News AMD deprioritizing flagship gaming GPUs: Jack Hyunh talks new strategy against Nvidia in gaming market

[Bamda]

AMD is not accountable to the consumer, which is the nature of an open market. I believe AMD is doing the best they can, and it's hard to fault them for pursuing strategies that work in their favor.

 

[n810]

I've seen small business such as coffee shop / cafes decide their last hour has few customers, so they start to close early, thinking that's better. What happens then is the last hour then becomes a dead hour as well... I've seen business trying to save on expenses, close themselves this way.

No you cannot win sacrificing the top end in a moving field. In fact the only place they should develop is at the top end - eventually the top end becomes middle, and middle becomes low. That is how it goes... if they aim for the middle, they'll hit the low end ... no one will want it, and they'll be out of business, and rightfully so.

The problems they have is that they have already abandoned top end. Like for example the 7000 series doesn't have multi GPU support in the drivers on Windows, like 6000 had....

The developers go where it's interesting, let us remember,
this famous quote from Alexander Hamilton ( the US founder )
"Power over a man's subsistence, is power over his will."
The developers go where they're offered MONEY, that is how it works. Perhaps the execs could pay more developers as an investment?
It would also help to sponsor developers who want to port things - like for example, if someone has a project that is popular - give them some hardware, especially top-end stuff, so they'll support it.

That is how it works, and they've clearly missed that point, and chosen failure.

 

[jlake3]

No you cannot win sacrificing the top end in a moving field. In fact the only place they should develop is at the top end - eventually the top end becomes middle, and middle becomes low. That is how it goes... if they aim for the middle, they'll hit the low end ... no one will want it, and they'll be out of business, and rightfully so.

That’s not how GPUs work. The Radeon VII didn’t trickle down the product stack, nor did the 1080Ti. They got cancelled and disappeared from shelves once better architectures were designed and replaced with products that achieved that performance level at lower manufacturing cost and power.

If AMD doesn’t develop a flagship that goes toe-to-toe with the 5090 today, and that level of compute power eventually trickles down to the mainstream… the 50 and 8000-series will both be off store shelves by the time that happens, and whatever AMD is selling in the midrange at the time wouldn’t use the same core or the same node as the hypothetical 8950XTX they didn’t develop.

The problems they have is that they have already abandoned top end. Like for example the 7000 series doesn't have multi GPU support in the drivers on Windows, like 6000 had....

SLI and Crossfire have both been dead for a while due to poor developer support, poor frame time consistency, and low uptake. DX12 has a way to do multi-GPU rendering at the graphics API level, but no one implements it for more than tech demos.

 

[Papusan]

AMD is not accountable to the consumer, which is the nature of an open market. I believe AMD is doing the best they can, and it's hard to fault them for pursuing strategies that work in their favor.

No one know for sure if that will work out for them. Not even so sure they know it themself. They may fail in this strategy and go back. And that won't come for cheap.

 

[SyCoREAPER]

I'm more interested in rumors of Qualcomm trying to buy Intels CPU division.

Power Efficient and Powerful CPU from them coupled with affordable cards from AMD would be a win for consumers, not so much AMD though since they have a CPU division.

I forsee if there purchase ever went through, the AMD will lock features of their GPUs to their CPUs. Hell they might even do that moving forward to boost revenue from no high end cards. Add "exclusive features with an all AMD system".

 

[King_V]

If anything, 7900XTX's weak performance vs 4090 did more to hurt the Radeon brand than to help it. If you can't win a fight, it's better to avoid the fight in the first place, than let the world know you're a loser.

So, 83% (at 4k ultra) to 97% (at 1080p medium) of the 4090's performance, with less than 80% of the 4090's power draw, and at about 50% of the 4090's price, is considered "weak performance?"

 

[SyCoREAPER]

So, 83% (at 4k ultra) to 97% (at 1080p medium) of the 4090's performance, with less than 80% of the 4090's power draw, and at about 50% of the 4090's price, is considered "weak performance?"

Performance is relative and what they are going by precisely and if it's a controlled lab setting or real world, achievable numbers.

AMD hyped up this/last gens GPUs to destroy Nvidia and while they performed admirably, they weren't close to an Nvidia killer.

So it all depends on interpretation and claims. This happens every time something is announced (not referring to you but companies).

The hype train or promise train, whatever it may be gets fed fire, the community speculates and it isn't what we expected. I wish we'd go back to announcements of actual products with actual results and real world outcomes. Same goes for games.

