AMD's a-comin!!!... and Intel's been a-dunnin!!

[Scotty]

It is a small subset of the market that would be willing to pay for that kind or processing power. That could included audio professionals, serious hobbyists with high track counts/sound design (and to a lesser extent people that can afford whatever system they want because they have means).

The performance matters to that population. Having your system keel over after you load up some virtual instruments would be a deal breaker as would not being able to track a drum kit at 128 samples across 10 or more channels.

Perhaps 95 percent of computer based musicians may not be able to appreciated the differences between a TR2 and an I9 7980XE but these processors are more for the workstation power user crowd.

Even within the audio niche, 95% of users won't notice the difference between top AMD and Intel processors.

You have mentioned that you got a new Ryzen system, right?

Are you satisfied with it?

Well, I have very modest needs, I don't use hundreds of instances at the same time as they do in certain benchmarks. And I bet neither do the majority of computer music makers.
I deliberately bought only a 4-core, the 1500X. It works fine and very efficiently with Reaper.

[fluffy_little_something]

It is a small subset of the market that would be willing to pay for that kind or processing power. That could included audio professionals, serious hobbyists with high track counts/sound design (and to a lesser extent people that can afford whatever system they want because they have means).

The performance matters to that population. Having your system keel over after you load up some virtual instruments would be a deal breaker as would not being able to track a drum kit at 128 samples across 10 or more channels.

Perhaps 95 percent of computer based musicians may not be able to appreciated the differences between a TR2 and a n I8 7980XE but these processors are more for the workstation power user crowd.

The latest TR will be fine for 99% of music requirements, let alone the usual non-music applications.
The discussion here is like whether to buy a Ferrari or a Lamborghini

[Scotty]

Consider the expertise of scanaudio and the dawbench work that they have done across buffer settings and latencies. The benchmarks are clear.. If you do a lot of low latency audio with high track counts the TR2 isn't the best choice... the data proves it out. That is what the focus of the discussion has been about not about who has the biggest dick. We are talking about workstation class performance. TR2 is an amazing processor but its design doesn't lend itself to low latency audio work which if this is where you live is not a good choice.

I'll be building two systems in the coming months one will be TR2 based for my son who is a film editor and another which is I9 based for my studio. There is no bias there... just choosing the best tools for the given job.

[fluffy_little_something]

And for 99% of people on KVR TR is the right tool as well

[Scotty]

By your own logic a TR2 or I9 would be overkill for most users and would most definitely not be the right tool for the job unless they had money to burn. I'll leave it there. I think I understand your central point - I just don't agree with it.

[fluffy_little_something]

Indeed, I think the latest super processors are overkill for most computer musicians. After all, up until recently i7 and i5 were just fine for most musicians. And there hasn't been any real change in the DAW and plugin world.

[Straight2Vinyl]

I think I'll have to finally upgrade my old rig next year when I have the funds available. The Phenom quad core 2.7 processor just isn't enough for plugins like Reaktor and Iris 2, which I'd like to get a lot more use out of.
I'm guessing my price range for the CPU will leave me somewhere in the I7 8700K or Ryzen range. I'll wait and see what the rumored early 2019 processors look like.
Anyone else using Reaktor and/or Iris 2? Any suggestions for a PC build that could handle those without getting into the TR price range?

[Scotty]

Reaktor seems to bring powerful computers to their knees. I have friend in the Ambient scene..he runs an i7 3930K overclocked to 4.2ghz - near identical to my current system - 6 cores - 12 threads. He runs a lot of Reaktor stuff and his track counts are low - but still Reaktor crushes his machine.I am not sure how Reaktor scales across cores or if he is running it in oversample mode but it is really demanding on CPU resources.

I think I'll have to finally upgrade my old rig next year when I have the funds available. The Phenom quad core 2.7 processor just isn't enough for plugins like Reaktor and Iris 2, which I'd like to get a lot more use out of.
I'm guessing my price range for the CPU will leave me somewhere in the I7 8700K or Ryzen range. I'll wait and see what the rumored early 2019 processors look like.
Anyone else using Reaktor and/or Iris 2? Any suggestions for a PC build that could handle those without getting into the TR price range?

