Is the I9 9900K overkill for music production?

[BananaJoe]

Synthetic benchmarks are at fault but still show something,

My retired i3 2120(second generation i3) could handle any big project as long as i had enough ram memory, it just could not deal multiple instances of known CPU hogs VSTi .

[legendCNCD]

an i3-8100 can be as good of a choice as an i9 9000k depending on what you do.

Exactly - like surfing the web and checking email, both are great at.

Are you running 300 notes of polyphony? Most people are not.
Are you running more than 50 VST instances? Most people are not.

This is just a synthetic benchmark, real world usage typically brings much better results(for all compared processors).

Yeah, not running those plugins, as in real world I have fex. Diva, Omnisphere2 + something, battery4.. then loads of hq effects depenging on the song, multiply the synths too..

I'd take the headroom of 9900K if having money for it, it'll give more in the long run I think.

[EnochLight]

I'd take the headroom of 9900K if having money for it, it'll give more in the long run I think.

Agreed 100%.

[Kaine]

The multi thread performance is not relevant for audio in almost all cases

Why? Is your software not capable of addressing more than one thread?

The i9 vs the i3 is insanely more powerful when it comes to total track count within a given project.

If multi core performance was a thing everyone should be using dual or quad Xeon setups that always existed.

People do and they used to do so a lot more until a few years back where single chips with tons of cores became a thing.

The reason I don't really advocate Xeon currently is the overall value on offer. You pay 20% or more on the kit over the desktop equal and don't get to overclock. Sure, you can go dual chip, but then you do take a memory addressing hit of up to 20% from the available performance overhead, much the same as you do with the 2 dies on one chip design with Ryzen.

The other main reason for not doing it right now is that none of the software is perticualarly well optimized past 32 threads, so it's a bit of a moot point for running twin chips, when you can get everything one chip with 16 cores and call it a day.

Still, 4 years ago, it was the only way to be able to fully rinse out your sequencer.

[cleverr1]

Quick question for Kaine - With the TZ390 i9 9900K system in its overclocked state is the Passmark CPU score increased from the stock?

I'm confused about the "turbo" feature and how that would kick in. Does your overclocked system maintain the max frequency or is there a higher "turbo" frequency relative to the new base frequency of 4.9 GHz.

Sorry if this is a stupid question!

[Kaine]

Regular turbo settings will stagger its core boosts, so one core will hit the advertised "turbo" clock and the rest are normally staggered by 100MHz each. So it might be 5.0, 4.9, 4.8 etc...

ASIO is fairly unforgiving, in that if any given core hits 100% load, then it'll crackle and glitch and it'll affect the whole signal. So, to some degree, it's only as strong as the weakest core.

I try to lock all the cores to the Turbo clock speed, so all cores are equal. In testing, it can offer somewhere around 10% gain where all plugins are equally sized and the loads are built up in an equal manner. I suspect the figure might prove a little higher in real terms, where one large plugin or complex effect might have a bigger impact in skewing the load balance if it hits a slow core, but that sort of testing is fairly hard to quantify with the monitoring tools available.

And no, there is no new turbo clock above it. If it's set it to given value then it's locked to that speed.

[cleverr1]

Thanks for this. My i7 4790K system died so I'm about to order your TZ390 i9 9900K system. There are a few items in the configurator that I don't need that don't have null options so I've used the online form to request a quote for the system without them.

[lwj]

@Kaine, that's an interesting suggestion. It makes sense if ASIO works like that. I'll give it a try and see if it makes a difference. I think the way it's set now, my CPU isn't locked but it should get equal turbo speeds on all cores. 4,8ghz I think.

It might not get a result though, since I start hearing dropouts well before it's near 100% utilization, so it's clearly still bottlenecking somewhere else first.

[EnochLight]

@Kaine, that's an interesting suggestion. It makes sense if ASIO works like that. I'll give it a try and see if it makes a difference. I think the way it's set now, my CPU isn't locked but it should get equal turbo speeds on all cores. 4,8ghz I think.

It might not get a result though, since I start hearing dropouts well before it's near 100% utilization, so it's clearly still bottlenecking somewhere else first.

Which DAW?

[Kalamata Kid]

On the Gigabyte - Z390 AORUS MASTER Motherboard there are two USB headers:

1. Type-C header with USB 3.1 Gen 2 support. This is the small 20 pin header
2. USB 3.1 Gen 1 header. This single header can provide two USB ports (need a splitter?) type A, I assume. This is the big 20 pin (or is it 19 pin?) header.

