AMD Ryzen 9 3000 is a 16-core Socket AM4 Beast

I will believe this only once AMD actually releases a product already, I am starting to think that the only way to purchase a 3000 series will be with the Half Life 3 bundle kit.
 
I will believe this only once AMD actually releases a product already, I am starting to think that the only way to purchase a 3000 series will be with the Half Life 3 bundle kit.

Your logic makes no sense. Just because they're launching the chips a few months later than the last release you are equating them to something that will never release and was never announced?
 
Your logic makes no sense. Just because they're launching the chips a few months later than the last release you are equating them to something that will never release and was never announced?
When you want something it doesn't matter if the delay is a day a week or a decade, it hurts all the same. But seriously though has AMD actually stated anything officially about clock speeds or core counts or power consumption or has it all been leaks?
 
Not much software uses 4 cores, let alone 8. The 16 threads on my 2700X mostly sit idle. I occasionally do some encoding that will use every available thread, but on daily use clock rate/IPC is much more important than 8 bazillion cores.

yes and no, One single task may not scale well with more cores, but having tons of tasks hit your PC all at once can slow it down considerably even if it's "more than enough" for standard operation. Take a look at how when you start up your PC and see Steam, Discord, Adobe cloud, etc all start up and even decent, highly-threaded PCs struggle to keep up. More cores means the potential for more software instances without slowing down your system, even if individual pieces of software don't run any faster.

When you want something it doesn't matter if the delay is a day a week or a decade, it hurts all the same. But seriously though has AMD actually stated anything officially about clock speeds or core counts or power consumption or has it all been leaks?

AMD showed a Ryzen 3000 8-core engineering sample matching the performance a 9900K in Cinebench (which is a best-case scenario benchmark for AMD) using considerably less power (in the 65w TDP range). Lisa Su also heavily hinted that we would see more than 8 cores on Ryzen 3000.
 
yes and no, One single task may not scale well with more cores, but having tons of tasks hit your PC all at once can slow it down considerably even if it's "more than enough" for standard operation. Take a look at how when you start up your PC and see Steam, Discord, Adobe cloud, etc all start up and even decent, highly-threaded PCs struggle to keep up. More cores means the potential for more software instances without slowing down your system, even if individual pieces of software don't run any faster.



AMD showed a Ryzen 3000 8-core engineering sample matching the performance a 9900K in Cinebench (which is a best-case scenario benchmark for AMD) using considerably less power (in the 65w TDP range). Lisa Su also heavily hinted that we would see more than 8 cores on Ryzen 3000.

Eh... I'm pretty sure 90% of the time the lag from having multiple programs load at start is I/O bottlenecks with the hard drive rather than an actual CPU processing bottleneck.
 
and to think just 2 years ago we were having to sit through release of quad core after quad core on the mainstream.
The bigger question is what todo with this massive power with so many cores, are they going to be doing next to nothing all the time.
Or will the software industry to be able to finally get us programs which can make full use of all this multi core power
The main stream programs are actually not even able to wake a single core for more than a few percents, yes there are a few who do make use of multi threading but they are rare ( i do not talk about raytracing and proffesional stuff here)
So my fear is that i end up with a multi core beast which will run my games fine, with whatever cores it uses. But the majority of time its gulping power with all those extra cores doing actually nothing.
We can all act like we do so much with the pc, but fact is most of the time these power houses are actually doing nothing most of the time the machine is on.
I checked my usage on my machine about 14 hours usage daily, i actually turn out to be using my i7 4c/8threads pc only about 2 to 4 hours with only 4 cores doing 70-95%(locked max cpu power to 95%) load most of my games do not even use the other threads at all. And those who do hardly ever show an actual load between 1 to 25% max
So even this cpu does not use the extra cores much at all.
 
Joel Hruska:

There has been a surfeit of what Alan Greenspan might have called "irrational exuberance" surrounding AMD and 7nm technology for both Ryzen and Navi. It appears to be fed by fanboys with no concept of how over-hyping the technology cycle behind a company can lead to fans being angry and even vengeful when AMD "fails" to deliver on promises they never made. Widespread coverage of these rumors can lead to them being treated as facts or near-facts, despite AMD doing absolutely nothing to confirm them.

The basic argument is the same, and goes like this:

1). AMD is about to do something extraordinary.

