S/PDIF encodes the audio using PCM frames, which are electrically encoded using Biphase Mark Code
A S/PDIF representation of a "zero-crossing" would be a series of PCM frames where the binary value in a channel goes from >0 to 100 or so "electrical" zero-crossings. This is by design. Even a completely silent audio signal (all numerical 0) is encoded by a repetitive, square-wavish mark/space sequence on the channel.
There is ZERO harmonic correlation between the encoded S/PDIF signal and the logical analog signal. This is by design... to ensure a noise-resistant channel by which clocking and data frames can be transmitted with resiliency.
Moreover, most output DACs don't actually use the incoming clock signal for playback. They use an internal oscillator. The input clock is merely used to delineate frames and get data into intermediate buffers.
But the playback clock in the DAC in my amplifier is not actually slaved to the S/PDIF bitstream. That clock is only used for unpacking the frames themselves -- it gets thrown into a buffer before playback. That's where the digital EQ and 5.1 matrixing happens... A separate, internal clock is used for analog output.
This is standard for all but the shittiest designs available now.
Stopband is up near 18kHz. You don't push it up to 20 because the rolloff is too sharp and messes audibly with the phase, and you really can't hear those frequencies anyway.
So when you do this, and then sample with an ADC, then the audio doesn't contain "missing" frequencies or any of the aliased energy in the output. You removed them in the first place. Now if somehow there was a magically clean signal path, all the way from microphone, through the studio, and analog reproductive medium, into your playback hardware, and out to your monitors, where the 20-40kHz band was completely unmolested, I would have to say this to you:
Only your dog would notice. You'd hear the EXACT same thing that you would if you listened to the filtered audio that was sampled digitally and reproduced. And you and I both know you'd fail the double blind test.
Besides... it's pointless anyway, when you consider mic preamps, effect boxes, and mixing desks all potentially contain their own lowpass filters for various performance and quality reasons.
But you're supposed to run the signal through a low-pass filter before you sample. That's how you avoid aliasing artifacts and slew-rate issues. Oh wait, you do the EXACT same filtering when you go to vinyl! (If the mastering engineer doesn't do this, which I'm sure he will, the lathe at the plant does this at a higher cutoff to protect the equipment)
If these supposedly missed frequencies aren't even in the passband, well...
The correct way to mix audio in fixed point is as follows:
1) Convert your gain or envelope from a floating point number to a fraction (G/256 or G/65536) 2) Multiply the track by the instantaneous gain/attenuation factor G (but don't divide yet). 3) Add masking noise 4) Sum across all mixed tracks 5) Divide by (N*256 or N*65536) where N is the number of mixed tracks
You can do this accurately with all 32-bit quantities if your tracks are 16-bit. If you need 24 or 32-bit fidelity, then you're already considering floats which are probably 64-bit, and 64-bit integer math works just as well.
OTH, a totally 32-bit FP has other benefits, but it's more interesting if it can come straight from the sampling equipment that way.
If you spent $3500 on an amplifier for driving 4 or fewer speakers, then you spent $2500 too much. I don't care what audiophiles say, every recording engineer who made the songs you listened to through it will tell you the same. A, let's say, Mackie m1400 is $300, retail. It can drive 1400 watts with 0.025% THD from 20-20000Hz. That's pretty fucking impressive...
I'm not saying you should have bought a rackmount power amplifier with cooling fans, but your money goes a long way for quality if you stay away from the money pit that is prosumer and boutique audio.
Bluetooth, dock/undock, hibernate, gaming, it all works. And IE is completely declawed so you don't even if you accidentally open an untrusted URL in it, you're not going to get adware toolbars installed and your NDIS stack rejiggered.
Because that was one of the few unique features of the game that I thought made it fun. It was like you were a rocket man, and the levels were designed to work with that (big open spaces).
We will take the book and its metaphors under consideration, but we will not speculate as to what "happens" when you are dead, except in venues of philosophy, neuroscience psychology, where we take reasoned approaches to understanding the nature of consciousness, existence, and so forth.
These are topics this particular book doesn't really get into in any consistent detail. So you'll forgive us if we're investigating in some different directions.
You can do application isolation in one of four ways. The lighter-weight of the three require installing kernel patches and/or additional management tools but are not particularly invasive.
Linux VServer is the closest to FreeBSD jails; it uses a few kernel patches to add isolation contexts to processes and views of network interfaces. Unlike jail(8) you need to start a new init(8) in each context. But like jail all processes share the original scheduler and VM, and use the host system's filesystem natively, only chroot'd.
