Why are DD signals compressed? A digital 6-channel audio signal is still relatively low-bandwidth compared to video/data. You would think that even a wireless connection could transmit this in real-time, and certainly a wired-solution like S/PDIF shouldOriginally posted by: Accord99
Why would you want to worsen your audio quality by compressing two channel audio such as CDs/MP3s/non AC3 DivX into lossy DD when you can simply send them uncompressed using a SPDIF output?Originally posted by: batmanuel
The reason most HTPC users like Soundstorm is that it will encode ALL of the PCs audio streams into DD, so CDs/MP3s/DivX can be sent to the receiver in AC3 instead of being output in analog form. This is especially useful if you are using your PC as a PVR box and use it to playback recorded television on a regular basis.
DD is maximum 640 Kilobits/s, typically 448Kilobits/s, for 6 channels of audio using techniques similar to MP3. 2-channel CD audio is something like 1.4 megabits/s. While the cable should be able to handle higher bandwidths, it was never part of the standard and compression algorithms like DD/DTS were used for multi-channel audio. Until recently, Creative's method was the only way to carry 6 channels of uncompressed digital audio over a single cable.Originally posted by: Ichinisan
Why are DD signals compressed? A digital 6-channel audio signal is still relatively low-bandwidth compared to video/data. You would think that even a wireless connection could transmit this in real-time, and certainly a wired-solution like S/PDIF should![]()
Why bother compressing, when you can simply pass the original two channel audio untouched through the same cable.Originally posted by: flashbacck
Mostly for the convenience of having 1 cable.
Which many people will find annoying, especially for music. And any half-recent receiver will have various DSP effects or DDPL2 or DTS Neo to upmix the audio anyhow if you want.Also, all your audio streams will be upmixed to 5.1, whether you're watching TV, listening to mp3's or whatever.
Originally posted by: Insane3D
Originally posted by: mechBgon
Bigger picture: nVidia's APU isn't a PCI part, it's always lived on the southbridge natively. Making a PCI card out of it might not be as straightforward as it sounds.
Excellent point. I also wonder how much it would hogthe PCI bus if it was in a traditional PCI card format vs being mapped to the Hypertransport connection it's on now. I imagine real time DD encoding might eat some bandwidth...
Originally posted by: batmanuel
Originally posted by: Insane3D
Originally posted by: mechBgon
Bigger picture: nVidia's APU isn't a PCI part, it's always lived on the southbridge natively. Making a PCI card out of it might not be as straightforward as it sounds.
Excellent point. I also wonder how much it would hogthe PCI bus if it was in a traditional PCI card format vs being mapped to the Hypertransport connection it's on now. I imagine real time DD encoding might eat some bandwidth...
Might be exactly the reason a Soundstorm card hasn't come out yet, as nVidia needs PCI-E to make it work.
Originally posted by: Ichinisan
Originally posted by: batmanuel
Originally posted by: Insane3D
Originally posted by: mechBgon
Bigger picture: nVidia's APU isn't a PCI part, it's always lived on the southbridge natively. Making a PCI card out of it might not be as straightforward as it sounds.
Excellent point. I also wonder how much it would hogthe PCI bus if it was in a traditional PCI card format vs being mapped to the Hypertransport connection it's on now. I imagine real time DD encoding might eat some bandwidth...
Might be exactly the reason a Soundstorm card hasn't come out yet, as nVidia needs PCI-E to make it work.
No. It would not require any more bandwidth than the Audigy 2. It would be hardware-based, so the encoding would be done on the card itself.
Originally posted by: Ichinisan
Why are DD signals compressed? A digital 6-channel audio signal is still relatively low-bandwidth compared to video/data. You would think that even a wireless connection could transmit this in real-time, and certainly a wired-solution like S/PDIF shouldOriginally posted by: Accord99
Why would you want to worsen your audio quality by compressing two channel audio such as CDs/MP3s/non AC3 DivX into lossy DD when you can simply send them uncompressed using a SPDIF output?Originally posted by: batmanuel
The reason most HTPC users like Soundstorm is that it will encode ALL of the PCs audio streams into DD, so CDs/MP3s/DivX can be sent to the receiver in AC3 instead of being output in analog form. This is especially useful if you are using your PC as a PVR box and use it to playback recorded television on a regular basis.![]()
I tried using the "digital" RCA mini jack from my friend's Audigy(1) with his Yamaha 4.1 system and got nothing because it was COMPLETELY PROPRIETARY!![]()
Mostly for the convenience of having 1 cable. Also, all your audio streams will be upmixed to 5.1, whether you're watching TV, listening to mp3's or whatever.
