193,985 users have contributed to 42,905 threads and 257,892 posts.
In the past 24 hours, we have 4 new thread(s), 17 new post(s) and 88 new user(s).
Maya:
Latency in Cubase is 128 samples and Vepro is 2 buffers. The hard drive that the samples are on is a Seagate 720o rpm sata Terrabyte drive. As I said, VEpro uses it's own gigabit lan connection (via a direct crossover cable) and I've even triied disabling the onboard lan (on the server, used for internet) and it makes no difference.
Would there be any advantage to using a gigabit router rather than a crossover cable even though there's only one server?
Hello BeatPete
A Latency of 128 Samples is very low for all the jobs you would do with your Computers
... Even if you have two fast systems a latency of 128 samples seems to be a very low value.
You can go up to 512 samples and you have still an acceptable delay (~12ms) for the "live play-in-situation".
So I recommend to increase the latency on 512. Check whether you still get the crackles or not.
If you still have them increase the value on 1024.
Your system should play the music with all the effects etc. with a latency of 1024 in any case!
If you still have crackles with this value we have to look for the matter which "put on the brakes".
Good luck
@beatpete said:
Thanks, but one of the reasons i got a new i7 is to use lower latencies. I can really feel a difference (especially with pianos) fro 128 to 256 let alone 512 samples...
Hello again
Please don't take me wrong. But I don't know whether you are specialist or not.
Nevertheless I try to explain some matters which are connected to Latencies - CPU - Delays and so on. Maybe there are other interested readers...
First of all: The sound quality is the same with a latency of 128 or with one of 2048.
But there is a difference: The higher the value of the latency the higher the delay.
You can recognize this Delay while you are playing a piano for example.
The time between "Pressing the Key" and the later starting sound in the monitor is the mentioned delay.
So let us take a low latency. Yes - but...
The Computer has a lot more time to handle its things when it gets a data package of 2048 instead of 16 x 128 (=2048 as well).
Let me explain the situation with the situation in a kitchen of MacDonald.
Example One: You order 2048 Burgers at 8:00 and you say: "I will be back at 8:46".
1 Order, 1 Output, 1 Start of the deep fryer, the chef could manage the order between other tasks. At 8:46 you will get your 2048 Burgers.
Example Two: You Order 128 Burgers at 8:00 as well > "I will be back in 2,8 Minutes". 2,8 Minutes later you take the 128 Burgers and in the mean time you order another 128 Burgers...
After 46 Minutes you also will have 2048 Burgers in total. But the stress of the kitchen crew was enormeous.
They had to treat 16 Orders , 16 Outputs 16 Starts of the deep fryer and the chef hadn't any time to do other jobs between the tasks...
And the Burgers in our Computer?
Example One: Latency time 2048 = Delay of 46 ms > less stress for the system
Example Two: Latency time 128 = Delay of ~2,8ms > a lot of stress for the system because the CPU has to do its job 16times in the same time (getting signals, calculating, sending the results to the output...) just as the kitchen crew above.
So it is our task to find the balance between our wish (low latency) and the ability of our system.
_____________________________________________________________________________
Your problem: Crackels!
I still suspect, that a Latency of 128 Samples is too lo.
Once more: Even if you have a fast system: Playing a project with 36 Tracks, each with effects, 2 - 3 Convoluten Reverbs for depths, some Master-effects...
I guess that you should forget your dream of a 128-Samples-Latency.
You could be happy when the system works with 256 or also with 512 - which makes a delay of ~11 ms.
If you are playing on a stage: 11 ms means 3,7m away from the next musician. Still not really a problem...
Another solution:
If you want to play with low delays: Take a very low latency for the situation of playing and adding melodies with your keyboard.
Mute most of the tracks, don't use effects while this part of work.
When it comes to mixing down the project - with all the effects and tracks - then you can choose a higher latency.
Best
Beat Kaufmann
Cheers BK for that very informative post!
Just a quick question...
Would RAM speed come into the equation at all?? Or is it mainly to do with the CPU calculations?
For instance:
I have 12GB of DDR3 RAM running at 1066Mhz CL7 latency, but still get pops n' crackles with MIR on my i7 920 at 512 samples latency.
If I used the same set up, but instead used faster or overclocked RAM, would I still get the same crackerly results, do you think? As the CPU still has to make exactly the same calculations anyway??
I notice the new 6 core Intel i7 980X only 'officially' supports 1066Mhz RAM anyways... so I'm thinking RAM speed has little bearing on this issue?
knievel
Yes, I do consider myself a bit of a PC "expert" as I've been using Cubase since it's inception and I used FXteleport since it's inception. Until I moved to W7, I could get 128/256 samples with FXT no problem, even on older computers. I tried 256 and the pops are still there. I can use Bidule (via rewire) with the same samples loaded on my main daw computer and get less pops and clicks at 128 than VEPro running on a separate computer.
Is there any advantage over using a gigabit switch instead of a crossover cable?
knievel:
I just saw your post after I posted. My I7 920 is overclocked to 3.2ghz and my ram is running at 1600mhz, so I don't think ram has much to do with it.
