AMD shores up its budget laptop CPUs by renaming more years-old silicon
- Thread starter JournalBot
- Start date
The Ryzen Z2 go is the same type of thing and is probably the same Mendocino silicon. The both do it far too often on both the laptop and desktop side of things. AMD has at least been good on the GPU side of things lately and hasn't rebranded the 6600 as a 9050 yet.
Upvote
15
(15
/
0)
Yep. A 1998 Intel Celeron had the Mendocino codename. I don't think Intel is using it anymore, though.
Upvote
53
(53
/
0)
As much as I agree that it's mildly scummy to just rename the processor, I can't hate this TOO much. People buying in this budget tier generally aren't buying based on performance, and the processors are suitable to being in a basic email / youtube / word / facebook / etc device.
I'd be a lot more incensed if they were doing this with parts sold stand alone, where you could reasonably expect new or under-informed builders to fall into the trap based on the processor name alone.
I'd be a lot more incensed if they were doing this with parts sold stand alone, where you could reasonably expect new or under-informed builders to fall into the trap based on the processor name alone.
Upvote
32
(41
/
-9)
And when software support goes away they'll tell us all how long the design has been supported in the market, rather than how long it's been since it was last sold.
Upvote
60
(60
/
0)
I suppose this can be interpreted as AMD acknowledging that their best offer in certain parts of the market is in fact these mature products. In that light, its helpful enough.
Upvote
20
(21
/
-1)
I don't think it's that - both AMD and Intel rename products in the laptops generally. I think this is more laptop OEMs going "we need to have bigger numbers every year otherwise no-one will buy our new models", AMD and Intel going "but we don't have a new product every year for every segment, especially budget", and so rebadged processors come into existence.
Upvote
84
(85
/
-1)
If I got it right if I want the latest silicon tech I have to go into their AI 300 thing even if I don't want AI?
Upvote
8
(9
/
-1)
Who else feels like we've pretty much reached the end of the road? The pivot to AI / npu is probably another symptom. From now on, they'll have to string together a series of concepts/ideas/tricks that are external to the semiconductor side of things, and the free lunch is over. And then it comes to a stop, and the whole industry will need to reinvent itself?
Upvote
0
(11
/
-11)
How could you not want AI? Everyone wants AI!
I know this because the AI told me so. And the AI wouldn't lie to me, would it?
But in case you think it's just some kind of buzzword, here's a detailed investigation into AI that ChatGPT told me you should watch:
View: https://www.youtube.com/watch?v=-qbylbEek-M
Upvote
16
(19
/
-3)
I remember when I bought my first computer. Saved money for two years. Bought a Pentium 75MHz. A big step up from our 80286. The speed of the listing of the dir command. Impressive. Then windows 95 came out. The machine was too slow a few months later.
These days though? I have an old duo core i7 on the attic. I guess it is getting close to two decades old. It interfaces with my small electronics lab. After switching the hard-drive with an SSD, the machine runs great... on windows 10.
For 95% of the work, old processors are more than enough. Even if you are a tech hobbyist.
Been looking for a good excuse to upgrade my non state of the art AMD 3600. Can't find any. Maybe a local Ai, but that would rationalize a better GPU.
Reselling/renaming old cpus? Makes sense to me. The tech has matured.
Upvote
38
(44
/
-6)
Both things can be true at the same time. It is to give OEMs something new looking to put in the next generation of laptops and it is still the best thing Intel/AMD have for that market since price point is a big factor there. For AMD one factor for Mendocino seems to be that it's only quad core while Zen 3 only ever had 8 core CCXs so never had a way to make a quad core max part. Zen 4 and 5 seem to be more flexible with that in their laptop APUs now but we still haven't seen a straight 4 core Zen 4 or 5 design yet. We've seen a 4xZen 4P + 4x Zen 4C design. I do wonder if we'll see a 4x Zen 4C design eventually as a replacement for these Mendocino quad core laptops. That should keep the chip size small for the CPUs but still offer the newer feature set and probably pretty good efficiency unless try to push the C cores too far. Something similar to Intel's N100 chips with just a cluster of 4E cores.
