Curious "Navi 48 XTX" Graphics Card Prototype Detected in Regulatory Filings

[btarunr]

A curiously described graphics card was detected by Olrak29 as it was making it through international shipping. The shipment description for the card reads "GRAPHIC CARD NAVI48 G28201 DT XTX REVB-PRE-CORRELATION AO PLATSI TT(SAMSUNG)-Q2 2024-3A-102-G28201." This can be decoded as a graphics card with the board number "G28201," for the desktop platform. It features a maxed out version of the "Navi 48" silicon, and is based on the B revision of the PCB. It features Samsung-made memory chips, and is dated Q2-2024.

AMD is planning to retreat from the enthusiast segment of gaming graphics cards with the RDNA 4 generation. The company originally entered this segment with the RX 6800 series and RX 6900 series RDNA 2 generation, where it saw unexpected success with the crypto-mining market boom, besides being competitive with the RTX 3080 and RTX 3090. This bust by the time RDNA 3 and the RX 7900 series arrived, and the chip wasn't competitive with NVIDIA's top-end. Around this time, the AI acceleration boom squeezed foundry allocation of all major chipmakers, including AMD, making large chips based on the latest process nodes even less viable for a market such as enthusiast graphics—the company would rather make CDNA AI accelerators with its allocation. Given all this, the company's fastest GPUs from the RDNA 4 generation could be the ones that succeed the current RX 7800 XT and RX 7700 XT, so AMD could capture a slice of the performance segment.



View at TechPowerUp Main Site | Source

 

[AusWolf]

AMD missed a spot by not saying a single word about this at Computex, imo.

Still, I'm probably gonna be in the market for a GPU by the time it lands in stores, so I hope it'll be good.

 

[ARF]

Will AMD's shareholders finally see the epic failures of Lisa Su, and ask her to take the responsibility, and finally resign?
Appoint a new CEO and restructure the company to be more similar to what nvidia is doing?

Fire Lisa Su now!

 

[AusWolf]

Will AMD's shareholders finally see the epic failures of Lisa Su, and ask her to take the responsibility, and finally resign?
Appoint a new CEO and restructure the company to be more similar to what nvidia is doing?

Fire Lisa Su now!

What failures? Ryzen is soaring, and RDNA 3 is not bad for what it is. If RDNA 4 ends up being another 5700 XT, I'll write it off as another semi-success.

 

[ARF]

What failures? Ryzen is soaring, and RDNA 3 is not bad for what it is. If RDNA 4 ends up being another 5700 XT, I'll write it off as another semi-success.

Ryzen looks "not bad", because intel is even worse. No comment about RDNA 3. I think your claim can't be serious.

 

[AusWolf]

Ryzen looks "not bad", because intel is even worse. No comment about RDNA 3. I think your claim can't be serious.

If both Intel and AMD are bad, then what do you define as "good"?

I had a 7800 XT for a couple of months. Despite some minor issues with idle power, it's fine.

 

[Vayra86]

Will AMD's shareholders finally see the epic failures of Lisa Su, and ask her to take the responsibility, and finally resign?
Appoint a new CEO and restructure the company to be more similar to what nvidia is doing?

Fire Lisa Su now!

What are you smoking?

Lisa Su made AMD soar to unforeseen heights.

 

[mikesg]

When high-end cards are over $1000, competing on price is the real measure of performance

 

[john_]

What failures? Ryzen is soaring, and RDNA 3 is not bad for what it is. If RDNA 4 ends up being another 5700 XT, I'll write it off as another semi-success.

They are doing some mistakes lately.
RX 7000 not having significantly better RT performance than RX 6000 lead to lost market share.
Zen 5 X3D chips coming to AM5 much latter than the platform introduction, which resulted in a lukewarm reception of the platform by press and public.
AM5 not supporting DDR4 because someone predicted that DDR5 will be ultra cheap soon.
Still not much from their software department.
Just yesterday reading that the new Ryzen AI will only support Windows 11 and not Windows 10, that's also stupid.
Not securing enough capacity at TSMC to take advantage of all the opportunities they had.
Things I forget right now.

But Su resigning? Nope. AMD doesn't need turbulences right now. Just better decision/prediction making.

 

[AusWolf]

RX 7000 not having significantly better RT performance than RX 6000 lead to lost market share.

That's because RX 7000 has essentially the same RT engine as RX 6000. And that's because they didn't have anything better at the moment. Let's see if that changes with RDNA 4.

Zen 5 X3D chips coming to AM5 much latter than the platform introduction, which resulted in a lukewarm reception of the platform by press and public.
AM5 not supporting DDR4 because someone predicted that DDR5 will be ultra cheap soon.