BTW Even though this project pops and clicks in realtime, if I export the mix (faster than realtime), the pops dissapear, weird huh?
@knievel said:
Cheers BK for that very informative post!
Just a quick question...
Would RAM speed come into the equation at all?? Or is it mainly to do with the CPU calculations?
For instance:
I have 12GB of DDR3 RAM running at 1066Mhz CL7 latency, but still get pops n' crackles with MIR on my i7 920 at 512 samples latency.
If I used the same set up, but instead used faster or overclocked RAM, would I still get the same crackerly results, do you think? As the CPU still has to make exactly the same calculations anyway??
I notice the new 6 core Intel i7 980X only 'officially' supports 1066Mhz RAM anyways... so I'm thinking RAM speed has little bearing on this issue?
knievel
Hi Knievel
For making an "assessment" a of a "music station" you should take each component into account.
12GB DDR3 1066MHz could be a good RAM compared with the rest of the system (GHz, amount of cores etc.) but in another system it would be a weak part.
For us - the average consumer - 12 GB, DDR3, 1066Mhz a good average... even if 1333Mhz would be better...[;)]
But as I said: We have to observe the whole system. Sometimes we have components, programs or settings outside the "Audio-Line" which can disturb a proper handling with the audio datas.
1. Good to know, which part of your PC or MAC is the weakest one. If you know that you can try to fit the part in an adequate way to the rest of the system.
One possibility to get first hints: Use the WN7/Vista "Leistungsbewertung" (Power Assessment?)
Example: I know that my grafic card is too weak compared with the rest. Its job: Showing datas on two screens of 2 x 1650 x 1050 pixels.
A first step could be: Switching off all the nice effects like shadows etc. Nevertheless, I should exchange the component for better results.
There are other programs for getting information and for finding the weakest part of the system: A freeware example is "Sandra Light"
2. But there are other "powerkillers". Try to keep your Computer free from any program which isn't necessary for producing music.
So switch off - if possible - all security programs, sidebare wheather forcasts, connections with ... etc.
A really bad situation could be:
You are working with MIR - and at the same time: The security program downloads and installs the newest antivirus files - in the background.
Such processes can lead to a crash.
Further, try to stop all autostart programs, auto tasks etc. as much as possible.
A lot of words but a short summary: It is not one single part which makes a fast system...[;)]
Best
Beat Kaufmann
@beatpete said:
BTW Even though this project pops and clicks in realtime, if I export the mix (faster than realtime), the pops dissapear, weird huh?
Hello Beatpete
As you mentioned your system seems to be a fast one. The hint, that the crackles desappear with exporting files shows that it isn't (only) a question of the latency.
If you have crackles (not pops, "gaps of sound") it could be a conflict with the graphic card. I estimate 75%.
As a first step you can observe whether the crackles have someting to do with movements in the display or not. More movements = other crackles?
In any case I recommend to check whether the DIGIFACE and your graphic-card share an IRQ or not (Hardware-Manager/im Geräte Manager).
If yes: try to arrange a separate IRQ for each component. In any case you should arrange it, that the graphic card gets an own IRQ.
If not: Try to change the PCIe slots for one of the two components. Maybe it make sense to solve this problem together with a computer specialist.
All the best
Beat Kaufmann
@beatpete said:
BTW Even though this project pops and clicks in realtime, if I export the mix (faster than realtime), the pops dissapear, weird huh?
Hello Beatpete
As you mentioned your system seems to be a fast one. The hint, that the crackles desappear with exporting files shows that it isn't (only) a question of the latency.
If you have crackles (not pops, "gaps of sound") it could be a conflict with the graphic card. I estimate 75%.
As a first step you can observe whether the crackles have someting to do with movements in the display or not. More movements = other crackles?
In any case I recommend to check whether the DIGIFACE and your graphic-card share an IRQ or not (Hardware-Manager/im Geräte Manager).
If yes: try to arrange a separate IRQ for each component. In any case you should arrange it, that the graphic card gets an own IRQ.
If not: Try to change the PCIe slots for one of the two components. Maybe it make sense to solve this problem together with a computer specialist.
All the best
Beat Kaufmann
Beat:
I think you're on the right track, moving the Cubase arrange gui around increases the pops. The graphics card is sharing irqs with a lot of stuff (including the 2 UAD-1 PcIe cards) but Not with the Digiface.
DG:
The VEPro gui is not open on the slave.
Hey DG - is the performance better with the VEP GUI minimized? This caught my eye.
@Mahlon said:
beatpete,
A suggestion which others made that helped me was to replace my network card with an intel card. That 90% of clicks and pops go away on my system (2 slaves and a master). I replaced all cards with an intel pro card and it did make the difference. Just a thought.
M.
Mahlon;
Can you explain more about how the new lan cards helped? i read your other post but you never replied with a "solved" answer. Did installing the cards completely get rid of your clicks and pops? What was the model # of the Intel cards?
OK - I have three slave units - just minimize all instances of VEP. I'll give it a shot.
