***OFFICIAL*** Ryzen 5000 / Zen 3 Launch Thread REVIEWS BEGIN PAGE 39

[GaiaHunter]

What a win against a five years old uArch running on a much inferior node! Congrats! AMD fans should celebrate it until the rest of their lives. What next, should we compare Ryzen 5000 to Sandy Bridge CPUs for good measure? Alright! An amazing win!

It is up to Intel to get a better uarch and node. If AMD took 5 years, Intel is taking longer...

 

[yeshua]

It is up to Intel to get a better uarch and node. If AMD took 5 years, Intel is taking longer...

Intel has already got two new uArchs (ICL/TGL) but they have no means to produce them (their 10nm node absolutely sucks - yields are horrible, power consumption at high frequencies is even more horrible). People have magically forgotten that AMD has been a fabless company for over a decade now while Intel uses its own fabs. And I'm not even sure outsourcing their CPUs to TSMC is even possible - TSMC doesn't even have enough capacity for AMD alone.

Also it's a question of mismanagement and pride. Intel needs an engineer at the helm, not a marketing person. Charlie Demerjian (or someone else I don't remember) has also hinted that Intel's internal culture under Brian Krzanich has turned into an utter hell for its employees. No wonder they continue to struggle instead of swallowing their pride and cooperating with e.g. Samsung, IBM or even TSMC (everyone loves money) to advance fabrication.

 

[HutchinsonJC]

Never understood the super low resolutions testing

Low resolution testing is meant to move bottlenecks away from the GPU. That's really all. I see folks saying they'd rather see results more in line with resolutions they'd play at, and that's fine, too. However, the lower resolution benchmarks are showing you the performance of the CPU in gaming. Which helps in forecasting gaming performance in the years coming on that CPU - as GPU performance increases, you will be able to see how the CPU keeps up (or doesn't) as compared to other CPUs.

Because some of us have moved to say 4k, already, there's usually no point in looking at these lower resolutions strictly as it relates to gaming your rig now, but rather how the CPU holds up in the future when the GPU power is stronger. At 4k, for a lot of people, the CPU isn't the bottle neck, which is why people in this thread are saying even with last gen or the gen before on Intel, you really don't need a CPU upgrade to game on 4k... the CPU isn't holding you back, yet. These lower res tests are showing how much growing room the CPU has as it relates to keeping up with the progressing increases in GPU power.

 

[GaiaHunter]

Intel has already got two new uArchs (ICL/TGL) but they have no means to produce them (their 10nm node absolutely sucks - yields are horrible, power consumption at high frequencies is even more horrible). People have magically forgotten that AMD has been a fabless company for over a decade now while Intel uses its own fabs. And I'm not even sure outsourcing their CPUs to TSMC is even possible - TSMC doesn't even have enough capacity for AMD alone.

People magically forget that Intel is more than 10x bigger than AMD and AMD was pretty much bankrupt for a decade...

It is what it is...

 

[yeshua]

People magically forget that Intel is more than 10x bigger than AMD and AMD was pretty much bankrupt for a decade...

It is what it is...

I haven't and Intel is a prime example of a monopoly which has almost destroyed itself - according to their roadmaps they will have competitive CPUs at the end of 2022 at the earliest. Only now Intel's monopoly is being replaced by another one and everyone is cheering.

 

[GaiaHunter]

Low resolution testing is meant to move bottlenecks away from the GPU. That's really all. I see folks saying they'd rather see results more in line with resolutions they'd play at, and that's fine, too. However, the lower resolution benchmarks are showing you the performance of the CPU in gaming. Which helps in forecasting gaming performance in the years coming on that CPU - as GPU performance increases, you will be able to see how the CPU keeps up (or doesn't) as compared to other CPUs.

Because some of us have moved to say 4k, already, there's usually no point in looking at these lower resolutions strictly as it relates to gaming your rig now, but rather how the CPU holds up in the future when the GPU power is stronger. At 4k, for a lot of people, the CPU isn't the bottle neck, which is why people in this thread are saying even with last gen or the gen before on Intel, you really don't need a CPU upgrade to game on 4k... the CPU isn't holding you back, yet. These lower res tests are showing how much growing room the CPU has as it relates to keeping up with the progressing increases in GPU power.