Do you mean Zen 4? If so, I agree. I guess they just wanted to cash in some non-X3D sales first before throwing the big guns in. They just didn't expect people not jumping on the platform costs.

Still not much from their software department.

Like what? As far as I'm concerned, their software is fine.

Just yesterday reading that the new Ryzen AI will only support Windows 11 and not Windows 10, that's also stupid.

I don't think most home users give a flying **** about AI in any shape or form.

Not securing enough capacity at TSMC to take advantage of all the opportunities they had.

Is it really a mistake? Or is it pure chance of being outbid by much wealthier companies like Nvidia?

 

[Vayra86]

Strategy isn't a daily changing perspective, but some people here seem to think that's how it works.

There is more to this world than social media conjecture.

 

[john_]

That's because RX 7000 has essentially the same RT engine as RX 6000. And that's because they didn't have anything better at the moment. Let's see if that changes with RDNA 4.

Not predicting that RT performance would be already a strong marketing card and that consumers would avoid buying expensive RX 7000 cards because of the lack of RT performance, it's a serious misjudgement.

Do you mean Zen 4? If so, I agree. I guess they just wanted to cash in some non-X3D sales first before throwing the big guns in. They just didn't expect people not jumping on the platform costs.

Ah, yes, thanks. Zen 4. They where keeping X3D for later, not realizing that tables had turned from the moment Intel was selling higher number of cores. And considering that Ryzen 1000 was a success because it had more cores than the competition, it is stupid they didn't realized they needed something special for the AM5 platform, meaning the X3D chips, sooner rather than latter. Latest rumors suggest that we will see the Zen 5 X3D chips in 2 months, not having to wait 7 months as with Zen 4. So they are probably fixing this one.

Like what? As far as I'm concerned, their software is fine.

They have to start competing Nvidia in software optimizations for their cards. OK, we talk about gaming cards, but looking how Nvidia cards get some extra features/optimizations it software, either that is rendering, AI, photo processing or whatever, this is important from a marketing at least perspective. People pay serious money today for graphics cards and the more reasons you give them to justify the money they spent, the better you will sell.

I don't think most home users give a flying **** about AI in any shape or form.

I am not talking about AI here, but the fact that if you buy a Ryzen AI 300 series laptop with Windows 11 and you want to "downgrade" it to Windows 10 to avoid all that AI crap and AI monitoring and start menu advertising and whatever other crap comes out from those brains in MS, you just can't. AMD wouldn't offer support for Windows 10 with their Ryzen AI 300 series APUs. This is just stupid.

Is it really a mistake? Or is it pure chance of being outbid by much wealthier companies like Nvidia?

Looking at AMD's financials the last 10 years and seeing how often they are usually spot on about their sales and income for the next quarter, I would say that they order from TSMC just about as much wafers as much they KNOW they will sell. Meaning if something starts selling much better than expected, because of whatever reasons, they just don't have enough capacity to cover that extra support immediately. So, when Ryzen 5000 series was a success, Ryzen prices skyrocketed because demand was much higher than supply. When crypto exploded, prices of RX cards skyrocketed, but if I remember correctly AMD didn't made the money Nvidia made, I am assuming because they haven't ordered extra wafers from TSMC. They did it latter, too late in the game and had to come out and say that they where going to make a billion less than anticipated. Now with AI they still seem to not have enough wafers to support the gaming market alongside the AI market. Nvidia is selling more AI GPUs they can produce, but we don't see shortages of Nvidia gaming GPUs in the market.

A company like AMD today should be more diverse. They could be doing business with Samsung's fabs if they can't find enough wafers at TSMC. Not all products need the efficiency of TSMC's nodes. AM4 Ryzen CPUs and APUs and some Zen 4 Ryzen CPUs could also be made at Samsung for example.

 

[AusWolf]

Not predicting that RT performance would be already a strong marketing card and that consumers would avoid buying expensive RX 7000 cards because of the lack of RT performance, it's a serious misjudgement.

Misjudgement, or lack of better RT hardware available? Nvidia jumped into the game a lot sooner than AMD, so I'm not blaming AMD for not having much to compete with for two generations. RDNA 2's RT unit is basically an "oh shit, we better do something fast" solution, and the one in RDNA 3 is basically a copy-paste. I don't think AMD would have done the same thing twice if they had anything else at the time. It's probably not because they didn't think better RT was necessary. They just didn't have anything else.