Upvote
7
(8
/
-1)
Not just disingenuous, it's arguably false advertising, depending on how they are advertising them.
Upvote
-2
(9
/
-11)
I can. They could be purchasing newly-released products with the same or worse performance as older/preowned products. It's still lying.
For basic email / youtube / word / facebook / etc, anything will do these days. That's been the case for years. No one needs to pretend otherwise.
Upvote
15
(16
/
-1)
I really agree with you.
The only place I'd differ is I'd quite like a local npu because I'd love to have a reasonably fast intelligent photo search. "[person] in [city] with trees" and I can't begin to justify a standalone GPU.
Upvote
11
(11
/
0)
Hold up! For real? I had no idea AMD were dipping back into the whole 'Athlon' branding again. On further investigation, apparently so - and the manufacturing process spans from 6nm to 12nm.
I thought I was fairly savvy about all this, as in the eternal 'have a mental list of CPUs and GPUs for my next mid-range PC build' that will never actually get built. But TIL that Athlon is still very much a Thing!
Upvote
5
(5
/
0)
I wrote something a while ago about a new update-and-append only branding scheme for how models were updated over time. Perhaps we need a parallel for rebranding existing product?
I'm going to go against the grain a bit and give Intel/AMD perhaps a bit more of a break than they deserve. I generally think the branding is at least somewhat deliberately confusing, some of the time. But I also think they face a series of hurdles that make even good-faith efforts at appropriate branding extremely difficult.
A big problem in my mind is how most people perceive technology. Older silicon is just fine - we've been at the point for a while that older architectures are perfectly performant for most tasks folks are going to use them for. But in the world of tech, most folks have been trained heavily into the "newer, better" mindset. More specifically, when it comes to computers people have been trained into "old is slow and unreliable" - reasonably so given the practical reality of digital cruft that can build up when a less-savvy person uses a computer over time. If such a user sees a computer being sold with a Ryzen 5600X or a i5-11400, they're probably more likely to think "wow, those chips are five years old - that computer must be garbage" even though those chips are good for just about anything most folks are going to throw at them.
I don't really want to forgive them their shenanigans. But I also think the problems they're attempting to address are genuinely difficult and even what I see as the best efforts have been met with abundant criticism. The "decoder ring" methodology felt like decent-faith effort at updating branding while providing some degree of transparency, and even that was roundly derided (in my memory, anyway). At least, for all its issues I could tell at a glance largely what I was getting, and I could just look up specs for remaining details. It sure didn't feel like it lasted long though.
I'm going to go against the grain a bit and give Intel/AMD perhaps a bit more of a break than they deserve. I generally think the branding is at least somewhat deliberately confusing, some of the time. But I also think they face a series of hurdles that make even good-faith efforts at appropriate branding extremely difficult.
A big problem in my mind is how most people perceive technology. Older silicon is just fine - we've been at the point for a while that older architectures are perfectly performant for most tasks folks are going to use them for. But in the world of tech, most folks have been trained heavily into the "newer, better" mindset. More specifically, when it comes to computers people have been trained into "old is slow and unreliable" - reasonably so given the practical reality of digital cruft that can build up when a less-savvy person uses a computer over time. If such a user sees a computer being sold with a Ryzen 5600X or a i5-11400, they're probably more likely to think "wow, those chips are five years old - that computer must be garbage" even though those chips are good for just about anything most folks are going to throw at them.
In the vein of continuing to be a bit easier on the chip-makers than I likely should be, but it also used to be that processors had fewer knobs and dials that you had to brand. When I was a teen, it was an architecture and a processor frequency - and little overlap between frequency ranges for a given architecture. Now you have at least architecture, base frequency, boost frequency, core count, iGPUs, mixes of core architectures/speeds, and TDP to account for, and that's not taking into account packaging differences for different platforms or any of a number of things I'm sure I'm missing. Granted, some of these things are related (TDP in particular might not be fair to add given its relationship to architecture and speed, but it IS a more conscious point of design decision these days so I'm going to leave it), but it's still a lot of detail to try and differentiate between in a model number - especially given that a part of branding is not just to communicate the technical details but intended market segment. I'd argue the latter shouldn't be important, but decades of branding decisions have made it important, so...
