Topic: USB2 vs PCIe Which will allow me the most tracks/vst effects/vst instruments?

T

thefyn

Member
Jun 3, 2006
823
0
16
My M-Audio fast track is starting to chug, I think USB 1.1 is catching up with it as CPU and disk % is in single figures even with 20 tracks and effects applied.

So I think the USB 1.1 is the bottleneck or it just does not like my AMD setup.

I asked around and not many people can give me a straight answer of USB 2 vs PCIe. They say on paper PCIe should be faster, but it relies on drivers etc etc.

I am thinking of getting this:

http://www.emu.com/products/product.asp?category=5 05&subcategory=491&product=20022

I already have a half decent mic pre.

I have tried adjusting buffers, drivers, VST troubleshoorting etc but I am still getting dropouts.

I have about 20 audio tracks, v VST synths and 10 of the tracks have VST effects applied.

My system should be fine. It is an AMD Phenom II X6 1090T with 8 gig and SS disk. But I have heard a few grumbles about north bridge and USB 2 shenanigans not playign nice.

http://forum.cakewalk.com/tm.aspx?m=2097264
 
USB/FW/PCIe has nothing to do with how many VSTs or VST-Instruments you can run. Well, the buffer size has.. but that's the same for every kind of interface.
for Mixing/Mastering etc you should increase the buffersize to 1024 then there will be a lot less dropouts and the system is more stable overall.

When it comes to recording a lot of tracks at the same time.. PCI and PCIe *can* sometimes offer more stability and lower buffer sizes. It all depends on the interface, the driver and your computer, though. For example I can easily record 18 tracks at the same time over USB with my Fireface UFX. Even at 32 samples if I want :D FW also works.
I don't know whether the EMU Cards are any good.
 
jipchen said:
USB/FW/PCIe has nothing to do with how many VSTs or VST-Instruments you can run. Well, the buffer size has.. but that's the same for every kind of interface.
for Mixing/Mastering etc you should increase the buffersize to 1024 then there will be a lot less dropouts and the system is more stable overall.

When it comes to recording a lot of tracks at the same time.. PCI and PCIe *can* sometimes offer more stability and lower buffer sizes. It all depends on the interface, the driver and your computer, though. For example I can easily record 18 tracks at the same time over USB with my Fireface UFX. Even at 32 samples if I want :D FW also works.
I don't know whether the EMU Cards are any good.
Click to expand...


Great info, thanks.

From an aspect of using a midi guitar to record a VST synth, I will need a low latency buffer for that.

When it comes to that, or VST amp sims via direct guitar, will the PCIe be better than USB 2 for less of a delay?

Also, I have 8 gig of ram and Win 7 64. Do 64 bit daws (using that extra ram) help in the real world compared to 32 bit Sonar? I know on paper it should, but has everything being 64 bit translated into REAL results?
 
jipchen said:
USB/FW/PCIe has nothing to do with how many VSTs or VST-Instruments you can run. Well, the buffer size has.. but that's the same for every kind of interface.
for Mixing/Mastering etc you should increase the buffersize to 1024 then there will be a lot less dropouts and the system is more stable overall.

When it comes to recording a lot of tracks at the same time.. PCI and PCIe *can* sometimes offer more stability and lower buffer sizes. It all depends on the interface, the driver and your computer, though. For example I can easily record 18 tracks at the same time over USB with my Fireface UFX. Even at 32 samples if I want :D FW also works.
I don't know whether the EMU Cards are any good.
Click to expand...

Not quite. The information (audio) has to go from the HDD to the memory/CPU to be directed to the interface (whether it be an expansion card or external interface). The bandwidth determines how much information can be delivered to the interface, and if the method of data transfer can't keep up (even thought the HDD, Memory, CPU and DSP can) you will have a dropout.

PCIe is faster than USB 1.1 hell its still quite a bit faster than USB 2.0.

Wouldn't go for an EMU card though, have one myself, drivers are a nightmare and they still do not have Vista or Win7 support, their driver suck so hard that ASIO4ALL works one thousand times better.


thefyn said:
Great info, thanks.

From an aspect of using a midi guitar to record a VST synth, I will need a low latency buffer for that.

When it comes to that, or VST amp sims via direct guitar, will the PCIe be better than USB 2 for less of a delay?

Also, I have 8 gig of ram and Win 7 64. Do 64 bit daws (using that extra ram) help in the real world compared to 32 bit Sonar? I know on paper it should, but has everything being 64 bit translated into REAL results?
Click to expand...

DSP power is the controlling issue with monitoring latency for the most part, it also comes down to the CPU as well because it has to send the streaming audio in a timely fashion so that the DSP has time to render the data, but the DSP power is really what you need to keep latency down.

Memory doe not help with latency and stability, it helps for having large VSTi libraries and having multiple VSTs up, which tend to take up memory, particular sample libraries.
 
