Discussion AMD SoC Halo series GPU discussion

Page 35 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

moinmoin

Diamond Member
Jun 1, 2017
5,203
8,365
136

Kronos1996

Member
Dec 28, 2022
57
94
61
Yeah that's how you win.
They built Rome with 2% server marketshare.
You're welcome.
Server customers are a very different market vs consumers. They have nearly unlimited piles of money if you make the best product for them. Gamers buying luxury goods do not. That’s a False Equivalence and always has been. Zen 2/3 took tons of marketshare without beating Intel in gaming because they offered great value. Until Zen 3X3D of course.

Bankrupting your company trying to build a halo part is an asinine way to get marketshare and it doesn’t work very well. Instead of making a few very good products you spread resources too thin like RDNA 3. That did more harm than good. Start small with 1-2 market Segments and go from there as profit grows. That’s how sane business strategy works.
 
Reactions: Tlh97

Kronos1996

Member
Dec 28, 2022
57
94
61
This number is an urban legend that doesnt pass the most elementary analysis,
it s just so much spread by the usual viral marketers that it s taken at face value by the uninformed and then echo chambered ad nauseam.
Not according to Jon Peddie Research. AMD is usually more like 15-20% but 2023 was bad for them. Only picked up in Q4 when Nvidia started running out of Lovelace.
 

Attachments

  • IMG_2727.jpeg
    560.2 KB · Views: 29
Reactions: scineram

adroc_thurston

Diamond Member
Jul 2, 2023
5,382
7,559
96
Server customers are a very different market vs consumers. They have nearly unlimited piles of money if you make the best product for them. Gamers buying luxury goods do not. That’s a False Equivalence and always has been
Uh, nope.
Sames rules!
Zen 2/3 took tons of marketshare without beating Intel in gaming because they offered great value
a) 3950X smashed Intel in everything but vidya (and was alright there too)
b) 5950X (Z3 parts in general) won gaming too. Saying otherwise is historic revisionism.
 
Reactions: Thunder 57

Abwx

Lifer
Apr 2, 2011
11,727
4,646
136
Not according to Jon Peddie Research. AMD is usually more like 15-20% but 2023 was bad for them. Only picked up in Q4 when Nvidia started running out of Lovelace.
JPR numbers are only for a few OEMs PCs, they dont even include secondary manufacturers and dont account at all the retail GPU market.
 

Kronos1996

Member
Dec 28, 2022
57
94
61
Short answer, yes I do. A diverse product portfolio is important in case one segment collapses. As I said originally, this hypothetical super-halo product would be 2-4 years out, so current wafer constraints aren't particularly relevant. And I think it would take a more than 0.1% of the dGPU market (local AI alone would be more than that).
Diverse product portfolio is good if you can afford it. You don’t start out trying to do everything at once. Start with one market at a time and slowly grow as your marketshare, revenue and profit do. Wasting resources making products you can barely afford isn’t wise.

David Wang said they wanna target the 80% volume GPU market with RDNA 4. That implies that AMD’s internal metrics show only 20% of the market buys 80 or 90 class cards. Per these charts from JPR, AMD sold 4.4 Million GPU’s in 2024. 800,000 isn’t a lot of units for the R&D cost to make. I suspect 80 class cards outsell 90 class 5-10 to 1. So a true halo part has even lower volume.

JPR charges for access to full reports but this article shares some of their data for 2024.

 

Attachments

  • Image.jpeg
    472.8 KB · Views: 19
  • IMG_2730.png
    70.2 KB · Views: 14
  • IMG_2732.png
    195.2 KB · Views: 16
  • IMG_2733.png
    265.8 KB · Views: 27
Reactions: scineram

Kronos1996

Member
Dec 28, 2022
57
94
61
Uh, nope.
Sames rules!

a) 3950X smashed Intel in everything but vidya (and was alright there too)
b) 5950X (Z3 parts in general) won gaming too. Saying otherwise is historic revisionism.
The rules for consumer markets are that you make products the majority of customers actually want to buy at the best price/performance. I’ve never made a purchasing decision based on whether the company in question makes a more powerful product. That’s completely irrelevant to me and most consumers.

I bought a Mazda 3 because it’s an incredibly reliable and affordable car. Not because Mazda also makes the Miata. That’s just stupid. While there is some value to the halo affect in the gaming market, I think it’s WILDLY blown out of proportion. Most markets don’t give a single crap about it at all. A few gamers do simply because culturally they turn everything into in epeen contest.

Zen 3 did not beat 10th/11th gen in gaming until X3D arrived late in the cycle. It was roughly equal.
 

Attachments

  • IMG_2734.jpeg
    1.1 MB · Views: 26
Last edited:

adroc_thurston

Diamond Member
Jul 2, 2023
5,382
7,559
96
The rules for consumer markets are that you make products the majority of customers actually want to buy at the best price/performance.
No that's what you do after you won.
I bought a Mazda 3 because it’s an incredibly reliable and affordable car.
car analogy opinion discarded.
Zen 3 did not beat 10th/11th gen in gaming until X3D arrived late in the cycle. It was almost equal.
no it did, HWUB testing just sucks (1080p ultra AAA slop is *not* how you test gaming CPU perf. Especially not in 2020, before the 9th gen bawkses made games generally more CPU-heavy.).
 

Kronos1996

Member
Dec 28, 2022
57
94
61
No that's what you do after you won.

car analogy opinion discarded.

no it did, HWUB testing just sucks (1080p ultra AAA slop is *not* how you test gaming CPU perf. Especially not in 2020, before the 9th gen bawkses made games generally more CPU-heavy.).
You think making bad value, high priced, niche products is how companies grow and get established? That shows a staggering lack of knowledge on the subject. You don’t enter a market making luxury goods, you start with the commodity segment.

