AMD is making laptop CPU model numbers simultaneously less and more confusing

Sarty

Ars Tribunus Angusticlavius
8,004
How are they expecting this will work? The only people who are going to remember this are giant nerds who would have just looked it up anyway!

P.S. Yes I know WE'RE all giant nerds here, but what about other people?
If we're being perfectly realistic, as long as you're not buying some Imtel Inside (R) bargain basement Walmart trash, regular users haven't needed to know the nitty gritty about their CPU in at least a decade, maybe two.
 
Upvote
53 (54 / -1)

50me12

Ars Tribunus Angusticlavius
7,660
Our branding is easy to understand, just watch this 20-minute presentation.

How are they expecting this will work? The only people who are going to remember this are giant nerds who would have just looked it up anyway!

P.S. Yes I know WE'RE all giant nerds here, but what about other people?

I worked at a place once where some hardware they made had code names and "parent code names" depending on the product. People refereed to these names constantly even after the product was released. And sometimes the code names were CHANGED in development and people would reference both old and new code name interchangeably.

Then you had the actual product names used after release ...

I created a quick translator to help figure out what people were talking about.
 
Upvote
20 (20 / 0)

Deathmonkey

Ars Tribunus Militum
2,950
So which digit will roll over into hexadecimal first?

There's no sense in waiting, we want the AAAAHX chip now!

Not a Ryzen 9 LEET-HX?

In reality the new naming scheme is a mess. I'm also a little shocked that this seemingly confirms that AMD will continue to sell 'new' Zen/Zen+ chips.
 
Upvote
8 (9 / -1)

Little-Zen

Ars Praefectus
3,234
Subscriptor
There was a brief stretch of time where it was fairly straightforward with Intel:

Celeron = low end
Pentium = slightly less low end
i3 = mainstream
i5 = slightly better mainstream
i7 = high end

You could count on the numbers following those to be a little confusing, but for the most part it stuck. It started to get confusing (to me at least) once they started adding more suffixes to them. K was for enthusiasts, sure. But there was T for low power (35W), and S for... middling power (65W). And no suffix for regular power (77W? 84W?).

Then they got rid of processor graphics so they added an "F", and then you had regular processors at 65W but it wasn't really 65W anymore, and high end processors at something but there was a 9900, 9900F, 9900KF, 9900K... and then laptop i7s went from being a little better than laptop i5s to full quad cores, then so did laptop i5s, and i3s were quad core briefly and then not, and now there's processors with an H, a U, an HK... and that's before you even get to the model numbers denoting which ones had Iris graphics and which ones didn't, then they changed up the naming scheme and added G numbers at the end of the laptop processors, but an i5 G7 had different graphics configuration (80EU) than an i7 G7 (96EU), and...

Yeah. I got lost.

Comparatively speaking, the Ryzen lineup seemed much easier to parse, up until they started mixing generations with the 5000 series laptop processors. And now I see they've decided to one up Intel with the further confusing numbers and letters, so... good on them. Maybe they can beat Intel there too?
 
Upvote
62 (62 / 0)
Upvote
3 (4 / -1)
Does this mean that a 8640u released in 2023 and 9640u released in 2024 are the same chip? Higher number so should mean better right? Nope, thanks amd.

That's a good point. It's up to AMD to at least make sure there's some sort of clock bump or process node advantage to make something newer (or higher numbered) perform better, and I'm not confident that will be the case.

I will caveat though that at least we can see what Zen version is in use, whereas the existing scheme doesn't do that for use and you just need to Google everything to see what the core architecture actually is.
 
Upvote
11 (11 / 0)

nehinks

Ars Tribunus Angusticlavius
7,423
Even as an enthusiast...I'm still going to need a cheat sheet/reference table to remember what each number means. Which isn't really superior.

And pretty annoyed that they are even planning on still producing old architectures. Effectively the first number is meaningless and only there to keep increasing year over year. I'd argue Architecture should be first, followed by segment. If they wanted to be accurate, rather than tricking people into thinking they got more than they did.

"Why is my computer so slow? It's a higher number than my last one."
 
Upvote
31 (32 / -1)

ERIFNOMI

Ars Legatus Legionis
18,134
Our branding is easy to understand, just watch this 20-minute presentation.

How are they expecting this will work? The only people who are going to remember this are giant nerds who would have just looked it up anyway!