2). AMD, being run by idiots, will choose to sell their extraordinary new product for roughly half the price as the competition, despite the fact that what AMD needs, more than anything, is stable, long-term profits and strong revenue gain across multiple market shares.

3). Even though the only way to establish #2 is by investing in one's own products and growing revenue, people expect that AMD will starve itself in the name of gaining market share, even though "Lose money on every product and make it up with volume," is not actually a winning move.

4). This practical issue will be solved with chiplets, because chiplets are magic, and 7nm wafers are not more expensive, and design costs have not risen, and AMD is not trying to break into markets like AI and deep learning where Nvidia has an enormous institutional advantage. AMD certainly isn't facing an entrenched competitor like Intel, whose quarterly profits dwarf AMD's by orders of magnitude.

5). The fact that 10nm has slipped so badly is proof that Intel can no longer compete and will slowly be destroyed by ARM and AMD while AMD takes over its market and rules the Earth.

The most annoying thing about all of this is that you could hit "Rewind" and turn the clock back to early 2006. They're basically the same arguments with updated product names (and, of course, the fact that AMD didn't own ATI in early 2006).

I expect AMD to take advantage of 7nm to build a much more competitive Navi than Vega or Polaris have been.I think they will offer a much higher level of performance per dollar and performance per watt. I have not made specific predictions past that because the rumor mill has done a lot of churning about Navi and most of it has been stupid. AMD will not launch an RTX 2070 killer at $250 because AMD isn't going to leave all that money on the table when it desperately needs revenue to fuel its own R&D. AMD wants to play in AI and DL. Nvidia owns those markets so completely, AMD is basically fighting to be a footnote. So clearly, the right solution is to make as much money as possible and plow that back into the business as quickly as possible, in order to build more aggressive AI-focused products on 7nm and steal a march on Nvidia.

Just kidding.

What I meant was, "The smart thing to do is to sell each GPU for one penny above cost, to make the fanboys happy."

(To be absolutely clear, I am not annoyed with you or any commenter specifically. I am tired of chasing down and debunking bad rumors based on dumb data).

I think Navi will be good. I share your concern about how good it will be because AMD has had a hard time securing a straight win against Nvidia in most market segments (the RX 570 is a blowout win against both the GTX 1050 Ti and the GTX 1650, but that's the exception that proves the rule). I think the $330 price tag on an RTX 2070 competitor is probably low, but it's not unbelievably, insanely low. The $250 rumor was.

The rumor mill all-too-often confuses “AMD will make a very competitive / superior play in terms of performance per dollar” with “AMD will gut its own profit margins in the name of offering an unsustainably good deal.”



2070 performance at 2060 price makes perfect sense to me, as does 2060 performance at 1660 price and so on down the stack.
 
If you didn't buy the x3 and unlock to x4 you were doing something wrong. :).
Well lol it also was based on luck, but yeah many people where lucky to get it running with the fourth enabled :D.
But i had a few friends who actually called me a liar untill they where invited at my home and saw i really had 4 cores running.

Anyway i am looking forward to the launch of the AMD newest series
And am glad that they are going to bring us some nice extra's and last but not least are able to really compete with the other products on the market again.
Maybe its going to bring the prices of hardware back to normal prices, so i can buy a fast gpu again.
However AMD is very hard to get in my country most of the products are often not even on the pricelists for many months after the releases. And if they do they have often ridiculous long delivery times often over 2 months
 
Last edited:
AMD showed a Ryzen 3000 8-core engineering sample matching the performance a 9900K in Cinebench (which is a best-case scenario benchmark for AMD) using considerably less power (in the 65w TDP range). Lisa Su also heavily hinted that we would see more than 8 cores on Ryzen 3000.

Not only a best case (2700X score is ~90% of 9900k), but Cinebench only measures non-AVX throughput.
 
I hope that the highest clocked chip ends up being one of the 12 core ones. That's what I'm looking for.
AMD has shown with Ryzen over and over again, they all clock the same (give a few 100MHz for the top bin, if there is such a thing). Love the chip, but I always recommend buy the slowest of the cores you want. Add some decent cooling and you can max the chip. Never over-volt, or only slightly and enjoy your top clocks. Remember kids, 200MHz means F'all!
 