I understand that.

What I am saying is that CPUs, GPUs and games engines are awfully complex these days.

The difference between CPUs architectures remain similar when a new GPU gen arrives at resolutions people play, even though low resolution testing predicts differently.

Unless I missed an article saying how zen2 is now worse compared to coffee lake with the move from turing to ampere.

 

[yeshua]

Speaking of testing resolutions and all of this.

Absolute most reviewers do a very bad job of testing because:

1) Rich people/tech enthusiasts will pair high end CPUs with high end GPUs while running their games at high resolutions (1440p/4K). This is such a small minority it's just laughable - less than 7% of people. And this group is also very very vocal because most average people neither assemble their PCs, nor read reviews or participate in forums (or comments sections) like this one.
2) A large number of people buy midrange CPUs and GPUs ($150-300) but it's covered only by the 5600X and zero GPUs by AMD/NVIDIA for their current gen products.
3) Another large group buys low-end CPUs with iGPUs or low-end GPUs. We have nothing from AMD for this group for this Ryzen generation and again zero GPUs for this price range (less than $200) from NVIDIA/AMD.

So, while the interests of the first group are satisfied, the second and the third ones barely receive any relevant information.

How many reviews of the 5600x paired with GTX 1660 Super/Ti, RX 5500/5600XT have you seen? I haven't seen a single one so far. Most reviewers (and people here) are absolutely out of touch with the world outside.

 

[GaiaHunter]

I haven't and Intel is a prime example of a monopoly which has almost destroyed itself - according to their roadmaps they will have competitive CPUs at the end of 2022 at the earliest. Only now Intel's monopoly is being replaced by another one and everyone is cheering.

AMD is still way smaller. I'm not happy with the price of the 5800X but the big jump in computational power available in the last 3 years was due to AMD and they are still pushing.

 

[yeshua]

AMD is still way smaller. I'm not happy with the price of the 5800X but the big jump in computational power available in the last 3 years was due to AMD and they are still pushing.

Two generations of Intel CPU uArchs used to give comparative jumps in performance and it took on average around three years. Intel did not raise prices for the luxury - AMD does and AMD fans here on AT are so happy it's cringe worthy.

AMD did make a splash by offering lots more cores but they did that with Ryzen 1000. Now we are on the fourth generation and you have to pay a lot more for the same number of cores - something Intel has never dared to do. For AMD fans it's OK though. Frankly speaking I'm appalled by what I hear. Had to ignore at least seven distinguished people here after their exceptional comments on the matter, like COVID-19 is responsible for the price hike. Yeah, and Trump. The unemployment rate has skyrocketed, people earn less but ... PC components get more expensive (except NVIDIA which has backtracked after the failure of RTX 2000).

 

[GaiaHunter]

1) Rich people/tech enthusiasts will pair high end CPUs with high end GPUs while running their games at high resolutions (1440p/4K). This is such a small minority it's just laughable - less than 7% of people. And this group is also very very vocal because most average people neither assemble their PCs, nor read reviews or participate in forums (or comments sections) like this one.

To be fair if even the 6 cores will extract all the performance of a 3090, any lower performance GPU will be a good pairing choice.

 

[GaiaHunter]

Two generations of Intel CPU uArchs used to give comparative jumps in performance and it took on average around three years. Intel did not raise prices for the luxury - AMD does and AMD fans here on AT are so happy it's cringe worthy.

AMD did make a splash by offering lots more cores but they did that with Ryzen 1000. Now we are on the fourth generation and you have to pay a lot more for the same number of cores - something Intel has never dared to do. For AMD fans it's OK though. Frankly speaking I'm appalled by what I hear.

To be fair Intel old generation CPUs never dropped price so when people compare 5600X prices to current 3600 prices that would never happen in the past, ie, the 6700K was the same price all its life until the 7700K came out and then phased out (bar some microcenter deal that did not happen for the rest of the world), while a 3600X started at $299 and went down to $200.

Also Intel kept cutting features.

