CPU recording: Processor or RAM more important?

  • Thread starter Thread starter electronicdreaming
  • Start date Start date
E

electronicdreaming

New member
hi everyone.
I was just wondering, when it comes to computer recording, is the Processor speed more important or the amount of RAM?
I'm customizing a PC right now, and have it configured at 3.0 processer (p4) and 512 RAM. I just record from a synthesizer and do vocals, so do you think this is sufficient? thank you.
 
More than sufficient.

I think that processor is of first importance, and RAM comes in second.
 
the ram comes more into effect when you start using software effects plugins

The P4 3.0 is best utilised with two sticks of matching DDR ram what MOBO are you using?
 
CPU is for real time sound effects. RAM is for sampler banks (VSTi, DXi). If you use neither of those, even very weak system will be suitable. If you use samples extensively, you need enormous amounts of RAM - starting with 1GB. For effects, you need as fast a CPU as there is on the market, in order for everything to work smooth.
 
neon glow said:
For effects, you need as fast a CPU as there is on the market, in order for everything to work smooth.
Click to expand...
That's rediculous.
People have been able to use lots of realtime effects smoothly for quite some time now. Ergo: You don't need the fastest cpu on the market. Anything that's 2GHz or higher is fast enough for most people. Buying the fastest cpu is a waste of money.

I've said it before and I'll say it again: For recording and mixing you don't need the fastest cpu. You need one that's fast enough.


Electronicdreaming: I'm pretty sure that you'll be fine for now with what you got.
 
You need one that's fast enough.
Click to expand...
How can a CPU be fast enough, if you are not limited by the amount of effects you're applying, and each one (for example Ozone - 64bit processing) takes up average of 10% from the CPU (that's how it is with my 2Ghz CPU). I have 2Ghz and still that's the problem I'm facing.
 
How can a CPU be fast enough, if you are not limited by the amount of effects you're applying, and each one (for example Ozone - 64bit processing) takes up average of 10% from the CPU (that's how it is with my 2Ghz CPU). I have 2Ghz and still that's the problem I'm facing.
Click to expand...

First off, that CPU meter thing really means processing resources -- the CPU and the RAM available (cache on the CPU plus the RAM sticks on the board).

The amount of CPU resources required by a plug-in is not affected by the speed of the CPU. The speed only means that instuctions are executed faster and so it can do more operations in a given amount of time -- so you can give it more to do before it starts to affect the audio performance.

The amount of RAM sets the amount of audio data that the system can hold onto and move around quickly. The more plug-ins, the more data is being generated and stashed in memory and the more RAM you need.
 
Maybe we're working with different tools but that's how it is with me. I use Sonar. The sound effects (reverb, distortion, delay, eq, etc) that i apply are REAL TIME, meaning that when I apply them, they don't change the original track. They get applied only when I press "play", right in the real time. ( I like it, because it makes me able to play guitar and hear all the applied effects from the speakers right away. ) It means that, once the sound is produced, it goes to CPU, and the CPU applies ALL THE EFFECTS that I selected ONE BY ONE. And there are MANY tracks playing at the same time. After that the sound goes to speakers. So when the sound is played, all the effects for every track have to be applied by the CPU in real time, and it shouldn't create any delays. This requires a hell of a CPU, especially when you use really complicated HQ effects, like iZotope Ozone. The effects you're talking about are probably the ones that you apply before the music's played, and they change the original sound, into new one, with effect. That's different. You don't need any great CPU speed for that. I was talking about the real time effects in my first suggestion here. This is what happens when I press play (just in case):

_______|_______This is where CPU works on sound________________|
Track1 -> effect1, effect2, effect3 -> m
Track2 -> effect1, effect2, effect3 -> i -> effect1, effect2, effect3 -> sound card
Track3 -> effect1, effect2, effect3 -> x

In case of RAM, it's not only audio data. How about Steinberg The Grand for example? 512 mb bank. Or Edirol HQ Orchestral? - 120 Mb bank. Or Hypersonic? - 100mb or so. Or battery drum samples? 20-50mb. Where are you gonna store all this? It can play from HDD (partially), but not when you try to achieve some decent sound with all of them playing at once. And because of the stupid way software is made, they remain in RAM even when not used. And it's not blah blah. I'm experiencing much troubles with all this currently. Waiting for the new laptop to arrive.
 