P.S. Yes I know WE'RE all giant nerds here, but what about other people?
Other people don't care anyway. This is all information for us nerds. A normal person is just going to buy whatever's on sale, or they're going to get what their employer gives them.
 
Upvote
25 (25 / 0)

Ten Wind

Ars Tribunus Militum
1,911
Does this mean that a 8640u released in 2023 and 9640u released in 2024 are the same chip? Higher number so should mean better right? Nope, thanks amd.

It'd indicate minor architectural changes with a new revision probably. They could stick the revision in the name too, I'd be fine with that, but that'd also make the names much longer and more confusing.

They kind of did that with the 1600AF in the past when they moved to the AF revision (Intel calls this stepping) and switched nodes. YD1600BBAEBOX was 14nm while the YD1600BBAFBOX was 12nm.

In the new naming scheme we would have had the 1600 revisions up to ae named 1610, while the revisions af and after would have been 2610. Then there would have also been a 2620, which would have been the Zen 2 2600 since both the 12nm 2600 and 12nm 1600AF came out in the same year.

Personally I find the new scheme way more informative, but it'll take a bit of getting used to probably. I do like being able to know the architecture and rough revision number from the name alone. Seems way more transparent.
 
Upvote
12 (12 / 0)
Does this mean that a 8640u released in 2023 and 9640u released in 2024 are the same chip? Higher number so should mean better right? Nope, thanks amd.

It'd indicate minor architectural changes with a new revision probably. They could stick the revision in the name too, I'd be fine with that, but that'd also make the names much longer and more confusing.

They kind of did that with the 1600AF in the past when they moved to the AF revision (Intel calls this stepping) and switched nodes. YD1600BBAEBOX was 14nm while the YD1600BBAFBOX was 12nm.

In the new naming scheme we would have had the 1600 revisions up to ae named 1610, while the revisions af and after would have been 2610. Then there would have also been a 2620, which would have been the Zen 2 2600 since both the 12nm 2600 and 12nm 1600AF came out in the same year.

Personally I find the new scheme way more informative, but it'll take a bit of getting used to probably. I do like being able to know the architecture and rough revision number from the name alone. Seems way more transparent.
I approve of the "AF" suffix. They need to produce more chips deserving of that.
 
Upvote
4 (4 / 0)

Tinolyn

Ars Scholae Palatinae
1,115
Subscriptor
Even as an enthusiast...I'm still going to need a cheat sheet/reference table to remember what each number means. Which isn't really superior.

And pretty annoyed that they are even planning on still producing old architectures. Effectively the first number is meaningless and only there to keep increasing year over year. I'd argue Architecture should be first, followed by segment. If they wanted to be accurate, rather than tricking people into thinking they got more than they did.

"Why is my computer so slow? It's a higher number than my last one."

I have to wonder if this new naming scheme is caused by the semiconductor/supply issues from COVID? It stands to reason having the "older" architecture continue to be used would, theoretically, help alleviate some of the price pain for getting an even marginal upgrade.

Or they could just be following Intel's playbook. Which is not a good thing.
 
Upvote
6 (6 / 0)

uno2tres

Ars Scholae Palatinae
1,114
Subscriptor
Given that AMD doesn’t have access to enough capacity at the latest nodes to meet demand (or divide it into ryzen=current, athlon=old), and that OEMs are loathe to be a model year behind, this seems like a reasonable, if far from perfect, solution. At least it’s reasonably clear without needing to look up every sku individually.
 
Upvote
6 (7 / -1)

Axl

Ars Scholae Palatinae
644
Does this mean that a 8640u released in 2023 and 9640u released in 2024 are the same chip? Higher number so should mean better right? Nope, thanks amd.

It'd indicate minor architectural changes with a new revision probably. They could stick the revision in the name too, I'd be fine with that, but that'd also make the names much longer and more confusing.

They kind of did that with the 1600AF in the past when they moved to the AF revision (Intel calls this stepping) and switched nodes. YD1600BBAEBOX was 14nm while the YD1600BBAFBOX was 12nm.

In the new naming scheme we would have had the 1600 revisions up to ae named 1610, while the revisions af and after would have been 2610. Then there would have also been a 2620, which would have been the Zen 2 2600 since both the 12nm 2600 and 12nm 1600AF came out in the same year.

Personally I find the new scheme way more informative, but it'll take a bit of getting used to probably. I do like being able to know the architecture and rough revision number from the name alone. Seems way more transparent.
I approve of the "AF" suffix. They need to produce more chips deserving of that.