The only reason Ryzen 9 made sense is because of chiplets. AMD never would’ve made halo parts like that with monolithic dies. The same strategy isn’t possible yet with consumer GPU’s because of the expensive packaging required. Therefore, I’m inclined to believe that the entire executive team at AMD knows more about business than you do.
 
Reactions: Tlh97 and marees

gdansk

Diamond Member
Feb 8, 2011
4,036
6,646
136
The only reason Ryzen 9 made sense is because of chiplets. AMD never would’ve made halo parts like that with monolithic dies.
What on earth do you think the plan for To Mega Navion was? Isn't every big AMD multi-chip already?
 

MS_AT

Senior member
Jul 15, 2024
556
1,170
96
Is it possible to preorder any of the other models already?
Not to my knowledge, so yes, might be it but I thought giving pricing information is not always equal to starting preorders. Might be I am misremembering things.
 

inquiss

Senior member
Oct 13, 2010
352
527
136
You think making bad value, high priced, niche products is how companies grow and get established? That shows a staggering lack of knowledge on the subject. You don’t enter a market making luxury goods, you start with the commodity segment.

The only reason Ryzen 9 made sense is because of chiplets. AMD never would’ve made halo parts like that with monolithic dies. The same strategy isn’t possible yet with consumer GPU’s because of the expensive packaging required. Therefore, I’m inclined to believe that the entire executive team at AMD knows more about business than you do.
Niche expensive products that are the best at everything is how iPhone won the mobile market. It's what Qualcomm tried to do with their entry into the laptop space, premium first then trickle down the mainstream parts once a brand image has been established.

I'm not saying your suggested method hasn't also had success, but I don't think you should be as sure about your opinion here as you seem to be.
 

adroc_thurston

Diamond Member
Jul 2, 2023
5,382
7,559
96
You think making bad value, high priced, niche products is how companies grow and get established?
yeah that's how Zen3 worked.
Horrible value ($300 for 6 cores. lmao). Wins everything.
Isn't every big AMD multi-chip already?
yeah. lmao.
The only reason Ryzen 9 made sense is because of chiplets
no it's made sense because it won.
Just like it made sense to win in server.
gfx guys just need a small loan of BALLS from mr. Forrest Norrod.
 

inquiss

Senior member
Oct 13, 2010
352
527
136
yeah that's how Zen3 worked.
Horrible value ($300 for 6 cores. lmao). Wins everything.

yeah. lmao.

no it's made sense because it won.
Just like it made sense to win in server.
gfx guys just need a small loan of BALLS from mr. Forrest Norrod.
Yeah, if only AMD knew how to infuse that ambition into the GFX. The started the process with the cross pollination if CPU knowhow into GPU, now make the most of it. Do it.
 

Abwx

Lifer
Apr 2, 2011
11,727
4,646
136
Per these charts from JPR, AMD sold 4.4 Million GPU’s in 2024.
Sold to OEM, not to the retail GPU market, prove is that even at 100$/GPU this would make only 440 millions for a whole year, while AMD s worse quarter was 429 millions without almost no sales to console manufacturers like Sony.

Given AMD s quartely sales JPR numbers amount to barely one quarter sales and hence using them as a geneality is underestimating their markeshare by at least a 3x ratio, the maths don lie.
 

jpiniero

Lifer
Oct 1, 2010
15,997
6,477
136
Sold to OEM, not to the retail GPU market, prove is that even at 100$/GPU this would make only 440 millions for a whole year, while AMD s worse quarter was 429 millions without almost no sales to console manufacturers like Sony.

Given AMD s quartely sales JPR numbers amount to barely one quarter sales and hence using them as a geneality is underestimating their markeshare by at least a 3x ratio, the maths don lie.

Figure most of that was inventory that AMD had already sold to distributors/OEMs before 2024.
 

reaperrr3

Member
May 31, 2024
82
267
86
Sold to OEM, not to the retail GPU market, prove is that even at 100$/GPU this would make only 440 millions for a whole year, while AMD s worse quarter was 429 millions without almost no sales to console manufacturers like Sony.
100$/GPU?
That's probably what AMD is asking for N33. For N31XTX, it was probably ~5x that or even more.
Also, what OEM?
AMD mobile GPUs are dead, and desktop "OEM" cards are usually by some AIB.

Given AMD s quartely sales JPR numbers amount to barely one quarter sales and hence using them as a geneality is underestimating their markeshare by at least a 3x ratio, the maths don lie.
No offense, but I think you're rather underestimating the average price AMD asked for their GPUs by a factor of at least 3x.

If a GPU costs ~25-30$ to make on N6, AMD will want to ask at least like 100-150$ per chip of AIBs, after all you got years of R&D to make back and want to make a decent profit on top of that.
 

TESKATLIPOKA

Platinum Member
May 1, 2020
2,627
3,193
136
AMD mobile GPUs are dead, and desktop "OEM" cards are usually by some AIB.
Actually, with RDNA4 they could have won some OEM design wins.
Although Nvidia is better positioned there by having GB206(128-bit), GB205(192-bit) and GB203(256-bit) GPUs.
Still, performance wise It would compete well with everything except 5090 Laptop.

As for Strix Halo, It's simply too expensive for what It offers to gamers. At least with RDNA4 It could have been a more interesting product, but AMD chose to use RDNA3.5 instead.
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |