Question AMD 2Q23 Earnings

[adroc_thurston]

but it is definitely one of their lower priority products now.

Still on track to mercy kill SPR-WS relatively soon(tm).

 

[Vattila]

4% YoY for Consoles kind of sucks because this is supposed to be the peak year.

Note that the Gaming segment includes console and gaming GPU sales. Apparently, console sales did well ("higher semi-custom revenue") but was dragged down by weak GPU sales ("lower gaming graphic sales").

"Revenue declined 4% year-over-year to $1.6 billion as higher semi-custom revenue was more than offset by lower gaming graphic sales. [...] Semi-custom SoC sales were strong in the quarter as Microsoft and Sony had healthy console demand based on improved retail availability globally, and the launches of new AAA games."

Earnings call transcript

 

[A///]

Still on track to mercy kill SPR-WS relatively soon(tm).

all of intel could collectively pass gass and they'd kill spr-ws. it was a useless launch but only made sense if you needed the io available which doesn't matter with tr7000 coming in hotter than the stinger of a hornet soon.

you want a killer system for cheaper and want the io? 5995wx pro through an odm, or the 5990x if you can't afford it but can't convince yourself to waste your hard earned bux on outclassed junk like the 3495x.

 

[tamz_msc]

RTG is still faltering, as usual.

 

[adroc_thurston]

RTG is still faltering, as usual.

They're doing well.
Odd to see Q3 being weak despite it being a seasonal strength for consoles.

 

[tamz_msc]

They're doing well.
Odd to see Q3 being weak despite it being a seasonal strength for consoles.

I mean actual dGPU sales, not consoles, when I say RTG. They are faltering on the software side as well. No FSR 3, RT not working on a major release, competitor upscaling surpassing FSR 2 in quality, yada yada.

 

[adroc_thurston]

I mean actual dGPU sales, not consoles, when I say RTG

Those are ripperoni for both vendors (until ML hypebros start scooping everything with 16 gigs of VRAM and up).

No FSR 3

Irrelevant.

RT not working on a major release

2yo Sony game port that gives NV GPUs a neverending struggle moment in asset decomp isn't really major, nor it is functional enough to blame AMD for that.

competitor upscaling surpassing FSR 2 in quality

Upscaling quality is always a function of how much %% of your frame time does one want to sacrifice.
FSR1/2 are by design cheap and portable.

 

[tamz_msc]

Those are ripperoni for both vendors (until ML hypebros start scooping everything with 16 gigs of VRAM and up).

ML hypebros like Elon are interested in NVIDIA, not AMD.

Irrelevant

Still points to the glacial pace of new software development at RTG.

2yo Sony game port that gives NV GPUs a neverending struggle moment in asset decomp isn't really major, nor it is functional enough to blame AMD for that.

That's not all. Wait till you learn that they released non-OEM drivers for Phoenix seven months after announcement.

FSR1/2 are by design cheap and portable.

Also terrible in quality and stagnant in terms of improvements. Nothing surprising though, from AMD. Make it open source and then have someone else figure it out.

 

[Timorous]

Also terrible in quality and stagnant in terms of improvements. Nothing surprising though, from AMD. Make it open source and then have someone else figure it out.

In game TAA is usually terrible hence why DLSS Q can match native 4K. DLSS Q vs DLAA though is totally different beast and shows how crap in game TAA solutions are.

I don't want fake frames, I don't want upscaling. I want hardware that is powerful enough to run at native for a relatively reasonable price (to me $~300 = 1080p60, ~ $500 = 1440p60 and ~ $700 = 4K60) and if necessary to use the much better TAA that DLSS or FSR use rather than the crappy in game stuff.

