Ryzen 3000 series might have more than 8 cores

[cj09beira]

1 hour ago, PianoPlayer88Key said:

Well, I *would* like more cores as well, but YES please chase IPC now!!  I think even a couple years ago we should have already had CPUs that would do over 200 single-threaded Cinebench R15 at under 4 GHz.  By now we should be doing 300 CB at 4 GHz single-threaded, if not more.

 

My current desktop's i7-4790K has about double the score in single-threaded Cinebench, as I'd think my previous desktop's multi-threaded score on the Athlon 64 X2 4000+ would have been.  I'd at least like my next CPU to make a similar jump over the 4790K, without having to resort to like 20 GHz clockspeeds which are unattainable with current tech, as I understand it.

 

Or for an even better improvement ... my dad's 486-120 he bought in October 1995 was I believe about seventy TIMES or so faster than the 286-10 (bought in January 1989) it replaced, and was about 1/3 the cost.  To start, the CPUs should catch up to where they would be if that pace of improvement had not slowed down.

 

Also I really want to be able to simultaneously encode multiple 4K HEVC lossless video in "placebo" preset at >120 fps each in Handbrake (BEFORE considering GPU assistance), not the ~ 0.0 fps (or taking FOUR(!) HOURS 16 MINUTES or longer to render a 17-SECOND video, and this on SSDs too!) my 4790K in desktop and 6700K in laptop currently d, without having to spend a fortune (as in more than $200-250) on a CPU.

 

 

I wonder why people think consumers don't need many cores?  Is it because we're not expected to run more than one or two things at once?

 

I do a little light multitasking here and there (and have a couple VMs running as well), although I'd really like to be able to step up the multitasking *WITHOUT* having to get an HEDT or server board.
 

  Reveal hidden contents

 

 

 

 

 

 

 

 

 

 

 

Also I'd like to see RAM support greatly increased, and not just doubling - for example my next non-HEDT (4-DIMM) system needs to be able to support 1 TB of RAM at least before considering RDIMMs.

 

 

Anyway ... I'm hoping that performance will be met with my next planned upgrade, which for now is planned either when DDR5 and PCIe 5.0 are out on mainstream, or the 7-year warranty on my Corsair AX760 (bought in January 2015) expires.

those kinds of ipc gains can't really be done now, at least not without moving away from x86-64 or dropping a bunch of old instructions, or adding new caches to the cpus, so dont expect 30+% ipc improvement anytime soon

[cj09beira]

1 hour ago, dizmo said:

I'd be happy with perfect memory support and parity with Intel's IPC.

Don't go overboard on cores AMD. We get it. Fix the other shit first.

thanks to amds design adding more cores does not prevent amd from having enough resources on improving the core, more cores is a simple bonus of moving to 7nm plus acording to oc3d zen 2 will have 10-15% ipc gain which is huge.

 

btw we seen to be getting more leaks of zen 2 recently, is it because of server parts coming first (oposite of what happened with zen 1) or maybe the launch is earlier than i expected (february- march next year) 

[Coaxialgamer]

10 hours ago, Rune said:

An 8 core soldered cannonlake chip is all I want.

Make it so, Intel.

Only 9999......$

Who are we kidding, intel can't yield that until 2020

[Stefan Payne]

4 hours ago, leadeater said:

And no software developer would develop for a hardware market that doesn't exist or is worth catering for. It's rather common for hardware to come before software support for it.

You forget the Consoles (again), they have more than 4 Cores for ages. 

And that is why there already is some Software that is optimized for more than 4 Cores in the Wild. 

 

 

And that mobile phones are "more advanced" than the PC is something that should never have happened. The PC was always at the top of new technology - in the 90s and early 2000s, but in the 2010s it lost this advantage...

[MyName13]

3 minutes ago, Stefan Payne said:

And that mobile phones are "more advanced" than the PC is something that should never have happened. The PC was always at the top of new technology - in the 90s and early 2000s, but in the 2010s it lost this advantage...

