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

[Kaine]

I hope for everyone who is buying in that they will see some serious drops on the DAW CPU metering cause I cannot recall when any CPU/machine upgrade provided the real performance boosts I was expecting, and I don't update that much leaving a nice long time in between to get at least some value and fun from the investment. It always seemed like about a 25-30pct real world increase maximum even with new faster memory, faster FSB etc. Nothing freezing 2 hungry soft synths would not achieve. (Intel user and always have been here)

I remember a couple generations with more like 30% jump, but you are correct they are few and far between.

Athlon 64 & Core 2 Quad coming from P4, and X58 coming from Core 2 Quad were both major jumps.

But yeah, since then we've seen nothing more than 10% on each generation step up, so if you replace every 5 years (3 generations) or so, then 25% - 30% in real terms isn't going to be far off.

The excitement for me and I suspect a lot of people isn't "game changing performance", rather the potential for it to kickstart the arms race again by eating some market share at Intels regular price points. The big jumps I note above came from the time where AMD and Intel actually competed performance wise and since those days progress has slowed.

People want to see the market shaken up, they want to see it progress again because over the last few generations, Intel to those only looking at CPU benchmarks it appears that they are doing nothing much more than spinning the wheels.

To be fair to them, they've spent much of that time expanding on the chipset feature sets and attempting to get the temps under control after the insanity of the last CPU war, but even so those improvements don't really help those of us who want to add a few more plugs to our projects.

As a side note remember also requirements depend on the plugs your using too. I'm running a 3960K at home (6 core around the same generation as yours) and I'm constantly running out of CPU overhead with less tracks than you're running. I'm heavy on Serum, but also plugs like VPS and some of the Acustica stuff which can eat processors for amusement.

[fluffy_little_something]

Not sure what your last sentence means...

Quick Google search:

"CPU locking is the process of permanently setting a CPU's clock multiplier. AMD CPUs are unlocked in early editions of a model and locked in later editions, but nearly all Intel CPUs are locked and recent models are very resistant to unlocking to prevent overclocking by users."

As it is stated on NewEgg site that processor is unlocked, I take it that it can be overclocked to perform even harder than the intial specs.

I was referring to your sentence, not the technology. It didn't make sense linguistically.

[Numanoid]

$255 for a top of the line Asus motherboard, fit for the new processors

http://www.tomshardware.com/news/asus-a ... 33714.html

[fluffy_little_something]

I find it crazy that some people are already pre-ordering the processors and motherboards now, without there being reliable reviews or anything. And of course there might be some issues typical of new products and platforms. So, I think I will wait till April or May. By then we will also know whether or not the new products run well on W7 and W8.

[DJ Warmonger]

I find it crazy that some people are already pre-ordering the processors and motherboards now, without there being reliable reviews or anything. And of course there might be some issues typical of new products and platforms. So, I think I will wait till April or May. By then we will also know whether or not the new products run well on W7 and W8.

1. There have been enough tests all around to internet to get consistent results.
2. There is nothing comparable in this price range, so there is a litle alternative for me anyway. If it's 280% performance boost or just 200%, it's still the best I can get
3. If something doesn't work, you can ask for refund.

[Numanoid]

I find it crazy that some people are already pre-ordering the processors and motherboards now, without there being reliable reviews or anything.

Do you not trust AMD

[fluffy_little_something]

I find it crazy that some people are already pre-ordering the processors and motherboards now, without there being reliable reviews or anything. And of course there might be some issues typical of new products and platforms. So, I think I will wait till April or May. By then we will also know whether or not the new products run well on W7 and W8.

1. There have been enough tests all around to internet to get consistent results.
2. There is nothing comparable in this price range, so there is a litle alternative for me anyway. If it's 280% performance boost or just 200%, it's still the best I can get
3. If something doesn't work, you can ask for refund.

1. Well, those data back weeks and have been done with engineering samples. The final products are not even on the shelves, yet.
2. Sure. Same here. But why not wait a few more weeks? Won't cost anything, but might save trouble.
3. Not necessarily. When something is flat broken, yes, but if there are compatibility issues, they might not take the chip back, especially since you have already applied the thermal paste thingy.

[DJ Warmonger]

I don't really worry about "compatibility", as I'm getting whole new rig. If the system boots up, it will work.

