Got the best processor, didn't see difference in Live

Discuss anything related to audio or music production.
Post Reply
IvanChocron
Posts: 2
Joined: Sun Dec 28, 2014 9:12 am

Got the best processor, didn't see difference in Live

Post by IvanChocron » Mon May 08, 2017 9:20 am

Hi! This is more a question for computer geeks.

I was using an i7-5930k (6 core @ 3.5 GHz). Basically a $500 processor. I took notes of Live's processor use while composing a tune.
Now I upgraded my processor to a Xeon E5-2695v3 (14 core @ 2.3 GHz). A $2000 processor, one of the best ones in the market. I noticed absolutely no difference in Live's processor gauge.
I do have the "multicore processing" option activated.
Does anyone know why? Is there any specific thing I should configure in the BIOS or in Ableton?

If interested, other specs of the PC is a Asus X99-A mother, 32GB DDR4 and an SSD drive.

Thanks.

TomKern
Posts: 358
Joined: Mon Dec 05, 2016 7:08 pm

Re: Got the best processor, didn't see difference in Live

Post by TomKern » Mon May 08, 2017 1:22 pm

Also keep in mind that Live distributes workload to cores on a per track basis. So full use of your 14 core monster would only come when you use 28 tracks (because of HyperThreading).

IvanChocron
Posts: 2
Joined: Sun Dec 28, 2014 9:12 am

Re: Got the best processor, didn't see difference in Live

Post by IvanChocron » Mon May 08, 2017 6:00 pm

Thank you for the link above about the same discussion.
Money is not an issue for me, so taking money out of the equation... imagining they would both be "free", would you go with the i7 5930k, 6 core @ 3.5, or with the xeon 2695v3, 14 core @ 2.3?

when comparing the same track, for moments of the song the i7 has better performance and for others the Xeon...

scheffkoch
Posts: 593
Joined: Thu Jan 08, 2015 12:37 pm

Re: Got the best processor, didn't see difference in Live

Post by scheffkoch » Tue May 09, 2017 7:24 am

...i would go for the higher clock rate instead of more cores!...
macbook pro m1pro, macos monterey, rme multiface via sonnet echo express se I, push 2, faderfox mx12, xone:k2

TomKern
Posts: 358
Joined: Mon Dec 05, 2016 7:08 pm

Re: Got the best processor, didn't see difference in Live

Post by TomKern » Tue May 09, 2017 9:51 am

scheffkoch wrote:...i would go for the higher clock rate instead of more cores!...
For music production? Why?!

I would do the opposite (if money were no issue at all)

It means less worry about adding more tracks to your set without chocking the system

scheffkoch
Posts: 593
Joined: Thu Jan 08, 2015 12:37 pm

Re: Got the best processor, didn't see difference in Live

Post by scheffkoch » Tue May 09, 2017 11:12 am

...but aren't e.g. synth vsts calculated by one core first so that a higher clock rate would give you more "headroom" (before you get drop outs)...?...
macbook pro m1pro, macos monterey, rme multiface via sonnet echo express se I, push 2, faderfox mx12, xone:k2

Richie Witch
Posts: 1018
Joined: Mon Feb 10, 2014 10:10 pm
Location: Washington, DC
Contact:

Re: Got the best processor, didn't see difference in Live

Post by Richie Witch » Tue May 09, 2017 11:40 am

I look at it like this....

You've taken the engine out of your car and replaced it with one that has more cylinders, although those cylinders are somewhat smaller--still, there are more of them. The problem is that you can double or triple your horsepower, but the top speed of the car is not going to change unless you also change the transmission.

That brawny processor is hobbled by the wrong architecture for its design. I think that sums up the previous thread too.

But even if you build a big, multi-processor server-style rig, there's no guarantee Live will run much faster since it's probably not designed to harness the power of that type of architecture. So your fuel injection system only has 8 ports and you've got 18 cylinders.... :D
"Watching the Sky" ~ A 4-track EP of piano, strings, and Native American flute

TomKern
Posts: 358
Joined: Mon Dec 05, 2016 7:08 pm

Re: Got the best processor, didn't see difference in Live

Post by TomKern » Tue May 09, 2017 11:54 am

scheffkoch wrote:...but aren't e.g. synth vsts calculated by one core first so that a higher clock rate would give you more "headroom" (before you get drop outs)...?...
Yes if one super heavy synth on one track chokes your system then the higher clock would be better.
That is not the usual use case though. Normally you would have many tracks all of them with some sort of heavy processing that doesn't take down a single core by itself.(even at only 2.3GHz)
Here the multicore can process each of these tracks on one core.