 

[yeshua]

To be fair Intel old generation CPUs never dropped price so when people compare 5600X prices to current 3600 prices that would never happen in the past.

Also Intel kept cutting features.

Those features were relevant mostly for tech-enthusiasts/corporations to be honest, so for me it's not such a big deal. And recently Intel has come to their senses as Tiger Lake Pentiums and Celerons have finally received full AVX2 support. Again, competition is absolutely essential and what AMD is doing now and how it's perceived shows how quickly people have forgotten what monopolies lead to. Actually AMD has already started acting like a monopoly: a company in a good faith would have lowered the pricing of their previous generation products and maybe introduced a new lineup at a tad higher prices (inflation, COVID-19 which actually is relevant since shipping has become more expensive but not a whole lot more expensive) but certainly not 50%/33% higher (3600->5600X, 3700X->5800X since no one knows if 5600/5700X will ever be released). Here I've been ridiculed and asked to shut up for this comparison which is weird and scary but what do I know. Maybe there are AMD shareholders here - dunno.

 

[DownTheSky]

How many reviews of the 5600x paired with GTX 1660 Super/Ti, RX 5500/5600XT have you seen? I haven't seen a single one so far. Most reviewers (and people here) are absolutely out of touch with the world outside.

None. Because those people will try to buy best bang for the buck cheap processor and pair it with a decent graphic card. That cpu is not 5600x.

When they test CPUs, reviewers try to pair them with the strongest GPU available to remove any bottlenecks. If you look at the charts, most of them have lower end CPUs too in there. So the average gamer can go on the internet, ask what CPU should pair with x GPU and get an answer.

There are also GPU scaling tests done by youtubers like Hardware Unboxed, but I don't really see the point in them. You can ballpark the that data from normal reviews.

 

[jpiniero]

How many reviews of the 5600x paired with GTX 1660 Super/Ti, RX 5500/5600XT have you seen? I haven't seen a single one so far. Most reviewers (and people here) are absolutely out of touch with the world outside.

Because you would be GPU limited unless you had a really old CPU.

 

[AnandThenMan]

Intel has already got two new uArchs (ICL/TGL) but they have no means to produce them

Then what good are they? Could on paper come up with an architecture that does 300GHZ.

 

[uzzi38]

Two generations of Intel CPU uArchs used to give comparative jumps in performance and it took on average around three years.

What was the last two back to back generations from Intel that simultaneously brought a ~19% and a 10% sustained single and multithreaded clock advantage at the same time whilst also sticking to the same power envelopes (which is an effective 30% improvement over two generations).

Intel did not raise prices for the luxury - AMD does and AMD fans here on AT are so happy it's cringe worthy.

Nehalem had a few quad cores.

i7 860/i7 920 - $284

i7 870/i7 940 - $562

i7 965/975 Extreme - $999

i7 2600K - $317

i7 3770K - $313

i7 4770K - $339

i7 5775C - $366 (understandable given the eDRAM cache)

i7 6700K - $350

i7 7700K - $305 (Zen launched 2 months later)

This is what happened with Intel's prices. Not sure where you got the idea that Intel never changed prices for the luxury, their prices were on an upwards trend until Kaby Lake.

Now we are on the fourth generation and you have to pay a lot more for the same number of cores - something Intel has never dared to do. For AMD fans it's OK though. Frankly speaking I'm appalled by what I hear.

Because whether you like it or not, with the exception of the 5800X they all have a place in the market. They are very competitive in multiple ways. Hence they are looked upon positively. Not a big fan of the price hike myself, but I can wait for prices to fall. I'm in no rush. In the mean time, I can sit on the sidelines and appreciate at least some companies actually able to innovate without their egos getting in the way.

 

[GaiaHunter]

Actually AMD has already started acting like a monopoly: a company in a good faith would have lowered the pricing of their previous generation products and maybe introduced a new lineup at a tad higher prices (inflation, COVID-19 which actually is relevant since shipping has become more expensive but not a whole lot more expensive) but certainly not 50%/33% higher (3600->5600X, 3700X->5800X since no one knows if 5600/5700X will ever be released). Here I've been ridiculed and asked to shut up for this comparison which is weird and scary but what do I know. Maybe there are AMD shareholders here - dunno.

