Monday, July 29th 2024

AMD Zen 5 Recall Caused by a Typo?

AMD Ryzen 9000 series "Granite Ridge" desktop processors were supposed to start selling on July 31, 2024, but the launch has since been delayed. Since then, social media has been abuzz with theory crafting behind what could be the cause of the delay. AMD's announcement of the delay mentions: "During final checks, we found the initial production units that were shipped to our channel partners did not meet our full quality expectations," causing some to speculate if there are design flaws such as the ones affecting Intel's 13th Gen and 14th Gen Core desktop processors. A picture doing rounds on social media has a more goofy explanation: there is a glaring typo on the product label printed on the integrated heatspreader (IHS) of the processors.

Apparently, some of the first batches of Ryzen 9000 processors see the brand extensions mislabeled. Ryzen 7 9700X is printed as "Ryzen 9 9700X." This error in the brand extension may have been easily "patched" if it was on the retail packaging (the box), where hardware manufacturers tend to fix typos by simply pasting stickers on them. You can't do this with the IHS, which is a key component of the processor's cooling mechanism. Also, since times immemorial, chip labels (information printed on the chip) have served as crucial last resorts for accuracy of information such as the chip's exact model number, steppings or revisions (if any), and production serial numbers, besides the chip's national origin, which determines the applicable import tariffs. A typo here could prove problematic. We're not entirely sure how AMD is fixing these errors with mere 1-2 week delays. It's likely that they're recalling the affected batch and simply replacing inventory in the channel with "good" batches. The recalled chips will simply have their IHS reprinted.
Source: Ian Cutress (Twitter)
Add your own comment

87 Comments on AMD Zen 5 Recall Caused by a Typo?

#26
azrael
ChaitanyaLink please.
There's this for instance. I believe I saw at least another one, but this is what I could quickly find.
Posted on Reply
#27
SL2
piloponthThis issue could easily be solved by blaming MB manufacturers, rounds of BIOS & microcode updates, and finally hiding the problem under the carpet.
Best comment so far
Posted on Reply
#28
LittleBro
ARFThe final assembly is done in Malaysia.
That's actually irrelevant. It is far far away to get it done within one extra week.
SL2First of all, why would they send them back? Can't they just put new lids in Taiwan/Malaysia on other CPU's from manufacturing/testing? I'm pretty sure it doesn't have to be the same individual CPU's, AMD has plenty over there.

Second, why a ship? This has priority, and everyone on AMD knows it. One box is 0.3 kg, so 1000 CPU's is 300 kg, that's nothing.
They said something about doing a product recall. You don't do a recall on something that was not delivered. Given the amount of time, your theory is more feasible, though.

AMD needs to get home all the CPUs from all the OEM's and retailers. One package per each, thousands of retailers/OEMs, hundreds of thousands of packages.
High priority air shipping is extremely expensive compared to freight shipping. It's not unreal that they went this way, though. It's all bout the money, after all.

They need to remove the incorrectly-printed heatspreaders and put on the new ones (newly laser engraved). Now I doubt that this is doable within that time-window even with air transport.
What does not make sense is why did they put different release date for Ryzen 9s. For those, for some reason, they need one extra week on top extra week for R5s and R7s.
You don't ship twice in order to save on shipping.
Posted on Reply
#29
Wirko
LittleBroOkay, the poor image quality may distort letter-spacing. But that 7 in 9700X looks very odd - spaces between 9, 7, 0 are just strange, like someone cut something from there and stamped another thing there.
Yes, I noticed that too. The "7" also looks rotated and can almost be confused with a "1".
Posted on Reply
#30
SL2
azraelThere's this for instance.
What a tool. He realized making reacting-to vids is the lowest of of low in youtube world and doesn't make any money, so he thought "hey this computer thing is kind of popular, let's try that instead"
Posted on Reply
#31
gffermari
Typo!!!!!!!!!
That’s a real problem. Not like the one Intel faces.
Anyway. I hope they don’t delay the release of the beloved 3Ds.
Posted on Reply
#32
azrael
SL2What a tool. He realized making reacting-to vids is the lowest of of low in youtube world and doesn't make any money, so he thought "hey this computer thing is kind of popular, let's try that instead"
He's gotten quite a bit of flak in the comments for his statement.
Posted on Reply
#33
Dr. Dro
LittleBroAs for the copyright, it seems okay to have (c) 2023 there.
Ryzen 9 7900X has some legit photos on the internet and there's (c) 2021 on them, despite 7900X being released on Sep 27th, 2022.
www.techpowerup.com/cpu-specs/ryzen-9-7900x.c2847

