Is the I9 9900K overkill for music production?

[v1o]

Reading the reviews and benchmarks. The AMD Threadripper chips are faster in productivity applications and for media encoding. Only get the 9900k if you want the best gaming rig.

[Kalamata Kid]

two weeks ago I ordered the i9-9900k form Compsource for $515. Might get it in mid November or as late as mid December.

Now need to get the motherboard cooler and ram. Will use the current case, PSU and Video card,

A bit of research I did I got the impression that this CPU will be great for music creation. Had considered even the top AMD 2950x 12 core CPU for $900 but decided against it for the fewer but faster cores of the 9900k plus to save $375.

[Kaine]

Well, 2990WX is also a 32c / 64t so I wouldn't think 250W TDP would be much for such a CPU.

It's also physically over a third more surface area and is running with less IPC than the Intel chip.

TR = 58.5 mm × 75.4 mm x 7.6mm
9900K = 37.5 mm x 37.5 mm x 4.4 mm

The Intel also behaves quite well at stock, it just gets warm quickly above that when pushed, as you say it's being sold a bit closer to it's limit than we've seen over recent generations.

So yes, any chip over 150w's is a hot chip and any chip bouncing around 250w's is a really hot chip. I do agree, but it doesn't mean they can't be cooled, although I wouldn't blame someone for choosing to wait util a refined refresh came along.

In every review I've seen, 9900k is a really hot CPU, it's quite clear Intel can't go higher frequency than this with their current generation and for 10nm there is a long way for them.

Yeah, I don't deny that for a second, I thought it was the case whilst overclocking too. Both of those chips are being run at close to their respective "ramp up" points (the point where voltage pull goes through the roof when overclocking... for want of a better term) and this is normally what happens when competition is tight between them.

The Zen's currently are poor overclockers and now so are the current I series, at least both in comparison other more recent models. All it really means is that both of them are being pushed really hard currently and the spare overhead is no longer there.

Thinking back to 10 - 15 years ago, if you'd have told me that 8 cores chips were out and you could overclock all of them for 300MHz each for a total of 2.5GHz of overclocking.... whilst I was still running a Prescott 1.8GHz chip... well, we've been spoiled in comparison.

As a note, nobody at these chip firms wants to release the chips being pushed this hard from the factory. They prefer the overhead being there, but once both firms get into the hyper-competitive mode they tend to slash the end users performance overhead to shreds as it looks better printed on the box where prospective customers can see it, rather than in a magazine read by a more limited audience.

[Kalamata Kid]

Reading the reviews and benchmarks. The AMD Threadripper chips are faster in productivity applications and for media encoding. Only get the 9900k if you want the best gaming rig.

I read some post and am glad I ordered the 9900k but then I read yours and I get buyers remorse. I am not savvy enough to know who is right. If I had gotten up from the left side of the bed last week I may not have ordered the 9900k but there seems to have been an urgency to get near the top of the waiting list so I ordered.

[Ichad.c]

I went with the 8700k.

I'm sure you'll be satisfied - congrats!

Ditto, I ugraded from a i5 2500k to the 8700k, worth every penny! Well, except the extortionist RAM prices.

Great to hear! I am running a i5 2500k as well and planning on getting a 8700k or maybe go with a Laptop 8750H.

I'd rather go with the 8700K unless you really need the laptop for stage work, if the laptop is just for home, you'll probally have it plugged in the whole time(performance = power usage) so it just like a desktop except that you can't choose your own parts or expand storage/RAM easily, and the 8700K is faster and has more cache.

https://cpu.userbenchmark.com/Compare/I ... 0418vs3937

[adl]

I went with the 8700k.

I'm sure you'll be satisfied - congrats!

Ditto, I ugraded from a i5 2500k to the 8700k, worth every penny! Well, except the extortionist RAM prices.

Great to hear! I am running a i5 2500k as well and planning on getting a 8700k or maybe go with a Laptop 8750H.

I'd rather go with the 8700K unless you really need the laptop for stage work, if the laptop is just for home, you'll probally have it plugged in the whole time(performance = power usage) so it just like a desktop except that you can't choose your own parts or expand storage/RAM easily, and the 8700K is faster and has more cache.

https://cpu.userbenchmark.com/Compare/I ... 0418vs3937

I am looking into getting a Thinkpad P52, which comes with a 8750H. Seems like a real workhorse plus i get the mobility. Not really planning on going to the stage, but visiting buddys every now to do music. The specs for the Thinkpad P52 are really good and regarding expanding: the only thing i upgraded during the last 6,5 years (where i got my current Desktop) was expanding RAM and HD storage with (SSD). Those are the exact components, which easily can be upgraded on a Laptop as well.