[Pictus]

Working(hack) Thunderbolt on Threadripper!
https://youtu.be/uOlQbP63lDQ

[Kaine]

The 2950X?
Would not be the 2990WX?

It would... brain fart.

Reaktor seems to bring powerful computers to their knees. I have friend in the Ambient scene..he runs an i7 3930K overclocked to 4.2ghz - near identical to my current system - 6 cores - 12 threads. He runs a lot of Reaktor stuff and his track counts are low - but still Reaktor crushes his machine.I am not sure how Reaktor scales across cores or if he is running it in oversample mode but it is really demanding on CPU resources.

I'm starting to see a similar sort of issue running a 3960K from that generation and certain packages like that. Given you've noted that the track count is low, I'll note that his single core score is around 2000 and chances are something as complex as Reaktor could be overloading a single core. He could check this next time it starts to glitch out by checking system monitor and seeing if any given perticular core is maxing out.

By comparison a current 8700K is running with close to 3000 on the single core score, so whilst both chips might look like they are running with 6 cores at around 4.2GHz, the newer generations have an additonal 40% or 50% worth IPC overhead to play with, which can make a hell of difference with the more intensive plugs.

[nIGhT-SoN]

When you take into consideration upgrade path, performance per dollar, platform price (motherboard), I think AMD wins big time. For example you can run a B350/450 motherboard with a Ryzen 7 without any problems, even overclock it, and those motherboards are quite good for the price. I have a MSI B350 Gaming Plus and I have really good latency overall and it's paired with an Ryzen 7 1700. For the money I spent I couldn't get a 7700k because a motherboard for that kind of cpu is quite expensive. Sure if money is not a problem, Intel wins at performance, for now. Looking how the things are going, next AMD cpu generation will be a headache for Intel, especially since AMD will use 7nm and Intel isn't able to produce at that level, they will still be using 12nm for a few years.

We gotta see the big picture, competition is good for us, consumers. Intel has kept us with 10% performance increase with every generation for some time. Now with AMD being a true competitor they will have to do more than that. We are living interesting times for cpu market.

[Scotty]

Once again your expertise is invaluable. I scour the web for good information on these topics and with few exceptions you are it. Thanks for taking the time to comment.

The 2950X?
Would not be the 2990WX?

It would... brain fart.

Reaktor seems to bring powerful computers to their knees. I have friend in the Ambient scene..he runs an i7 3930K overclocked to 4.2ghz - near identical to my current system - 6 cores - 12 threads. He runs a lot of Reaktor stuff and his track counts are low - but still Reaktor crushes his machine.I am not sure how Reaktor scales across cores or if he is running it in oversample mode but it is really demanding on CPU resources.

I'm starting to see a similar sort of issue running a 3960K from that generation and certain packages like that. Given you've noted that the track count is low, I'll note that his single core score is around 2000 and chances are something as complex as Reaktor could be overloading a single core. He could check this next time it starts to glitch out by checking system monitor and seeing if any given perticular core is maxing out.

By comparison a current 8700K is running with close to 3000 on the single core score, so whilst both chips might look like they are running with 6 cores at around 4.2GHz, the newer generations have an additonal 40% or 50% worth IPC overhead to play with, which can make a hell of difference with the more intensive plugs.

[bmrzycki]

Once again your expertise is invaluable. I scour the web for good information on these topics and with few exceptions you are it.

Agreed. Thank you Kaine, your ongoing testing posts are why I continue to read this thread.

[Debutante]

https://bit-tech.net/news/tech/cpus/cor ... formance/1

https://www.guru3d.com/news-story/amd-r ... -tool.html

https://hexus.net/tech/news/cpu/125819- ... prio-tool/

...so this is the latest... Windows scheduler largely to blame for poor TR performance, or is it? Why did it take a third party to do AMDs job like that? Great news all the same.

Who has a TR and how've you fared?

[camsr]

Windows has a weird way of switching threads between cores. It's always baffled me as to why it does constant switching of even lightweight processes (with no propensity to max out) between cores.