The big issues with upscaling and crappy TAA is that it allows devs or publishers to be cheap. Rather than doing extra optimisation passes they can just launch it and expect people with a 4090 to use DLSS to play at 4K60 because it looks as good as their crappy 4K native TAA anyway. It is going from being a positive tech that allowed people with lower tier hardware to enjoy some nice effects or to extend the life of an older GPU (FSR mainly for that one) to one that is allowing even more poorly optimised titles onto the market. To me those techs are a net negative so the less time AMD spends on them and the more time they spend on better hardware the better. All AMD need is a great TAA solution that can replace the in game ones and then release their version of DLAA.

 

[Tup3x]

All AMD need is a great TAA solution that can replace the in game ones and then release their version of DLAA.

In Genshin Impact devs replaced the default TAA with FSR2. The problem? It's worse and looks quite bad in motion. They also removed the really good SMAA implementation but people were angry and they restored it. 1,5x resolution scale with SMAA looks superior especially in motion to 1,5x resolution scale with FSR2. They have a lot of work to do. The fun thing with DLSS is that you can easily hack DLAA in if it for some reason is not already supported. Honestly, releasing a game with DLSS but without DLAA should be illegal.



I don't think their situation in consumer GPUs is going to improve any time soon. Strix Halo is going to be interesting but for now... It's the console business that creates the money.

 

[FangBLade]

AMD is doing quite well; they just need to start selling Zen 4 server processors, which will significantly improve their financial situation. On the other hand, Intel is facing much bigger problems than it seems, and I don't see how they will get out of them anytime soon.

 

[Hitman928]

AMD is doing quite well; they just need to start selling Zen 4 server processors, which will significantly improve their financial situation. On the other hand, Intel is facing much bigger problems than it seems, and I don't see how they will get out of them anytime soon.

They said Genoa is ramping in volume already and they expect Genoa(x) and Bergamo to continue to ramp in 2H. That’s part of the forecast for high 2H data center revenue.

 

[Ajay]

So, barring China making some drastic cuts backs in their end of the semiconductor supply chain (chemistry), the first half of 2024 should see more revenue recognition on the server side?

 

[Markfw]

They said Genoa is ramping in volume already and they expect Genoa(x) and Bergamo to continue to ramp in 2H. That’s part of the forecast for high 2H data center revenue.

I should say so, as I have had my 3 Genoa, the 9654 is RETAIL for months.

 

[adroc_thurston]

ML hypebros like Elon are interested in NVIDIA, not AMD.

You're saying stuff you have no idea of.
Don't.

Still points to the glacial pace of new software development at RTG

Stuff least relevant is...
...least relevant.

Wait till you learn that they released non-OEM drivers for Phoenix seven months after announcement.

Phoenix has issues specific not to software at all.

Also terrible in quality and stagnant in terms of improvements.

The good implementations look perfectly fine.
Which is the point.

 

[tajoh111]

AMD keeping prices high to increase margins is showing losing potentially more marketshare than it is worth.

If we consider AMD console revenue grew while their overall revenue shrunk for gaming, AMD discrete graphic revenue may be less than 100 million dollars this quarter. That is not enough to sustain R and D and AMD needs volume. AMD has to accept people will never consider their brand as equal to Nvidia without actions to justify this. RDNA3 did not help AMD efforts(fell below expectations) and their lack of software engineers in the RTG group show they are the budget brand. Hate marketing will only get them so far and has a side effect of poisoning the market for everyone. Only AMD fans will overlook AMD software short coming and pay Nvidia prices for AMD hardware.

Neglected driver releases for RDNA2 post RDNA3 launch, late phoenix graphic drivers, failure on time some of their promised software features show that they are not spending enough on software engineers and the lack of sales is going to worsen this. I think this poor execution on the software side is the result of engineers being moved to datacenter side. But this gamble might not pay off.

I think MI300 is coming a bit too late. Q3 outlook doesn't look good and if Q4 is where we start seeing some Mi300 revenue as a result of El Capitan, I don't think the supply of Mi300 will be that strong this year. And if q1 of 2024 is when supply of MI300 becomes sufficient, AMD will soon have to contend with Blackwell. Nvidia has a tendency to make announcements near the end of Q1 for their next gen compute products. AMD need to release their data center products early since they have a software disadvantage, releasing it late mean companies are losing valuable time developing software to work on MI300.

