The Ultimate Zen: AMD's Zen 3 Achieves 89% Higher Performance Than First-generation Zen

[ratirt]

Are people actually buying this crap? TPU had the the 2700x as 6% better in gaming at 720p and that number decreases as you increase the resolution. TPU has the 3800x 13% better at 720p gaming than the 2700x. I'm not knocking the improvements but lets use some reality here.

Why everything has to be about gaming only? Nobody in a right state of mind would measure IPC only by gaming performance. The reality here is that the numbers you see here are based on a suite of benchmarks not just gaming.

 

[HD64G]

Zen arch was good in the 1st iteration but its main advantage was the path it was set on with great scalability and improvements for at least 3-4 gens. Compatibility of AM4 platform was a big thing for consumers also. TSMC manufacturing progress helped a lot for sure but AMD's engineering teams did a great job.

 

[AnarchoPrimitiv]

Too bad they’re leaving their fanboys in the slums with their nonstop price hikes.

Nonstop price hikes? The 1800x debuted at $500....the 5800x debuted at $450....and from zen1 to Zen2 the prices went down... what are you talking about, I'm seriously asking

 

[Mouth of Sauron]

Are people actually buying this crap? TPU had the the 2700x as 6% better in gaming at 720p and that number decreases as you increase the resolution. TPU has the 3800x 13% better at 720p gaming than the 2700x. I'm not knocking the improvements but lets use some reality here.

Hmmmm, some travel back in time is required... Back when Zen1 was introduced, Intel had like 90% market share and virtually everything was optimized for their CPUs. Games, software, benchmarks - you name it.

To be honest, optimization for something like Bulldozer or Piledriver was, well, stupid.

Times had changed, and now AMD sells more new CPUs (or one-before-new) than Intel does. Everything is optimized for both. The same set of tests, *if* the mentioned piece of software was still updated and optimized, will result in different numbers if repeated today.

Much the same goes for games, they are mostly NVIDIA-optimized at top level - which may change, too.

Apart from that, "final and deciding parameter" - of course, "the gaming crown" is such a ridiculous measurement - done at Ultra settings which serve for reducing speed and mostly nothing else and at the resolution which is already outdated, and does hundreds of FPS. Really useful.

Anyhow, regarding reality - things change, new firmwares, new drivers, new software optimization, Windows itself...

 

[londiste]

Nonstop price hikes? The 1800x debuted at $500....the 5800x debuted at $450....and from zen1 to Zen2 the prices went down... what are you talking about, I'm seriously asking

Nobody really bought 1800X. 1700 ($329) or 1700X ($399) was where it was at.
Zen+ was the one with good prices - 2700 ($299) and 2700X ($329).
Zen2 went backwards - 3700X ($329) and 3800X ($399).
Zen3 brought another price bump - 5800X ($449).

With clocks under load being somewhat variable and TDP a moving target, fixed-clock IPC tests paint a better picture of architectural progress. Sweclockers for example:

AMD Ryzen 7 5800X och Ryzen 5 5600X - Test - Test: Vid samma klockfrekvens

Ryzen 5000-festen forsätter i testlabbet. Näst på tur står de billigare processorerna Ryzen 7 5800X och Ryzen 5 5600X.

www.sweclockers.com

Yeah, it's pretty awesome, but also badly needed to stay competitive. Anyone can compare the statistics to Intel's advances in same time, number of generations, node progress?

What progress? End of 2018 brought 9900K and with little to no architectural changes 9900K/10700K are a pretty fixed target for most of this timeframe.

 

[dirtyferret]

Why everything has to be about gaming only? Nobody in a right state of mind would measure IPC only by gaming performance

I don't know why ask AMD as they literally post it in their slide.

Hmmmm, some travel back in time is required... Back when Zen1 was introduced, Intel had like 90% market share and virtually everything was optimized for their CPUs. Games, software, benchmarks - you name it.

To be honest, optimization for something like Bulldozer or Piledriver was, well, stupid.

Times had changed, and now AMD sells more new CPUs (or one-before-new) than Intel does. Everything is optimized for both. The same set of tests, *if* the mentioned piece of software was still updated and optimized, will result in different numbers if repeated today.

Much the same goes for games, they are mostly NVIDIA-optimized at top level - which may change, too.

Apart from that, "final and deciding parameter" - of course, "the gaming crown" is such a ridiculous measurement - done at Ultra settings which serve for reducing speed and mostly nothing else and at the resolution which is already outdated, and does hundreds of FPS. Really useful.

Anyhow, regarding reality - things change, new firmwares, new drivers, new software optimization, Windows itself...

Your time machine is broken as Intel's share was never that high, it peaked at just over 80% and looking at modern games like cyberpunk and valhalla those initial ryzen performance haven't changed over time so you may want to change your incorrect theory to something more accurate supported by facts.

yes, because when you increasing resolution, its become more gpu bound and useless as CPU bechmark.

Thank you captain obvious for repeating what I just stated in my original post.

 

[ratirt]

I don't know why ask AMD as they literally post it in their slide.

That was a rhetorical question but oh well.
Amd is probably clinging to the fact that past gens Ryzen CPUs weren't good for gaming but this has improved for sure.