All it leads to is as mentioned speculation and the community getting fired up for good or worse without concrete evidence of anything.

The day a product is released and reviewed by reputable non-Shills (like GN and DB).

 

[AndrewJacksonZA]

This entire thing seems to me, right now, to be "data centre first, and consumers get whatever's left."

CEO Lisa Su says AMD is now a data center-first company — DC is now 4X larger than its gaming sales​

https://www.tomshardware.com/tech-i...over-4x-higher-than-its-gaming-business-sales

 

[ballsysteml0rd]

It makes sense for AMD to do reasonably priced GPUs with reasonable performance. But can they? So far, all I've heard is that iGPUs will take that lower end market.

 

[ballsysteml0rd]

As an example of what I'm talking about, here's a link from when the RX580 fell to $160: https://www.pcgamer.com/this-overclocked-radeon-rx-580-8gb-is-the-cheapest-around-at-dollar160/
And I saw the 4GB model fall to $100 during that time.

Then they sold like hot-cakes. But now I don't see AMD trying to go for such razor thin margins.

 

[ballsysteml0rd]

And while we are on the topic, whatever happened to AMD's claim that they power more gaming than any other option? Like, they have PS, XBox , some handhelds, and part of the PC and Laptop space.

If that's not enough market share to convince devs than what is?

 

[King_V]

Performance is relative and what they are going by precisely and if it's a controlled lab setting or real world, achievable numbers.

AMD hyped up this/last gens GPUs to destroy Nvidia and while they performed admirably, they weren't close to an Nvidia killer.

So it all depends on interpretation and claims. This happens every time something is announced (not referring to you but companies).

The hype train or promise train, whatever it may be gets fed fire, the community speculates and it isn't what we expected. I wish we'd go back to announcements of actual products with actual results and real world outcomes. Same goes for games.

All it leads to is as mentioned speculation and the community getting fired up for good or worse without concrete evidence of anything.

The day a product is released and reviewed by reputable non-Shills (like GN and DB).

Precisely. Hype is hype. Advertising.

The numbers I pulled are straight off the GPU hierarchy here.

I guess if AMD wanted to, they could make a bigger, more power hungry card to compete with or maybe even outperform the 4090, probably without even having to go as high as the 4090's price. But, why bother? How many would they make? How many would they sell?

I don't have numbers, but are 4090 sales contributing much to Nvidia's bottom line?

The 7900 XTX edges out the 4080 Super. The 4090 is really more of a Titan/Titan V type of card than a regular product. It performs amazingly, but is stupidly expensive and, with quick back-of-the-napkin math for the Titan RTX and 4090, stupidly inefficient.

But, somehow, failure to outperform Nvidia's low-volume halo product makes AMD a failure. Because of reasons.

 

[-Fran-]

Precisely. Hype is hype. Advertising.

The numbers I pulled are straight off the GPU hierarchy here.

I guess if AMD wanted to, they could make a bigger, more power hungry card to compete with or maybe even outperform the 4090, probably without even having to go as high as the 4090's price. But, why bother? How many would they make? How many would they sell?

I don't have numbers, but are 4090 sales contributing much to Nvidia's bottom line?

The 7900 XTX edges out the 4080 Super. The 4090 is really more of a Titan/Titan V type of card than a regular product. It performs amazingly, but is stupidly expensive and, with quick back-of-the-napkin math for the Titan RTX and 4090, stupidly inefficient.

But, somehow, failure to outperform Nvidia's low-volume halo product makes AMD a failure. Because of reasons.

You're not wrong, but you have to admit AMD, for some bizarre reason, as a talent to sabotage itself when it counts. They've had several opportunities to get both market share and reputation by just being "decent" and fumble the ball hard:

- Vega -> Mostly crapshoot.
- RDNA1 - 5700 series launch -> Crapshoot.
- RDNA2 - 6K series -> very good, hence they go rep back; still sold it at a premium because reasons.
- RDNA3 - 7K series -> A lot of gremlins and AMD fumbling their internal testing numbers, much like with Zen5 recently. VR was not working at launch. Like... How can you screw that up?

I could go even further back just by memory alone, but 1 is an isolated event. 2 start a trend. AMD is just a constant at this point and that is their biggest problem.

They've improved for sure over the years on areas that matter, but for some reason regress at times in those same areas.

This is good for them as well, because less SKUs will or should allow them to laser-focus on what (should) matter to consumers: it just works out of the box with no hassle in all aspects.

Regards.

