Why everyone say Zen 5 is bad ?

[R0H1T]

Asking a $100 premium for a 4/6nm node and not being a native 16-core is bad.

Would you pay a $200 premium if it were 16c "native" zen5c cores?

AMD will need to overall the memory/cache hierarchy before they can get to 16-32c on desktops, there's just no way they won't be heavily bottlenecked without it!

 

[SL2]

Second mistake: Being vague with reviewers about what performance they should expect.

That's not what I've seen so far.

Some reviewers are getting lower results than others, jump to 51:10. AMD seems to have a clear idea of what to expect, and I haven't heard about any reviewer saying otherwise.

 

[Knight47]

Because this feels more like a Zen+ launch or like the 7800XT which is a 6800XT in performance.

 

[kapone32]

Where the hell did that publication come from in this argument? I say you give us a break.

You are the one that asserted that people who review this positive are not being objective.

 

[P4-630]

Need a poll, "Who's going to upgrade" and should not include the x3d, because they don't exist yet.

 

[ARF]

Those 9600X and 9700X should be badged as non-X models. X should be 105W TDP. Or maybe we can expect 9650X/9600XT and 9750X/9700XT in near future with higher TDP.

Higher TDP doesn't help. It's actually quite stupid, because you go out of the sweet voltage-power curve, for miserable 1-2% performance improvement.
You should ask for undervolted parts, not for unsustainably overvolted like the entire Ryzen 7000 series.

Ryzen 7 9700X should have been labeled Ryzen 3 9300 XT.

While a true Ryzen 7 9700X should have 12 cores with a 88-91 W TDP.
Ryzen 5 X should be 65 W, while non-X parts should be with 47-51 W TDP.

Why AMD gave a 65 W part an "X" designation is beyond me... Wait, it isn't. It's for markleting, to be able to say how much power they saved compared to the 7700X, even though they didn't compared to the non-X.

Probably the non-X parts will be 47-51 W

 

[Dr. Dro]

You are the one that asserted that people who review this positive are not being objective.

It's hilarious how you're in full-on damage control mode. I actually have a very positive view of this product

Ryzen 7 9700X should have been labeled Ryzen 3 9300 XT.
Probably the non-X parts will be 47-51 W

This is beyond wishful thinking, it's straight up delusional

 

[ARF]

This is beyond wishful thinking, it's straight up delusional

It's not delusional, it's a smart sales strategy, which will motivate the clients to spend

 

[Dr. Dro]

It's not delusional, it's a smart sales strategy, which will motivate the clients to spend

We've repeatedly told you otherwise in this very thread

 

[ARF]

We've repeatedly told you otherwise in this very thread

I know the corporate greed, tell me something new, something that you take from the users' feedbacks, and overall reception. I know you can't.

 

[SL2]

Ryzen 7 9700X should have been labeled Ryzen 3 9300 XT.

How about Phenom LXIX 12345 AI¹⁴⁺⁺, I don't think that name is taken yet.

Also, please note that I didn't put Raptor lakes under the fires, because I'm a good person.

 

[dj-electric]

Hi, I have tested the 9700X for about a week now and am involved in the media circle with AMD.

A few personal notes, extrapolate what you will from it:

1) Documentation about the expectations of performance from AMD's side was relatively poor. We were not given the Ryzen 7 7700 or 7700X as reference points, and I was only asked to give them Cinebench scores. I was not answered with "that's too low" or "That's too high". AMD is aware of my testing's findings and upon receiving those, all I have seen was a nod of sorts. There were not attempts to "correct" any of it. The results did include a number of both performance upgrades, but also regressions compared to the 7700X, even when PPT matched, essentially overclocked. There's no arguing that despite all of this, Zen 5 is really, really power efficient at everything it does.

2) System stability and maturity of existing AM5 boards for Zen 5 seems incomplete. Using 2 ASUS boards, one being the X670E HERO and one being the X670E Extreme, there are obvious oddities. Boot times are longer, PBO functionality with Ryzen Master is essentially broken and causes unbootable situations. This situation may be a red flag regarding any other stability of operation

3) AMD has once again chosen to not give media and testers enough time with Zen 5, often giving them a single day or just two. An AMD CPU that arrives 8-10 days before its release is a miracle. This is bad for results, bad for performance research and overall bad for media people's mental and physical health. This is just generally a bad business practice. For comparison, Intel chips often arrive 2 whole weeks before their release, letting you have proper back and forth with engineers, properly probe, verify and validate results. One of the reasons is a much more robust logistics system. AMD often relies on outsourcing their PR and logistics, which damages review quality. We know that on top of all that, Zen 5 had a serious logistics hiccup that cause the official delays.