They have to start competing Nvidia in software optimizations for their cards. OK, we talk about gaming cards, but looking how Nvidia cards get some extra features/optimizations it software, either that is rendering, AI, photo processing or whatever, this is important from a marketing at least perspective. People pay serious money today for graphics cards and the more reasons you give them to justify the money they spent, the better you will sell.

I'm not sure how relevant these things are to the average consumer. I think it's more of an "Nvidia=good, AMD=bad" bandwagon than anything. I, for example, don't need any of those that you listed, and for pure gaming and general use, AMD's software is fine.

I am not talking about AI here, but the fact that if you buy a Ryzen AI 300 series laptop with Windows 11 and you want to "downgrade" it to Windows 10 to avoid all that AI crap and AI monitoring and start menu advertising and whatever other crap comes out from those brains in MS, you just can't. AMD wouldn't offer support for Windows 10 with their Ryzen AI 300 series APUs. This is just stupid.

If one is tech-savvy enough to downgrade from Windows 11 to 10, then that person probably knows that Windows 11 is basically 10 with a new graphical skin. With that in mind, I expect no issues getting Windows 11 drivers running on 10. The rest of the potential customer base (the majority) won't care, they'll just flow with 11.

Looking at AMD's financials the last 10 years and seeing how often they are usually spot on about their sales and income for the next quarter, I would say that they order from TSMC just about as much wafers as much they KNOW they will sell. Meaning if something starts selling much better than expected, because of whatever reasons, they just don't have enough capacity to cover that extra support immediately. So, when Ryzen 5000 series was a success, Ryzen prices skyrocketed because demand was much higher than supply. When crypto exploded, prices of RX cards skyrocketed, but if I remember correctly AMD didn't made the money Nvidia made, I am assuming because they haven't ordered extra wafers from TSMC. They did it latter, too late in the game and had to come out and say that they where going to make a billion less than anticipated. Now with AI they still seem to not have enough wafers to support the gaming market alongside the AI market. Nvidia is selling more AI GPUs they can produce, but we don't see shortages of Nvidia gaming GPUs in the market.

A company like AMD today should be more diverse. They could be doing business with Samsung's fabs if they can't find enough wafers at TSMC. Not all products need the efficiency of TSMC's nodes. AM4 Ryzen CPUs and APUs and some Zen 4 Ryzen CPUs could also be made at Samsung for example.

I don't know. These are too big things for a mere mortal like myself to comment on. I'm just waiting for whatever RDNA 4 will put on the table.

 

[Daven]

AMD missed a spot by not saying a single word about this at Computex, imo.

Still, I'm probably gonna be in the market for a GPU by the time it lands in stores, so I hope it'll be good.

None of the three companies said anything about next gen GPUs. No Battlemage, no RDNA4 and no Blackwell information was shared.

 

[AusWolf]

None of the three companies said anything about next gen GPUs. No Battlemage, no RDNA4 and no Blackwell information was shared.

Yeah, shame.

 

[Daven]

A company like AMD today should be more diverse. They could be doing business with Samsung's fabs if they can't find enough wafers at TSMC. Not all products need the efficiency of TSMC's nodes. AM4 Ryzen CPUs and APUs and some Zen 4 Ryzen CPUs could also be made at Samsung for example.

Even Intel is making many to most of its chips at TSMC. Lunar Lake is all TSMC.

 

[john_]

Misjudgement, or lack of better RT hardware available? Nvidia jumped into the game a lot sooner than AMD, so I'm not blaming AMD for not having much to compete with for two generations. RDNA 2's RT unit is basically an "oh shit, we better do something fast" solution, and the one in RDNA 3 is basically a copy-paste. I don't think AMD would have done the same thing twice if they had anything else at the time. It's probably not because they didn't think better RT was necessary. They just didn't have anything else.

Intel did a better job in their first try. AMD didn't really had to do much with RX 7000. Just increase the Ray accelerators. Maybe it was difficult, or was taking too much size on the die? Don't know. But that's what limit their sales in my opinion.

I'm not sure how relevant these things are to the average consumer. I think it's more of an "Nvidia=good, AMD=bad" bandwagon than anything. I, for example, don't need any of those that you listed, and for pure gaming and general use, AMD's software is fine.

They are probably not. But they paint a picture where Nvidia GPU can do 10 things and AMD GPU can do 5 things and in those 5 things an Nvidia GPU is in general better. With that idea in mind the average consumer happily pays more money to get an RTX 3050 instead of an RX 6600. Obviously that's the wrong choice, but in the end the money end to Nvidia's pockets.

If one is tech-savvy enough to downgrade from Windows 11 to 10, then that person probably knows that Windows 11 is basically 10 with a new graphical skin. With that in mind, I expect no issues getting Windows 11 drivers running on 10. The rest of the potential customer base (the majority) won't care, they'll just flow with 11.

