AMD Zen 2 (Ryzen 3000) - *** NO COMPETITOR HINTING ***

[drunkenmaster]

It doesn't mean they actually achieve that frequency though. With Intel defining TDPs at base frequency, if the manufacturers don't account for that in their design, it's unlikely to ever sustain any meaningful turbo. What Intel once described as "guaranteed turbo" now no longer applies. It's a d**k move, in my opinion.

Sunny Cove will have a much more powerful integrated graphics solution and better FPU capabilities, so it makes sense for clocks to fall.

Also, 10 nm is no 14 nm++++++ lol.

Well you've contradicted yourself a bunch there, the clock speeds are not guaranteed, which means they don't have to hold for heavier FPU loads and it doesn't matter if they are sustainable at that clock speed. What matters is that the make speeds they've listed for base clocks and for max boost clocks are lower on 10nm chips than those before them. This means that even in the lightest integer loads on a single core, the new chips can't hit the same clock speeds the previous 14nm+++++++ chips could. Intel have also stated quite plainly many times that the node won't clock as well as 14nm, at least at first. It was supposed to be 10nm + before they hit 14nm base speeds, and further iterations to hit 14nm+ speeds, and 14nm has been boosted in frequency several times further since then.

Faster graphics and better FPU capabilities have no direct bearing on clock speed potential at all. AVX512 has lower clock speeds, and Icelake has AVX512, but as with current chips with AVX512 capability, clock speeds are only lower when using those instructions and once again, we're not talking about guaranteed boost speeds in all loads, but a max boost speed achievable in any type of load and again, it's lower compared to the previous generation.

it only makes sense that clockspeeds fall because Intel have told everyone 10nm can't obtain the same clock speeds, it makes no sense that clock speeds would drop because they added AVX512 or a bigger igpu.

 

[Jacky60]

8600K, or maeby just no?

My 5960X does 5397 35186 with 3000Mhz memory (4500 CPU 4250 cache)
I hope high memory speeds will give a big boost to the 3800X and know they do to the 9900K but either way I'm almost certain to wait for (edited) Zen 3 from the way things are looking.
I want 50% more performance not 5%

 

[Participant]

My 5960X does 5397 35186 with 3000Mhz memory (4500 CPU 4250 cache)
I hope high memory speeds will give a big boost to the 3800X and know they do to the 9900K but either way I'm almost certain to wait for Ryzen 3 from the way things are looking.
I want 50% more performance not 5%

Do you mean Zen 3?

 

[DylRicho]

Well you've contradicted yourself a bunch there, the clock speeds are not guaranteed [...]

Not true. Turbo Boost 2.0 frequencies have historically been marketed as "guaranteed turbo frequencies," further made relevant when Turbo Boost Max 3.0 was introduced as the 'opportunistic' variant.

It is also the distinction given between Turbo Boost technology and Burst Performance technology.

[...] which means they don't have to hold for heavier FPU loads and it doesn't matter if they are sustainable at that clock speed. What matters is that the make speeds they've listed for base clocks and for max boost clocks are lower on 10nm chips than those before them. This means that even in the lightest integer loads on a single core, the new chips can't hit the same clock speeds the previous 14nm+++++++ chips could.

Assuming a decent cooling system, it's only the AVX workloads on the CPU side of things that will heavily impact those max clocks, and who is going to be doing any heavy AVX workloads on a 4C/8T 15 W mobile processor? That sounds like a disaster waiting to happen.

Intel have also stated quite plainly many times that the node won't clock as well as 14nm, at least at first. It was supposed to be 10nm + before they hit 14nm base speeds, and further iterations to hit 14nm+ speeds, and 14nm has been boosted in frequency several times further since then.

Yes, they have.

Faster graphics and better FPU capabilities have no direct bearing on clock speed potential at all.

That's false. The clock speed is dependent on more than the type of workload. You're heavily restricted by what the cooling system is capable of maintaining at equilibrium. FPU workloads output more heat because the FPUs themselves do so.

Faster graphics, no. But that isn't what I said. I said "more powerful" in the sense of more execution units with roughly equivalent clocks. It's a well-known fact that the TDP is a balancing act between CPU and GPU for chips like this. That was always the problem with AMD's APUs as well.

The only reason that Intel counterparts weren't affected as much was because their GPU solutions were abysmal in comparison.

 

[Jacky60]

Do you mean Zen 3?

Yes

 

[Grim5]

New benchmarks time

This time, done on Userbench

CPU: Ryzen 3600 4.1ghz
RAM: DDR4 4200mhz CL17
MOBO: Asrock x570 Phantom

https://www.userbenchmark.com/UserRun/17880340

Aggregated Ryzen 3600 results so far on Userbench place it 5% under the 8700k in Single Core scores and 0.4% ahead of the 8700k in Multi Core scores

https://cpu.userbenchmark.com/Compare/AMD-Ryzen-5-3600-vs-Intel-Core-i7-8700K/4040vs3937

 

[humbug]

New benchmarks time

This time, done on Userbench

CPU: Ryzen 3600 4.1ghz
RAM: DDR4 4200mhz CL17
MOBO: Asrock x570 Phantom

https://www.userbenchmark.com/UserRun/17880340

Aggregated Ryzen 3600 results so far on Userbench place it 5% under the 8700k in Single Core scores and 0.4% ahead of the 8700k in Multi Core scores

https://cpu.userbenchmark.com/Compare/AMD-Ryzen-5-3600-vs-Intel-Core-i7-8700K/4040vs3937

vs 2600 +22%

https://cpu.userbenchmark.com/Compare/AMD-Ryzen-5-3600-vs-AMD-Ryzen-5-2600/4040vs3955

vs 2600X +15%

https://cpu.userbenchmark.com/Compare/AMD-Ryzen-5-3600-vs-AMD-Ryzen-5-2600X/4040vs3956

 

[Ree]

man it feels like it's taking forever for this to come out, money is just sitting here burning a hole.

