john_AI is all the rage now. AMD pushed mobile AI CPUs faster (already here this month) while it delays GPUs where it doesn't have any real success and that makes total sense. Not just because of the hardware itself, but also because of the hostile nature of the GPU market towards AMD, where everyone is praising/choosing Nvidia, while trouncing AMD.

Su has shown from it's first days at AMD that she will cancel, put on hold or at least downgrade it to a non priority any product line that doesn't bring money to the company. If consumers and press only want Nvidia, that's what they will get.

The only positive thing Su has done was that she was brave enough to greenlight Zen. But she has taken no risks since then, and that's why AMD has stagnated and even regressing today. She needs to take a risk or two again.

stimpy88The only positive thing Su has done is she was brave enough to greenlight Zen. She has taken no risks since then, and that's why AMD has stagnated and even regressing today. She needs to take a risk or two again.

Yeah, moving RDNA 3 onto a chiplet design was not a risk at all, and neither was the dual-issue shader core. ;) /s

Radeon is regressing because RDNA 2 was good, and RDNA 3 offers nothing extra on top, if you don't count the extra performance of the 7900 series. The aforementioned chiplet design and dual-issue shaders didn't work out as well as planned. As for Ryzen, I see no regression.

AusWolfYeah, moving RDNA 3 onto a chiplet design was not a risk at all, and neither was the dual-issue shader core. ;) /s

Radeon is regressing because RDNA 2 was good, and RDNA 3 offers nothing extra on top, if you don't count the extra performance of the 7900 series. As for Ryzen, I see no regression.

It's not much of a risk when you overprice the hardware so much that the extra $10 per chip is payed back with an MSRP higher than the sky.

The real risk should have been to delay RDNA 3 and fix the dogcrap they knew it was. Now Su has caused irreparable damage to the Radeon brand. Gamers already had issues trusting the brand to deliver, but this finished it off. And this clamouring surrounding the "amazing" "new" RDNA 4 chips is BS too, as all it is, is a bugfix release of what RDNA 3 was supposed to be in the first place.

RDNA 5 better be amazing, or Radeon will be dead.

stimpy88It's not much of a risk when you overprice the hardware so much that that extra $10 per chip is payed back with an MSRP higher than the sky.

What's overpriced? Almost every tier of RDNA 3 GPU comes at a lower price than the competing Nvidia product. Only the 7600 is overpriced for what it is, but so is the 4060, so neither company gets a point there.

Edit: I'm not gonna respond to the "boo hoo Radeon is dead" alarmist BS the rest of your post is. It takes far more than a single mediocre generation to kill a company like AMD.

stimpy88The only positive thing Su has done was that she was brave enough to greenlight Zen. But she has taken no risks since then, and that's why AMD has stagnated and even regressing today. She needs to take a risk or two again.

You see glass half empty, I see it half full, meaning, I wouldn't use the phrase "the only positive" but "the only negative". I am also pointing at that situation where AMD is reluctant to take risks in manufacturing, try to get more wafers from TSMC and flood the market when they have the superior product. Looking at AMD's financial reports for each quarter, the last few years, being spot on with their predictions at the start of the quarter, it gives the idea of them ordering exactly from TSMC as much as they know they will sell, meaning they are losing all the opportunities to become a bigger company by offering higher quantities of products in the market. They are avoiding to start price wars with Nvidia and Intel, but that also means that they can't get market share fast enough in CPUs or any in GPUs. On the other hand Su did took a huge risk spending dozens of billions for Xilinx and that might start paying out now with the AI craze.
As for Zen. Well she didn't had too many products back then to choose from. Greenlighting Zen wasn't a brave step or a risk, it was just AMD's last and only chance.

stimpy88And this clamouring surrounding the "amazing" "new" RDNA 4 chips is BS too, as all it is, is a bugfix release of what RDNA 3 was supposed to be in the first place.

It's probably what they should have done with RDNA3. Focus on Raytracing performance because that's where Nvidia is driving the market and the whole marketing narrative. RT performance. And it's probably what they are focusing now, RT performance, because probably SONY asked for it.

john_It's probably what they should have done with RDNA3. Focus on Raytracing performance because that's where Nvidia is driving the market and the whole marketing narrative. RT performance. And it's probably what they are focusing now, RT performance, because probably SONY asked for it.

RDNA 3 has essentially the same RT cores as RDNA 2. RDNA 4 will get an update because it is now that AMD is ready with it. If they wanted to include the updates with RDNA 3, then it would still be in development right now, while the rest of the architecture has been ready for a long time. It wouldn't have made sense.