 

[dirtyferret]

Amd is probably clinging to the fact that past gens Ryzen CPUs weren't good for gaming but this has improved for sure.

There is no argument there but the slide makes it look like one steady progression after another

 

[noel_fs]

doesnt sound right to me but whatever

 

[lexluthermiester]

Why everything has to be about gaming only? Nobody in a right state of mind would measure IPC only by gaming performance. The reality here is that the numbers you see here are based on a suite of benchmarks not just gaming.

True. However, gaming performance very often translates directly to non-gaming tasks. Also consider that many review sites do non-gaming testing on CPU's being reviewed, like TPU.

Nobody really bought 1800X

Total nonsense! I sold a ton of them. Not everyone whats to OC. In fact most users what a system to run well in a stock config. So yeah, rubbish statement.

1700 ($329) or 1700X ($399) was where it was at.
Zen+ was the one with good prices - 2700 ($299) and 2700X ($329).
Zen2 went backwards - 3700X ($329) and 3800X ($399).
Zen3 brought another price bump - 5800X ($449).

Not sure where you're quoting those prices, but they're not on par with the prices I've seen and dealt with. But that is in the USA, you region might be different.

 

[ratirt]

True. However, gaming performance very often translates directly to non-gaming tasks. Also consider that many review sites do non-gaming testing on CPU's being reviewed, like TPU.

Is that true? Because how I see it the processors power depends on so many things and if you put a GPU into that equation then the CPU gets screwed up and the processing power of the processor is skewed. Like with the low high resolution of a game. Some games run better on one CPU and don't run well on others. I mean you can see the difference. Games' coding sometimes it runs better on the one CPU that's considered weaker. I think games are not exactly a good indicator of how good a CPU is. They show some stuff but there is too many things that can skew the results or at least, don't show the actual picture. I mean, if you measure CPU performance you need to get other stuff, like GPU at least out of the equation.

 

[lexluthermiester]

Is that true?

Absolutely!

Because how I see it the processors power depends on so many things and if you put a GPU into that equation then the CPU gets screwed up and the processing power of the processor is skewed.

Not at all. One does not slow down or tax the other just for being next to each other in the same system.

Some games run better on one CPU and don't run well on others. I mean you can see the difference.

While that's true, there is a limit to how great that difference will be. If you take an Intel multi-threaded 8core CPU and compare it to an AMD multi-threaded 8core CPU, there will of course be differences based on what tasks you throw at it, but they will be in each others neighborhood of performance. What makes one more appealing to any given user is which one excels at what and how much each cost in relation to each other. Determining value is highly subjective and depends greatly in individual needs and plans. There is a reason why the common advice in the industry is if you need a power-house PC for power-user type tasks, buy a high-end gaming rig and fill it with storage drives because you will then have your power-house system.

What makes Ryzen so great right now is that they have the performance advantage as well as the price advantage. This makes Ryzen an excellent value for a great many ATM. However, Intel still has the absolute performance crown because Core i9 & Xeon CPUs can OC like mad without a lot of fidgeting required, which is why i9 and Xeon CPU's are used by everyone for benchmarking and are still the preferred choice for gamers who can spend the money.

 

[londiste]

Not sure where you're quoting those prices, but they're not on par with the prices I've seen and dealt with. But that is in the USA, you region might be different.

MSRPs. They have been selling for prices close to that for at least first half of the lifetime.

 

[dirtyferret]

This chart shows how little improvement zen + was to zen architecture.

 

[ratirt]

This chart shows how little improvement zen + was to zen architecture.

True but it was not considered new gen Ryzen but an update to 1st gen called Ryzen+.
If you consider 1st gen Ryzen and 3000 series which in fact is a 2nd gen Ryzen, the increase in performance is substantial.

What I would like to know is, what's the difference at 1440p but preferably 4k for the Ryzen+ 2700x vs 5800x (if any noticeable) and how will it look like with future games. 1080p for me is too low of a resolution to be honest. I don't want to upgrade my 2700x if there will be no difference playing at 4k or 1440p. From what I see in the reviews of the 5800x the difference between 2700x and 5800x is like nothing. I seriously doubt that this will change in the near future but who knows.

 

[dirtyferret]

True but it was not considered new gen Ryzen but an update to 1st gen called Ryzen+.
If you consider 1st gen Ryzen and 3000 series which in fact is a 2nd gen Ryzen, the increase in performance is substantial.

I'm not considering anything but what AMD posted on their slide.

 

[Mouth of Sauron]

I don't know why ask AMD as they literally post it in their slide.

Your time machine is broken as Intel's share was never that high, it peaked at just over 80% and looking at modern games like cyberpunk and valhalla those initial ryzen performance haven't changed over time so you may want to change your incorrect theory to something more accurate supported by facts.

Thank you captain obvious for repeating what I just stated in my original post.

You, like, have a problem with everything and everybody? Do you ever read what other people say, hell, do you ever read what *you* say? You're a really unpleasant person which brings nothing to discussion and insults everyone. I don't intend to answer you, you know?

 

[dirtyferret]

You, like, have a problem with everything and everybody? Do you ever read what other people say, hell, do you ever read what *you* say? You're a really unpleasant person which brings nothing to discussion and insults everyone. I don't intend to answer you, you know?