[EnochLight]

I saw his review a few days ago. I'm calling BS on clocking the 9900K to 5Ghz on all cores - every other review I've read indicates to get that high, you're baking the die to dangerously high temps. The TDW is just too much at 5Ghz overclock on all cores.

https://www.tomshardware.com/reviews/in ... 47-11.html

The extreme torture test, which far exceeds any kind of workload an audio system created a maximum tdp pull of 249w, where the cooler I was using is rated to 250w TDP and I've seen it handle more.

We're seeing some variance in chips now they are released where some won't clock above 5GHz, some do 5.2GHz all day long, so currently we're binning here and this isn't uncommon in the first few batches, the silicon quality will stabilize over the coming months, it normally does.

Also, just to note, the 2990WX is also a chip that pulls 250W TDP, we're not seeing anything here that is shockingly extreme.

In this instance, Prime got just over a 3 hour test and AIDA stress test got a 12 hour one and nothing throttled within that time frame. Chips don't pull constant loads, AVX instructions send them through the roof and nothing audio hammers that instruction set remotely in the fashion the stress tests do. The chip will pull the current that it needs and I'm sure in both cases if you throw a 24-hour torture test onto them then yes, you probably will bake them to some degree. That's an argument against doing crazy assed testing with synthetics more than anything else though, as that isn't remotely close to being a real-world scenario.

The biggest problem I've seen commented on so far is the VRM's cooking, which everyone's noted is solved by a bit of airflow over them. This has always been one of the reasons I prefer tower air coolers over AIO's, as air is forced by them over those crucial VRM's when overclocking.

What were the die temps on all 8 cores when you were pushing them to 5 Ghz? Also, exactly what cooler did you end up using to reach the 5 Ghz overclock? I'm seriously considering the 9900K.

Thanks

[Kaine]

BeQuiet Dark Rock Pro 4. The testing will tend to rotate through the cores when running, so at any given time you'll have a highest and lowest pair and of course this will swing depending upon the test being used in the cycle.

The highest point of the prime cycle saw 1 core hit 87 whilst the lowest was sat around 10 degrees lower. It was brief and only on 2 parts of the 10 part test. Rest of the time it was bouncing between 70 and 80 degrees per core and those peaks were only really appearing with AVX running hard, otherwise, it drops further.

Throttle point is 100 degrees and I remained 10% away from that at the highest point, under the heaviest loads from the synthetic, so I thought that promising. It was a launch chip through, so I can't rule out further batch variations, I'll be alerted if things start to fail, but I'm wondering if it worth a further retest anyway once the dust clears and we start seeing decent runs coming through. I honestly don't know how these yields are doing in light of the shortage we're seeing right now.

[Scotty]

Kaine... any idea how roughly the i9 9900K will do relative to a 3930K at 4 ghz . I use RME Raydats and track live at 64 samples and then mix to as high as 2048 samples if my 3930K is struggling. I am hoping for a 50% to 70% increase in ASIO bandwidth if I go with an I9 9900k as compared to my 3930K. Do you think that is reasonable. I use a lot of effect chains as I do sound design as I mix... That is my workflow with may 6 to 9 instances of Kontakt and/or East West for light orchestral backing.

[EnochLight]

BeQuiet Dark Rock Pro 4. The testing will tend to rotate through the cores when running, so at any given time you'll have a highest and lowest pair and of course this will swing depending upon the test being used in the cycle.

The highest point of the prime cycle saw 1 core hit 87 whilst the lowest was sat around 10 degrees lower. It was brief and only on 2 parts of the 10 part test. Rest of the time it was bouncing between 70 and 80 degrees per core and those peaks were only really appearing with AVX running hard, otherwise, it drops further.

Throttle point is 100 degrees and I remained 10% away from that at the highest point, under the heaviest loads from the synthetic, so I thought that promising. It was a launch chip through, so I can't rule out further batch variations, I'll be alerted if things start to fail, but I'm wondering if it worth a further retest anyway once the dust clears and we start seeing decent runs coming through. I honestly don't know how these yields are doing in light of the shortage we're seeing right now.