My first Ryzen 5 1600 sucked (launch model) so it had to be replaced by a Ryzen 5 1600 AF. I always appreciated the humor. And Mendocino is back, in red this time!
 
Upvote
2 (2 / 0)

Hattin Gokbori

Wise, Aged Ars Veteran
105
Laptop/ average Joe PC buyers browsing the market-

Intel: If you are part of the average 90% of the buyers, you'll only need to make decision about i3, 5 and 7. And I bet you already know about our generation numbering. We have long been maintained an easy and consistent system to make choice easy for you.

AMD: You are either have to be a huge nerd/gamer to decipher the meaning of our processors, then you will be worthy to buy.

Result: Intel (with inferior, overpriced processors) dominates consumer market.
 
Upvote
-12 (4 / -16)

watermeloncup

Ars Tribunus Angusticlavius
8,882
Laptop/ average Joe PC buyers browsing the market-

Intel: If you are part of the average 90% of the buyers, you'll only need to make decision about i3, 5 and 7. And I bet you already know about our generation numbering, we are doing the same thing for 20 years to make life easy for you.

AMD: You are either have to be a huge nerd/gamer to decipher our processors.

Intel's numbers are even more arbitrary than this. AMD also has marketing names e.g. Ryzen 3/5/7/9 which are probably all the average user needs to know.
 
Upvote
12 (12 / 0)

Ten Wind

Ars Tribunus Militum
1,911
Even as an enthusiast...I'm still going to need a cheat sheet/reference table to remember what each number means. Which isn't really superior.

And pretty annoyed that they are even planning on still producing old architectures. Effectively the first number is meaningless and only there to keep increasing year over year. I'd argue Architecture should be first, followed by segment. If they wanted to be accurate, rather than tricking people into thinking they got more than they did.

"Why is my computer so slow? It's a higher number than my last one."

I have to wonder if this new naming scheme is caused by the semiconductor/supply issues from COVID? It stands to reason having the "older" architecture continue to be used would, theoretically, help alleviate some of the price pain for getting an even marginal upgrade.

Or they could just be following Intel's playbook. Which is not a good thing.

That is sort of what happened with Mendocino, it's a neat development. It's pretty much the cheapest SoC they can produce on TSMC 6nm. It's a Zen 2 ccx with some of the Zen 3 improvements backported, and the latest uncore blocks like VP9 decoding added. Since it's fabricated on TSMC 6nm, it's cheap to produce and it's not cutting into Zen 4 production either. It is still Zen 2, but it's quite different from the Zen 2 parts in the 3000 series.
 
Upvote
15 (15 / 0)

Hattin Gokbori

Wise, Aged Ars Veteran
105
Laptop/ average Joe PC buyers browsing the market-

Intel: If you are part of the average 90% of the buyers, you'll only need to make decision about i3, 5 and 7. And I bet you already know about our generation numbering, we are doing the same thing for 20 years to make life easy for you.

AMD: You are either have to be a huge nerd/gamer to decipher our processors.

Intel's numbers are even more arbitrary than this. AMD also has marketing names e.g. Ryzen 3/5/7/9 which are probably all the average user needs to know.
AMD copied this 3/5/7 system from Intel which was a good step. But their no. of similar products and number based processor marketing (intel usually only advertise power and the name. ex.- core i3. Users don't even have to see the number like 7120G etc) confuses average buyers. This is one of the reasons Laptop OEMs still generally avoids AMD for general models and Laptops with AMD are hard to find in developing markets. I'm sure most Ars readers have no problem with such thing as most are long-time so called nerds. But 90% of the buyers are like their 60 year old dads. And they don't trust store salesmen and still like to make their own informed decisions.
 
Upvote
-11 (2 / -13)
I got all the way down to the second-to-last paragraph before realizing this is only for laptop chips.

Scrolled to the top, read the headline, yep there it is. Scrolled back down to make this comment:

Yeah, this is really stupid and confusing and consumer-unfriendly. :facepalm: And I'm an AMD fan who's only ever bought Athlons and Ryzens (never bought a single Intel chip).
 
Upvote
6 (7 / -1)

phoenix_rizzen

Ars Praefectus
4,926
Subscriptor
Bah! The second digit should be the architecture.

7630 vs 7620 is now meaningless.
7630 vs 7720 is now backwards.

