is also 64 bit. Your
>only real bottleneck would be HDD access and, well, any other parts of the
>DAW software which couldn't use this feature but still had to deal with
the
>extra data. I could envisage this might be minimal though.
>
>So I could see that maybe, on 64 bit hardware with specifically written
host
>software to suit, you may come close to the same performance as 44.1...
>
>...maybe...
>
>...did I say maybe?
>
>Best I can come up with. Anybody else?
>
>Cheers,
>Kim.
>
>"Kim" <hiddensounds@hotmail.com> wrote:
>>
>>
>>Well yes you're dead right. Even in the scenario I described we're talking
>>about a "less than double" scenario, not a "less than status quo" situation.
>>Like you say, hardware based stuff will pretty much always use precisely
>>double the resources. Indeed I recall some sketch ideas one of the experts
>>(chuck maybe?) put forward about the highly unlikely possibility of running
>>a Paris system at 88.2 and that, with EDS at least, you could probably
program
>>it to simply process double the stuff, but you'd end up with half the tracks...
>>ignoring fundamentals like changing the converters etc.
>>
>>And effects calculations for double the frequency are not like, say, increasing
>>from 24 bit to 48 bit audio. With 88.2 you literally simply have twice
as
>>many numbers rather than numbers twice as big. Sometimes numbers twice
as
>>big can be pushed through more efficiently, especially if you're on 64
bit
>>hardware, but if you've got twice as many numbers it pretty much just takes
>>twice as many cycles.
>>
>>It's fairly much identical in every respect to trying to play a 44.1 project
>>twice as quickly on your DAW. It seems pretty clear that it will just take
>>twice as much grunt.
>>
>>I'll be interested to hear more details. It does seem that either there
>is
>>a FU of some sort, or that there is some missing info here. It wouldn't
>surprise
>>me if there are hidden efficiencies somewhere, but I'd be amazingly surprised
>>if you used less grunt overall. Bamboozled in fact. ;o)
>>
>>Cheers,
>>Kim.
>>
>>>Perhaps so, but we're talking about overall system resource use at 88.2
>>as
>>>well. You certainly don't get additional capabilities to run plugins,
for
>>
>>>instance, and when using dedicated DSP cards like the UAD-1, system
>>>resources are half what's available at 44.1. I'm going to have to get
this
>>
>>>issue and read the article. SOS is usually pretty on top of things when
>>it
>>>comes to accuracy in their articles. This sounds to me like a big FU as
>>far
>>>as taking the overall big picture into account.
>>>
>>>
>>>
>>
>
|
|
|
|
|
|
|
|
| Re: OT: Waster CPU Cycles when tracking @ 44.1/48k? [message #93137 is a reply to message #93134] |
Tue, 27 November 2007 17:08 |
Kim
Messages: 1246
Registered: October 2005
|
Senior Member |
|
|
OK I think I've got it, but could you just go over one more time the bit
after "Now listen closely..." ;o)
"TCB" <nobody@ishere.com> wrote:
>
>INTEGER! 64 bit INTEGER PROCESSORS!!!!!! PLEASE FOR THE LOVE OF GOD DON'T
>MAKE ME GO OVER THIS AGAIN. FPU PROCESSING IS ALREADY 128 BIT WIDE ON MODERN
>PROCESSORS, AND ALMOST ALWAYS PARALLEL. SO THE FPU (IF THE ALGORITHM IS
PROPERLY
>CODED) COULD DO TWO 64 BIT WORDS, FOUR 32 BIT WORDS, EIGHT 16 BIT WORDS,
>ETC. AS LONG AS IT'S THE SAME PROCESS.
>
>INTEGER! INTEGER! INTEGER!
>
>Sorry,
>
>TCB
>
>"Kim" <hiddensounds@hotmail.com> wrote:
>>
>>
>>Having said all that...
>>
>>I do believe noting somewhere that they included a function with 64 bit
>processors
>>whereby they can actually act as two 32 bit processors working in parallel.
>>I think it required that the exact same operation was being performed in
>>each half, but this would appear to apply in audio. If you were able to
>use
>>that particular feature of the processor within a 64 bit OS (which I'm
not
>>sure about) you just MIGHT almost break even... ignoring UAD scenarios
>>etc, but on the host only side, you should theoretically come close. The
>>host app would obviously need to be written specifically to do this on
64
>>bit hardware, but it may be a possibility.