I don't really want to forgive them their shenanigans. But I also think the problems they're attempting to address are genuinely difficult and even what I see as the best efforts have been met with abundant criticism. The "decoder ring" methodology felt like decent-faith effort at updating branding while providing some degree of transparency, and even that was roundly derided (in my memory, anyway). At least, for all its issues I could tell at a glance largely what I was getting, and I could just look up specs for remaining details. It sure didn't feel like it lasted long though.
Last edited:
Upvote
7
(8
/
-1)
But is it deceptive or unhelpful?
I'm of two minds on this. A mature architecture that has had firmware applied and many/most of the rough edges figured out being down graded to a budget laptop a device that previously either wouldn't have gotten any service at all or a "new" chip that was so cheaply made or badly produced it's the off cuts of better chips.
So a company doesn't want to devalue their primary product line and renames the chip. You're still arguably getting a better result if they aren't screwing with anything else, but people look at this and think that this cheap laptop has a "net new" processor in it. So they are being mislead, but arguably they might not be getting this device at all, or at a worse price if the rebadge doesn't happen.
For grandparents just getting online to see some pictures and read some emails keeping it cheap is probably your best bet, will they care if a formerly mid tier processor from years ago now running their new device? I'd wager not.
With that said I wish the rebadges were more clear, but we see in cars even many shops will pull parts in from sometimes even other companies to fill a gap if they need. So if we are willing to accept that with a car is it unreasonable to demand more clarity with a Processor?
I think in this case no it's not because usually the engine is just using a part or 12 not a wholesale drop in of say a Volvo engine in a Saab or whatever. Where as here the entire process is just take a Zen ... 3 5040XIL or whatever and just change the name to a non premium brand name now that it's "old". It's not a mixing of parts to make the thing go, it's just a drop and swap and go.
Overall I suspect the basic premise is a net positive, but the way it's implemented and the fact we live in the worst timeline ensures it probably won't stay that way for long.
Upvote
6
(7
/
-1)
While the rebranding is scummy, the performance of the Zen2 and Zen3 CPUs is perfectly cromulent. I know people using Zen2 and Zen1 CPUs today and they do everything requested of them, even gaming. A lot of people are perfectly happy with 60fps at 1920x1080.
Upvote
8
(13
/
-5)
I did a deep dive into my hand held options and discovered there is alot if old designs out there being marketed with newish names, designed to confuse. Super lame
Upvote
1
(1
/
0)
Yes, though the NPU isn't the only improvement you get from doing so relative to previous AMD product lines. The Ryzen AI 300 Series uses Zen 5 instead of Zen 4, potentially offers more GPU cores (depending on model) and uses RDNA 3.5 instead of RDNA3.
IIRC, Every new AMD laptop CPU released since Ryzen 7000 (Phoenix) in 2023 contains an NPU with a theoretical max performance of 10, 16, or 50 TOPS. I think they did a 55 TOPS variant with one OEM, but I can't remember which it was. The chips they still sell sans NPU are earlier designs, AFAIK.
Upvote
3
(3
/
0)
OEMs actively want yearly rebrands to create the illusion of progress. Refusing to give them what they want isn't in a silicon vendor's best interest.
Upvote
5
(6
/
-1)
I thought someone was just having fun and photoshopping Ryzen over an old Athlon chip in the header image.
I did go and look at what the actual Athlon chips looked like, and they're nothing alike beyond being on a square. Memory is a messy messy thing.
I did go and look at what the actual Athlon chips looked like, and they're nothing alike beyond being on a square. Memory is a messy messy thing.
Upvote
3
(3
/
0)
If we're not there, then we're pretty close. Intel are talking about 14A and even if the numbers aren't indicative of the actual transistor size, we all know that there's a limit to the smallness, no matter how much NA you have with your EUV. We've had multicore CPUs with and without SMT, GPUs and NPUs. Maybe the only significant change to be made is to use something other than silicon which will require a huge and expensive upheaval on the production side.