That being said, I could wait a while to see if someone successfully ran it

But why not wait a few more weeks?

Been waiting since early January to get a new PC, just to see if Ryzen cuts it.

[Ichad.c]

If that diagram is correct...

http://cdn.wccftech.com/wp-content/uplo ... 40x370.jpg

It seems AMD will be particularly good at SSE instructions, which I take it are rather important for music software (the Legend synth for instance).

It's either incorrect, or inconsequential. Can it really be 8-10 times faster (or, maybe in this case, run certain instructions) than comparable CPUs?

Interesting that SSE seems to fare so well as I believe AVX is somewhat hobbled in Ryzen compared to Intel's offerings. Some AVX instructions will take 2 cycles to perform rather than Intel's 1. Any developers care to weigh on in the implications there? This stuff is way above my head.

I'm no expert dev(I'm a hobbyist) but most of those tests with "x" threads and instruction sets "xyz" are pretty useless for general real-time audio tasks. Threading will help the DAW a bit, but is not great for real-time audio. Though the biggest reason why I say those tests are useless is because they use all the fancy instruction sets:

1) Those instruction sets like nice evenly spaced and repetitive data and real-time audio is not particularly SIMD(instruction-set) friendly, anything with feedback is not SIMD friendly in general, maybe only 10% of audio algorithms are SIMD friendly, generally plug-ins that have latency will be SIMD friendly, but some other things are too. Even when a plug-in says "optimized with SSE", it's usually only partially.

2)"The Instruction Set War" - basically because Intel & AMD adds their own 'unique' and therefore incompatible instruction sets, it's costs too much time, money and labor to write code that supports all these innovative yet useless instruction sets. The latest screw-up is FMA3 & FMA4 (fused-multiply-add). SSE2 is kind of the standard for audio stuff because it is so widely supported. The real part where these companies should be competing is throughput(how many cycles to do what), latency, branch prediction and cache efficiency. If they stopped messing around in this war, our PCs in general could be between 4%-10% faster without any hardware change! And high performance programs/plug-ins would be cheaper because it would be faster/easier to program/optimize. Very few audio companies that I know support multiple instruction sets; time = money.

For those technically inclined:

http://www.agner.org/optimize/blog/read.php?i=25

That being said, I'm looking forward to these new AMD chips - if they live up to most of the hype, I will surely consider it for my next PC build in two years time when most of the hype has died down and real-life tests are done.

[Dominus]

Here's what I'm looking at, the $300-ish CPUs. (AMD vs Intel)

AMD Ryzen 7 1700 CPU Performance:

Base Clock: 3.0GHz / Boost Clock: 3.7GHz / TDP: 65W
Price: $ 329[Cinebench R15] Multi-Threading: 1410
46% more performance than Core i7-7700K ($349): 967

Would really like to see how well it stacks up in DAW performance. (Tracks, plugins, etc.)

I was done with overclocking years ago, spending time and money trying to eke out the best performance just to get a few extra percentage points isn't worth it to me. No one is ever going to top the Celeron 300A running at 450MHZ with a stock fan.

[V0RT3X]

If the specs for video encoding and playback beat what intel has to offer, i think I might have to reconsider my recent intel build. Of course how these handle multiple instances of Diva and Bazille also count too

[cron]

If that diagram is correct...

http://cdn.wccftech.com/wp-content/uplo ... 40x370.jpg

It seems AMD will be particularly good at SSE instructions, which I take it are rather important for music software (the Legend synth for instance).

It's either incorrect, or inconsequential. Can it really be 8-10 times faster (or, maybe in this case, run certain instructions) than comparable CPUs?

Interesting that SSE seems to fare so well as I believe AVX is somewhat hobbled in Ryzen compared to Intel's offerings. Some AVX instructions will take 2 cycles to perform rather than Intel's 1. Any developers care to weigh on in the implications there? This stuff is way above my head.

I'm no expert dev(I'm a hobbyist) but most of those tests with "x" threads and instruction sets "xyz" are pretty useless for general real-time audio tasks. Threading will help the DAW a bit, but is not great for real-time audio. Though the biggest reason why I say those tests are useless is because they use all the fancy instruction sets:

1) Those instruction sets like nice evenly spaced and repetitive data and real-time audio is not particularly SIMD(instruction-set) friendly, anything with feedback is not SIMD friendly in general, maybe only 10% of audio algorithms are SIMD friendly, generally plug-ins that have latency will be SIMD friendly, but some other things are too. Even when a plug-in says "optimized with SSE", it's usually only partially.