>>
>>You still wouldn't use less of course, but you might only use, say, 5-10%
>>more grunt than otherwise given that memory bandwidth is also 64 bit. Your
>>only real bottleneck would be HDD access and, well, any other parts of
the
>>DAW software which couldn't use this feature but still had to deal with
>the
>>extra data. I could envisage this might be minimal though.
>>
>>So I could see that maybe, on 64 bit hardware with specifically written
>host
>>software to suit, you may come close to the same performance as 44.1...
>>
>>...maybe...
>>
>>...did I say maybe?
>>
>>Best I can come up with. Anybody else?
>>
>>Cheers,
>>Kim.
>>
>>"Kim" <hiddensounds@hotmail.com> wrote:
>>>
>>>
>>>Well yes you're dead right. Even in the scenario I described we're talking
>>>about a "less than double" scenario, not a "less than status quo" situation.
>>>Like you say, hardware based stuff will pretty much always use precisely
>>>double the resources. Indeed I recall some sketch ideas one of the experts
>>>(chuck maybe?) put forward about the highly unlikely possibility of running
>>>a Paris system at 88.2 and that, with EDS at least, you could probably
>program
>>>it to simply process double the stuff, but you'd end up with half the
tracks...
>>>ignoring fundamentals like changing the converters etc.
>>>
>>>And effects calculations for double the frequency are not like, say, increasing
>>>from 24 bit to 48 bit audio. With 88.2 you literally simply have twice
>as
>>>many numbers rather than numbers twice as big. Sometimes numbers twice
>as
>>>big can be pushed through more efficiently, especially if you're on 64
>bit
>>>hardware, but if you've got twice as many numbers it pretty much just
takes
>>>twice as many cycles.
>>>
>>>It's fairly much identical in every respect to trying to play a 44.1 project
>>>twice as quickly on your DAW. It seems pretty clear that it will just
take
>>>twice as much grunt.
>>>
>>>I'll be interested to hear more details. It does seem that either there
>>is
>>>a FU of some sort, or that there is some missing info here. It wouldn't
>>surprise
>>>me if there are hidden efficiencies somewhere, but I'd be amazingly surprised
>>>if you used less grunt overall. Bamboozled in fact. ;o)
>>>
>>>Cheers,
>>>Kim.
>>>
>>>>Perhaps so, but we're talking about overall system resource use at 88.2
>>>as
>>>>well. You certainly don't get additional capabilities to run plugins,
>for
>>>
>>>>instance, and when using dedicated DSP cards like the UAD-1, system
>>>>resources are half what's available at 44.1. I'm going to have to get
>this
>>>
>>>>issue and read the article. SOS is usually pretty on top of things when
>>>it
>>>>comes to accuracy in their articles. This sounds to me like a big FU
as
>>>far
>>>>as taking the overall big picture into account.
>>>>
>>>>
>>>>
>>>
>>
>
|
|
|
|
| Re: OT: Wasted CPU Cycles when tracking @ 44.1/48k? [message #93150 is a reply to message #93135] |
Wed, 28 November 2007 08:56 |
Paul Artola
Messages: 161
Registered: November 2005
|
Senior Member |
|
|
I have found that mixes saved to disk-based storage sound brittle and
bright. That's why I prefer using tape-based storage for a warmer and
more digital saturation type sound. Digital data clipping is not
pretty, but with tape, those missing bits seem to show up somehow.
9-track, baby, that's the way to go.
- Paul Artola
Ellicott City, Maryland
On 28 Nov 2007 10:38:17 +1000, "TCB" <nobody@ishere.com> wrote:
>
>Why do I do I even put up with you fucking idiots? The command queuing on
>SCSI controllers is the only way to get even close to a pro sound. But only
>on U160 controllers because the disk saturates better than on the U320 ones.
>Trust me, the only thing that sounds anywhere near as warm and punchy as
>SCSU U160 is SAS fibre channel and it's even more expensive.
>
>It's all in the command queuing, but you're all to deaf to fucking hear it.