Another more comprehensive solution is OpenVZ. This is OS-level virtualization that is similar to Solaris Zones. It requires more extensive kernel patches and provisioning tools. It is designed primarily for dedicated server hosting scenarios.
Then there is UML (User Mode Linux). This is linux on linux emulation that runs a special version of the kernel as a userspace process. You give the UML guest a file which behaves like a hard disk from the perspective of that guest, or you can choose to use a special filesystem driver that uses a "real" directory up on the host. This is close to VMWare or Xen-like solutions except it can't run an unmodified OS, only one given a special UML kernel which doesn't really have hardware drivers -- it shunts the actual work in syscalls up to the host OS.
Finally there is KVM, which is included in the Linux kernel as of recently without patches. This is a lightweight hardware-assisted virtualization solution in the style of Xen that can run unmodified guest operating systems.
Anyone can do that... that's kid stuff. The most effective trolls are (at least periphially) on topic, shy to the left or the right of the most vocal posters in the thread, push the right buttons, subtle in logical inconsistency, and posted from accounts with UIDs less than one million.
The bulk transfer mode should have allowed endpoints to send unsolicited packets in the open timeslots, however, rather than using a one-shot "okay let's see if anyone wants to transfer" approach. All that's needed is a support for a HUB-supplied NAK when an existing transfer from a different endpoint is occuring -- and a nice HUB might even store and forward.
a) Time Warner (Roadrunner) is the last ISP I would want to use. Ask any of their customers. b) I still don't understand why you would mesh 10+ nodes and then try routing above layer 3. Look at the dismal performance of any non-supernode based P2P network to see what I'm talking about. c) No, I just felt like saying 'Fuck You'. I've never been to Linux World and I don't have a subscription to Linux Journal anymore, so sue me. d) You were trying to impress me. You were trying to put me in your place with your bullshit lists. e) I deploy Windows ADs and hybrid Windows/Unix networks all the time. You don't think I know the pros and cons of each operating environment? I've never shat on a Microsoft product that I haven't used and didn't deserve it. XP Tablet is a whole lot less of a clusterfuck than Vista on the tablets I've tried. Not that XP is that hot either. But Vista? Come on. Not at least until SP1. I'd put linux on one instead just for the sheer audacity of it -- trying different handwriting recognition engines to see what hilarious engrish comes out.
Oh, isn't that cute, a wannabe penguinhead who thinks they know more than me. Try again, young'in. When I say "worked professionally with it", I meant the... [snip]...
Okay, good for you. a) What ISP doesn't use Linux or FreeBSD. b) Why did you feel the need to use FreeSWAN to make a mesh network when you could use protection at layer-4 and let hardware routing do its job c) Just between us, Linux World is a circle jerk and Linux Journal is largely a joke except for 'diff -u' which is cribbed from kernel traffic, so no, maybe I had you figured wrong but this still doesn't impress me all that much.
I did, but it was the laptop I wanted, blah blah blah, I want a pony, blah blah blah kinda stupid to pay almost 50% more just so I can use a "free" OS.
Fine. You had a poor subjective experience, but you're making really poor characterizations based on insufficient anecdotal evidence. Enjoy Vista, you deserve it.
http://www.debuntu.org/how-to-connect-to-a-cisco-vpn-using-vpnc You shouldn't have to install the shitty Cisco VPN software and binary kernel modules just to make an IPSEC VPN tunnel... and the kernel breaking Citrix sounds like a bunch of bullshit really. I can see updates to X.org breaking icaclient if the libraries changed versions, but AFAIK it's statically linked with Motif so...
Note that the NSA is a division of the DoD that deals primarly with COMSEC and HUMINT issues, so they would be using the same type of precautions (probably fewer) than they require of their contractors who work with those materials.
Sensitive parts of the facility are EM shielded, certainly, and utilize white noise generators. But it is unlikely the use any encryption on systems and network links that are entirely within the secured facilities. I mean, they've got MPs armed with machine guns roaming around the base, so I think that's about as secure as you can get. Access control will be primarily physical (man traps, guards, photo IDs). User accounts restrictions will be on the same order as the rest DoD (8>characters, 90 or 180 day rotation, none of your last 5 passwords, 3 classes). Biometrics are unlikey, but NSA-designed smart cards are almost sure to provide two-factor authentication.
Key material (which the NSA is responsible for generating, disseminating, and destroying) will be encrypted at rest and in transit, however, as it is very sensitive. Also, links that go from the NSA secure areas to other DoD agencies, JWICS, etc. are encrypted and closely controlled and monitored by DISA.
(they use Cisco routers, firewalls, and other network equipment, for what that's worth, re: montioring and logging capabilities)
There is no connection from the classified network to the public internet at all, so there is (clearly) no logging going on there. Internet access would be provided elsewhere at Forte Meade using separate equipment.
