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

[Scougar]

AMD have matched Intel's best performance at an unknown clock speed, and its an engineering sample with unfinalized clocks.

With that in mind doesn it matter what the clock speeds are? do you agree it doesn't, they only thing that matters is the performance?

It doesn't matter on the speed per se, that is arbitrary. It could be 1mhz for all it matters and if it matched Intel chip you would be dumb to say it is a bad chip because it runs at 1mhz. What does matter is where this test CPU sits within the product stack range of speeds. Can it go to 5ghz? Was it only 4.2ghz during demo. Are they having 12 and 16 cores, and on what configs with disabled die sections or not. Will 8 cores be 4+4 or a true 8 etc.

My guess is that they very closely matched it and was between 4.2 to 4.4 and this 7nm has legs to go much further.

I am extremely excited about what they will offer this time and really hope they nailed 5ghz tbh to become dangerous to Intel again. Intel are a massive company and will respond in time with products. They are not dead, just like AMD isn't dead with all their time playing the massive underdog.

 

[Perfect_Chaos]

https://www.thestreet.com/investing/amd-cto-discusses-upcoming-products-14833554

Improving the Gaming Performance of AMD's CPUs

While AMD's first and second-gen Ryzen desktop CPUs have often -- with the help of advantages in the number of CPU cores and threads they support -- outperformed comparably-priced Intel CPUs in multi-threaded workloads, Intel often retained an edge for single-threaded workloads. And that, in turn, gave its CPUs superior results in many gaming benchmarks, particularly ones involving gameplay at lower resolutions.

Papermaster said AMD is mindful of the single-threaded performance gap that has remained for Ryzen, and promised his company will deliver "very exciting gains" in this area while maintaining its multi-threaded performance lead. "What you will see with our third-generation Ryzen really is simply outstanding gaming performance," he declared.

 

[humbug]

they proven that in cinebench its quicker. dont be making up the rest. previously in cinebench amd had quicker chips. guess what that meant in games ? slower than a 8400. cinebench is just cinebench. not gaming.

as to humbug . i honestly dont know what to say.without making stuff up like many. if they have matched 9900k at cinebench with whatever clock they have vs a stock 9900k erm..im kinda so so. as said above. previous amd chips did this with intel and look how that turned out. plus you can oc the intel more. so its mainly about price now.

You said your self the 2700X is 15 to 20% slower in games, this engendering sample is 15% faster than the 2700X, it proves its that much faster, you can argue "oh but that doesn't mean it makes any difference is games" i think i will agree it can't really be proven one way or the other, but i think its probably pretty likely, Zen vs Zen+ gained the same performance in Cinebench as it did in games, i think to force "But it doesn't mean it transfers to games" is a straw-man argument.

 

[humbug]

You said your self the 2700X is 15 to 20% slower in games, this engendering sample is 15% faster than the 2700X, it proves its that much faster, you can argue "oh but that doesn't mean it makes any difference is games" i think i will agree it can't really be proven one way or the other, but i think its probably pretty likely, Zen vs Zen+ gained the same performance in Cinebench as it did in games, i think to force "But it doesn't mean it transfers to games" is a straw-man argument.

PS: 8400 vs 2600.

8400 all core Boost 4Ghz
2600 All core Boost 3.4Ghz
and the 4.2Ghz overclock

Clearly clock for clock Zen+ is just as fast as Coffeelake in games.

 

[KungFuSpaghetti]

PS: 8400 vs 2600.

8400 all core Boost 4Ghz
2600 All core Boost 3.4Ghz
and the 4.2Ghz overclock

Clearly clock for clock Zen+ is just as fast as Coffeelake in games.

That's the problem though... clock for clock... But an Intel will be around 1ghz higher over all the cores, which is pretty significant. Hopefully this is rectified in the new chips.

 

[humbug]

Clock speed is irrelevant ^^^^^ the FX-9590 was an all core 5Ghz part, it was still crap, IPC is all that matters.

I'll add to this...

2700X stock (4.025Ghz) vs 9900K Stock (4.7Ghz) clock displayed in the OSD. GTX 1080TI 1080P

The 9900K is clocked 17% higher.