>
>
>TCB
>
>P.S. Don't forget to make a small partition on the outer edge of the drive
>for your time code, and cut a small slice between it and the main data partition
>so the SMPTE doesn't bleed into the rest of your tracks. Now I've given away
>all of my secrets, and for what? You morons are all going to keep using SATA
>anyway . . .
|
|
|
|
| Re: OT: Wasted CPU Cycles when tracking @ 44.1/48k? [message #93152 is a reply to message #93150] |
Wed, 28 November 2007 09:18 |
DJ
Messages: 1124
Registered: July 2005
|
Senior Member |
|
|
My DATs are Phat, but I usually bake them to get that "warm" vibe.
"Paul Artola" <artola@comcast.net> wrote in message
news:j57rk3123h1ntqbtda14uo4cvogo2cmu59@4ax.com...
>I have found that mixes saved to disk-based storage sound brittle and
> bright. That's why I prefer using tape-based storage for a warmer and
> more digital saturation type sound. Digital data clipping is not
> pretty, but with tape, those missing bits seem to show up somehow.
> 9-track, baby, that's the way to go.
>
> - Paul Artola
> Ellicott City, Maryland
>
> On 28 Nov 2007 10:38:17 +1000, "TCB" <nobody@ishere.com> wrote:
>
>>
>>Why do I do I even put up with you fucking idiots? The command queuing on
>>SCSI controllers is the only way to get even close to a pro sound. But
>>only
>>on U160 controllers because the disk saturates better than on the U320
>>ones.
>>Trust me, the only thing that sounds anywhere near as warm and punchy as
>>SCSU U160 is SAS fibre channel and it's even more expensive.
>>
>>It's all in the command queuing, but you're all to deaf to fucking hear
>>it.
>>
>>
>>TCB
>>
>>P.S. Don't forget to make a small partition on the outer edge of the drive
>>for your time code, and cut a small slice between it and the main data
>>partition
>>so the SMPTE doesn't bleed into the rest of your tracks. Now I've given
>>away
>>all of my secrets, and for what? You morons are all going to keep using
>>SATA
>>anyway . . .
>
|
|
|
|
| Re: OT: Wasted CPU Cycles when tracking @ 44.1/48k? [message #93156 is a reply to message #93128] |
Wed, 28 November 2007 12:46 |
|
|
:)
Chris Ludwig <chrisl@adkproaudio.com> wrote:
>Hi Lamont
> I think you gonna need to clean out the bong AND get new glasses. :)
>That said it sounds like some that would be passed around on the
>Gearsluts forum. Hey did you know that audio sounds different playing
>from a firewire than it does from a internal drive !!! NO really I can
>hear the difference. But thats because of the pure silver oxygen free
>firewire cables and SATA cables. :)
>
>Chris
>
>
>
>LaMont wrote:
>> Opps, upun further review.. I mis-read the print. And it stated that If
you
>> don't want to get half or quarter of the performance out of your DAW when
>> streaming samples, then work @ 44.1/48 k..
>> Sorry about that..Gota get a new eye glass persciption..
>>
>>
>> "LaMont" <jjdpro@gmail.com> wrote:
>>> Guys, I just read (Sound on Sound DEC 07 ) that those of us who track
@
>> 44.1
>>> & 48k are actually using "More" CPU cycles than those who track & Mix
at
>>> 88.2/96/192..WOW!!
>>>
>>> Meaning (If I'm getting this right) The computer works Harder to track
and
>>> mix sessions at those low rates (44.1/48k)..
>>>
>>> If this is correcxt, then starting fresh in 08(after this last mixing
session)
>>> , all of my session will be @ 88.2.
>>>
>>> Since hard driver are going for very little theses days (500 gig seagate
>>> ATA 3.0 $99.00 bucks everywhere, this makes going to the higher rates
a
>> must..
>>> Please (all) chime in..DJ, Neil, Thad, John,Chris, Bill, Kim, Don, Sarah,
>>> Chuck, Doug W etc..
>>
>
>--
>Chris Ludwig
>
>ADK Pro Audio
>(859) 635-5762
>www.adkproaudio.com
>chrisl@adkproaudio.com
|
|
|
|
Goto Forum:
Current Time: Tue Nov 12 09:45:59 PST 2024
Total time taken to generate the page: 0.03720 seconds
|
The PARIS Forums
Search
Help
Members
Register
Login
Home