2)"The Instruction Set War" - basically because Intel & AMD adds their own 'unique' and therefore incompatible instruction sets, it's costs too much time, money and labor to write code that supports all these innovative yet useless instruction sets. The latest screw-up is FMA3 & FMA4 (fused-multiply-add). SSE2 is kind of the standard for audio stuff because it is so widely supported. The real part where these companies should be competing is throughput(how many cycles to do what), latency, branch prediction and cache efficiency. If they stopped messing around in this war, our PCs in general could be between 4%-10% faster without any hardware change! And high performance programs/plug-ins would be cheaper because it would be faster/easier to program/optimize. Very few audio companies that I know support multiple instruction sets; time = money.

For those technically inclined:

http://www.agner.org/optimize/blog/read.php?i=25

That being said, I'm looking forward to these new AMD chips - if they live up to most of the hype, I will surely consider it for my next PC build in two years time when most of the hype has died down and real-life tests are done.

Thanks for the input. Single thread performance is still an important metric for me, so I'm gonna see how Ryzen overclocks compared to Intel before I jump in.

[Debutante]

I hope for everyone who is buying in that they will see some serious drops on the DAW CPU metering cause I cannot recall when any CPU/machine upgrade provided the real performance boosts I was expecting, and I don't update that much leaving a nice long time in between to get at least some value and fun from the investment. It always seemed like about a 25-30pct real world increase maximum even with new faster memory, faster FSB etc. Nothing freezing 2 hungry soft synths would not achieve. (Intel user and always have been here)

I remember a couple generations with more like 30% jump, but you are correct they are few and far between.

Athlon 64 & Core 2 Quad coming from P4, and X58 coming from Core 2 Quad were both major jumps.

But yeah, since then we've seen nothing more than 10% on each generation step up, so if you replace every 5 years (3 generations) or so, then 25% - 30% in real terms isn't going to be far off.

The excitement for me and I suspect a lot of people isn't "game changing performance", rather the potential for it to kickstart the arms race again by eating some market share at Intels regular price points. The big jumps I note above came from the time where AMD and Intel actually competed performance wise and since those days progress has slowed.

People want to see the market shaken up, they want to see it progress again because over the last few generations, Intel to those only looking at CPU benchmarks it appears that they are doing nothing much more than spinning the wheels.

To be fair to them, they've spent much of that time expanding on the chipset feature sets and attempting to get the temps under control after the insanity of the last CPU war, but even so those improvements don't really help those of us who want to add a few more plugs to our projects.

As a side note remember also requirements depend on the plugs your using too. I'm running a 3960K at home (6 core around the same generation as yours) and I'm constantly running out of CPU overhead with less tracks than you're running. I'm heavy on Serum, but also plugs like VPS and some of the Acustica stuff which can eat processors for amusement.

I remember the jump from x2 to x4 to x8 being pretty much on point expectation wise... Pretty much the same hertzage until my 8350 sat at 4ghz. What I'm expecting now is at least 1.5x my 8350.

I might hold off again to get a dual CPU machine. I was planning on one with 2x E5-2640. At that price an Zen based CPU would be far better. The CADing likes that too.

[fedexnman]

This is a win win for the consumer . Intel now has competition they haven't had since AMD went X64 . So sit back and enjoy ...

[glokraw]

The value of stocks is often not objective, but subject to certain human conditions such as greed, gambling, etc.
Luckily, I am not an investor, I have no interest in becoming rich.

Casino gambling is different than investing in thousands of AMD employees
showing up for work each day. And greed is not the same as working to
build wealth to benefit your family, friends, and perhaps victims of tradgedy.

Ironically, the young American snowflakes that are satisfied with sharing
a studio apartment, a morning starbucks, and a cellphone,
are actually hindering economic progress by their complacence,
and Vince the bill collector will come, in one form or another,
for those who by their youth and nation, feel inVincible.

Vince don't want us buyin no stinkin vsts