AusWolfRDNA 3 has essentially the same RT cores as RDNA 2.

I know.

AusWolfRDNA 4 will get an update because it is now that AMD is ready with it.

I think I read somewhere that SONY asked for higher RT performance and AMD can't say no to SONY. If this is true, "Thank You SONY".

AusWolfIf they wanted to include the updates with RDNA 3, then it would still be in development right now, while the rest of the architecture has been ready for a long time. It wouldn't have made sense.

They where not focusing on RT when developing RDNA3. They thought that Navi 31 was big enough to offer that extra performance for the time being. They thought that raster was still enough. While it is for gaming, it's pathetic for marketing and eventually sales.

john_I know.

I think I read somewhere that SONY asked for higher RT performance and AMD can't say no to SONY. If this is true, "Thank You SONY".

They where not focusing on RT when developing RDNA3. They thought that Navi 31 was big enough to offer that extra performance for the time being. They thought that raster was still enough. While it is for gaming, it's pathetic for marketing and eventually sales.

It was high time for AMD to do something with RT regardless of Sony. If they thought they could get away with using the same cores for RDNA 3 that they did for RDNA 2 because the 7900 XTX can brute force its better performance, the same can't be said about RDNA 4 which probably won't reach the performance of the 7900 XTX in any case.

AusWolfIt was high time for AMD to do something with RT regardless of Sony. If they thought they could get away with using the same cores for RDNA 3 that they did for RDNA 2 because the 7900 XTX can brute force its better performance, the same can't be said about RDNA 4 which probably won't reach the performance of the 7900 XTX in any case.

And we seen that brute forcing RT performance out of a bigger more expensive chip, didn't worked. When 7900XT/X came out i was screaming about RT performance. They where calling me "Nvidia shill" back then. But we have seen that consumers go where marketing tells them to go and Nvidia managed to make everyone talking about RT and DLSS.

john_And we seen that brute forcing RT performance out of a bigger more expensive chip, didn't worked. When 7900XT/X came out i was screaming about RT performance. They where calling me "Nvidia shill" back then. But we have seen that consumers go where marketing tells them to go and Nvidia managed to make everyone talking about RT and DLSS.

Yup. Besides, Nvidia has also been taking the brute force approach. The RT-to-raster performance ratio hasn't changed much since Turing, but for some reason, the masses are pissing their pants with joy looking at Ada, and I keep wondering whether it's the sheer power of marketing, or if they really see something that I don't.

Translated by Google (Moderator edit)

This information that AMD will no longer offer graphics cards for the high-end sector is really sad. And it only strengthens my desire to push forward with my old project. Namely the comparison of AMD and Nvidia graphics cards in the enthusiast sector. Of course, graphics card prices are really very expensive at the moment. But of course that is primarily thanks to Nvidia. It was probably predictable that AMD would also join in. Nevertheless, it would have been smarter to reduce the prices by €100 or €150 compared to the Nvidia models. I agree that the two high-end graphics cards from AMD, the 7900 XT and XTX, should have been offered at a lower price when they were introduced. But they have now fallen again. Which of course makes them more attractive compared to the Nvidia graphics cards. You can now get a 7900 XT for around €780. And a 7900 XTX for less than €1000. This of course makes it much more attractive compared to the direct competition from the green camp. In this case, that would be the Nvidia RTX 4080 TI Super and the RTX 4080 Super.

SailorMan-PTThis information that AMD will no longer offer graphics cards for the high-end sector is really sad.

And it encourages me even more in my endeavor to move forward with my old project. Namely the comparison of AMD and Nvidia graphics cards in the enthusiast sector. It's clear that graphics card prices are really expensive at the moment. But this is of course primarily thanks to Nvidia. It was probably foreseeable that AMD would also join in.

Nevertheless, it would have been more intelligent to lower the prices by €100 or €150 compared to the Nvidia models.

I agree that the two high-end graphics cards from AMD, 7900 XT and XTX, should have been offered at a lower price when they were launched. However, they have also dropped again at the moment. Which of course makes them more attractive than the Nvidia graphics cards.

You can now get a 7900 XT for around €780. And a 7900 XTX for less than €1000. This of course makes them far more attractive compared to the direct competition from the green camp. In this case, that would be the Nvidia RTX 4080 TI Super and the RTX 4080 Super.

I think it's more a matter of pride than anything else. It's better to have a great midranger than a steaming heap of a "flagship" that is only good at getting pounded by its direct competition, struggling to show any kind of benefit against the second tier of product from the other brand. If you look at the share numbers, the market doesn't consider any of AMD's cards attractive at the moment, even though the 7900 XTX has been selling rather well - about half as much as the worst-selling NVIDIA card this generation (the original 4080).