Mobile CPUs have better performance than x86 CPUs?

[Stefan Payne]

7 hours ago, Belgarathian said:

AMD going down the FX route again. 

1) Imagine Bulldozer a year or better two earlier. At best, lets say may 2011.

2) Bulldozer wasn't that bad, just that late. Wich totally ruined it.

3) the Cores were OK, the reused K10 Northbridge not so much.

 

So the biggest Problems with Bulldozer was the delay and the same garbage I/O Part that the Phenom also had. 200MHz more Northbridge Clock can give you the same performance (or more) than 200MHz more core...

 

So if everything would run as good as Ryzen now, the Competitor to Bulldozer would have been the i7-800 and 900 series and i5-700. Sandy Bridge came at the end of 2011...

 

7 hours ago, Belgarathian said:

8c/16t is plenty, please chase IPC now. 

Depending on the Workload, but we had a standstill in Core Count for over 10 Years. So no, that's not plenty.

We need more cores to get the software for that. There is no real incentive to make good multithreaded Software...

 

7 hours ago, Belgarathian said:

Increase core count for your HEDT platform and server chips. 

No, because HEDT is not (just) about the cores but also Memory and PCIe Lanes. 

There are enough people who don't want/need the extra benefits of the HEDT Plattform and just need the Cores...

[Stefan Payne]

7 hours ago, leadeater said:

To me it would make more sense to use that new die space for more/better FMA units to close the AVX gap with Intel

Isn't that AVX stuff similar to graphics architectures??

 

More AVX requires a complete redesign wich also will increase power consumption and decrease efficiency...

 

And you have to keep the compatibility with older processors in mind...

 

7 hours ago, leadeater said:

and maybe go with an extra memory channel per die to feed that beast. 3 channel Ryzen, 6 channel Threadripper, 12 channel EPYC sounds go to me and helps all 3 product lines equally (EPYC the most in regards to NVDIMMs becoming more a thing).

Well, that would require a new socket, wich AMD probably doesn't want (yet)...

 

Although I agree that the AM4 socket is a bit shit for the full ryzen chip because each one comes with 32 lanes and there is no way to use those 32 lanes on an AM4 Board.

 

7 hours ago, leadeater said:

Even though Zen doesn't clock as high as Intel most of the areas Intel CPUs are faster aren't actually directly due to the high clocks but more complex CPU cores.

Right now a good portion is due to the higher clockrates. And AMD already could increase that with the adaptation of the better "12nm" manufacturing process. 

 

The next step will be huge and they also had a bit more time to do something on the cores.

 

So I'd expect higher clockrates and maybe better cache performance (lower latency, higher throughput). That is traditionally a weakpoint of AMD Designs since forever...

[asus killer]

7 hours ago, bcredeur97 said:

we're going beyond the # of cores a consumer will reasonably fully utilize 

 

which is cool I guess.. why not? if it doesn't affect anything else. 

thermals, cost electricity and even performance. We already seen how it went wrong with Apples latest macbook pro.

Clock speeds will probably have to be sacrificed for this many cores to work properly. Need for insane motherboards. Etc.. all that and in the end the majority of software doesn't even fully utilize 4 cores.

[leadeater]

1 hour ago, Stefan Payne said:

Isn't that AVX stuff similar to graphics architectures??

 

More AVX requires a complete redesign wich also will increase power consumption and decrease efficiency...

 

And you have to keep the compatibility with older processors in mind...

Not really, a CPU core is made up of INT units and FP units (plus other black magic) which can be sort of be any makeup you like because that side of the CPU is abstracted from view from higher points in the chain like the OS. CPU microcode is used to describe to the OS what sort of features the CPU supports more in data structure and bit size terms i.e. AVX2 (256bit), how that AVX2 (256bit) is actually done on the hardware/silicon side is up to the CPU architects and is where differences in performance actually comes from.

