Taperssection.com

Gear / Technical Help => Post-Processing, Computer / Streaming / Internet Devices & Related Activity => Topic started by: blu666z on February 23, 2004, 05:14:17 PM

Title: Testing Speed Improvements
Post by: blu666z on February 23, 2004, 05:14:17 PM
New RAM stick should be here today.  I want to test how much quicker I can resample a large file.  Is an hour of pink noise a good test to resample from 48 -> 44.1 or should I use an actual show?  Not sure if pink noise might be easier or harder to resample and then skew my results a little.

-Kevin
Title: Re:Testing Speed Improvements
Post by: mirth on February 23, 2004, 05:43:54 PM
Kevin, why not just do a before & after?
Title: Re:Testing Speed Improvements
Post by: blu666z on February 23, 2004, 06:08:57 PM
I am.  Just trying to figure out if pink noise is a good control to use.  If pink noise is always easy to downsample it might not give the most accurate results.

-Kevin
Title: Re:Testing Speed Improvements
Post by: Simp-Dawg on February 23, 2004, 06:47:26 PM
do a before with pink noise and one with a show, then do the same after installing ram
Title: Re:Testing Speed Improvements
Post by: John Kelly on February 23, 2004, 08:14:27 PM
Wouldn't resampling be more of a test of processor speed than it would of the amount of RAM?
Title: Re:Testing Speed Improvements
Post by: nic on February 23, 2004, 09:51:19 PM
Wouldn't resampling be more of a test of processor speed than it would of the amount of RAM?

it is more of a cpu workout, but the data still has to get to the cpu.
data coming from ram is faster than data coming from a hard-disk
Title: Re:Testing Speed Improvements
Post by: blu666z on February 23, 2004, 10:08:05 PM
Not sure what to think of this.

48 -> 44.1
16 bit
Anti-alias filter
Highest level of interpolation accuracy
Soundforge 6.0
78 minute file

512MB RAM - 14:15
1 GB    RAM - 14:34

Took longer...what the hell?!?!

-Kevin
Title: Re:Testing Speed Improvements
Post by: nic on February 23, 2004, 10:17:50 PM
is all of the memory the same speed?
what platform(cpu, mobo, ram type/speed) are you using?
Title: Re:Testing Speed Improvements
Post by: blu666z on February 23, 2004, 10:34:02 PM
is all of the memory the same speed?
what platform(cpu, mobo, ram type/speed) are you using?

I couldn't figure out what speed my old stick, 512mb,  was running at so I just bought the fastest my mobo was rated for under the impression it would run at a slower speed if the old stick was slower.

XP
Asus A7V333
AMD Athlon XP 1800+ 1.53 GHz
512MB 184-PIN DIMM 64MX64 DDR PC2700 <-New Stick

Did I do something wrong?  XP is at least recognizing the added memory.

-Kevin
Title: Re:Testing Speed Improvements
Post by: nic on February 23, 2004, 11:02:47 PM
I'm not 100% positive, but I'm thinking the XP 1800 is limited to a FSB(front side bus) of 133 which is equivalent to DDR 2100(DDR naming conventions are even worse than AMD's conventions for their cpus!)

you may also want to check to see what your virtual memory config is....with 1GB of ram, I tested/ran my system with *0* virtual memory and it ran considerably faster but wasnt 100% stable for all applications.
play with the memory settings of your operating system to find the best balance of RAM vs virtual....normally you want about 1.5x the amount of ram as your virtual, but I've found that audio processing with large amounts of memory you may actually want to set the virtual to about 1/4 that of physical ram...nothing scientific to back this up, just personal observations based on MY workstation:
PIII 1.12 GHz
1GB PC133 RAM
ASUS mobo w/ VIA chipset -??? model
onboard IDE plus (2x) Promise 133tx IDE controllers  -only 1 drive per channell
all hard drives are ATA133 @ 7200 rpm with DMA on

also noticed that with Sound Forge, if the SF temp folder is on a seperate drive from the operating system, it speeds up processing and save time
Title: Re:Testing Speed Improvements
Post by: blu666z on February 23, 2004, 11:11:27 PM
I'm not 100% positive, but I'm thinking the XP 1800 is limited to a FSB(front side bus) of 133 which is equivalent to DDR 2100(DDR naming conventions are even worse than AMD's conventions for their cpus!)

you may also want to check to see what your virtual memory config is....with 1GB of ram, I tested/ran my system with *0* virtual memory and it ran considerably faster but wasnt 100% stable for all applications.
play with the memory settings of your operating system to find the best balance of RAM vs virtual....normally you want about 1.5x the amount of ram as your virtual, but I've found that audio processing with large amounts of memory you may actually want to set the virtual to about 1/4 that of physical ram...nothing scientific to back this up, just personal observations based on MY workstation:
PIII 1.12 GHz
1GB PC133 RAM
ASUS mobo w/ VIA chipset -??? model
onboard IDE plus (2x) Promise 133tx IDE controllers  -only 1 drive per channell
all hard drives are ATA133 @ 7200 rpm with DMA on

also noticed that with Sound Forge, if the SF temp folder is on a seperate drive from the operating system, it speeds up processing and save time

Memory crucial said would work with my mobo:
http://www.crucial.com/store/listparts.asp?model=A7V333&DetailMB=Y&cat=RAM

At the bottom it says:
Important information about your system
If three PC2700 modules are used, the system will only run at 266MHz.

That is bus speed right?  

I'll play around with the virtual memory settings. Thanks.

-Kevin
Title: Re:Testing Speed Improvements
Post by: nic on February 23, 2004, 11:22:46 PM
the motherboard may take DDR2700, but the XP 1800 wont run at that bus speed...this is still fine as the memory should underclock to the bus speed of the cpu.

with the XP line, AMD went to a "double speed" bus architecture; 2 bus per processor. thats why I said 133 FSB and Crucial is saying 266, Crucial is giving you the "double speed" value for the 133 FSB.
DDR also performs best when it is installed in pairs like old-school sims(but with sims, you HAD to install in pairs, with DDR you dont have to)
if your original DDR memory is 2100 then your newer memory wont sync up for best performance because of differing speeds, CAL settings, etc...this performance increase isnt always very noticeable though...it is noticeable under heavy video rendering and 3d modeling which is much more ram intensive than audio processing is
Title: Re:Testing Speed Improvements
Post by: blu666z on February 23, 2004, 11:39:56 PM
Bummer.  I really thought I was going to see a nice increase.

-Kevin