Thanks. On the 9900K specs page, it states T Junction is 100 C (Intel states their T Junction is the max temp allowed at the die). It very likely throttles at a lower temp than 100 C...

https://www.intel.com/content/www/us/en ... 9900k.html

I'm still considering the 9900K, and would most definitely overclock. I'm just trying to determine whether or not pushing it to 5 Ghz is worth shaving life off of my purchase. I realize I won't push those temps in my DAW except for short bursts, but I also do video transcoding and run a 24/7 media server with the same machine, so it worries me that I might be transcoding several Plex video streams at the same time for 2-3 hours, possibly frying the chip.

I'll consider that BeQuiet Dark Rock Pro 4 - looks like a decent cooler.

[EnochLight]

Kaine... any idea how roughly the i9 9900K will do relative to a 3930K at 4 ghz . I use RME Raydats and track live at 64 samples and then mix to as high as 2048 samples if my 3930K is struggling. I am hoping for a 50% to 70% increase in ASIO bandwidth if I go with an I9 9900k as compared to my 3930K. Do you think that is reasonable. I use a lot of effect chains as I do sound design as I mix... That is my workflow with may 6 to 9 instances of Kontakt and/or East West for light orchestral backing.

Um, I think it's pretty clear that the 9900K will crush your old 3930K. It's almost 2X the speed in multi, and single thread looks roughly 50% faster:

[Kaine]

Not with DB results I'm afraid. I've done both, but on vastly different builds of DB, so not really comparable.

CPU mark raw benches for the 3930K was 12000 and the 9900K looks to be around 20500, so it's probably the best comparison I can pull from anywhere.

*Edit* I Should probably learn to refresh the page before replying....

[Kaine]

Thanks. On the 9900K specs page, it states T Junction is 100 C (Intel states their T Junction is the max temp allowed at the die). It very likely throttles at a lower temp than 100 C...

Ahh, ok, I had it in mind that the throttle kicked in there, but having just Googled up people referencing the Intel tech guides it's normally 5 degrees below that point, sorry, my error.

So, if indeed accurate it's 95 degrees. I was getting throttle warnings around there, but honestly couldn't really tell the difference once it started to spike up. There's very little distance for a meter to move between 95 and 100 degrees and once you're in thermal run away it's a rather rapid process for the test warnings to start appearing as you've probably seen before.

I'm still considering the 9900K, and would most definitely overclock. I'm just trying to determine whether or not pushing it to 5 GHz is worth shaving life off of my purchase. I realize I won't push those temps in my DAW except for short bursts, but I also do video transcoding and run a 24/7 media server with the same machine, so it worries me that I might be transcoding several Plex video streams at the same time for 2-3 hours, possibly frying the chip.

My current Plex server (and it has been 24/7 for a couple of years now) is my 10-year-old 960X chip that did 5 years as my studio system first. It's always had a 10% overclock and still holds up well, even after a decade of abuse. Admittedly, it was never as hot at launch as a current i9 has proven to be, but unless you're streaming more than 4 or 5 people concurrently I'd hope that the new chip wouldn't be under quite as much pressure given the headroom you'll have to play with.

[EnochLight]

My current Plex server (and it has been 24/7 for a couple of years now) is my 10-year-old 960X chip that did 5 years as my studio system first. It's always had a 10% overclock and still holds up well, even after a decade of abuse. Admittedly, it was never as hot at launch as a current i9 has proven to be, but unless you're streaming more than 4 or 5 people concurrently I'd hope that the new chip wouldn't be under quite as much pressure given the headroom you'll have to play with.

Cool, thanks for that! I actually hadn't even considered just keeping my existing 3770K as my sole Plex server. Hmmmm... Now you've got my gears turning - LOL!

That would save me from being in the middle of a heavy writing/production setting and wondering why my CPU is getting eaten up (family member's TV habits at night are always unpredictable). Might be worth the additional energy consumption.

Cheers!

[Scotty]

Yeah so it is really hard to pinpoint where I will land. Based on what I understand I might be able to get 50-% tp 70% more ASIO through-put on average buffers (say 256) but it won't blow the doors off the 3930K. It already has a high clock (4.2ghz to 4.5ghz stable on water) and quad channel memory with 12 logical cores. Not quite enough juice to justify a new build quite yet at least at the price point I hope to come in at.

Not with DB results I'm afraid. I've done both, but on vastly different builds of DB, so not really comparable.

CPU mark raw benches for the 3930K was 12000 and the 9900K looks to be around 20500, so it's probably the best comparison I can pull from anywhere.

*Edit* I Should probably learn to refresh the page before replying....