 

[thestryker]

- RDNA1 - 5700 series launch -> Crapshoot.
- RDNA2 - 6K series -> very good, hence they go rep back; still sold it at a premium because reasons.
- RDNA3 - 7K series -> A lot of gremlins and AMD fumbling their internal testing numbers, much like with Zen5 recently. VR was not working at launch. Like... How can you screw that up?

...

This is good for them as well, because less SKUs will or should allow them to laser-focus on what (should) matter to consumers: it just works out of the box with no hassle in all aspects.

Regards.

While the random issues don't help I don't think they particularly hurt anywhere near as much as the pricing has. I'm not sure fewer SKUs will help with regards to launch type issues because time and QA are what solves those. It is still important to have a smooth launch though of course.

Even if we just look at RDNA they've had good products, but not priced to gain market (pricing all launch pricing):
RX 5700 slightly faster than RTX 2060 both at $349
RX 5700 XT slightly faster than RTX 2060 Super both at $399

RX 6700 XT $479 faster than RTX 3060 Ti $399
RX 6800 $579 faster than RTX 3070 $499
RX 6800 XT $649 slightly slower than RTX 3080 $699 (much slower in RT but equal/slightly faster in raster)
RX 6900 XT $999 slightly slower than RTX 3090 $1499 (worse RT performance than RTX 3080)

RX 7900 XT $899 splits the difference between RTX 4070 Ti $799 and RTX 4080 $1199 (RT performance slightly faster than RTX 4070 $599)
RX 7900 XTX $999 slightly faster than RTX 4080 $1199 (RT performance slightly slower than RTX 4070 Ti $799)
RX 7600 $279 slightly slower than RTX 4060 (RT performance over 20% slower, but RT less relevant at this perf level)

AMD's products aren't necessarily poorly priced if they were in a competitive market, but they're not in one so they shouldn't be priced that way. If they knock 20% off the price of everything I've listed AMD is probably sitting at 30%+ of the market no matter what other issues they may have had at launch. So if this new strategy keeps the die size down and they're able to drop the pricing at each performance tier they bring to the market it has a good chance of working.

 

[-Fran-]

While the random issues don't help I don't think they particularly hurt anywhere near as much as the pricing has. I'm not sure fewer SKUs will help with regards to launch type issues because time and QA are what solves those. It is still important to have a smooth launch though of course.

Even if we just look at RDNA they've had good products, but not priced to gain market (pricing all launch pricing):
RX 5700 slightly faster than RTX 2060 both at $349
RX 5700 XT slightly faster than RTX 2060 Super both at $399

RX 6700 XT $479 faster than RTX 3060 Ti $399
RX 6800 $579 faster than RTX 3070 $499
RX 6800 XT $649 slightly slower than RTX 3080 $699 (much slower in RT but equal/slightly faster in raster)
RX 6900 XT $999 slightly slower than RTX 3090 $1499 (worse RT performance than RTX 3080)

RX 7900 XT $899 splits the difference between RTX 4070 Ti $799 and RTX 4080 $1199 (RT performance slightly faster than RTX 4070 $599)
RX 7900 XTX $999 slightly faster than RTX 4080 $1199 (RT performance slightly slower than RTX 4070 Ti $799)
RX 7600 $279 slightly slower than RTX 4060 (RT performance over 20% slower, but RT less relevant at this perf level)

AMD's products aren't necessarily poorly priced if they were in a competitive market, but they're not in one so they shouldn't be priced that way. If they knock 20% off the price of everything I've listed AMD is probably sitting at 30%+ of the market no matter what other issues they may have had at launch. So if this new strategy keeps the die size down and they're able to drop the pricing at each performance tier they bring to the market it has a good chance of working.

I didn't get too much into pricing, because perception of "value" varies wildly from person to person.

You can have many objectives metrics on where one is better than the other, but different knobs for different folks. I do believe price is a reflection of where AMD thought the cards would be at launch, but that is also part of the reason why they sabotage themselves. Thing is: if they don't build the base confidence, they can have the better product and no one will buy it anyway. Which has happened in the past for whatever reason.

EDIT: Just to give you my own example of what knobs I look for in a GPU:
1- VRAM amount for longevity and VR games and Downsampling (higher internal res and resizing for quality).
2- Great raster performance under ~$500 (my cap used to be $300, but here we are).
3- AV1 support (now and going forward) or general competent encoder outside of H264.
4- Power under 300W.
5- Monitor support. This is not trivial as nVidia has a somewhat crappy track record with monitor support. I like my displays and arranging them in weird ways. AMD has always delivered there, consistently, which is their biggest saving grace for me. Or at least, the monitors and TVs I've owned and used have had zero problems or weirdness. My laptop just works with any crappy TV or monitor I hook into it.