That usually leads to a much more stable system.

I have to admit though that 3.5 GHz and 2.3GHz are far enough apart in speed that the outright endorsement for the system with more cores is far from a slam dunk and would depend on the exact usage scenario OP has.

And since I'm never in the "price is absolutely insignificant" situation, I personally would definitely go with the 6 cores at 3.5GHz and not even feel a slight bit jealous about it.

Emanresu0891
Posts: 154
Joined: Wed Apr 02, 2014 5:29 pm
Location: Ann Arbor, Michigan U.S.

Re: Got the best processor, didn't see difference in Live

Post by Emanresu0891 » Tue May 09, 2017 8:04 pm

I also would go for the highest clocked cpu you can get. There are very few programs that utilize all the cores effectively beyond software that is custom made for things like servers running a lot of virtual machines usually used in business or school settings. When you get into the server league of cpu's you need server ram which gets very hot so then you need proper cooling. This all can get very loud which is not ideal in a studio.

I feel the best idea for studios that need a lot of native plugins is to use two or more high clocked computers running Vienna Ensemble Pro. https://www.vsl.co.at/en/Vienna_Softwar ... semble_PRO

Vep basically creates a computer cluster dedicated to running plugins.

This is what I would do if money was not an issue.

TomKern
Posts: 358
Joined: Mon Dec 05, 2016 7:08 pm

Re: Got the best processor, didn't see difference in Live

Post by TomKern » Tue May 09, 2017 8:35 pm

Emanresu0891 wrote:I also would go for the highest clocked cpu you can get. There are very few programs that utilize all the cores effectively beyond software that is custom made for things like servers running a lot of virtual machines usually used in business or school settings.
That sounds like random copy pasta someone might have picked up on some tech forum without really understanding it.

Live is a program that uses multi core pretty effectively (one core per track). While it may not be as effective as it could be, with one core per chain/device/process, it's certainly effective enough for most real world use cases.

Stromkraft
Posts: 7033
Joined: Wed Jun 25, 2014 11:34 am

Re: Got the best processor, didn't see difference in Live

Post by Stromkraft » Tue May 09, 2017 9:09 pm

TomKern wrote:
Emanresu0891 wrote:I also would go for the highest clocked cpu you can get. There are very few programs that utilize all the cores effectively beyond software that is custom made for things like servers running a lot of virtual machines usually used in business or school settings.
That sounds like random copy pasta someone might have picked up on some tech forum without really understanding it.

Live is a program that uses multi core pretty effectively (one core per track). While it may not be as effective as it could be, with one core per chain/device/process, it's certainly effective enough for most real world use cases.
I'm agreeing with Tom here. Work is stacked on each core. Number of tracks divided by (real) cores indicate the work per core that needs to be done. This normally varies per core of course (in real songs anyway) and there is also overhead.

The results as presented here does not indicate necessarily that expected gains of more cores when using more tracks isn't a reality. The assumption that more cores should improve performance in any other way than thinning out work on more (real) cores is to the best of my knowledge not reasonable.

Also, even if a certain CPU is is faster per core, that core has to calculate more tracks!. Granted if you overload a core you're overloading it.

One should also take into account that not all plug-ins with native multi-core support should be run with that active in Live. Always try to toggle this if there are performance issues.
Last edited by Stromkraft on Wed May 10, 2017 7:01 am, edited 2 times in total.
Make some music!

Stromkraft
Posts: 7033
Joined: Wed Jun 25, 2014 11:34 am

Re: Got the best processor, didn't see difference in Live

Post by Stromkraft » Tue May 09, 2017 9:16 pm

IvanChocron wrote:Hi! This is more a question for computer geeks.

I was using an i7-5930k (6 core @ 3.5 GHz). Basically a $500 processor. I took notes of Live's processor use while composing a tune.
Now I upgraded my processor to a Xeon E5-2695v3 (14 core @ 2.3 GHz). A $2000 processor, one of the best ones in the market. I noticed absolutely no difference in Live's processor gauge.
I do have the "multicore processing" option activated.
Does anyone know why? Is there any specific thing I should configure in the BIOS or in Ableton?

If interested, other specs of the PC is a Asus X99-A mother, 32GB DDR4 and an SSD drive.

Thanks.
You need to assess actual core usage with 14-28 tracks and not look at the percentage for audio engine load inside Live. Add 14 duplicates at a time until audio is breaking up and watch actual CPU usage. To make it easy make all tracks replikas so work is the same everywhere.
Make some music!

Post Reply