None of this stuff is classified. Note that the NSA together with NIST make a lot of policy and procedures recommendations on this stuff to the rest of the DoD agencies and the US government in general. We get their best practices 10 years after they figure it out (so they always have an edge), but they put out good stuff when they feel like it.
Because I see that a lot. You get an SSH login to your web presence and suddenly you're the "linux guru" at your place of work.
And in that 14 years, you still haven't learned to check to see if your hardware is going to be supported in a recent kernel before you buy it? I guess you bought the wrong tablet PC, because we've deployed it quite successfully for use in idiot-proof AV systems, among other things.
As a linux user of 8 years, I think desktop distributions have come a long way in that time period, and I don't find them any more or less difficult to get working than a Windows OS.
I think the only reason that you were able to get Vista to work on your laptop is because it was supplied with an OEM support disk especially for your machine; they did all the hard work of getting it on there and correctly configured for you.
There are vendors who do this for linux if you care to look for them and pay the premium.
Evolution isn't necessarily a function of biological birth and death anymore, at least for humans. We are as much a product of our genes as we are of technology and society which are side-effects of our existance yet are primarily the basis for our food-chain superiority and (relative) well being/happiness.
That we can identify sources of risk and proactively eliminate them (rather than waiting for behaviorial conditioning over generations to do that work; i.e. fear of snakes, heights, etc.) means that social institutions can make us adapt our behavior much faster than biological means; this is critical to our further proliferation.
Insects can adapt to niches quickly because they have short lifespans and lots of offspring. Humans do the opposite. We assimilate and analyze data quickly to adapt to change using technology, while rearing few children with long lifespans (to better steward and disseminate the human-life-span outliving accumulated human knowledge and culture).
Our social systems are an evolutionary product; our continued evolution occurs in a cultural space, not so much the meat space, if you will.
So uh, yeah. No particular need to subject our children to sure bodily harm. Although learning-by-mistake is a powerful mechanism which I still think is valuable. A balance needs to be achieved, clearly.
Slashdot Mirror
That serial port can interface storage through , thereby expanding the virtual address space of the emulator to an arbitrary size.
S/PDIF encodes the audio using PCM frames, which are electrically encoded using Biphase Mark Code
A S/PDIF representation of a "zero-crossing" would be a series of PCM frames where the binary value in a channel goes from >0 to 100 or so "electrical" zero-crossings. This is by design. Even a completely silent audio signal (all numerical 0) is encoded by a repetitive, square-wavish mark/space sequence on the channel.
There is ZERO harmonic correlation between the encoded S/PDIF signal and the logical analog signal. This is by design... to ensure a noise-resistant channel by which clocking and data frames can be transmitted with resiliency.
Moreover, most output DACs don't actually use the incoming clock signal for playback. They use an internal oscillator. The input clock is merely used to delineate frames and get data into intermediate buffers.
But the playback clock in the DAC in my amplifier is not actually slaved to the S/PDIF bitstream. That clock is only used for unpacking the frames themselves -- it gets thrown into a buffer before playback. That's where the digital EQ and 5.1 matrixing happens... A separate, internal clock is used for analog output.
This is standard for all but the shittiest designs available now.
Stopband is up near 18kHz. You don't push it up to 20 because the rolloff is too sharp and messes audibly with the phase, and you really can't hear those frequencies anyway.
So when you do this, and then sample with an ADC, then the audio doesn't contain "missing" frequencies or any of the aliased energy in the output. You removed them in the first place.
Now if somehow there was a magically clean signal path, all the way from microphone, through the studio, and analog reproductive medium, into your playback hardware, and out to your monitors, where the 20-40kHz band was completely unmolested, I would have to say this to you:
Only your dog would notice. You'd hear the EXACT same thing that you would if you listened to the filtered audio that was sampled digitally and reproduced. And you and I both know you'd fail the double blind test.
Besides... it's pointless anyway, when you consider mic preamps, effect boxes, and mixing desks all potentially contain their own lowpass filters for various performance and quality reasons.
But you're supposed to run the signal through a low-pass filter before you sample. That's how you avoid aliasing artifacts and slew-rate issues.
Oh wait, you do the EXACT same filtering when you go to vinyl! (If the mastering engineer doesn't do this, which I'm sure he will, the lathe at the plant does this at a higher cutoff to protect the equipment)
If these supposedly missed frequencies aren't even in the passband, well...
The correct way to mix audio in fixed point is as follows:
1) Convert your gain or envelope from a floating point number to a fraction (G/256 or G/65536)
2) Multiply the track by the instantaneous gain/attenuation factor G (but don't divide yet).
