Page 1 of 1

Lets compare resource usages between 4 & beta 5!

Posted: Sun Jun 19, 2005 12:37 am
by timothyallan
Okay, I realize that 5 is still beta so there is going to be some bloat. However, the most unexpected thing has happened with 5.. and IT'S GOOD!

I loaded up a Live 4 project I'm working on, 25+ tracks with midi and audio.

In Live 4.1.4 : 80% processor usage, 450mb RAM used
In Live 5b1: Get this... 50-55% processor usage!!! but 860mb RAM used.

The RAM i figure is for debug code etc etc, plus RAM is cheap and CPU's aint! How often does this happen, where an update uses LESS CPU than the previous version. Not just less, but like 20-25% less!!

Truely, this is Yaybleton Live 5!

Any others with similar experiences?

-Tim

Posted: Sun Jun 19, 2005 12:44 am
by Machinate
yeah, my template is about 20 % percent lighter and reaktor 4 is also pretty lean. As long as I steer clear of pitchshifting "complex" samples, I'm good!

Posted: Sun Jun 19, 2005 1:00 am
by AdamJay
i really wouldn't do any comparisons till the final version.
its gonna change with every beta, and there's bound to be OS specific debugging code in the betas.

but yea, things do look positive ;)

Posted: Sun Jun 19, 2005 1:01 am
by AdamJay
Machinate wrote:As long as I steer clear of pitchshifting "complex" samples, I'm good!
freeze those bastards! ;)

Complex sounds soooo good. and its great to know that Freezing a complex warped clip reduces its cpu usage.

Posted: Sun Jun 19, 2005 1:14 am
by timothyallan
AdamJay wrote:i really wouldn't do any comparisons till the final version.
its gonna change with every beta, and there's bound to be OS specific debugging code in the betas.
Yes, the reason i originally posted this was because even with the added debug code and bloat, it still performed better than the previous version. I wanted to see if everyone was experiencing the same phenomenon. I had written that I realized it was a beta and contained some extra code.

I'm not trying to take away from your performance thread at all with a new beta version one, just voicing my suprise at the lesser processor usage. I'm a software developer myself, and it's a rare thing for this to happen without a total rewrite.

-Tim

Posted: Sun Jun 19, 2005 1:45 am
by AdamJay
for what its worth, the Live 4 Performance test does perform an overall 4% better in v.5b1 (went from 49% to 47%) now, i kinda hope there is a bunch of debug code in there. cause as you said, it could get even better.

i'll take that 4+%! and freeze, and saturator, and track resizing, and , and , and.....


oh, btw, the day of the final release i'll release a NEW Live 5 Performance Test that will be totally diffferent from v.4's test. it will be independent of how v.4 compares to v.5, as it will serve the same purpose of the last test: benchmarking machines against each other.

Re: Lets compare resource usages between 4 & beta 5!

Posted: Sun Jun 19, 2005 2:29 am
by supster
timothyallan wrote:
In Live 4.1.4 : 80% processor usage, 450mb RAM used
In Live 5b1: Get this... 50-55% processor usage!!! but 860mb RAM used.

Any others with similar experiences?

-Tim

ya, i noticed this too. ive been so busy i havent gotten around to finishing a compartive test of my own .. but .. no doubt about it:

even two, three instances of reaktor are way ligther on my cpu.

also, ive had a problem with Live since vs 2: a "stickiness" when the resource meters get above even 40% at times! everything slowed down to a crawl on the interface using VSTs, even when i wasnt pushing it to the max

its silky smooth now :) first time in 3 versions. so... they definitly changed some fundamental code somewhere.
.

Re: Lets compare resource usages between 4 & beta 5!

Posted: Sun Jun 19, 2005 2:45 am
by ilia
You guys do realize that L5 behaves differently with respect to inactive plugins? E.g. if you have a return track that nothing is being sent to, the live5 CPU meter doesn't count the plugins on that track (and, I am assuming, the CPU is being conserved).

Posted: Sun Jun 19, 2005 4:17 am
by timothyallan
im using task manager for cpu %, Live never reports actual usage for me...

Posted: Sun Jun 19, 2005 4:20 am
by AdamJay
illa, yea i think that can be an attribute to the folks who are seeing the larger considerable cpu differences in their templates.

but when you compare the playback of pre-existing material in L4 vs. L5, and watch the CPU, you'll see that there is some faster performance there. certainly not as great a difference, but it is at least a little faster.

and if its using less cpu on inactives... great too!

Posted: Sun Jun 19, 2005 9:15 am
by poshook
My projects have lower CPU consumption under Live4 (difference cca 10-15%).

Posted: Sun Jun 19, 2005 10:49 am
by DJRetard
Same here, better performance on ver four.

Posted: Sun Jun 19, 2005 11:02 am
by timothyallan
I'd like to go a little OT and say that DJRetard is a fuggin funny name. Please tell me you actually play out with that name!? I think I'd go see you even if you played horribly mixed german death metal trance.

Posted: Sun Jun 19, 2005 3:04 pm
by DJRetard
Thanks Timothy. Website coming soon. Full of Retarded mixes and tunes.

Posted: Sun Jun 19, 2005 3:54 pm
by David_T
Tim's comment about RAM is interesting:

I had a huge graphics slowdown followed by bluescreen crash right after hitting the spacebar twice in arrange view. I had it happen a second time, but was able to shut down the program before it crashed the system. I can't reproduce it deliberately, so kinda hard to report it as a bug, but it seems like it might have been sort of a RAM overload.

Pent. M 1.6Ghz, 512 RAM.