AMD Navi 32 RDNA 3 GPU Spotted in Forbes Video

[T0@st]

Forbes published its video interview with AMD CEO and President Lisa Su at the end of May, but it has taken two weeks for hardware news sites to realize that unreleased silicon was in plain view within the spotlight piece. Folks likely regarded it as a simple puff piece due to the title reading "This CEO Made AMD Billions - Now She Wants To Dominate The Market With AI." Hoang Anh Phu, a Vietnamese technology enthusiast, managed to pay close attention to a curious segment in the Forbes video and uploaded AI-upscaled screengrabs to Twitter along with the comment/question: "Navi 32 die shot(?!)."

RDNA3 Navi 31 and Navi 33 GPU products have already reached the retail market—AMD's high-end (chiplet design) Radeon RX 7900-series is based on the former and it launched last December. The latter arrived in the (monolithic N33 XL) form of Radeon RX 7600 cards at the end of May 2023. Even board partners are seemingly becoming impatient about a lack of new offerings in the mid-range—Sapphire is very likely to release another previous gen Radeon RX 6750 XT custom card this week in China. Team Red has not publicly acknowledged that Navi 32 is a work in-progress, so it is slightly odd that an example sat next to EPYC Genoa, Raphael, and Raphael X3D dies on a table—as spotted in the Forbes feature. Screenshots show an Infinity Cache setup with four memory stacks on a previously unseen die. Leaks have indicated that Navi 32 will be a chiplet design with a GCD (200 mm²) in the middle, surrounded by the four MCDs (37.5 mm²). The full package area size is eyeball estimated to occupy around 350 mm² of space, which corroborates info uncovered in the past.




The Forbes introduction states: "AMD's CEO orchestrated one of the greatest turn arounds in Silicon Valley history. Next, she's preparing for battle in the AI revolution-and she expects to keep winning."

It continues: "From a conference room atop AMD's headquarters in Santa Clara, California, a stretch of highway 101 running outside, Lisa Su presides over a company older than the term "Silicon Valley." Down the road is a link to the company's past, an old foundry in Sunnyvale where AMD used to press its chips. But from her window she can see a recent milestone in the company's fast-evolving present: the offices of arch-nemesis Intel, whose market capitalization ($120.3 billion) AMD's now eclipses ($153.5 billion)."

View at TechPowerUp Main Site | Source

 

[MrDweezil]

So is it:
1. Still sitting on too much 6xxx inventory to release now.
2. Isn't enabling them to hit better price/performance targets than discounted existing tech, so not worth releasing at all.

 

[btarunr]

So is it:
1. Still sitting on too much 6xxx inventory to release now.
2. Isn't enabling them to hit better price/performance targets than discounted existing tech, so not worth releasing at all.

That thing needs to beat NVIDIA AD104 in cost. AD104 is a 295 mm2 monolithic silicon. Navi 32 is a 200 mm2 5 nm GCD with four 6 nm MCDs worth 38 mm2 each, plus the exotic MCM packaging and Infinity Fanout links, and the fact that its 256-bit memory bus demands 8 memory chips and a more complex PCB.

I have a hunch that AD104 is more economical than Navi 32, and faster. The RTX 4070 Ti (maxed out AD104) is within 8-5% of the RX 7900 XT, and naturally AMD would've tried to space the top Navi 32 SKU at least 10-15% perf away from 7900 XT.

This makes the Navi 32 a non-starter compared to AD104 and RTX 4070 Ti. This is probably why AMD launched RX 7600 after 7900 series, its performance segment for this generation is a mess.

 

[ymdhis]

So is it:
1. Still sitting on too much 6xxx inventory to release now.
2. Isn't enabling them to hit better price/performance targets than discounted existing tech, so not worth releasing at all.

3. they have much better profit margins on CPUs (especially for server chips), so no point in wasting their limited silicon production capability on GPUs.

 

[HD64G]

Combo of points 1 and 3 above might be the main reason of this delay alongside some hardware level fixes needed.

 

[MrDweezil]

That thing needs to beat NVIDIA AD104 in cost. AD104 is a 295 mm2 monolithic silicon. Navi 32 is a 200 mm2 5 nm GCD with four 6 nm MCDs worth 38 mm2 each, plus the exotic MCM packaging and Infinity Fanout links, and the fact that its 256-bit memory bus demands 8 memory chips and a more complex PCB.

I have a hunch that AD104 is more economical than Navi 32, and faster. The RTX 4070 Ti (maxed out AD104) is within 8-5% of the RX 7900 XT, and naturally AMD would've tried to space the top Navi 32 SKU at least 10-15% perf away from 7900 XT.

This makes the Navi 32 a non-starter compared to AD104 and RTX 4070 Ti. This is probably why AMD launched RX 7600 after 7900 series, its performance segment for this generation is a mess.

Forget about the AD104, a complex chiplet design might not be price competitive with their own 6800 cards.

 

[jpvalverde85]

Looked like a Fiji chip with HBM, but dies aren't same size. I hope latency doesn't kill the performance against a monolithic Navi 21.

 

[phanbuey]

A little late to the party and will face pressure from Apple, Amazon, Google and Intel among others... Going to be interesting.

 

[auxy]

That thing needs to beat NVIDIA AD104 in cost. AD104 is a 295 mm2 monolithic silicon. Navi 32 is a 200 mm2 5 nm GCD with four 6 nm MCDs worth 38 mm2 each, plus the exotic MCM packaging and Infinity Fanout links, and the fact that its 256-bit memory bus demands 8 memory chips and a more complex PCB.

I have a hunch that AD104 is more economical than Navi 32, and faster. The RTX 4070 Ti (maxed out AD104) is within 8-5% of the RX 7900 XT, and naturally AMD would've tried to space the top Navi 32 SKU at least 10-15% perf away from 7900 XT.

This makes the Navi 32 a non-starter compared to AD104 and RTX 4070 Ti. This is probably why AMD launched RX 7600 after 7900 series, its performance segment for this generation is a mess.

Only four MCDs.
Also, rumors say that Navi 32 was sent for a respin to fix problems found in Navi 31/33. That's why we haven't seen it yet.

 

[Denver]

3. they have much better profit margins on CPUs (especially for server chips), so no point in wasting their limited silicon production capability on GPUs.

Exactly. And it doesn't require heavy and continuous investment in software (drivers)

 

[Minus Infinity]

Honestly why don't they skip the rest of RDNA3 and just release RDNA4 next year and I'm not joking.

 

[Pumper]

AMD managed to ruin the rest of the lineup the moment they decided to release the overpriced 7900XT instead of naming it the 7800XT from the start. Now the "real" 7800XT can't even compete with last gen stuff.