I have zero interest in RT performance and upscaling technologies due to my use cases. AMD is the better pick in most price segments, except 4090. Unfortunately for nVidia, I'm not swapping 2 months of my mortgage for a 4090.

Regards.

 

[thestryker]

I didn't get too much into pricing, because perception of "value" varies wildly from person to person.

You can have many objectives metrics on where one is better than the other, but different knobs for different folks. I do believe price is a reflection of where AMD thought the cards would be at launch, but that is also part of the reason why they sabotage themselves. Thing is: if they don't build the base confidence, they can have the better product and no one will buy it anyway. Which has happened in the past for whatever reason.

None of this matters in the greater market when you're so far down compared to the competition. For enthusiasts of course people will make a well informed decision based on what will work best for them, but that's not reflective of the market as a whole. AMD didn't start to regain market share in CPUs by providing a better product with better experience simply a cheaper product that was good enough. This meant that by the time they released a better product they could charge a premium (Zen 3 until ADL came out) and continue to gain market share.

 

[SyCoREAPER]

None of this matters in the greater market when you're so far down compared to the competition. For enthusiasts of course people will make a well informed decision based on what will work best for them, but that's not reflective of the market as a whole. AMD didn't start to regain market share in CPUs by providing a better product with better experience simply a cheaper product that was good enough. This meant that by the time they released a better product they could charge a premium (Zen 3 until ADL came out) and continue to gain market share.

AMD didn't become "better" only at Zen 3, a threat to Intel yes, but not better. That happened way before Zen 3. AMD used to be the poor mans Intel and seen as an inferior chip but that's going back to what 2 decades now?

While marketshare didnt magically pickup, popularity even if not reflected by sales did. As said above in another post, AMD does consistently sabotage itself and has had multiple opportunities to kick Intel and Nvidia while they were face down in the dirt**. They chose not to cease the opportunity but instead be complacent with where they were and play it safe.

**I am not a fanboy. Being a shill for a company that owes you nothing is idiotic. That said I I own Intel, Nvidia and AMD and like each for their own reasons so I'm not picking sides, just giving my opinion objectively.

 

[-Fran-]

None of this matters in the greater market when you're so far down compared to the competition. For enthusiasts of course people will make a well informed decision based on what will work best for them, but that's not reflective of the market as a whole. AMD didn't start to regain market share in CPUs by providing a better product with better experience simply a cheaper product that was good enough. This meant that by the time they released a better product they could charge a premium (Zen 3 until ADL came out) and continue to gain market share.

I can agree it's less relevant if you go further into the "less savvy consumer" spectrum and that is where perception is everything and AMD has little of it. They haven't been able to make a dent because nVidia is king due to completely unrelated reasons to "value". They're in the news now a lot, they have the top dog product and people just blindly recommends nVidia because of those things mainly. My point is: even if AMD was indeed better, nVidia still sells more. Intel now has been in the news for other reasons and AMD has encroached the enthusiast mind a tad more, so that has trickled down to the masses a bit. Plus now you see more AMD CPUs around in your BestBuys and other "brick and mortar" shops where people sees what is around.

It's definitely not one single issue AMD has, but for the masses is just having the name in ther brains as a valid alternative and that is won via attacking the mainstream segment.

Regards.

 

[ballsysteml0rd]

While the random issues don't help I don't think they particularly hurt anywhere near as much as the pricing has. I'm not sure fewer SKUs will help with regards to launch type issues because time and QA are what solves those. It is still important to have a smooth launch though of course.

Even if we just look at RDNA they've had good products, but not priced to gain market (pricing all launch pricing):
RX 5700 slightly faster than RTX 2060 both at $349
RX 5700 XT slightly faster than RTX 2060 Super both at $399

RX 6700 XT $479 faster than RTX 3060 Ti $399
RX 6800 $579 faster than RTX 3070 $499
RX 6800 XT $649 slightly slower than RTX 3080 $699 (much slower in RT but equal/slightly faster in raster)
RX 6900 XT $999 slightly slower than RTX 3090 $1499 (worse RT performance than RTX 3080)

RX 7900 XT $899 splits the difference between RTX 4070 Ti $799 and RTX 4080 $1199 (RT performance slightly faster than RTX 4070 $599)
RX 7900 XTX $999 slightly faster than RTX 4080 $1199 (RT performance slightly slower than RTX 4070 Ti $799)
RX 7600 $279 slightly slower than RTX 4060 (RT performance over 20% slower, but RT less relevant at this perf level)

AMD's products aren't necessarily poorly priced if they were in a competitive market, but they're not in one so they shouldn't be priced that way. If they knock 20% off the price of everything I've listed AMD is probably sitting at 30%+ of the market no matter what other issues they may have had at launch. So if this new strategy keeps the die size down and they're able to drop the pricing at each performance tier they bring to the market it has a good chance of working.