Which is better: 7340 or 7920?

They're incrementing digits out of order.

7430 vs 7290 is much easier to parse. Both released in same year, former using Zen 4 while latter using Zen 2, low end vs high end.

They need to fire their marketing team.
 
Upvote
16 (16 / 0)

Rainywolf

Ars Tribunus Militum
1,931
So they made a number scheme that will literally break in 8 years when its *gasp* 2030 and the whole number scheme roles over.

example
in 2029 it will be 9640U

in 2030 it will be 0640U

do you honestly think marketing will allow them to ship new products that have a lower number than last years model? So buckle up for another rebranding probably in 2029.
 
Upvote
1 (6 / -5)

watermeloncup

Ars Tribunus Angusticlavius
8,882
Laptop/ average Joe PC buyers browsing the market-

Intel: If you are part of the average 90% of the buyers, you'll only need to make decision about i3, 5 and 7. And I bet you already know about our generation numbering, we are doing the same thing for 20 years to make life easy for you.

AMD: You are either have to be a huge nerd/gamer to decipher our processors.

Intel's numbers are even more arbitrary than this. AMD also has marketing names e.g. Ryzen 3/5/7/9 which are probably all the average user needs to know.
AMD just copied this 3/5/7 system from Intel which was a good step. But their no. of similar products and number based processor marketing (intel usually only advertise power and the name. ex.- core i3. Users don't even have to see the number like 7120G etc) confuses average buyers.

What are you talking about? Intel CPUs are almost always advertised with the model number, e.g. i5-1135G7. If they aren't, I assume that the seller is hiding that it's using an older generation of CPU. With AMD the full name will be like Ryzen 5 7640U (example from the lead image). It's functionally very similar but with the added benefit of being able to figure out more details just from the number.
 
Upvote
8 (8 / 0)

Ten Wind

Ars Tribunus Militum
1,911
So really the important number is the gen (second to last), segment (should indicate core count and whatnot) then features (last), then the power use letter.

The rest is going to be :shrug:

The first number is still going to be important because there's a huge difference between a 2019 Matisse Zen 2 part and a 2023 Mendocino Zen 2 part and you'd want to know which you're getting.
 
Upvote
-1 (2 / -3)

SeanJW

Ars Legatus Legionis
11,977
Subscriptor++
The bigger question to me than the model numbering is why they are using anything other than their latest architecture in newly released chips.

Generally because they don't need to use the latest fab for them, and it's "good enough"

The Athlon 3xxx APUs were Zen (not Zen+, not Zen 2...original Zen)..... so they used the 14nm fabs, and they soak up the lower end up market with "good enough" competitors for the really crappy Celeron and Pentium that match.
 
Upvote
11 (11 / 0)

evan_s

Ars Tribunus Angusticlavius
7,517
Subscriptor
Does this mean that a 8640u released in 2023 and 9640u released in 2024 are the same chip? Higher number so should mean better right? Nope, thanks amd.

That's a good point. It's up to AMD to at least make sure there's some sort of clock bump or process node advantage to make something newer (or higher numbered) perform better, and I'm not confident that will be the case.

I will caveat though that at least we can see what Zen version is in use, whereas the existing scheme doesn't do that for use and you just need to Google everything to see what the core architecture actually is.

They'll increase the max turbo by 50mhz that you may or may not ever actually hit and call it good. They say it's to make things clearer but it really seems to be the opposite. By making the first digit a year and hiding the core type in the second to last digit it really seems designed to let them release minor tweaks of very old chips and have them seem new again.

Also as others have pointed it this doesn't seem to give any way to differentiate between the old 5000 and 6000 chips since they are both based on the same Zen 3/3+ cores for the CPUs but the 6000 gpu is much better and overall it is much more efficient. The 6000 series chips will probably be the higher performance part in that tier so should probably end in 5 but that remains to be seen if they are actually consistent on that at all.
 
Upvote
5 (5 / 0)

mgc8

Ars Praetorian
437
Subscriptor++
Another win for marketing, another loss for humanity... <sigh>

But one thing bugs me about this: in a world where you can "protect" rounded rectangles, how come Intel didn't get mad over the Mendocino name? I mean, sure, it's a "codename", but it's publicised right there in the official slides!

I am a card-carying AMD fan these days (been for a few years), but for me there will always be only One Mendocino -- what a wonderful little processor that was! <3
 
Upvote
0 (2 / -2)