TheWinterSnow said:
Not quite. The information (audio) has to go from the HDD to the memory/CPU to be directed to the interface (whether it be an expansion card or external interface). The bandwidth determines how much information can be delivered to the interface, and if the method of data transfer can't keep up (even thought the HDD, Memory, CPU and DSP can) you will have a dropout.
Click to expand...
True, what I said was pretty imprecise.. I referred to the thread title "which will allow me the most (...) vst instruments" and I can not imagine a situation where the interface determines the numer of VST-I instances? I've seen the CPU / RAM crapping out, but VSTi dropouts because of the interface are new to me. Even USB 1 can handle that (at least it could for me). Also, that's why I said it kinda depends on the interface's transfer protocoll (USB/FW/PCI/e) because of the buffersize. With a high buffersize (while mixing) I can open a pretty much unlimited number of VSTs and VST-Instruments even with my old USB1 interface.
I can't imagine a situation where you "record" 20 tracks of VST-instruments at the same time :eek:
PCIe is faster than USB 1.1 hell its still quite a bit faster than USB 2.0.
Click to expand...
I remember an article saying audio recording can't even exceed the possible bandwidth of USB, but I honestly don't know if that's true. It said even a high number of tracks is still below the maximum capacity of USB (2.0), and of course FW and PCI / PCI-E. I tried to find it, but I can't :Smug:
Do you perhaps know some details / background info on this?
By the way, I'm not trying to say USB is as good as PCI-e, rather saying "the driver matters, not the connection".

I can somehow understand the article, because there is not a single interface using FW800 for example (the Fireface 800 and the Metric Halo only have the 800 connection for daisy-chaining), and with the Fireface UFX, RME proofed that USB 2.0 can smoke Firewire 400 with both stablity and number of channels (I never would've thought that)
 
TheWinterSnow said:
Wouldn't go for an EMU card though, have one myself, drivers are a nightmare and they still do not have Vista or Win7 support, their driver suck so hard that ASIO4ALL works one thousand times better.
Click to expand...

Thanks. I had hell of a time with Mackie Blackjacks asio drivers and had to send it back.

Has anyones tried J bridge?

I have Sonar 7 hom (32 bit)

I just upgraded to X1 essentials and I hate the interface so I have gone back to Sonar 7.

Will J bridge access the extra ram for my virtual instruments like Native Instruments Kontakt etc?

I have Win 7 64. I did a quick search and saw a few people having trouble with things like Toontrack.
 
Also, I have USB 3 on my mobo and a 1tb usb3 drive for backups. Holy shite that thing is fast compared to USB 1.1 for backing up files!!!
 
jipchen said:
True, what I said was pretty imprecise.. I referred to the thread title "which will allow me the most (...) vst instruments" and I can not imagine a situation where the interface determines the numer of VST-I instances? I've seen the CPU / RAM crapping out, but VSTi dropouts because of the interface are new to me. Even USB 1 can handle that (at least it could for me). Also, that's why I said it kinda depends on the interface's transfer protocoll (USB/FW/PCI/e) because of the buffersize. With a high buffersize (while mixing) I can open a pretty much unlimited number of VSTs and VST-Instruments even with my old USB1 interface.
I can't imagine a situation where you "record" 20 tracks of VST-instruments at the same time :eek:

I remember an article saying audio recording can't even exceed the possible bandwidth of USB, but I honestly don't know if that's true. It said even a high number of tracks is still below the maximum capacity of USB (2.0), and of course FW and PCI / PCI-E. I tried to find it, but I can't :Smug:
Do you perhaps know some details / background info on this?
By the way, I'm not trying to say USB is as good as PCI-e, rather saying "the driver matters, not the connection".

I can somehow understand the article, because there is not a single interface using FW800 for example (the Fireface 800 and the Metric Halo only have the 800 connection for daisy-chaining), and with the Fireface UFX, RME proofed that USB 2.0 can smoke Firewire 400 with both stablity and number of channels (I never would've thought that)
Click to expand...

Highy depends on if the interface of choice has DSP or not, if it does not, then all plugins and VST's will be processed on the CPU and will be limited to how much it can process. This is why having VSTs up will take processing power and will generally take up CPU power even if you have a DSP but it won't be as intense. Buffer sizes increase the amount of information that is streamed in blocks so that it does not have to do processing close to real time, there is a delay in the audio but that delay allows the processor to request information from the HDD at a more natural pace. Since it is working less hard, the available CPU power left available is increased with higher buffer sizes.

So basically, buffer sizes always effect CPU power in terms of the CPU streaming information to the FSB (to USB or PCI), the faster you want that information there, the more overhead, and the CPU processing close to real time places more strain on it. However if you are trying to place more information down the FSB and it doesn't have the bandwidth, it will not be able to send, when the CPU realizes that the processed data is backed up and the time sensitive information is delayed, you have a dropout. This is and issue with the CPU not being able to send out information to its output port as fast as it would like to or needs to and is caused by having a lot of tracks with too small of a buffer. This would be OP's problem as he clearly said that his CPU isn't going up to 100% but is just dropping out. Thats a huge sign that the information is not consistent and since its not from low CPU resources is the result of a bottleneck post CPU (and USB 1.1 is WAY too slow).