There are no results in this so i have taken the averages from the OSD in random places, no shenanigans you can see the run troughs side by side for a true performance comparison.

9900K is faster (on the right) in all, percentage difference in (Brackets)

Ryse of the Tomb raider: 188 - 213 (+12%)
GTA-V: 132 - 153 (+16%)
AC Odyssey: 103 - 116 (+13%)
Forza 4: 156 - 177 (+13%)
Witcher III: 141 - 162 (+15%)
AC Origins: 105 - 113 (+8%)
FC5: 118 - 127 (+9%)
FA4: 106 - 122 (+16%)
Arma III 92 - 112 (+22%)

So that's 10 to 20% with most around 15% to the 9900K, with 17% higher clocks.

The fact is Zen+ gaming performance is excellent when measured by clock speed alone, 15% slower than the 9900K purely based on clock speed advantage 9900K, this engendering sample has 15% higher performance than the 2700X.

 

[KungFuSpaghetti]

Clock speed is irrelevant ^^^^^ the FX-9590 was an all core 5Ghz part, it was still crap, IPC is all that matters.

I'll add to this...

2700X stock (4.025Ghz) vs 9900K Stock (4.7Ghz) clock displayed in the OSD. GTX 1080TI 1080P

The 9900K is clocked 17% higher.

There are no results in this so i have taken the averages from the OSD in random places, no shenanigans you can see the run troughs side by side for a true performance comparison.

9900K is faster (on the right) in all, percentage difference in (Brackets)

Ryse of the Tomb raider: 188 - 213 (+12%)
GTA-V: 132 - 153 (+16%)
AC Odyssey: 103 - 116 (+13%)
Forza 4: 156 - 177 (+13%)
Witcher III: 141 - 162 (+15%)
AC Origins: 105 - 113 (+8%)
FC5: 118 - 127 (+9%)
FA4: 106 - 122 (+16%)
Arma III 92 - 112 (+22%)

So that's 10 to 20% with most around 15% to the 9900K, with 17% higher clocks.

The fact is Zen+ gaming performance is excellent when measured by clock speed alone, 15% slower than the 9900K purely based on clock speed, this engendering sample has 15% higher performance than the 2700X.

Bit of a sweeping statement lol. Clockspeed is hugely relevant, especially when you can only match your rivals IPC but are 1ghz behind on all core frequency, hence Intel being significantly better on gaming as you've just shown on all those examples.

Measuring gaming performance for Zen+ by clockspeed alone isn't really something to brag about. I'm really happy with my 2700x, but it's not really a great gaming CPU, it's about as quick as a 6/7 series Intel because whilst IPC is great, all core and single core clocks suck. I'm excited for the Ryzen 3000 though, as I don't think AMD will **** the bed on this one as these are by their own admission, gaming CPU's.

 

[Perfect_Chaos]

You also want the minimums not just averages imo.

 

[Murphy]

Clock speed is irrelevant ^^^^^ the FX-9590 was an all core 5Ghz part, it was still crap, IPC is all that matters.

Not really, Instructions per cycle are intrinsically linked to clock-speed (aka:cycles), you could have the most cycles per second of anything on earth but it's not going to help if the instructions (code) runs like doo-doo and you can have the most streamlined instruction pipeline in the world but if one cycles takes you 30sec that means diddly.

 

[KungFuSpaghetti]

You also want the minimums not just averages imo.

Yup, that's why I went for the 2700x over the 2600x, even though both are basically stop gap CPU's until Ryzen 3xxx, because they were similar on average, but 1% lows the 2700x was a lot stronger.

 

[humbug]

Bit of a sweeping statement lol. Clockspeed is hugely relevant, especially when you can only match your rivals IPC but are 1ghz behind on all core frequency, hence Intel being significantly better on gaming as you've just shown on all those examples.

Measuring gaming performance for Zen+ by clockspeed alone isn't really something to brag about. I'm really happy with my 2700x, but it's not really a great gaming CPU, it's about as quick as a 6/7 series Intel because whilst IPC is great, all core and single core clocks suck. I'm excited for the Ryzen 3000 though, as I don't think AMD will **** the bed on this one as these are by their own admission, gaming CPU's.