I doubt AMD will produce more zen2, other than for epycs. They just have too many products going on 7nm.

And the launch price is slightly irrelevant if you can't buy any product and the product you can is over the MSRP.

Hopefully thet can keep restocking and after a month or 2 stock stabilizes and then price adjusts.

 

[yeshua]

i7 860 - $305
...
i7 7700K - $305

A price increase of $0 for the span of ~10 years (if we remove all the intervening uArch between them) while IPC rose by more than 50%? According to AMD fans that would have made Ryzen CPUs at the very least ~$150 more expensive. Not convinced, sorry.

I doubt AMD will produce more zen2, other than for epycs. They just have too many products going on 7nm.


And the launch price is slightly irrelevant if you can't buy any product and the product you can is over the MSRP.


Hopefully thet can keep restocking and after a month or 2 stock stabilizes and then price adjusts.

I will patiently wait for a 65W 5700X for at most $350 regardless.

 

[uzzi38]

A price increase of $0 for the span of ~10 years (if we remove all the intervening uArch between them) while IPC rose by more than 50%? According to AMD fans that would have made Ryzen CPUs at the very least ~$150 more expensive. Not convinced, sorry.

This is hilarious.

I'm sorry, but this is simply hilarious.

 

[yeshua]

I don't know what to make of your comment thus I will simply ignore it.

 

[uzzi38]

Sorry I was too busy laughing.

You got the price of the i7 860 wrong, it was $284, and you spat out the weakest argument possible.

"Ah yes but over a 10 year gap the prices stayed the same (they didn't)"

You have no right to be saying anything about people being unreasonable AMD fanboys. Stop being hypocritical.

 

[rainy]

i7 7700K - $305 (Zen launched 2 months later)

To be exact it was 339 dollars for (OEM) and 350 for box version.
https://ark.intel.com/content/www/u...-7700k-processor-8m-cache-up-to-4-50-ghz.html
https://www.cpu-world.com/CPUs/Core_i7/Intel-Core i7 i7-7700K.html

Btw, yeshua, your previous nickname on AT forum was birdie?

 

[yeshua]

Sorry I was too busy laughing.

You got the price of the i7 860 wrong, it was $284, and you spat out the weakest argument possible.

"Ah yes but over a 10 year gap the prices stayed the same (they didn't)"

You have no right to be saying anything about people being unreasonable AMD fanboys. Stop being hypocritical.

Product Specifications

quick reference guide including specifications, features, pricing, compatibility, design documentation, ordering codes, spec codes and more.

ark.intel.com

Recommended Customer Price $305.00

Product Specifications

quick reference guide including specifications, features, pricing, compatibility, design documentation, ordering codes, spec codes and more.

ark.intel.com

Recommended Customer Price $339.00

You're right, a $34 price increase over the span of almost 10 years while AMD adds $50-100 for a single generation.

 

[uzzi38]

Product Specifications

quick reference guide including specifications, features, pricing, compatibility, design documentation, ordering codes, spec codes and more.

ark.intel.com

Recommended Customer Price $305.00

Anandtech launch article states $284. Intel's RCPs can change over the lifespan of the product thus I didn't refer to that.

EDIT: But if you do want to take Intel's site as proof, then the 7700K is listed as $339, so you're still wrong.

Product Specifications

quick reference guide including specifications, features, pricing, compatibility, design documentation, ordering codes, spec codes and more.

ark.intel.com

 

[yeshua]

Anandtech launch article states $284. Intel's RCPs can change over the lifespan of the product thus I didn't refer to that.

339-284 = $55 increase over the span of almost a decade. I'm not even sure if it accounts for the inflation for the same period.

Also:

* Lithography 45 nm vs 14nm.
* Number of transistors 774 million vs 2160 million.
* Several generations of iGPUs in between.
* Several new CPU intruction sets (AVX/AVX2)
* And CAPEX to the tune of dozens of billions of dollars for new fabs while AMD uses exactly the same node from TSMC and given their volumes must pay significantly less for wafers this time around because TSMC has improved their yields.

Everything points at Intel who must have significantly increased their pricing but they didn't.