3) Add masking noise
4) Sum across all mixed tracks
5) Divide by (N*256 or N*65536) where N is the number of mixed tracks
You can do this accurately with all 32-bit quantities if your tracks are 16-bit. If you need 24 or 32-bit fidelity, then you're already considering floats which are probably 64-bit, and 64-bit integer math works just as well.
OTH, a totally 32-bit FP has other benefits, but it's more interesting if it can come straight from the sampling equipment that way.
If you spent $3500 on an amplifier for driving 4 or fewer speakers, then you spent $2500 too much. I don't care what audiophiles say, every recording engineer who made the songs you listened to through it will tell you the same. A, let's say, Mackie m1400 is $300, retail. It can drive 1400 watts with 0.025% THD from 20-20000Hz. That's pretty fucking impressive...
I'm not saying you should have bought a rackmount power amplifier with cooling fans, but your money goes a long way for quality if you stay away from the money pit that is prosumer and boutique audio.
Bluetooth, dock/undock, hibernate, gaming, it all works. And IE is completely declawed so you don't even if you accidentally open an untrusted URL in it, you're not going to get adware toolbars installed and your NDIS stack rejiggered.
...if your home was infested with cockroaches.
There's a point in that battle where you seriously consider setting of a small neutron bomb, if only you could find one... OH SHIT, BRB: FBI!
NT
Because that was one of the few unique features of the game that I thought made it fun. It was like you were a rocket man, and the levels were designed to work with that (big open spaces).
We will take the book and its metaphors under consideration, but we will not speculate as to what "happens" when you are dead, except in venues of philosophy, neuroscience psychology, where we take reasoned approaches to understanding the nature of consciousness, existence, and so forth.
These are topics this particular book doesn't really get into in any consistent detail. So you'll forgive us if we're investigating in some different directions.
Need refs plz for us drawfags.
You can do application isolation in one of four ways. The lighter-weight of the three require installing kernel patches and/or additional management tools but are not particularly invasive.
Linux VServer is the closest to FreeBSD jails; it uses a few kernel patches to add isolation contexts to processes and views of network interfaces. Unlike jail(8) you need to start a new init(8) in each context. But like jail all processes share the original scheduler and VM, and use the host system's filesystem natively, only chroot'd.
Another more comprehensive solution is OpenVZ. This is OS-level virtualization that is similar to Solaris Zones. It requires more extensive kernel patches and provisioning tools. It is designed primarily for dedicated server hosting scenarios.
Then there is UML (User Mode Linux). This is linux on linux emulation that runs a special version of the kernel as a userspace process. You give the UML guest a file which behaves like a hard disk from the perspective of that guest, or you can choose to use a special filesystem driver that uses a "real" directory up on the host. This is close to VMWare or Xen-like solutions except it can't run an unmodified OS, only one given a special UML kernel which doesn't really have hardware drivers -- it shunts the actual work in syscalls up to the host OS.
Finally there is KVM, which is included in the Linux kernel as of recently without patches. This is a lightweight hardware-assisted virtualization solution in the style of Xen that can run unmodified guest operating systems.
I hope this helps.
Anyone can do that... that's kid stuff. The most effective trolls are (at least periphially) on topic, shy to the left or the right of the most vocal posters in the thread, push the right buttons, subtle in logical inconsistency, and posted from accounts with UIDs less than one million.
NT
I thought it was supposed to give you the best of all worlds - is google more accurate and targeted?
nt
The bulk transfer mode should have allowed endpoints to send unsolicited packets in the open timeslots, however, rather than using a one-shot "okay let's see if anyone wants to transfer" approach. All that's needed is a support for a HUB-supplied NAK when an existing transfer from a different endpoint is occuring -- and a nice HUB might even store and forward.
a) Time Warner (Roadrunner) is the last ISP I would want to use. Ask any of their customers.
b) I still don't understand why you would mesh 10+ nodes and then try routing above layer 3. Look at the dismal performance of any non-supernode based P2P network to see what I'm talking about.
c) No, I just felt like saying 'Fuck You'. I've never been to Linux World and I don't have a subscription to Linux Journal anymore, so sue me.
d) You were trying to impress me. You were trying to put me in your place with your bullshit lists.
e) I deploy Windows ADs and hybrid Windows/Unix networks all the time. You don't think I know the pros and cons of each operating environment? I've never shat on a Microsoft product that I haven't used and didn't deserve it.
XP Tablet is a whole lot less of a clusterfuck than Vista on the tablets I've tried. Not that XP is that hot either. But Vista? Come on. Not at least until SP1. I'd put linux on one instead just for the sheer audacity of it -- trying different handwriting recognition engines to see what hilarious engrish comes out.