Ok, I'll ask you again. Does the Audigy 2 ENCODE DD 5.1 real time....I'm NOT talking about DECODING.
It was configured properly. The speakers would not allow you to switch to digital mode because they couldn't find the signal, even though Creative's proprietary software was switched to "digital only". It was not acceptable to run a *long* optical cable from the Audigy Drive on the front. After further research (Audigy instructions), I found out that the digital non-optical signal from the Audigy is proprietary (only for Creative Inspire speakers).
ENCODING RAW AUDIO CHANNELS TO DOLBY WITH A PROPOSED HARDWARE-BASED CARD WILL NOT AFFECT PCI BANDWIDTH BECAUSE ENCODING OCCURS WITHIN THE CARD ITSELF.Originally posted by: Insane3D
Ok, I'll ask you again. Does the Audigy 2 ENCODE DD 5.1 real time....I'm NOT talking about DECODING.Originally posted by: Ichinisan
No. It would not require any more bandwidth than the Audigy 2. It would be hardware-based, so the encoding would be done on the card itself.Originally posted by: batmanuel
Might be exactly the reason a Soundstorm card hasn't come out yet, as nVidia needs PCI-E to make it work.Originally posted by: Insane3D
Excellent point. I also wonder how much it would hogthe PCI bus if it was in a traditional PCI card format vs being mapped to the Hypertransport connection it's on now. I imagine real time DD encoding might eat some bandwidth...Originally posted by: mechBgon
Bigger picture: nVidia's APU isn't a PCI part, it's always lived on the southbridge natively. Making a PCI card out of it might not be as straightforward as it sounds.
On-chip Gigabit LAN - NopeOriginally posted by: mechBgon
Ok, time to ask the obvious question: with SoundStorm-equipped motherboards starting in the $55 area (Asus A7N266-VM/AA), why not just smack a used $15 Duron and a $30 stick of PC2100 onto one, and build your HTPC around that?
Optical and coaxial S/PDIF-outs, check.
Optional S-Video-out thingamabob, check.
Full SoundStorm APU, check.
Low price, long life, stability, passive cooling, check.
Fits microATX cases, check.
Supports 200MHz-based and 266MHz-based SocketA processors including Applebred and Thorton, check (BIOS 1007 gives them formal recognition).
You want a SoundStorm-equipped HTPC, then track down an A7N266-VM and buy it while you still can.Even with the CPU, heatsink and memory, I bet it would be cost-competitive with the PCI-SoundStorm card you're envisioning.
If GB Lan is not on-chip, you'll need PCI-X to have enough bandwidth.
If you have PCI-X, then you really need a 64-bit processor to use that bandwidth.
Originally posted by: Pariah
If GB Lan is not on-chip, you'll need PCI-X to have enough bandwidth.
Theoretical GBit LAN can send 125MB/s which is below the the theoretical bandwidth of standard 32bit 33MHz PCI, 133MB/s. Real world GBit LAN is closer to 50 to 60MB/s max which is well within the capability of real world PCI which can be as high as 120MB/s.
In the first place, it's not an nF2 systemOriginally posted by: Ichinisan
On-chip Gigabit LAN - NopeOriginally posted by: mechBgon
Ok, time to ask the obvious question: with SoundStorm-equipped motherboards starting in the $55 area (Asus A7N266-VM/AA), why not just smack a used $15 Duron and a $30 stick of PC2100 onto one, and build your HTPC around that?
Optical and coaxial S/PDIF-outs, check.
Optional S-Video-out thingamabob, check.
Full SoundStorm APU, check.
Low price, long life, stability, passive cooling, check.
Fits microATX cases, check.
Supports 200MHz-based and 266MHz-based SocketA processors including Applebred and Thorton, check (BIOS 1007 gives them formal recognition).
You want a SoundStorm-equipped HTPC, then track down an A7N266-VM and buy it while you still can.Even with the CPU, heatsink and memory, I bet it would be cost-competitive with the PCI-SoundStorm card you're envisioning.
GB LAN is becoming a necessary for a heavily-used "media center", from which the rest of the devices in your household will store/access media.
If GB Lan is not on-chip, you'll need PCI-X to have enough bandwidth. If you have PCI-X, then you really need a 64-bit processor to use that bandwidth. If you have a 64-bit processor, then you have no SoundStorm
Also, how is it easy to passively-cool an AMD NF2 system?!