Anyway, if this typo proves to be legit, then ... AMD is not far behind the Intel lol.
Copyright date is always at least a year prior. Sometimes more, depends on the development cycle of the processor and how long it took for it to be finalized. My Xeon E5-4669 v3 QS processor has a copyright date of 2012, despite being released in mid 2015 (it's one of the last Haswell-EP CPUs to come out)



According to its batch number, it was also manufactured week 43, 2016, which is also super late for a QS processor. I guess this was a test unit before I bought it from that American dude in 2017. It seemed unused too. It's been mine ever since.
Posted on Reply
#34
AusWolf
CrackongNow it became collector's item.
Exactly what I thought.
ARFI am not going to buy any AMD processor until they redesign this awful IHS, and place the chiplets which emit too much heat under its central area. :kookoo:
I'd still much rather have this IHS than a rectangular one that bends under the cooler's pressure.

As for the CCD placed under the central area, I'd like to see that, too, but I'm not sure how much reality is in such expectation.
Posted on Reply
#35
Daven
LittleBroThat's actually irrelevant. It is far far away to get it done within one extra week.


They said something about doing a product recall. You don't do a recall on something that was not delivered. Given the amount of time, your theory is more feasible, though.

AMD needs to get home all the CPUs from all the OEM's and retailers. One package per each, thousands of retailers/OEMs, hundreds of thousands of packages.
High priority air shipping is extremely expensive compared to freight shipping. It's not unreal that they went this way, though. It's all bout the money, after all.

They need to remove the incorrectly-printed heatspreaders and put on the new ones (newly laser engraved). Now I doubt that this is doable within that time-window even with air transport.
What does not make sense is why did they put different release date for Ryzen 9s. For those, for some reason, they need one extra week on top extra week for R5s and R7s.
You don't ship twice in order to save on shipping.
Guys, you are over thinking this. AMD is probably recalling the typo parts while simultaneously sending out a new patch of parts without the typo. The delay is just the time it takes to make more.

Its not like the first patch is the only patch they can ever make. Hell, they could even throw the returned first patch with the typo into the trash if they wanted to. Lol!
Posted on Reply
#36
JWNoctis
AusWolfI'd still much rather have this IHS than a rectangular one that bends under the cooler's pressure.

As for the CCD placed under the central area, I'd like to see that, too, but I'm not sure how much reality is in such expectation.
Probably (and hopefully) closer than one might expect, if they were going to tackle the power consumption and the bandwidth limitation of the current IF infrastructure. The next generation may well have some sort of tile layout and at least bring them closer to centre.
Posted on Reply
#37
Melvis
randomTPUreaderThis would also explain why the delay is only 1-2 weeks.

If there were a hardware defect AMD wouldn't be able to spin up a corrected chip to manufacturing, package and ship it in that short amount of time.
Yeah I totally agree! anybody thats ever worked in manufacturing would know this and it would make perfect sense with exactly what you said, 1-2weeks turn around is to short of time to fix anything else. Id place money on this is exactly what the recall was for.
Posted on Reply
#38
SL2
LittleBroThey said something about doing a product recall.
I realized that I was just thinking about review samples and not the whole operation, and edited my post. Never saw your reply until now. I'm sorry.
Posted on Reply
#39
Pumper
OnasiIf it really turns out to be just a typo issue that would be hilarious. Especially in contrast to the current Intel debacle.
But it would end up worse for AMD in the end, as the early reviews show lower performance than advertised. Fixing a typo will not fix the performance.
Posted on Reply
#40
SL2
DavenGuys, you are over thinking this. AMD is probably recalling the typo parts while simultaneously sending out a new patch of parts without the typo. The delay is just the time it takes to make more.