 

[DragonQ]

I hate all these user benchmark aggregation websites, results are all over the place with tonnes of people not knowing what they're doing posting scores. I just want controlled benchmarks before release date, damn it! >_>

Either way, the i7-8700K is a donkey now. Good overclocking potential from its base 4.3-4.7 GHz but the Ryzen 5 3600 looks to be about the same stock performance at somewhere around 3.9-4.3 GHz (unknown all core boost, potential XFR) and it should cost nearly half as much, sans any price gouging. Throw in the fact that it's a drop-in replacement for anyone on Ryzen 1/2, it has fewer security issues, and has PCIe 4.0, and you'd have to have a very specific use case to even consider the Intel variant at this point.

The i5-9400 is actually cheaper than the R5 3600 right now but it can't be overclocked and would be held back by its relatively modest clock speeds and lack of SMT. Something like the i5-9500 or i5-9600 would be a better competitor for gaming but they don't seem to actually be available anywhere.

The "mid-range" 6 core market is looking great this generation, even if I feel like the 8 core market isn't as nice as I thought it'd be. Maybe in a year 8 cores will drop in price as the 12 and 16 core variants become more common.

 

[robfosters]

man it feels like it's taking forever for this to come out, money is just sitting here burning a hole.

I know, same here. It’s bloody jarring. I want it, and I want it now!

 

[Mensy]

Per-core performance is still the primary metric to be looking for, but if you're wanting to keep a system for half a decade, you will want at least the 6-core option, if not the 8-core. Personally, I would persuade you into the 8-core territory.

The next-generation consoles will be using 8-core Zen 2 processors at around 3.00 GHz, which means that anybody owning such CPUs in their PC should see a tangible benefit to their overall playability since the frequency deficit will be used by the console manufacturers to ensure that all of the cores are properly taken advantage of.

This then causes a domino effect on the desktop/laptop side of things, and that 8-core purchase will be worthwhile.

So you suggest i get the Ryzen 7 3700x over the 5 3600x. Is the difference gonna be that much better for gaming ? The price difference currently seems to be £50 so i dont mind streching to that but i dont wanna spend anymore than £280 on a cpu. Was looking closer to £230 but i can strech it as im gonna need a new motherboard, ram and i think i might need a new cooler as im not sure my corsair H100i will be compatible with the new motherboards

 

[Drollic]

So you suggest i get the Ryzen 7 3700x over the 5 3600x. Is the difference gonna be that much better for gaming ? The price difference currently seems to be £50 so i dont mind streching to that but i dont wanna spend anymore than £280 on a cpu. Was looking closer to £230 but i can strech it as im gonna need a new motherboard, ram and i think i might need a new cooler as im not sure my corsair H100i will be compatible with the new motherboards

There's an AM4 mounting plate you can buy from corsair I believe for slightly older coolers. They originally offered it for free.

 

[Murphy]

New benchmarks time

This time, done on Userbench

CPU: Ryzen 3600 4.1ghz
RAM: DDR4 4200mhz CL17
MOBO: Asrock x570 Phantom

https://www.userbenchmark.com/UserRun/17880340

Aggregated Ryzen 3600 results so far on Userbench place it 5% under the 8700k in Single Core scores and 0.4% ahead of the 8700k in Multi Core scores

https://cpu.userbenchmark.com/Compare/AMD-Ryzen-5-3600-vs-Intel-Core-i7-8700K/4040vs3937

DDR4 4200mhz, so i assume that means it was run on the new 1/2 divider, unless of course R3 can cope with much higher RAM speeds than the 3733 break point shown in their slides.

 

[Grim5]

Those AMD slides said that drr4 5100mhz was achieved on zen3 in their internal overclocking

 

[Robert896r1]

I have a feeling these chips will require a bit more thinking when trying to overclock. There's a lot of factors that need to be taken into consideration:

- RAM speed and timings
- Auto OC algorithms vs manual OC
- Defining scenarios in which auto is a better option
- base clock overclocking

I think just setting your multiplier and calling it a day might not work out well here. While it's exciting to have more things to play with as an enthusiast, more general users might find it frustrating.

 

[N19h7m4r3]

Those AMD slides said that drr4 5100mhz was achieved on zen3 in their internal overclocking

Zen 3 is at least a year away; they only mentioned it in a slide for 7nm+.



They showed 5100Mhz on Zen 2.

 

[Murphy]

Those AMD slides said that drr4 5100mhz was achieved on zen3 in their internal overclocking

And they showed 3733Mhz was the 1/2 divider break point.

 

[sajtion]

zen 2 feels like meh now. the anticipation and all this waiting has killed it. few months back i would have paid money for it but now i'm thinking intel is going to come back with something better instead.

 

[humbug]

zen 2 feels like meh now. the anticipation and all this waiting has killed it. few months back i would have paid money for it but now i'm thinking intel is going to come back with something better instead.

Like what?

 

[Plec]

zen 2 feels like meh now. the anticipation and all this waiting has killed it. few months back i would have paid money for it but now i'm thinking intel is going to come back with something better instead.

Sounds like INTEL need to hurry or you'll be disenchanted with their release, whatever, too...