The bigger question is what todo with this massive power with so many cores, are they going to be doing next to nothing all the time.
Or will the software industry to be able to finally get us programs which can make full use of all this multi core power
The main stream programs are actually not even able to wake a single core for more than a few percents, yes there are a few who do make use of multi threading but they are rare ( i do not talk about raytracing and proffesional stuff here)
So my fear is that i end up with a multi core beast which will run my games fine, with whatever cores it uses. But the majority of time its gulping power with all those extra cores doing actually nothing.
We can all act like we do so much with the pc, but fact is most of the time these power houses are actually doing nothing most of the time the machine is on.
I checked my usage on my machine about 14 hours usage daily, i actually turn out to be using my i7 4c/8threads pc only about 2 to 4 hours with only 4 cores doing 70-95%(locked max cpu power to 95%) load most of my games do not even use the other threads at all. And those who do hardly ever show an actual load between 1 to 25% max
So even this cpu does not use the extra cores much at all.
The next wave is an OS or some on-chip that decodes single thread into muti-thread. That is the next big breakthrough and we are about to have it. GPU is making the switch as far as the closer to metal APi's, CPU is next. Who, when, IDK but it is already developed and pursued by many companies. We will all find out together, but it will be huge $$ for whomever wins, and PC tech will finally have a leap forward after 20 years of baby steps.
Welcome to the age of a 1000 cores!
 
The next wave is an OS or some on-chip that decodes single thread into muti-thread. That is the next big breakthrough and we are about to have it. GPU is making the switch as far as the closer to metal APi's, CPU is next. Who, when, IDK but it is already developed and pursued by many companies. We will all find out together, but it will be huge $$ for whomever wins, and PC tech will finally have a leap forward after 20 years of baby steps.
Welcome to the age of a 1000 cores!

Nope.
 
When you want something it doesn't matter if the delay is a day a week or a decade, it hurts all the same. But seriously though has AMD actually stated anything officially about clock speeds or core counts or power consumption or has it all been leaks?
Dude really..
 
I know he is exaggerating for effect but, no, we will not be stuck on core one doing most of the work until the year 2250 or so. Things must change, simple as that.

Sure, in some SciFi Fantasy.

But In the foreseeable future, nothing is going to break up serial dependent code to run on multiple processors.

There seems to be a massive misunderstanding of how multiprocessing works among non-programmers. Every time there is talk of core count increase, people jump in and claim, now everything can be coded for 8 cores, or coded for 16 cores, etc...

That isn't how it works at all.

Code has serial dependent sections and parallel potential sections. Once the work has been done to split off the parallel potential sections for multiprocessing, it is done. It doesn't matter how many processor cores were used when the work was done, those now parallel sections will scale (within reason) to n cores. If you make a loop parallel, you don't spawn 4 threads to process because 4 cores are most prevalent, you spawn n threads, where n a value related to a system call telling you how many cores you have, so that parallel code will keep scaling automatically, the more cores you add.

Find a rendering/encoding program from 10 years ago, and run on a 16 or even 32 cores machine, and almost certainly it will scale to use all those cores.

People might like everything to work like Rendering/Encoding in responds to more cores, but it won't. Rendering and Encoding are known as Embarrassingly Parallel problems.

But hardly anything else that matters to home computer users is. Everything else is a combination parallel/serial code, and then Amdahl's law kicks the crap out of core scaling. Even if your code is 80% parallel, you quickly hit diminishing returns on current home CPUs.

dtp_462724_USER_CONTENT_0_html_m77592488.jpg


80% Parallel is the 20% serial line represented by the triangles. Diminishing returns hit early. You get about 2X speedup with 3 processors, 3X with 6 processors, but 4X takes 16 processors. No amount of processor cores will get you a 5X speedup.

Having more CPU cores at home, won't change Amdahl's Law, and it won't enable developers to somehow make Serial dependent code, into Parallel code.
 
Last edited:
Unless you have things loading from 2 or more NVME drives, I highly doubt any modern 6 core processor is going to struggle with startup tasks.

haha..............
Oh there are ways..... Granted some not useful.
For example, my work laptop (same for 300+ people). NVME coffeelake quad core i7. McAfee Endpoint 10.6 Eats the CPU for Breakfast lunch and dinner (Pegs 100% for a while), while disk sits at less than 50% and thats without any other startup tasks.......... Just McAfee.
Too bad I have no ability to change that.........
For big businesses McAfee is as standard as it gets...........
We get many tickets about 100% CPU usage lol..
Give me 16-32 threads and we'll peg the disk I/O easily
 
But the majority of time its gulping power with all those extra cores doing actually nothing.