 

They key difference in AVX performance of Zen vs Intel is directly in FP makeup and how each of them achieve processing of 128bit (AVX) and 256bit (AVX2). As explained below

 

Quote

AMD has four 128-bit units for floating point and vector operations. Two of these can do addition and two can do multiplication. Intel has two 256-bit units, both of which can do addition as well as multiplication. This means that floating point code with scalars or vectors of up to 128 bits will execute on the AMD processor at a maximum rate of four instructions per clock (two additions and two multiplications), while the Intel processor can do only two. With 256-bit vectors, AMD and Intel can both do two instructions per clock. Intel beats AMD on 256-bit fused multiply-and-add instructions, where AMD can do one while Intel can do two per clock. Intel is also better than AMD on 256-bit memory writes, where Intel has one 256-bit write port while the AMD processor has one 128-bit write port.

https://www.agner.org/optimize/blog/read.php?i=838

 

AMD could widen the FMA units to 256bit each and the write port to 256bit which would allow them to roughly draw equal to Intel for AVX2, however to do that requires more transistors so more die area and since AVX2 isn't critical to the general masses it's a good area to cut back on to lower cost and complexity. The down side is it makes Zen look weak in benchmarks because it is so significantly slower at AVX2 it taints the whole view of the processors so many would just stick to Intel to avoid a draw back that may not actually be applicable to them at all.

 

1 hour ago, Stefan Payne said:

Well, that would require a new socket, wich AMD probably doesn't want (yet)...

 

Although I agree that the AM4 socket is a bit shit for the full ryzen chip because each one comes with 32 lanes and there is no way to use those 32 lanes on an AM4 Board.

Depends how many reserved pins there are, would absolutely require new motherboards and firmware. Server users wouldn't care about a new socket though so they could add an extra channel to the die and just have it not connected on Ryzen then in 2020 ish release AM5 with it connected, or actually diverge the server and desktop architectures but it's too soon for that in my opinion.

 

1 hour ago, Stefan Payne said:

Right now a good portion is due to the higher clockrates. And AMD already could increase that with the adaptation of the better "12nm" manufacturing process. 

 

The next step will be huge and they also had a bit more time to do something on the cores.

 

So I'd expect higher clockrates and maybe better cache performance (lower latency, higher throughput). That is traditionally a weakpoint of AMD Designs since forever...

A lot of the performance difference actually come from the very small detail areas like how many cycles it takes to populate caches, stuff like that. I'd pull up specific information on that but I'm lazy right now so as a rough exemplar Intel might take 7 cycles to populate L2 cache and AMD might take 9 so just brining that down to 7 without a clock increase will increase performance. Clocks can only do so much, reducing time doing nothing/waiting can be more effective. 

 

Edit:

Some extra infro to look over to see sort of what I'm talking about.

https://www.anandtech.com/show/11170/the-amd-zen-and-ryzen-7-review-a-deep-dive-on-1800x-1700x-and-1700/12

https://www.anandtech.com/show/10591/amd-zen-microarchiture-part-2-extracting-instructionlevel-parallelism/3

[leadeater]

5 hours ago, Arika S said:

the issue is that we've gone from 4 to 6 to 8 and now 8+ in a very short amount of time, so they have been working with 4c/8t which i would argue still is not fully taken advantage of so we're now just pushing 6c and 8c to the wayside which will be even less optimized if we're going to 12 or 16 core what ever AMD decide to make. software is going to be a jumbled mess

Perfect world theory the faster jump would help all iterations of core counts and code should be written to be scalable so if you have 2 or 16 cores it should be able to utilize both as best as possible, theory aside reality is a bit harder.  

[leadeater]

1 hour ago, Stefan Payne said:

You forget the Consoles (again), they have more than 4 Cores for ages. 

And that is why there already is some Software that is optimized for more than 4 Cores in the Wild. 

I don't forget them at all, that's a completely different platform and ecosystem with different requirements and tools. A console is not a PC no matter how similar in hardware they may or may not be. A PC can be almost anything running any software with any plugin or modification with different operating systems with different compilers being utilized, it's such a mixed bag compared to a console.