Oops, heh.Originally posted by: mechBgon
In the first place, it's not an nF2 systemA7N266-VM/AA is nForce "classic".
Originally posted by: mechBgon
Secondly, where did we start talking about passively cooling the CPU, because I must've missed that.
I guess I thought you were talking about building an entire system...Originally posted by: mechBgon
Low price, long life, stability, passive cooling, check.
I think that my 6x120GB RAID-5 file server would easily absorb multiple gigabit clients if it had the right mobo/NIC+interface combination.Originally posted by: mechBgon
In my testing with a couple of systems equipped with 32-bit PCI-based gigabit NICs and 150000rpm SCSI drives (an nForce2 rig and an A64 rig), I found that 32-bit gigabit is good for about 25 megabytes per second, and your typical ATA bulk-storage drives aren't going to be able to sustain a great deal higher than that when doing data retrieval, even for just one single client. Add two or three clients pulling data simultaneously, and now your ATA drive is floundering around trying to pretend it's SCSI. You could have a 10Gbit connection but that won't make your ATA drives seek any better or sprint any faster.
Just my 2¢ worth![]()
Read XBit Labs' review of the Promise SX6000 here and see what you think.Originally posted by: Ichinisan
Oops, heh.Originally posted by: mechBgon
In the first place, it's not an nF2 systemA7N266-VM/AA is nForce "classic".
Originally posted by: mechBgon
Secondly, where did we start talking about passively cooling the CPU, because I must've missed that.I guess I thought you were talking about building an entire system...Originally posted by: mechBgon
Low price, long life, stability, passive cooling, check.
I think that my 6x120GB RAID-5 file server would easily absorb multiple gigabit clients if it had the right mobo/NIC+interface combination.Originally posted by: mechBgon
In my testing with a couple of systems equipped with 32-bit PCI-based gigabit NICs and 150000rpm SCSI drives (an nForce2 rig and an A64 rig), I found that 32-bit gigabit is good for about 25 megabytes per second, and your typical ATA bulk-storage drives aren't going to be able to sustain a great deal higher than that when doing data retrieval, even for just one single client. Add two or three clients pulling data simultaneously, and now your ATA drive is floundering around trying to pretend it's SCSI. You could have a 10Gbit connection but that won't make your ATA drives seek any better or sprint any faster.
Just my 2¢ worth![]()
Ouch!As we see, the highest speed was obtained in RAID 0 of 4 HDDs, and with the increase in the number of HDDs in the array, this value got a little bit lower. At the same time, we can't help pointing out that the array speed is lower than that of a single IBM DTLA 307015 with a regular UDMA100 controller.
Originally posted by: BFG10K
AFAIK nVidia's sound drivers still have a lot of issues in games so any petition should be aimed at fixing those first.
Think about it. Nearly all of the bandwidth is used by GB LAN. That leaves nothing for your other devices...not to mention DV CAPTURE HARDWARE!
Yeah, that's exactly the controller that I got...Originally posted by: mechBgon
Read XBit Labs' review of the Promise SX6000 here and see what you think.
Anyway, I didn't quite realize that you wanted your HTPC to also be the fileserver for the whole household. What about building your killer fileserver on your dream platform (nF3 250Gb, i875P-CSA, etc) and make your HTPC a separate computer that accesses the fileserver like the others do?
Originally posted by: Pariah
Think about it. Nearly all of the bandwidth is used by GB LAN. That leaves nothing for your other devices...not to mention DV CAPTURE HARDWARE!
How is 50-60MB/s nearly all the bandwidth of PCI? And that's under ideal conditions with high quality Gbe controllers, cabling and jumbo frames, which is an extremely rare setup for home users. I've used Gbe at home for years, so my hardware is a bit older and doesn't use jumbo frames, and I get performance around where mechBgon does, about 25-30MB/s. No where remotely close to saturating PCI, and well below the capabilities of one of my hard drives. DV capture hardware requires 3.6MB/s. A 24x CLV CDROM drive can handle that speed. I used to transfer DV movies from my camera to my PII 266 laptop with some dirt slow 4200RPM HD years ago. Big deal.
You've never actually used Gbe have you? And I can almost guarantee you that your RAID 5 array doesn't come anywhere close to saturating the PCI bus. ATA RAID 5 is garbage unless you have a recent 3Ware card.