Fuck.
Oh, isn't that cute, a wannabe penguinhead who thinks they know more than me. Try again, young'in. When I say "worked professionally with it", I meant the
Okay, good for you. a) What ISP doesn't use Linux or FreeBSD. b) Why did you feel the need to use FreeSWAN to make a mesh network when you could use protection at layer-4 and let hardware routing do its job c) Just between us, Linux World is a circle jerk and Linux Journal is largely a joke except for 'diff -u' which is cribbed from kernel traffic, so no, maybe I had you figured wrong but this still doesn't impress me all that much.
I did, but it was the laptop I wanted, blah blah blah, I want a pony, blah blah blah kinda stupid to pay almost 50% more just so I can use a "free" OS.
Fine. You had a poor subjective experience, but you're making really poor characterizations based on insufficient anecdotal evidence. Enjoy Vista, you deserve it.
http://www.debuntu.org/how-to-connect-to-a-cisco-vpn-using-vpnc
You shouldn't have to install the shitty Cisco VPN software and binary kernel modules just to make an IPSEC VPN tunnel... and the kernel breaking Citrix sounds like a bunch of bullshit really. I can see updates to X.org breaking icaclient if the libraries changed versions, but AFAIK it's statically linked with Motif so...
Most of that is not likely.
Note that the NSA is a division of the DoD that deals primarly with COMSEC and HUMINT issues, so they would be using the same type of precautions (probably fewer) than they require of their contractors who work with those materials.
Sensitive parts of the facility are EM shielded, certainly, and utilize white noise generators. But it is unlikely the use any encryption on systems and network links that are entirely within the secured facilities. I mean, they've got MPs armed with machine guns roaming around the base, so I think that's about as secure as you can get. Access control will be primarily physical (man traps, guards, photo IDs). User accounts restrictions will be on the same order as the rest DoD (8>characters, 90 or 180 day rotation, none of your last 5 passwords, 3 classes). Biometrics are unlikey, but NSA-designed smart cards are almost sure to provide two-factor authentication.
Key material (which the NSA is responsible for generating, disseminating, and destroying) will be encrypted at rest and in transit, however, as it is very sensitive. Also, links that go from the NSA secure areas to other DoD agencies, JWICS, etc. are encrypted and closely controlled and monitored by DISA.
(they use Cisco routers, firewalls, and other network equipment, for what that's worth, re: montioring and logging capabilities)
There is no connection from the classified network to the public internet at all, so there is (clearly) no logging going on there. Internet access would be provided elsewhere at Forte Meade using separate equipment.
None of this stuff is classified. Note that the NSA together with NIST make a lot of policy and procedures recommendations on this stuff to the rest of the DoD agencies and the US government in general. We get their best practices 10 years after they figure it out (so they always have an edge), but they put out good stuff when they feel like it.
Because I see that a lot. You get an SSH login to your web presence and suddenly you're the "linux guru" at your place of work.
And in that 14 years, you still haven't learned to check to see if your hardware is going to be supported in a recent kernel before you buy it? I guess you bought the wrong tablet PC, because we've deployed it quite successfully for use in idiot-proof AV systems, among other things.
As a linux user of 8 years, I think desktop distributions have come a long way in that time period, and I don't find them any more or less difficult to get working than a Windows OS.
I think the only reason that you were able to get Vista to work on your laptop is because it was supplied with an OEM support disk especially for your machine; they did all the hard work of getting it on there and correctly configured for you.
There are vendors who do this for linux if you care to look for them and pay the premium.
Evolution isn't necessarily a function of biological birth and death anymore, at least for humans. We are as much a product of our genes as we are of technology and society which are side-effects of our existance yet are primarily the basis for our food-chain superiority and (relative) well being/happiness.
That we can identify sources of risk and proactively eliminate them (rather than waiting for behaviorial conditioning over generations to do that work; i.e. fear of snakes, heights, etc.) means that social institutions can make us adapt our behavior much faster than biological means; this is critical to our further proliferation.
Insects can adapt to niches quickly because they have short lifespans and lots of offspring. Humans do the opposite. We assimilate and analyze data quickly to adapt to change using technology, while rearing few children with long lifespans (to better steward and disseminate the human-life-span outliving accumulated human knowledge and culture).
Our social systems are an evolutionary product; our continued evolution occurs in a cultural space, not so much the meat space, if you will.
So uh, yeah. No particular need to subject our children to sure bodily harm. Although learning-by-mistake is a powerful mechanism which I still think is valuable. A balance needs to be achieved, clearly.