 

Anyway the console hardware came first then software/games came later and got better over time which was the point anyway.

[Drak3]

7 hours ago, schwellmo92 said:

Look at mobile phones for instance, 8 core processors were common

Except they operate as quad cores. 4+4, because processors for mobile phones can't adjust clocks on the fly.

[asus killer]

this is just an old AMD play. Can't compete on clock or IPC, give them more cores. It was amazing at first, now i fell it is irrelevant for the main PC user. They should try and work on clock instead.

A bit like the old i7's. People will say "see it was worth it to buy" in a couple of years time, until them they will just have cores collect dust on almost all applications. Or PC's will just emulate phones, having all sorts of crap running in the background for no good reason.

[Doobeedoo]

It's 7nm and another core increase, that's really quite something. Excited to see the frequency.

[Lawliet93]

44 minutes ago, asus killer said:

this is just an old AMD play. Can't compete on clock or IPC, give them more cores. It was amazing at first, now i fell it is irrelevant for the main PC user. They should try and work on clock instead.

A bit like the old i7's. People will say "see it was worth it to buy" in a couple of years time, until them they will just have cores collect dust on almost all applications. Or PC's will just emulate phones, having all sorts of crap running in the background for no good reason.

You are wrong... The 7nm process offers clocks close to intels 14nm ++++ and they are actually increasing ipc by about 15%(according to Oc3d.com) so they might have an advantage over intel by about 10%. If they also change infinity fabrics latencies, they could have a winner in their hands. 

[The Benjamins]

43 minutes ago, asus killer said:

this is just an old AMD play. Can't compete on clock or IPC, give them more cores. It was amazing at first, now i fell it is irrelevant for the main PC user. They should try and work on clock instead.

A bit like the old i7's. People will say "see it was worth it to buy" in a couple of years time, until them they will just have cores collect dust on almost all applications. Or PC's will just emulate phones, having all sorts of crap running in the background for no good reason.

Where have you been, clocks are IPC have both hit walls, 7nm it expected to bring better clocks and zen2 is also expected to have a slight IPC improvment which will put it very close to intel. core/thread count is the area that yield the biggest improvements. The same amount of money it takes to gain 5% IPC would of gained 50% core improvements, why would anyone waste money on a poor investment.

[Jito463]

1 hour ago, asus killer said:

this is just an old AMD play. Can't compete on clock or IPC, give them more cores. It was amazing at first, now i fell it is irrelevant for the main PC user. They should try and work on clock instead.

A bit like the old i7's. People will say "see it was worth it to buy" in a couple of years time, until them they will just have cores collect dust on almost all applications. Or PC's will just emulate phones, having all sorts of crap running in the background for no good reason.

The thing is, we'd still be stuck on 4-core systems as mainstream if it weren't for AMD pressuring Intel (in fact, we'd probably be stuck on dual-core or evens single-core systems if not for AMD).  While it may not be immediately beneficial to all people, the hardware has to come first before the software catches up.

 

Also - as has already been pointed out - AMD is working on both IPC and core count for Zen 2.  It's not like they're only able to do one or the other.

[mynameisjuan]

28 minutes ago, Jito463 said:

the hardware has to come first before the software catches up

Not necessarily, you can code a program to use as many cores as it can, if the cores are not there no harm. 12+ cores has been available for a while but devs dont seem to care. 

 

Hell, even with quad cores most programs still use 1-2 cores. Keep blaming the hardware but no one other than video editors have been pushing cores. 

[MyName13]

1 hour ago, Lawliet93 said:

You are wrong... The 7nm process offers clocks close to intels 14nm ++++ and they are actually increasing ipc by about 15%(according to Oc3d.com) so they might have an advantage over intel by about 10%. If they also change infinity fabrics latencies, they could have a winner in their hands. 