Its not like the first patch is the only patch they can ever make. Hell, they could even throw the returned first patch with the typo into the trash if they wanted to. Lol!
I'm pretty sure the returned ones will be sold sooner or later, but there's no hurry doing that.

Also, we don't know if this affects retail only, making the amount much smaller. I mean, I'm not sure HP or Lenovo is prepared to recall all prebuilts just for this..
LittleBroWhat does not make sense is why did they put different release date for Ryzen 9s. For those, for some reason, they need one extra week on top extra week for R5s and R7s.
You don't ship twice in order to save on shipping.
I'm guessing they might give reviewers less headroom for Ryzen 5/7 this time, and to compensate they'll let the Ryzen 9 reviews come later.

On the other hand, they could just let Ryzen 9 get reviewed first like planned..
Posted on Reply
#41
Calmmo
WirkoAMD must be MAD now.
DAM , AMD must be real MAD right now

sorry couldnt resist
Posted on Reply
#42
ARF
AusWolfI'd still much rather have this IHS than a rectangular one that bends under the cooler's pressure.
You mean this?

Posted on Reply
#43
Daven
CalmmoDAM , AMD must be real MAD right now

sorry couldnt resist
So DAM MAD that AMD is taking MDA to cool down.
Posted on Reply
#44
Hecate91
ARFYou mean this?

What bending is there? It looks like excess thermal paste to me, which isn't a problem with the IHS.
Posted on Reply
#45
Chrispy_
randomTPUreaderThis would also explain why the delay is only 1-2 weeks.

If there were a hardware defect AMD wouldn't be able to spin up a corrected chip to manufacturing, package and ship it in that short amount of time.
Regardless of what the fault is, AMD officially stated it as a quality issue all along - which means the solution was always going to be to ship out new inventory that passed QC and have the retailers/integrators/distributors return the recalled stock - hence the 1-2 weeks which is the shipping time.

Any hardware defect would have delayed by months.
Posted on Reply
#46
ARF
Hecate91What bending is there? It looks like excess thermal paste to me, which isn't a problem with the IHS.
I don't see any paste there, just that part of the IHS is pressed downwards, looks like cut manually with a knife, and it is quite visible. I don't why you can't see it.
Posted on Reply
#47
theouto
They'll probably find their way onto ebay in the coming years, and then someone will post a picture of their new ryzen 9 9700X in here. I am willing to bet 50 cents.
Posted on Reply
#48
Darkholm
Generally speaking, this i QA/QC thing which AMD stated as flawed. Most of us has "IT engineering thinking" and in the wake of lots and lots unfinished and unpolished products launched lats 5 years or so, 1st conclusion was "something's wrong with CPU itself" :D It looks that certain someone had a few beers more than it should before started working on laser engraving machine :D
Posted on Reply
#49
Wirko
Hah, just a few days ago, I saw an article over at Tom's about a guy that made a Macintosh clone (a Hackintosh, right?) despite this:
One of the integrated circuits on the original Macintosh Plus board was mislabeled as a 74LS257 when it was a 74F253 IC.
Of course we're dealing with Apple here so this was not a mistake. The choice of the wrong label was very smart, the two chips are functionally similar, and I'm sure the man had a day or two of debugging hell to correct that.
Posted on Reply
#50
holyprof
LittleBroTake that picture with a grain of salt.

Okay, the poor image quality may distort letter-spacing. But that 7 in 9700X looks very odd - spaces between 9, 7, 0 are just strange, like someone cut something from there and stamped another thing there.
That 7 absolutely looks out of place. Different colour, pixel dithering is not the same, indicating it's been hand drawn or photoshopped over the, I'm presuming, original 9900X label which makes sense to be a legit Ryzen 9 9900X.
Posted on Reply
Add your own comment
Nov 5th, 2024 01:25 EST change timezone

New Forum Posts

Popular Reviews

Controversial News Posts