Heavy Jerk said:
For what it's worth, I have an EMU 0404 PCI in my old PC, and it's stable as a rock. Win XP though.
Click to expand...

You know for years I have been thinking that I actually got a faulty card, my outputs about a year ago dropped volume by 12db and obtained a massive amount of crossover distortion (like guitar amps). It has never been fully stable and was almost unusable until I tried ASIO4ALL and this was with XP 32-bit.
 
TheWinterSnow said:
Highy depends on if the interface of choice has DSP or not, if it does not, then all plugins and VST's will be processed on the CPU and will be limited to how much it can process. This is why having VSTs up will take processing power and will generally take up CPU power even if you have a DSP but it won't be as intense. Buffer sizes increase the amount of information that is streamed in blocks so that it does not have to do processing close to real time, there is a delay in the audio but that delay allows the processor to request information from the HDD at a more natural pace. Since it is working less hard, the available CPU power left available is increased with higher buffer sizes.

So basically, buffer sizes always effect CPU power in terms of the CPU streaming information to the FSB (to USB or PCI), the faster you want that information there, the more overhead, and the CPU processing close to real time places more strain on it. However if you are trying to place more information down the FSB and it doesn't have the bandwidth, it will not be able to send, when the CPU realizes that the processed data is backed up and the time sensitive information is delayed, you have a dropout. This is and issue with the CPU not being able to send out information to its output port as fast as it would like to or needs to and is caused by having a lot of tracks with too small of a buffer. This would be OP's problem as he clearly said that his CPU isn't going up to 100% but is just dropping out. Thats a huge sign that the information is not consistent and since its not from low CPU resources is the result of a bottleneck post CPU (and USB 1.1 is WAY too slow).
Click to expand...
Aaah, I see. Thanks man! :)
 
Some PCI-E card have latencies comparable with USB2, because PCI to PCI-E bridge (this cards it`s just "upgraded" to PCI-E) introduces additional latency, i.e. PCI-version is faster than PCI-E. All depends on particular implementation.
 
TheWinterSnow said:
Highy depends on if the interface of choice has DSP or not, if it does not, then all plugins and VST's will be processed on the CPU and will be limited to how much it can process. This is why having VSTs up will take processing power and will generally take up CPU power even if you have a DSP but it won't be as intense. Buffer sizes increase the amount of information that is streamed in blocks so that it does not have to do processing close to real time, there is a delay in the audio but that delay allows the processor to request information from the HDD at a more natural pace. Since it is working less hard, the available CPU power left available is increased with higher buffer sizes.

So basically, buffer sizes always effect CPU power in terms of the CPU streaming information to the FSB (to USB or PCI), the faster you want that information there, the more overhead, and the CPU processing close to real time places more strain on it. However if you are trying to place more information down the FSB and it doesn't have the bandwidth, it will not be able to send, when the CPU realizes that the processed data is backed up and the time sensitive information is delayed, you have a dropout. This is and issue with the CPU not being able to send out information to its output port as fast as it would like to or needs to and is caused by having a lot of tracks with too small of a buffer. This would be OP's problem as he clearly said that his CPU isn't going up to 100% but is just dropping out. Thats a huge sign that the information is not consistent and since its not from low CPU resources is the result of a bottleneck post CPU (and USB 1.1 is WAY too slow).



You know for years I have been thinking that I actually got a faulty card, my outputs about a year ago dropped volume by 12db and obtained a massive amount of crossover distortion (like guitar amps). It has never been fully stable and was almost unusable until I tried ASIO4ALL and this was with XP 32-bit.
Click to expand...

Thanks.

Awesome. I was wondering why I was getting dropouts since my CPU was in single digits.

Also, it looks like the EMU has some win 7 drivers updated a month ago.

http://www.emu.com/support/files/download2.asp?Centric=1029&Legacy=0&Platform=1

The only thing that makes me writhe on the floor is those analog cables attached to the device:

0404PCIe_zoom.jpg


I just spent all this cash on half decent cables, mics and a mic pre, and it all goes into some Hosa looking wank:
 
BTW all of this research made me discover RME.

I am def going that route if my first album gets good response.
 
Mutant said:
I am quite happy with zero problems here with my 1212M.

But i am on XP.
Click to expand...

1212m here too, it when I said it was almost unusable it was on XP 32-bit, once I got ASIO4ALL it actually started to work correctly. I have no problems with it on XP now with ASIO4ALL, in windows 7, both the native drivers will not play work at all and ASIO4ALL works half the time. Will have to try the Windows 7 Drivers, last time I was looking around EMu had not made drivers, said they had no intention for it and where ignoring all emails regarding vista/7 support.

Like I said in a previous post, I think I just got stuck with a defective card, I have a tendency of getting defective computer stuff, the motherboard of my first home build had micro stress cracks all across the PCB that would cause traces to open after the system got hot, took forever to find out what was wrong, just like it took my to find out what was going on with my audio interface issues.
 
You must log in or register to reply here.

Similar threads

AeternusEternus
What hardware upgrade will do the most for me?
Replies
14
Views
751
Backline
iHate
I
Top Bottom
Back