Upvote
4
(4
/
0)
It's true that the improvement train has slowed dramatically, but how much those improvements matter still depends on what you do with your PC. Some of the video encoding tools I use are CPU-only. The 9950X looks like it would be 40-50% faster than the 5950X in the desktop I built back in 2020. That's enough for me to think about upgrading, though I haven't pulled the trigger yet.
If what you use is Office + an internet browser + a little gaming -- and that's most of what I do outside video encodes -- then the gains are going to be a lot smaller.
Maybe 10-15 years ago, I saw a talk by Bob Colwell where he predicted that annual CPU single-threaded performance gains would fall to 5-6% as lithography improvements slowed. The industry isn't quite that low yet, but it's not far off.
Upvote
8
(8
/
0)
With the combination of warmed-over last gen CPUs and Windows 11 spyware, the low-end laptop space has become even more of a dumpster fire than ever.
Upvote
0
(3
/
-3)
I disabled Copilot and used a Copilot key remap app to restore right Ctrl functionality. At least the Ryzen AI laptop was 40% off, due to lack of sales.
Upvote
6
(6
/
0)
Cars are also products that get small incremental improvements, and to that end every car model gets yearly editions that are simply named after years. Can't we do the same with CPUs at this point?
Upvote
12
(13
/
-1)
Upvote
-2
(0
/
-2)
That's exactly what they're doing, and its been going on for years. This is just the next round in the brand name shell game. The gall of it is, this kind of thing actually works. What's in a name? Everything!
It's really not that simple with vehicles, either. Some year models get major updates and the manufacturers spend the next several years ironing the bugs out of that update till they issue another major update and the cycle starts all over again. The only people that can really keep up are the industry insiders who can keep up where the more competent and experienced design engineers went suggesting which year models at which manufacturers have the fewest bugs.
This isn't all that much different with silicon. You want to know where the better quality products are, follow the employment history of the better computer engineers. Right now the better hardware engineers are at Apple, AMD, and Nvidia. Intel has a serious brain drain going on and Qualcomm is still busy playing catch up to Apple. Keep in mind that there's always a design lead time. It takes several years to design, test, redesign, retest, then begin manufacturing, retest and tweak once more, before releasing a vehicle or processor. Engineering hires this year won't show more than barest influence till their hardware design contributions make it through the next product release cycle which starts waaaaaay (years) before the release announcement.
Last edited:
Upvote
5
(5
/
0)
It's just a parallel processing NPU on the chip - low power compute to take a load off the main CPU for specific tasks. For example, there's software that will use the NPU for game streaming, leaving the entirety of the GPU to handle the game itself, the NPU handing just the streaming, not getting bogged down the the primary CPU. Other examples is handing the background blurring function of online meetings. In short, there are uses beyond 'AI' interfaces.
Separately, considering that 'AI' (i.e., ML and neural networks) are showing up in just about every app imaginable, you'll find it hard to escape. Having specialized silicone to handle these side functions certainly won't hurt.
TLDR: it's just a specialized NPU on the chip, it doesn't do anything that your apps don't ask it to, and it can be used for more than 'AI' functions.
Upvote
6
(6
/
0)
They just sell silicon which had some defects with the defective cores disabled as quad core parts. See also the Zen 3 Ryzen 5300GE, etc etc etc.
Upvote
3
(3
/
0)
I'm not saying we couldn't, but I'm not sure how much it would help.
So under this concept, we'd need different nomenclature for each CPU family. The Ryzen AI 300 Series is different silicon from Ryzen 200 or Ryzen 100. Then you still need model numbers to differentiate between lower-end and higher-end SKUs.
if the Ryzen AI HX 370 becomes the Ryzen AI 370 2024, then Ryzen 200 silicon needs to become the Ryzen 2xx 2025, and these new rebadges would presumably be the Ryzen 1xx 2025. So which is better -- the Ryzen AI 370 2024 (because that's when Strix Point launched), or the Ryzen 200 2025? Which matters more -- the year or the SKU identifier?
We'd still need differentiating part numbers even if years were appended.
Upvote
2
(2
/
0)
Upvote
4
(4
/
0)