4) Pricing for the first wave of Zen 5 CPUs was revealed so close to their review NDA lift, that often reviews are finding themselves rewritten and their wording changed based on the value perspective compared to the assumed one at the time of initial testing. This can work for better or worse, but it's always worse having to redo large chunks of a review right before it has to go up. Often the best way to conclude and summarize test results is to give it some time to simmer. There's always a bit of testing result impressions dust that needs to settle before you can form your most balanced and well thought-through conclusions.

5) AMD's own marketing slides were very, very optimistic. They gave us all rough numbers, plotted them on nice charts and presented Zen 5 as a true next-generation, double digit performance adding core architecture people will rush to upgrade for. Usually, people are used to having marketing claims clash with reality, but with AMD there was more genuine optimism at the back of rockstars like the Ryzen 7000 series itself and the Ryzen 7 7800X3D which is a unique chip in the landscape of CPUs. Reviewers and testers of Ryzen 9000 felt the same kind of excitement before Zen 5's release, expecting a generational leap that was much bigger than the one observed in testing. This to my impression has heavily impacted the presentation and conclusion for Zen 5 based CPUs (so far) for many reviewers. The disappointment itself only added salt and pepper to the overall technical and value proposition situation of current Zen 5 CPUs

 

[SL2]

1) Documentation about the expectations of performance from AMD's side was relatively poor. We were not given the Ryzen 7 7700 or 7700X as reference points, and I was only asked to give them Cinebench scores. I was not answered with "that's too low" or "That's too high".

That's not good to hear. I guess PCWorld had a different experience, see my video above.

I hope you'll get more time for upcoming models!

 

[R0H1T]

Hi, I have tested the 9700X for about a week now and am involved in the media circle with AMD.

A few personal notes, extrapolate what you will from it:

1) Documentation about the expectations of performance from AMD's side was relatively poor. We were not given the Ryzen 7 7700 or 7700X as reference points, and I was only asked to give them Cinebench scores. I was not answered with "that's too low" or "That's too high". AMD is aware of my testing's findings and upon receiving those, all I have seen was a nod of sorts. There were not attempts to "correct" any of it. The results did include a number of both performance upgrades, but also regressions compared to the 7700X, even when PPT matched, essentially overclocked. There's no arguing that despite all of this, Zen 5 is really, really power efficient at everything it does.

2) System stability and maturity of existing AM5 boards for Zen 5 seems incomplete. Using 2 ASUS boards, one being the X670E HERO and one being the X670E Extreme, there are obvious oddities. Boot times are longer, PBO functionality with Ryzen Master is essentially broken and causes unbootable situations. This situation may be a red flag regarding any other stability of operation

3) AMD has once again chosen to not give media and testers enough time with Zen 5, often giving them a single day or just two. An AMD CPU that arrives 8-10 days before its release is a miracle. This is bad for results, bad for performance research and overall bad for media people's mental and physical health. This is just generally a bad business practice. For comparison, Intel chips often arrive 2 whole weeks before their release, letting you have proper back and forth with engineers, properly probe, verify and validate results. One of the reasons is a much more robust logistics system. AMD often relies on outsourcing their PR and logistics, which damages review quality. We know that on top of all that, Zen 5 had a serious logistics hiccup that cause the official delays.

4) Pricing for the first wave of Zen 5 CPUs was revealed so close to their review NDA lift, that often reviews are finding themselves rewritten and their wording changed based on the value perspective compared to the assumed one at the time of initial testing. This can work for better or worse, but it's always worse having to redo large chunks of a review right before it has to go up. Often the best way to conclude and summarize test results is to give it some time to simmer. There's always a bit of testing result impressions dust that needs to settle before you can form your most balanced and well thought-through conclusions.