If a tech savvy person can make Win 11 drivers work on 10, then AMD offering official support would have been extremely easy for them. But they chose to make their lives easier, by ignoring Win 10, even if that meant fewer sales.

I don't know. These are too big things for a mere mortal like myself to comment on. I'm just waiting for whatever RDNA 4 will put on the table.

You are a mortal? I thought I was talking to a god like myself!
Well they are putting the XTX logo on the top mid range GPU we expect to see based on RDNA4. Hope they have a good reason, like much better performance and it's not desperation from their marketing department.
-"We need to have an XTX model in the market. WE NEED TO HAVE AT LEAST ONE!"

 

[ARF]

They are doing some mistakes lately.
RX 7000 not having significantly better RT performance than RX 6000 lead to lost market share.

RX 7000 is a failure no matter how you look at it.
1. Chiplet design was abandoned.
2. Performance targets were not met.
3. Sales are extremely low.
4. Prices are very high.
5. Ray-tracing poor performance.
6. FSR a joke, extremely low quality image.
7. Drivers not released regularly, instead bugs stay for quarters without anyone paying attention about fixing them.
8. Maybe lost the interest in the GPU department, and potentially leaving the market segment?

 

[Sound_Card]

Will AMD's shareholders finally see the epic failures of Lisa Su, and ask her to take the responsibility, and finally resign?
Appoint a new CEO and restructure the company to be more similar to what nvidia is doing?

Fire Lisa Su now!

She brought them from $2 stock price to $160 and probably going to hit $250 at the rate they are going. She should make AMD like Nvidia? I am not sure what exactly that means. It could mean: Make your name so strong, like Apple, you can over price anything and anyone is going to buy it. But to be fair, and judging by the other sensational and mostly misinformed post you made above, I am pretty sure you are a troll.

 

[AVATARAT]

1. Chiplet design was abandoned.

Are you sure?

 

[john_]

She brought them from $2 stock price to $160 and probably going to hit $250 at the rate they are going. She should make AMD like Nvidia? I am not sure what exactly that means. It could mean: Make your name so strong, like Apple, you can over price anything and anyone is going to buy it. But to be fair, and judging by the other sensational and mostly misinformed post you made above, I am pretty sure you are a troll.

Well, not everyone bought at $2. Anyway, if AMD doesn't start thinking big today, not tomorrow, today because they don't have any more time, that $227 that the share did a few months ago will become a one time bubble. For now even that $160 that the price hovers now doesn't look safe. It is more possible to go down to $100 than going over $200 again. And that's because MI300X/A doesn't make much headlines, x86 client CPUs could see competition from ARM soon, server CPUs are becoming secondary with everyone investing in AI, meaning custom chips and mostly Nvidia GPUs, Intel is coming back with Lunar Lake and Battlemage, meaning AMD could start losing handhelds(I think the first version of MSI Claw, was just a beta testing device, which will explain why MSI gone with Intel) and mobile sales, in gaming they are dropping way behind, meaning in the future Nvidia will have even stronger control of the market and pricing and seeing Intel building something good at TSMC (Lunar Lake does look really good), AMD could also lose the next XBOX.

 

[Sound_Card]

Most of what you wrote is wistful speculation. AMD's stock price is volatile, but volatile in an upwards direction.

 

[ARF]

She brought

Jim Keller brought, he joined AMD in 2012 and started to work on the new Zen micro-architecture.
Lisa Su was appointed two years later.

Bad attempt, you troll.

Computer Architect Jim Keller Joins AMD as Chief of Processor Group

Industry Veteran Architected Several Generations of Popular Apple, Broadcom and AMD Processors SUNNYVALE, CA -- (Marketwire) -- 08/01/12 -- AMD…...

ir.amd.com

 

[the54thvoid]

Jim Keller brought, he joined AMD in 2012 and started to work on the new Zen micro-architecture.
Lisa Su was appointed two years later.

Bad attempt, you troll.

Computer Architect Jim Keller Joins AMD as Chief of Processor Group

Industry Veteran Architected Several Generations of Popular Apple, Broadcom and AMD Processors SUNNYVALE, CA -- (Marketwire) -- 08/01/12 -- AMD…...

ir.amd.com

Lisa Su was senior VP at AMD in Jan 2012. Keller joined in August.

 

[ARF]

Lisa Su was senior VP at AMD in Jan 2012. Keller joined in August.

Ok, but even then it was Rory Read who appointed Jim Keller, while Su's role was to work for semi-custom solutions / consoles business.