Okay...this is the way I always see it. Your multitrack program uses RAM to run while using the CPU to do the actual processing. Then when you add a plugin...well, plugins are just another separate program. It uses RAM to keep running while using CPU to process the audio using the effect you choose. The more plugins you add the more CPU and RAM is used. It'd be the same way if you opened up Internet Explorer, and then opened up another internet explorer, and then another and another until you had like 20 windows. All of them require RAM to stay open and if you run out of RAM you get the warning that you're out of ram and chances are IE will crash. BUT if you had the 20 windows open and you had each page trying to load up a website at the same time, you'd also be adding CPU power on top of the RAM usage. This is exactly what plugins are doing. Using both.
So therefore, I see CPU as the most important thing to buy because if you want flawless fast processing for audio, a lot of RAM and a slow CPU isn't going to do much for you. However, I wouldn't discount getting lots of RAM too, but 512 should be enough for you. I guess in the computer world...the more the better. Again, hope this helps. :cool:
 
The CPU speed is speed; there's no "size" associated with it, unless you consider its cache. A 2 GHz CPU with a 512kB cache runs at a slower clock speed than a 2.4GHz CPU with a 512kB cache, but this only means that in one second of time the 2.4 GHz CPU can perform about 20% more operations. But that doesn't affect how many applications can be running -- it's RAM that affects that. If all the code that they need to run properly can't fit into RAM without having to swap too much stuff out to disk, then it really doesn't matter how fast of a CPU you have.

Therefore, if the CPU is fast enough to execute instructions fast enough to keep up with the audio stream and not cause problems, it's fast enough. Any faster won't effect the performance ceiling very much at all. And you could have a 10 GHz CPU and it wouldn't let you run four times more plug-ins successfully compared to a 2.4GHz CPU if they both only have 256MB of RAM.
 
Yeah, you are definitely talking about a different thing. Ok, do you know that CPU is the thing that calculates all the sound effects? CPU takes sound, calculates the sound effect, and gives it out changed to sound card and speakers. Now imagine it happening in realtime. I assume you also know that CPU can not calculate many sound effects at the same time. When there are many of them, CPU calculates them in the order, but does it so fast, that we hear no delays. Now, the FASTER the CPU, the FASTER it can go through MORE sound effects that are applied. Usually it has to deal with calculating more than 10 sound effects AT ONCE. Of course it doesn't do it AT ONCE. It does it in the order the effects were applied. In other words, before sound goes to speakers, it passes through CPU, and CPU calculates all the effects on the sound WHILE the sound is playing. In REAL TIME. IF the CPU is too slow, it can not smoothly apply all the effects to all the tracks so fast, without any delays or drop-outs. The plug-ins do surely take up some RAM, when they are loaded, but that's almost nothing. But do you know that any sound changing plugins use the CPU to change the sound? The percentage of CPU power shows how many more calculations CPU can handle at once (it is an unused speed, which can be utilized for more sound processing plugins). If you've never used Sonar with real-time effects (you can stack them up in a little box, that's on the left of every track, like a hardware effects stands, that are connected one to another), it's probably harder to understand what I'm talking about when saying "real-time effects". Real-time effects need FAST CPU to operate. One CPU does what the whole line of hardware sound effects usually does. but if you're usually limited with hardware sound effect boxes, you are not limited with software ones. That's why you need faster CPU to do them all at once in real time.

Yes, that's right. Plugin code gets loaded into RAM. But that's not even important, because you don't need much ram to load many applications. It doesn't take up much. What you mostly need much ram for is BANKS of sounds. The sound effects on the other hand, need the CPU speed, and even little RAM is usually enough for them. Their SIZE is not big. The CALCULATIONS that they perform are complicated...

I basically repeated what I said in the previous post. I don't know how else to explain.... If it's still not clear, I'll try to think of some analogies...
 
Once you have sufficient ram, 1GB is generally regarded as being sufficient, the cpu becomes the bottleneck.

Having enough ram means your whole project can be resident in memory, without needing disk access. A dual channel P4 mobo with DDR400 memory can deliver that data from ram at upto 6.4 GB/sec which is more than fast enough to stream audio.