I'll rephrase that then, Clock speed on it own is irrelevant, again the FX-9590 was a 5Ghz part and yet its gaming performance was junk.

This is the point, clock for clock Zen+ is = to Coffeelake in gaming, it is 15% slower in games because its clocked that much lower.

The Ryzen 3000 engineering sample was measured at 15% higher performance, what clock speed that was is irrelevant, it could have been 1.2Ghz for all anyone should care because be that as it may its still 15% faster than its predecessor.

Don't get so hung up on clock speed, a CPU with 100% performance vs 100% performance of the other CPU is 100% performance vs 100% performance if one is running at 1.2Ghz and the other at 8.5Ghz they still have the same performance.

 

[Journey]

Don't get so hung up on clock speed, a CPU with 100% performance vs 100% performance of the other CPU is 100% performance vs 100% performance if one is running at 1.2Ghz and the other at 8.5Ghz they still have the same performance.

Reminds me of the Athlon speed ratings from back in the 2K's an AMD Athlon 2500+ was 1.83GHz, vs. a Pentium 4 2.6 at 2666MHz, and the Athlon was faster for the most part.

 

[humbug]

Not really, Instructions per cycle are intrinsically linked to clock-speed (aka:cycles), you could have the most cycles per second of anything on earth but it's not going to help if the instructions (code) runs like doo-doo and you can have the most streamlined instruction pipeline in the world but if one cycles takes you 30sec that means diddly.

CPU A cycles 50 instruction per clock and is clocked at 1Ghz while CPU B cycles 25 instructions per clock but is clocked at 1.5Ghz.

Which CPU is the higher performance part?

CPU A?
CPU B?

 

[humbug]

Reminds me of the Athlon speed ratings from back in the 2K's an AMD Athlon 2500+ was 1.83GHz, vs. a Pentium 4 2.6 at 2666MHz, and the Athlon was faster for the most part.

You know why AMD gave the AthlonXP parts naming schemes with numbers higher than the clock speed? for example the 3800+ was 2.7Ghz, they did that to say "this is the equivalent to Intel's clock speed" they did this because they realised Intel's CPU having been clocked higher looked like they were faster to laymen.

Infact the lower clocked AMD CPU's were faster because they had much higher IPC's, they did more work for every clock cycle.

 

[KungFuSpaghetti]

I'll rephrase that then, Clock speed on it own is irrelevant, again the FX-9590 was a 5Ghz part and yet its gaming performance was junk.

This is the point, clock for clock Zen+ is = to Coffeelake in gaming, it is 15% slower in games because its clocked that much lower.

The Ryzen 3000 engineering sample was measured at 15% higher performance, what clock speed that was is irrelevant, it could have been 1.2Ghz for all anyone should care because be that as it may its still 15% faster than its predecessor.

Don't get so hung up on clock speed, a CPU with 100% performance vs 100% performance of the other CPU is 100% performance vs 100% performance if one is running at 1.2Ghz and the other at 8.5Ghz they still have the same performance.

Yeah not really sure what you're getting at to be honest, as I don't disagree that clock speed on it's own is irrelevant... pretty sure I never said it was. My point was simply that AMD may be able to match Intel for IPC clock for clock, but having a chip 20% lower clocking put the advantage firmly back with Intel again.

As for the engineering sample, it showed awesome promise, but that was a threaded test which favours AMD as SMT is simply better than HT. I've zero doubts that had that test instead been gaming benchmarks, then the results would have certainly favoured Intel. Probably not by much, so this is why I'm hyped for the new series, especially the 12 and 16 core high clocking parts (if they clock as per leaked guesses and can do that all core). As for it's clockspeed being irrelevant, well, that's your opinion, I disagree. If it was 4.7 or whatever would show that there's zero IPC improvement and AMD have simply increased clocks, then that's bad... if it was 4.0ghz and the higher end parts are going to be 5.1ghz, then that's absolutely awesome.