But then Intel will have their 10nm

[Brooksie359]

12 minutes ago, mynameisjuan said:

Not necessarily, you can code a program to use as many cores as it can, if the cores are not there no harm. 12+ cores has been available for a while but devs dont seem to care. 

 

Hell, even with quad cores most programs still use 1-2 cores. Keep blaming the hardware but no one other than video editors have been pushing cores. 

Nobody will take the time to code their program to use more cores if the vast majority of people have 2 to 4 cores. If the vast majority of people start having 4 to 6 or even 6 to 8 then it gives devs an actual reason to program to utilize more cores. 

[leadeater]

5 minutes ago, Brooksie359 said:

Nobody will take the time to code their program to use more cores if the vast majority of people have 2 to 4 cores. If the vast majority of people start having 4 to 6 or even 6 to 8 then it gives devs an actual reason to program to utilize more cores. 

When your software is only loading up 4 thread decently on a a typical system of the last 5 years it doesn't look so bad, when you have 16 threads then it actually starts to make the software look poorly written and you'll be branded as having garbage software, right or not. If anything devs will support more threads just to not look bad and for no other reason than that.

[AngryBeaver]

If AMD is making the move to more cores on the consumer platform that isn't a good sign. It means they are hitting a clock speed wall already. So they are just throwing more cores at the problem.

 

This is what we are seeing with intels consumer move from 4c to 6c now to 8c in the next release. They are starting to hit the frequency wall while also being able to only make minimal IPC gains... so in that scenario if you want to improve you add more cores. Now the question will be how long until the ring bus needs to be axed for a better solution on the consumer side.

 

Now back to my amd point. AMD is already struggling for single threat performance. So if they are hitting a frequency wall already at the 4.2ghz range, then they will be hard pressed to ever catch up to intel. They are also behind in IPC which will make that hurdle even taller.  So their solution appears to be back to where it was in the old days... Produce weaker cores, but add more of them to each chip. Gives them an edge in multi-threading, but hurts single-thread and by proxy gaming performance.

[mynameisjuan]

1 minute ago, Brooksie359 said:

Nobody will take the time to code their program to use more cores if the vast majority of people have 2 to 4 cores. If the vast majority of people start having 4 to 6 or even 6 to 8 then it gives devs an actual reason to program to utilize more cores. 

Dude many programs still use single core. Devs that dont give a shit about their program will not recode it, good devs will. I dont get your argument. You want full performance of your program. 

 

Games are pushing hardware, software that is not is just laziness. 

[AngryBeaver]

1 hour ago, Lawliet93 said:

You are wrong... The 7nm process offers clocks close to intels 14nm ++++ and they are actually increasing ipc by about 15%(according to Oc3d.com) so they might have an advantage over intel by about 10%. If they also change infinity fabrics latencies, they could have a winner in their hands. 

So you are taking rumors from a less than accurate source as gospel?

 

Did you know that the whole 7nm stuff is BS. Do you know why? Because there is no normalized way they are being measured. All the manufacturers measure them differently and that is one of the reasons intel wants there to be a set standard for determining the NM of a process.

 

Now as to the 15% IPC improvement, first off that is a pretty unrealistic number. Second they are much farther behind intel than a mere 5% which you are basically implying. They are anywhere from 15-18% behind intels IPC currently.

[C2HWarrior]

12 minutes ago, AngryBeaver said:

If AMD is making the move to more cores on the consumer platform that isn't a good sign. It means they are hitting a clock speed wall already. So they are just throwing more cores at the problem.

 

This is what we are seeing with intels consumer move from 4c to 6c now to 8c in the next release. They are starting to hit the frequency wall while also being able to only make minimal IPC gains... so in that scenario if you want to improve you add more cores. Now the question will be how long until the ring bus needs to be axed for a better solution on the consumer side.

I really don't think its "oh we can't improve IPC or clock speeds so let's add more cores" it's more than likely they just have more space and power consumption is more manageable due to the smaller process that they decided to offer more cores.