It now depends on how many real time plugins you want to run before the cpu starts to choke.

In my case I may run several instances of Drumagog and Amplitube plugins. These gobble cpu. When I overclocked my machine from 2.8Ghz to 3.5Ghz, I was able to run Drumagog on 2 more tracks in a big project than I could previously.

The faster the chip, the more you can do
 
My XP2600+ runs just fine. My current project has about 12 tracks, each with at least one effect (vocals have more, some are pitch-shifted), most have more effects, and my CPU usage (from the task manager) is never above 15% when playing back. Much higher during mixdown (because mixdown is at about 4x real-time) but no stuttering due to effects.

Oh yeah, I have 1 GB RAM and I only use IDE drives for my recording setup, no SATA. Plugins require memory to be resident and run. They need the CPU to render their effects. Both are important. However, a 2.4 to 2.8 PR rating CPU (AMD/INTEL) should be able to handle most projects easily in the home envronment, coupled with 1 GB of memory. 512 is good too, anything below that and you have to think of other ways to complete projects.

Oh yeah, a 550 MHz P3 can still be used to complete projects. If you use offline effects, submix a few tracks, and so on. The process will just take longer. But the listener will never know!!
 
neon, I understand what you are trying to say, in fact I am saying virtually the same thing. We are disagreeing on the relative effect of RAM and CPU speed.

For example, you say "The plug-ins do surely take up some RAM, when they are loaded, but that's almost nothing," I disagree. At the very least they need enough RAM to hold the altered copy of the audio data they have crunched down from the original audio data.

And you also say, "Plugin code gets loaded into RAM. But that's not even important, because you don't need much ram to load many applications. It doesn't take up much... The sound effects on the other hand, need the CPU speed, and even little RAM is usually enough for them. Their SIZE is not big." I think you must be thinking that the amount of RAM a plug-in requires is equal to the amount of disk space the plug-in code occupies. That is not at all the case. As I mentioned above, plug-ins in particular need to buffer the audio they are effecting through RAM.

But 'nuff said. I think neither of us are really qualified to answer this question here (I know I sure ain't). Maybe someone with a deeper knowledge of how these apps work at a low level can chime in here...
 
So you're saying real time effects need to buffer audio in RAM? That's what I'm not sure about either. I think that audio is read from hard drive, can go right to CPU. And there's no need to buffer it. But I don't know about that.

But I'm much more sure that audio data is not stored anywhere after it was affected by plugin. It just plays out of speakers. Only the originally recorded track is stored on HDD. Plugins have to affect it every time when you press play. Before sound gets to speakers, it goes through plugins.
 
neon glow said:
So you're saying real time effects need to buffer audio in RAM? That's what I'm not sure about either. I think that audio is read from hard drive, can go right to CPU. And there's no need to buffer it. But I don't know about that.

But I'm much more sure that audio data is not stored anywhere after it was affected by plugin. It just plays out of speakers. Only the originally recorded track is stored on HDD. Plugins have to affect it every time when you press play. Before sound gets to speakers, it goes through plugins.
Click to expand...

Data Processing:

CPU gets instructions from Software program
CPU requests Data from the system controller (northbridge)
If the memory is on memory, controller passes on the data.
If it is in disk, the controller requests the southbridge for the data
Southbridge passes on the request to the disk
Disk sends the Data to the memory via the SB, then the NB.
Then the controller passes the data to the CPU from memory.
HDD-Memory
Memory-CPU cache-CPU core-Memory, etc., for as long as calculations are required
Output device driver (resident in memory, having been loaded from disk)
PCI bus
Output device hardware

All processing happens only between memory and CPU. DMA and UDMA systems allow disk to read directly into memory, instead of having to read first to CPU.

Even when the disk does get involved in this process, it is when Windows starts hitting the disk for virtual memory. That is when slowdowns start to happen. Which is why a large memory size is essential for smooth functioning of the system.

But whether 'storage' happens depends entirely on your system and buffer settings. I have a nice fat 2 seconds playback buffer, so real time effects aren't really 'real time' but slightly pre-calculated. As of now there is no real lag when recording, but anyway I submix before additional tracks need to be recorded. Better safe than sorry.

No program runs from disk. The disk is only secondary to the entire computer recording process.
 
You must log in or register to reply here.
Back
Top