My point exactly. Why does AMD say they want to grab more market yet price their GPUs so high? Those two goals are antithetical to each other.

 

[udance4ever]

as someone who isnt about flagship GPUs, I think their strategy to capture the mini/palm PC & PC handheld markets is perfect - this is the future & akin to the strategy Apple took with Apple Silicon to maximize phone & laptop battery life (yet 2 other key segments) & still be "performant." PC handhelds is exactly where I see AMD's dominance (just look at the mass popularity ROG Ally & Legion Go & the reemergence of handheld gaming in general). Personally, these are the vectors I needed to get my feet wet in Windows gaming after a long 20 year absence. AMD has their share of work to do to keep reducing the power consumption of the x86 architecture when it was never inherently designed for low power & portability. AMD will surely figure it out & be king of the hill here!

 

[Bomblibu]

Seems that lot of commentators missed this part of the article:

"So the question I ask is, the PlayStation 5, do you think that’s hurting us? It’s $499. So, I ask, is it fun to go King of the Hill? Again, I'm looking for scale. Because when we get scale, then I bring developers with us.

So, my number one priority right now is to build scale, to get us to 40 to 50 percent of the market faster. Do I want to go after 10% of the TAM [Total Addressable Market] or 80%? I’m an 80% kind of guy because I don’t want AMD to be the company that only people who can afford Porsches and Ferraris can buy. We want to build gaming systems for millions of users.

Yes, we will have great, great, great products. But we tried that strategy [King of the Hill] — it hasn't really grown. ATI has tried this King of the Hill strategy, and the market share has kind of been...the market share. I want to build the best products at the right system price point. So, think about price point-wise; we’ll have leadership."

Everything is about pricing!

 

[TheHerald]

So, 83% (at 4k ultra) to 97% (at 1080p medium) of the 4090's performance, with less than 80% of the 4090's power draw, and at about 50% of the 4090's price, is considered "weak performance?"

Your numbers are way off partner. Also 1080p is meaningless, both gpus are bottlenecked. The xtx has 80% raster and like 40% rt performance while consuming more power than the 4090. No, that's not weak, that's disastrous.

 

[King_V]

Your numbers are way off partner. Also 1080p is meaningless, both gpus are bottlenecked. The xtx has 80% raster and like 40% rt performance while consuming more power than the 4090. No, that's not weak, that's disastrous.

I'm getting my numbers from the GPU Rasterization Hierarchy table, and I don't really care for RT, which I didn't reference anyway. Where are you getting your numbers from?

 

[TheHerald]

I'm getting my numbers from the GPU Rasterization Hierarchy table, and I don't really care for RT, which I didn't reference anyway. Where are you getting your numbers from?

There is nowhere in that chart that has power draw. It just has TDP.. Also I don't get why you are comparing it to the 4090. Why not compare it to the 4080 or the 4080s? Suddenly your power draw and price arguments evaporate to thin air.

 

[King_V]

There is nowhere in that chart that has power draw. It just has TDP.. Also I don't get why you are comparing it to the 4090. Why not compare it to the 4080 or the 4080s? Suddenly your power draw and price arguments evaporate to thin air.

Because I was responding to a comment that stated that the 7900XTX had weak performance compared to the 4090, and lost the fight to the 4090. To which you then replied, continuing the comparison and denying the numbers I'm using.

If you think it should've been compared to the 4080 or 4080 Super, then you need to take it up with @baboma and ask them why they made that comparison in the first place.

To hit on other points:

• TDP is roughly max power draw. I don't know if anyone was planning on spending money on a 4090 in order to run games at 1080p with medium settings. They want that, presumably, to run high resolutions and high frame rates.
• The 7900 XTX slightly outperforms both the 4080 and 4080 Super, but uses slightly more power than them as well. But that's a completely separate issue.
  • Also, the 4080 Super came out over a year later.

And finally, I noticed that you didn't seem to touch on my numbers (aside from TDP) once I showed you where I got them, AND you didn't mention where you got your numbers.