I prefer to have a front panel (controller panel) that supports at least three USB 3.1 of which on is type-C and the other two are type-A. But then again all can be type-A. I have searched but cannot find a panel that has all three USB post at 3.1. I may consider putting some or all in the back.

[BananaJoe]

I think some people got me wrong. I was just trying to make a point that that low end 8th\9th generation Intel Core is enough for most people, not all! Of course the better performer you get the best.

But answering your question, latency relies mostly on single core clock. And for everything else the lower end CPUs of today have four or six cores which is good enough and high enough clock speeds, more than that while typically(not always!) giving you better performance will reduce your clock speed per core that is why cores never increases performance by too much.

Regarding everything else, most software have limits on how many cores they can use or are not optimized as you said. Cubase for example is limited in 12 or 14 cores, i don't remember well. If you have more Cores than that Cubase will not make use of the extra ones. Having a 128 cores quad Xeon would be futile.

The multi thread performance is not relevant for audio in almost all cases

Why? Is your software not capable of addressing more than one thread?

The i9 vs the i3 is insanely more powerful when it comes to total track count within a given project.

If multi core performance was a thing everyone should be using dual or quad Xeon setups that always existed.

People do and they used to do so a lot more until a few years back where single chips with tons of cores became a thing.

The reason I don't really advocate Xeon currently is the overall value on offer. You pay 20% or more on the kit over the desktop equal and don't get to overclock. Sure, you can go dual chip, but then you do take a memory addressing hit of up to 20% from the available performance overhead, much the same as you do with the 2 dies on one chip design with Ryzen.

The other main reason for not doing it right now is that none of the software is perticualarly well optimized past 32 threads, so it's a bit of a moot point for running twin chips, when you can get everything one chip with 16 cores and call it a day.

Still, 4 years ago, it was the only way to be able to fully rinse out your sequencer.

[HM]

I may finally assemble my PC this coming Sunday.
I am concerned about the BIOS settings on the
Gigabyte - Z390 AORUS MASTER with i9-9900k.

I see various AORUS board being mentioned again and again, but when I check out there seems typically to be the dreaded "double 16X" PCIe slots that I was told to avoid back in this thread, (due to multiplexer/latency),..

A bit confused now

What is it that I fail to understand ?

[DarkERA]

When it comes to CPU its never overkill. Though 9900k is a beast and long lasting CPU

[Pictus]

I may finally assemble my PC this coming Sunday.
I am concerned about the BIOS settings on the
Gigabyte - Z390 AORUS MASTER with i9-9900k.

I see various AORUS board being mentioned again and again, but when I check out there seems typically to be the dreaded "double 16X" PCIe slots that I was told to avoid back in this thread, (due to multiplexer/latency),..

A bit confused now

What is it that I fail to understand ?

Only two Z390 motherboards have the PLX chip that is bad for latency, the
ASUS WS Z390 Pro and SuperMicro C9Z390-PGW.
In the other Z390 motherboards if you use a graphic card it will take 16 lanes, but
if you use two graphic cards each will take 8 lanes.
In the ASUS WS Z390 Pro and SuperMicro C9Z390-PGW if you use two graphic
cards, each will take 16 lanes because of the PLX chip, but with the side effect
of added latency...

[HM]

I may finally assemble my PC this coming Sunday.
I am concerned about the BIOS settings on the
Gigabyte - Z390 AORUS MASTER with i9-9900k.

I see various AORUS board being mentioned again and again, but when I check out there seems typically to be the dreaded "double 16X" PCIe slots that I was told to avoid back in this thread, (due to multiplexer/latency),..

A bit confused now

What is it that I fail to understand ?

Only two Z390 motherboards have the PLX chip that is bad for latency, the
ASUS WS Z390 Pro and SuperMicro C9Z390-PGW.
In the other Z390 motherboards if you use a graphic card it will take 16 lanes, but
if you use two graphic cards each will take 8 lanes.
In the ASUS WS Z390 Pro and SuperMicro C9Z390-PGW if you use two graphic
cards, each will take 16 lanes because of the PLX chip, but with the side effect
of added latency...

Thanks !

Guess I will just put the Aorus Pro rev 1.0 on my monitor-list for a good offer, the basic-price is not crazy

Recently I have seen various 2 x 16G DDR4 3000 relatively cheap also