 

[tamz_msc]

You're saying stuff you have no idea of.

Guys like George Hotz trashing ROCm and then posting pictures on Twitter of XFX 7900 XTX cards that he bought is but one example of AMD being bought in large numbers.

 

[adroc_thurston]

Guys like George Hotz trashing ROCm

He had no idea what he was doing.

and then posting pictures on Twitter of XFX 7900 XTX cards that he bought is but one example of AMD being bought in large numbers

That's small fry stuff.
Hyperscalers is where the money is at and MI300X (and even A) has quote some very solid traction there.
Again, ROCm issue is inter-GPU portability and not really the s/w itself.

 

[tamz_msc]

He had no idea what he was doing.

That's small fry stuff.
Hyperscalers is where the money is at and MI300X (and even A) has quote some very solid traction there.
Again, ROCm issue is inter-GPU portability and not really the s/w itself.

MI300 is not part of Gaming. I thought we were talking about consumer gaming GPUs being bought by AI startups?

 

[adroc_thurston]

MI300 is not part of Gaming.

It's part of Radeon.
Designed by very much GPU hweng people.

 

[A///]

fsr 3 lol. as much as I'd love to see amd make a comeback kid case for their video cards good grief do they need a serious shakedown. they need to bring in some industry leading architects to ge ttheir juices flowing. even intel realised they messed up hiring raja koudouri. people give that moron more credit than he deserves.

ML hypebros like Elon are interested in NVIDIA, not AMD.

youve spelt wanker wrong.

 

[JustViewing]

Nothing that I saw. Seems they are going hard into the AI craze as it is an exploding market. TR is quite small potatoes in their range of products. I’m sure they will still be bringing them to market, but it is definitely one of their lower priority products now.

Actually TR can be a good candidate for AI researchers because of its multiple PCIe slots. This would allow AMD to sell both TR and CDNA cards.

 

[DAPUNISHER]

The perpetually bearish position some have on AMD puts a big smile on my face. This time it's the dGPU market that they are putting a fork in. It's the surest sign it'll be fine in the long run.

In 2016 they were putting a fork in the whole company. Telling me how console wins were not going to help save them. Margins are razor thin you know. And the only reason they got the contracts is no one else was willing to take such tiny profits.

Enterprise? Fuhgeddaboudit! They would have to make good products, which they don't, and will likely never get the chance to. Even if they did make good products, they couldn't supply enough to make the money they need to keep the ship from sinking. All they can do is compete on price at the low end of consumer because again, the products are trash. Have been for way too long, and they don't have the R&D money to do anything about it. Intel would just cut prices to shovel the last scoop of dirt on them anyways. RIP AMD.

Fast forward to 2023 and Sony is their biggest customer. Enterprise looks better than it ever did. AM4 has been dominating the retail space for (hyperbole incoming) what seems like forever now. There are 100s of millions of AMD powered consoles sold and counting. They are dominating the growing handheld console and PC market.

When the doom and gloomers start brigading, I know everything will be okay. 🖕

 

[Hitman928]

Actually TR can be a good candidate for AI researchers because of its multiple PCIe slots. This would allow AMD to sell both TR and CDNA cards.

That may be true, but AMD is focused on data center customers right now which is much higher volume than they could get with workstation solutions. Again, not saying they won’t have workstation products, just that it is clearly not their priority.

 

[Hitman928]

So, barring China making some drastic cuts backs in their end of the semiconductor supply chain (chemistry), the first half of 2024 should see more revenue recognition on the server side?

AMD didn’t want to give much guidance into 2024, but based on the statements they did make, I would expect 1H24 data center revenue to be down sequentially from 2H23 (due to El Capitan revenue largely realized in 2H23) but up from 1H24 (due to higher Epyc sales).