thinking to upgrade my pc but have some doubt

Yes, a quad-core CPU will allow you to achieve lower latency. Your current latency won't be cut in half, but it will be reduced. Or at least allow you to reduce it by reducing buffer size. It will also allow you to have more tracks and more plugins in a project.
 
Don't forget to ask yourself whether reduced latency is worth the cost of the upgrade. If you don't play virtual instruments in real time, your 9ms RTL might be just fine. It's the equivalent of playing through an amplifier that's 9 feet away from you.
 
And yes, recording at higher sample rates does lower latency. But again, how important is lower latency to you? For myself, I do everything at high latency and it's never a problem. If I were to upgrade my CPU it would be to support more tracks, instruments and effects without having to freeze them. But not for reducing latency.

Don't get your latency mixed up. There is system latency and Round trip latency ( RTL).
My old Tascam system had 40 -60 ms of reported ( system) latency, but there is actually 0 seconds ( .05ms?)  of latency as far as tracking goes using direct monitoring. You should hear no latency if monitoring at your interface. Any interface and computer for that matter unless it's a POS.  
 
Sonar adjusts for the system latency and what your hearing at your interface via monitors or headphones. Unless the interface drivers are crap it will be bang on. And Tascams drivers are so so.. My Scarlett 6i6 has way better latency performance than the Tascam, but there is no difference in the 2 devices as far as monitoring goes.
 
Round trip latency is another matter and you'll only notice this if you try and use real time plug ins and monitor the back end of your system Input echo on etc.
It will add up no matter what computer or interface you are using, there's no way out of it.
Latency is introduced at your A/D converters.
USB/ firewire chips can introduce latency too, this is why PCIe cards are better at RTL performance.  
Then your software will introduce latency and this goes way up when you run the signal though a bunch of processors. Number crunching takes CPU cycles, only here can you cut down on some latency with more horsepower.
Insert the output through USB again--more latency here
Then you interface adds more latency as the signal goes back through the D/A converters.
 
So theres more to it than just a few pieces of your system.
But like I said, the worst round trip latency performance in the world will not be noticed by those who are not using real time processing and use direct monitoring of the input.
Only time I hear latency is if I record midi and forget to turn off certain efxs like the LP64 multi band.
 
Most important to your system I think is good fast hard drives.
Then comes RAM for all those plug ins.
Then comes CPU horsepower for running more stuff than would be possible 10 years ago.
Then comes a good power supply for all those hard drives!
 
You can certainly do a lot with 2 cores @ 3.5 and 4 Gigs of RAM,,, that's all I have but I'm 80% Audio.
 
PS I use 44.1 because I'm not a world class studio. Makes life easier all round if your burning CD's.
Higher sample rates do not make music sound better, performance and recording technique does.

It's important to have currant drivers that match your operating system, Is there ASIO drivers for the Edrol for W 7 or 8 64 bit? Or are you planing on staying with XP?
Nothing wrong with WDM drivers, ASIO is usually better latency performance, that's all. 
 
As I said , don't worry about latency unless your actually hearing it. If you are using the direct monitoring yes, you will hear none. Turn on input echo and you should hear a 9ms delay.
 
You can run the Centrance ASIO test for RTL, there's another utility as well but I'm not at home Google - "Round Trip Latency test"
It involves patching a cable from your output to your input. It sends a short burst of sound and analyzes the latency.
This test will say for my Tascam = 32ms RTL, at the lowest buffer setting which is unstable so it really  ends up around 40+ms RTL before it is stable (no dropouts).
My Scarlett does better @ 32ms R T latency at a stable buffer setting. ( 9 ) I can lower to (2) it and it stays stable which is why the Scarlett is a much better interface, But I leave it in it's default safe zone because I don't ever use input echo on audio.
So this test gives you  a better view, I'm not exactly sure what Sonars results really mean,
Sonar reports the Tascam @ 9ms and the Scarlett @ 5 ms. So those figures seem to be what you 'hear' if using INPUT ECHO and absolutely no efxs in the bins. But I'm pretty sure those are not the actual RTL figures.
 
While you have the cable patched, play a drum track, like a snare or kick, and re- record it to a new audio track. It can be the output of a soft synth or an audio track. Now put the 2 tracks together and zoom way in and observe the transient spikes. They should line up perfectly if Sonar and your audio drivers are working correctly.

ASIO versus WDM usually doesn't make a great deal of difference in terms of latency. For that matter, the interface isn't the primary determinator of latency, either. Rather, it's the buffer size that accounts for 90% of the latency. Reduce the buffer size and you reduce latency.
 
However, smaller buffers give the CPU less time to process data, so there is a limit to how small you can shrink the buffers. At some point the CPU won't be able to keep up. That's why a faster CPU lets you make the buffers smaller and thereby reduce latency.
 
Zero-latency monitoring exists so that you don't have to worry about latency while recording audio such as guitar or voice. You only have to deal with latency when playing software synthesizers in real time. 
 
I get around that problem by using a hardware ROMpler to record my MIDI parts. That way, I'm listening to the synthesizer in real time just like I do with guitars and voice, using my interface's zero-latency monitoring. After the MIDI has been recorded, I can then redirect it to a software instrument such as a soft synth or sampler. Even though my computer is ancient (3.4 GHz Pentium D), latency is almost never an issue.
 
As for upgrading the interface, the main benefit would be getting more inputs and outputs. Focusrite is not known for especially low latency, but they are solid products and the manufacturer's customer support is excellent. I recently moved to a Focusrite interface and am pretty happy with it.

http://www.gearslutz.com/board/music-computers/618474-audio-interface-low-latency-performance-data-base-20.html
 
The bookmark is the end of the thread your welcome to start at the beginning,,, everything you wanted to know about RTL ( latency) and were afraid to ask. Reading this thread , or the important parts, was a real education for me. It is of note that Roland has refused to send these people a unit to test for latency. 

ok. I did some additional testing...this time with asio instead of wdm. the latency got a lot better than with wdm. I am now at 6 ms working at 44.1khz. compared to the wdm's 9 ms I have a 30% improvement. tried the lowest setting, 3 ma, but there were several pops and clicks. 6 work just fine though and it is virtually not detectable.
 
now a question: the sound quality should remain the same, no matter what driver I use, right? I don't hear any differences but want to make sure.
In addition, since I use direct monitoring, I can hear my guitar PRIOR to start Sonar. this said, from what I understand, the asio or wdm drivers are activated only when I get into the DAW. am I wrong? this given, when I hear my guitar prior to entering sonar it means I'm using a Windows driver.....isn't it?
 
all the best
 
michael

ok. thanks again for the explanation. so far it looks and sounds good with asio. the only issue, if I can call it so, is that when I'm playing back the song and mess around, let's say, moving a audio clip from a place to another, I can hear pops and clips but only while moving it. if I do it with the song stopped, everything is ok. with wdm at 9 ms I could virtually do everything while playing back the song with no click whatsoever.
guess this has something to do with latency setting in sonar: the lower the more pops I get.....
 
I'll have to make additional testing. for example, I didn't do any resize of the wdm buffer of my audio module. let's see.
 
will let you know what I found out in the testing.
 
overall, let me thank you again for your help: you have been fantastic, together with cactus music.
 
all the best
 
michael