 

[nashathedog]

Do we even need PCIE 4, let alone 5? What benefits to your average computer user will they give?

If you need to use other slots, you have more bandwidth for it. Think NVMe and M.2 for SSD’s.. also future proof as PCIE3 has been around for 6 years now.

I for one welcome it.

Ah righto All improvements are welcome!

This is what I'm looking forward too, My current m/b has 2 M.2 slots which is great now SSD prices are dropping but my second slot appears to be throttled which 4.0 should fix, It'd be even better if we also got more lanes so we could get m/b's with 3 or 4 M.2 slots.

 

[humbug]

Yeah not really sure what you're getting at to be honest, as I don't disagree that clock speed on it's own is irrelevant... pretty sure I never said it was. My point was simply that AMD may be able to match Intel for IPC clock for clock, but having a chip 20% lower clocking put the advantage firmly back with Intel again.

As for the engineering sample, it showed awesome promise, but that was a threaded test which favours AMD as SMT is simply better than HT. I've zero doubts that had that test instead been gaming benchmarks, then the results would have certainly favoured Intel. Probably not by much, so this is why I'm hyped for the new series, especially the 12 and 16 core high clocking parts (if they clock as per leaked guesses and can do that all core). As for it's clockspeed being irrelevant, well, that's your opinion, I disagree. If it was 4.7 or whatever would show that there's zero IPC improvement and AMD have simply increased clocks, then that's bad... if it was 4.0ghz and the higher end parts are going to be 5.1ghz, then that's absolutely awesome.

The performance so far is 15% higher.

Its not bad if its getting that from 4.7Ghz, the 4.7Ghz 9900K is 15% faster than the 2700X so the 15% gained puts it on an even parr with the 9900K at the same clock speed, how would that be bad? When chip A gets 100% performance at 4.7Ghz and chip B also gets 100% performance at 4.7Ghz, that makes chip A bad, how?

 

[KungFuSpaghetti]

The performance so far is 15% higher.

Its not bad if its getting that from 4.7Ghz, the 4.7Ghz 9900K is 15% faster than the 2700X so the 15% gained puts it on an even parr with the 9900K at the same clock speed, how would that be bad? When chip A gets 100% performance at 4.7Ghz and chip B also gets 100% performance at 4.7Ghz, that makes chip A bad, how?

Double edged sword really, I'd expect IPC AND clock speed improvements for a 2nd/3rd generation of Zen. If it's 0% IPC improvements that's not great, as the 5.0/5.1ghz SKU's will be barely any quicker for gaming, making all of those extra cores unusable (from a gaming perspective, which again, is what these chips are aimed at), unless all of our existing games get overnight updates to use 12/16 cores properly. With the ES chip being a 65w part, I can't see how it will be 4.7ghz though, I'd have thought it will have to be a lot lower than that, otherwise there's little point of a high clocking 16 core or even 12 core CPU using double the power.

 

[humbug]

Double edged sword really, I'd expect IPC AND clock speed improvements for a 2nd/3rd generation of Zen. If it's 0% IPC improvements that's not great, as the 5.0/5.1ghz SKU's will be barely any quicker for gaming, making all of those extra cores unusable (from a gaming perspective, which again, is what these chips are aimed at), unless all of our existing games get overnight updates to use 12/16 cores properly. With the ES chip being a 65w part, I can't see how it will be 4.7ghz though, I'd have thought it will have to be a lot lower than that, otherwise there's little point of a high clocking 16 core or even 12 core CPU using double the power.

Here is how i look at it.

I'm not expecting any Ryzen 3000 to beat the 9900K by any significant margin in gaming, if at all in fact, what i am expecting to see is by any reasonable measure gaming performance parity, given there is only a 15% difference between the 9900K and the 2700X, its very achievable, by the looks of it they will achieve that, at least.

That's good, AMD don't have to beat Intel in gaming performance metrics, if they can match Intel then Intel can no longer boast about having "the fastest gaming CPU" what's more AMD are likely to be offering "9900K equality performance" for a whole lot less making the Intel part utterly irrelevant.

 

[nashathedog]

nvm, it's scary around here.