5) AMD's own marketing slides were very, very optimistic. They gave us all rough numbers, plotted them on nice charts and presented Zen 5 as a true next-generation, double digit performance adding core architecture people will rush to upgrade for. Usually, people are used to having marketing claims clash with reality, but with AMD there was more genuine optimism at the back of rockstars like the Ryzen 7000 series itself and the Ryzen 7 7800X3D which is a unique chip in the landscape of CPUs. Reviewers and testers of Ryzen 9000 felt the same kind of excitement before Zen 5's release, expecting a generational leap that was much bigger than the one observed in testing. This to my impression has heavily impacted the presentation and conclusion for Zen 5 based CPUs (so far) for many reviewers. The disappointment itself only added salt and pepper to the overall technical and value proposition situation of current Zen 5 CPUs

It also doesn't help when you go from 1800/1700x to 2700x to 3700x to 5700/5800x to 7700/7700x and now again 9700x

Just keep a lane will you

I don't agree with the Ryzen Master thing, I don't OC from within Windows & have never considered it stable although ideally your own software should be able to do it! But I have anywhere between 50-500 third party games/applications (i.e. non MS) installed at various times so I know there's a lot of conflict that may lead to system instability. This is why I do it from the BIOS, always!

 

[dj-electric]

That's not good to hear. I guess PCWorld had a different experience, se my video above.

I hope you'll get more time for upcoming models!

Understandable, with Zen 5 so far my impression is that 50 different reviewers had 50 different experiences.

@R0H1T Ryzen Master is a take it or leave it, but usually at the launch of a new desktop generation, you pick the first option in order to get all the monitoring tools existing software in the market haven't quite patched themselves into supporting just yet. This is, after all, a part of the user experience. You should know how AMD's official tools behave from a user's perspective. Naturally, the use of Ryzen Master was stopped once the unbootable system situation occurred. All tuning from there was performed BIOS side exclusively, and monitoring was done with external tools

 

[freeagent]

Zen 6 is where it’s supposed to be at anyways.. ahh well.

 

[Chomiq]

Personally I'm happy that I don't really have a reason to jump on AM5 yet, 5800X3D is sufficient for me right now. As for 9000 series from the reviews of 9600 and 9700 some things come in mind:
- AMD needed more time to prepare stable AGESA/BIOSes for existing AM5 motherboards, initial delay was not enough to get things 100% stable
- some QC issues were not as simple as mislabeled Ryzen 9 9700X, case in point 9600X reviewed by GN was defective (confirmed by AMD)
- while efficiency was improved one can wonder if power limits for the CPUs were gimped because AMD wanted to make them more efficient or simply because AMD couldn't get the CPUs stable at higher frequencies during higher loads.

All of this points me to something I've been considering ever since AM5 launched - completely skipping the AM5 and delaying next CPU update until AM6 generation drops.

 

[Solaris17]

They jump to these negative conclusions based on early bios revisions & even chipset driver maturity.

Which is fine. Don’t buy things based on what’s promised. Buy things based on what’s offered.

 

[P4-630]

Fine wine right?...

 

[phanbuey]

Fine wine right?...

That 12700k is like the Conti Grand Cru at this point. Congrats - you have 9700X gaming performance... 3 years later.

 

[R0H1T]

next CPU update until AM6 generation drops.

That's probably 2026/27 at the earliest. Believe it or not DIY is a dying hobby/niche & no amount of feel good sales numbers will change that.

 

[Solaris17]

That 12700k is like the Conti Grand Cru at this point. Congrats - you have 9700X gaming performance... 3 years later.

Iv seen him take it to dinner wearing his best blazer.

But it’s still a really great chip I don’t blame him at all.

 

[kapone32]

It's hilarious how you're in full-on damage control mode. I actually have a very positive view of this product



This is beyond wishful thinking, it's straight up delusional

Damage control? I have no opinion on this chip. It has no interest to me. All I was illustrating was that these chips are no different than going from 1700 to 2700. The reviews can support whatever position as we have seen.

 

[oZ65]

I disagree. There is very minimal (if any) application performance uplift without PBO, and with PBO, power consumption and heat are through the roof. Gaming performance improvement is non-existent either way.

In my book, this is a performance increase per watt of 13%.

As Alexander J. Yee (Mysticial) one of the authors of y-cruncher wrote on numberworld.org, it really depends on the workload.
Some instructions (AVX-512) saw a 2x increase (I'm not kidding) others 0%.

 

[mb194dc]

They're used to new chip generations providing a significant uplift for "normal" use case application and gaming. But that hasn't been true for the last couple of generations or even longer depending on what you're using for. Even with a 4090, which is a total unrealistic setup for 99%+ of people, gains are eked out in modern 1440p resolutions or higher.

For general computer use, even chips from 10 years ago are fine. Stagnation is the name of the game now.

Obviously there are some use cases where the newer chips are going to save people a lot of time and can do things much faster. It's all niche stuff.