They only burn power if you turn off energy management. No reason they have to be burning power at all at idle. Silicon has had this problem “fixed” for quite a while now.
 
typing this from the magazine......

One of the big impediments to Ryzen performance with the current design has been the dual CCX [core complex] layout where L3 cache accesses to a different CCX can be significantly slower. Games, in particular, seem to dislike this arrangement, and AMD is working to address this issue for mainstream models.

page 15 of Maximum PC May 2019
 
Part of the reason AMD CPUs are a bit slower in games is that games don't usually make use of more than four to six CPU cores, so the extra cores on Ryzen CPUs are often idle. That may change in the coming years, but the other part of the equation is latency—the time to access and process data. AMD's dual-CCX design has higher cache and memory latency than Intel's 6-core and 8-core parts, leading to slightly worse overall performance in latency sensitive workloads like games. Incidentally, that same latency problem exists with both Threadripper and Intel's X299 parts: worse latency and worse gaming performance than the i7-8700K.
Of course none of the above snippets from either pcgamer or maximumpc has anything to do with the unknown improvements we'll get from AMD's upcoming Ryzen 3000.
https://www.pcgamer.com/amd-vs-intel-whats-the-difference-between-cpus/
Sorry for a pcgamer link, you may have to turn off your brainy parts to read it. ;)

For what I do I will take the hit and purchase AMD. ymmv
 
Last edited:
Sure, in some SciFi Fantasy.

But In the foreseeable future, nothing is going to break up serial dependent code to run on multiple processors.

There seems to be a massive misunderstanding of how multiprocessing works among non-programmers. Every time there is talk of core count increase, people jump in and claim, now everything can be coded for 8 cores, or coded for 16 cores, etc...

That isn't how it works at all.

Code has serial dependent sections and parallel potential sections. Once the work has been done to split off the parallel potential sections for multiprocessing, it is done. It doesn't matter how many processor cores were used when the work was done, those now parallel sections will scale (within reason) to n cores. If you make a loop parallel, you don't spawn 4 threads to process because 4 cores are most prevalent, you spawn n threads, where n a value related to a system call telling you how many cores you have, so that parallel code will keep scaling automatically, the more cores you add.

Find a rendering/encoding program from 10 years ago, and run on a 16 or even 32 cores machine, and almost certainly it will scale to use all those cores.

People might like everything to work like Rendering/Encoding in responds to more cores, but it won't. Rendering and Encoding are known as Embarrassingly Parallel problems.

But hardly anything else that matters to home computer users is. Everything else is a combination parallel/serial code, and then Amdahl's law kicks the crap out of core scaling. Even if your code is 80% parallel, you quickly hit diminishing returns on current home CPUs.

View attachment 160644

80% Parallel is the 20% serial line represented by the triangles. Diminishing returns hit early. You get about 2X speedup with 3 processors, 3X with 6 processors, but 4X takes 16 processors. No amount of processor cores will get you a 5X speedup.

Having more CPU cores at home, won't change Amdahl's Law, and it won't enable developers to somehow make Serial dependent code, into Parallel code.

Things will change and they must, if computing is going to move forward. The fact that some are hanging onto the past, even recent past, does not negate the need for change. No, I will say it again and exaggerate for effect, like I did before, we will not be on single core full load until 2250, unlike many who thinks it is just going to stay as it has always been.
 
Things will change and they must, if computing is going to move forward. The fact that some are hanging onto the past, even recent past, does not negate the need for change. No, I will say it again and exaggerate for effect, like I did before, we will not be on single core full load until 2250, unlike many who thinks it is just going to stay as it has always been.

Repeating meaningless drivel, doesn't raise it's status beyond that.
 
Repeating meaningless drivel, doesn't raise it's status beyond that.

LOL! Keeping everything the same as it always was? Not going to happen, regardless of what you want to think or believe. You see, I am not an either or type of person, I understand that things need to change, even though it will not be today. AMD also understands the need for change, which is one of the reasons that are at the forefront of those changes. If no changes were needed, why is Intel in full on panic mode once AMD laid the hammer down in March of 2017 and on? Oh well, I prefer not having quad core refresh after quad core refresh after quad core refresh...... :)
 
Last edited:
I am more interested in IPC gains than more cores past 8. You can put 30 slow cores on a die but most applications are single threaded. Anyone who primarily uses their computer for games should as well. 8 cores is enough for games. IPC is where the improvement needs to happen.
 
Back
Top