Dr. DroGreat news.

:nutkick:

Dr. DroThe longer AMD keeps this in the oven, the better it will be

The longer AMD keeps this in the oven, the less sales it will get.

ARF:nutkick:



The longer AMD keeps this in the oven, the less sales it will get.

Nonsense. Polaris cards are almost a decade old, completely out of support and they still sell. The market has made it clear through the 4090 that it is interested in quality, well supported products and that it is willing to pay for that.

Dr. DroNonsense.

:kookoo:

Dr. DroPolaris cards are almost a decade old, completely out of support and they still sell

Navi 33 draws 20 watts less and is more than double as fast.
Buy Radeon RX 7600, if you have no clue what to buy :D

My understanding of the strategy for RT in RDNA 3 was that since AMD have a portion of the RT job done on compute units, having dual issues units and doubling should have increased the overall performance for RT while also improving the overall performance.

But the thing is high end compute performance of AMD GPU is somehow in the same ball park than Nvidia. Their strategy of leveraging compute units for a portion of the work would work if you have way more compute power than your opponent, but it set to fail if you don't have massively more compute power than your opponent.

If RNDA 3 had 192 CU instead of 96, it would have been plenty to perform well against Nvidia in RT while clearly beating Nvidia outside RT. But that chips is probably not doable. Nvidia strategy paid of since they can dedicate a small portion of these chips to RT units that are totally useless outside RT but kicks ass when there is a RT load.

While the RT cores of a nvidia card process the RT workload, the compute units just continue what they were doing. On AMD they have to switch between compute workload and RT workload.

Good to see that AMD is realizing that their strategy is a dead end in a era of very large silicon GPU.

We will see for the release date but later than Q4 would be a miss of opportunities.

ARFBuy Radeon RX 7600, if you have no clue what to buy :D

The RTX 4060 is a better card.

Dr. DroThe market has made it clear through the 4090 that it is interested in quality, well supported products and that it is willing to pay for that.

Do you seriously believe that are the reasons for the 4090's uptake?
Nothing to do with consumers mindset of wanting the latest and greatest despite costs, mainly for bragging rights and Epeen?
Where's the "quality" when Northridgefix stated he's repairing 20 a week due to faults and Nvidia is denying they should have done a recall .

Caring1Do you seriously believe that are the reasons for the 4090's uptake?
Nothing to do with consumers mindset of wanting the latest and greatest despite costs, mainly for bragging rights and Epeen?
Where's the "quality" when Northridgefix stated he's repairing 20 a week due to faults and Nvidia is denying they should have done a recall .

Repeat that enough times and you may come to believe that yourself. People at large like money and performance per dollar far more than bragging rights, there's a reason the xx60-tier cards are historically the most popular amongst Steam users.

Mind you, it's weird I have to point this out but like, the folks behind NorthridgeFix, NorthwestRepair and similar channels, they may be brilliant technicians, but the key takeaway is that they're still YouTubers and as such, they are after views. It gets people to click when they make thumbnails with insane statements like "The 4090 has a chronic problem AND YOUR CARD WILL DIE!", the only issue with that statement is that it's hyperbole at best and a lie at worst. But it gets you to click on their video, which to their defense, is usually very informative and high quality - the important factor is that it is good for their business (both the channel and new customers for the repair shop). See it for what it is, entertainment value first, informational value second.

Dr. DroMind you, it's weird I have to point this out but like, the folks behind NorthridgeFix, NorthwestRepair and similar channels, they may be brilliant technicians, but the key takeaway is that they're still YouTubers and as such, they are after views. It gets people to click when they make thumbnails with insane statements like "The 4090 has a chronic problem AND YOUR CARD WILL DIE!",

And yet they have NEVER resorted to claims like that. They do their job and people watch, they don't go out of their way to sensationalise anything.

Caring1And yet they have NEVER resorted to claims like that. They do their job and people watch, they don't go out of their way to sensationalise anything.

Sigh. I shouldn't need to literally go on YT type in their names to give you the search results, but alas. Again, they aren't phony, their videos are entertaining and of high informational value, but you can't deny the thumbnail drama isn't there. There's nothing wrong with that, it's just a smart way to get people to click on the video. Things like "50% of all 4090 will end up in trash", "top reason not to buy AMD 7900 series GPU", "three melted 4090 melted connectors. This is toxic" (lol at the gas mask at the thumbnail of that one) and "4090 connectors are melting like chocolate", it's really just clickbait drama to draw in people to watch an actually good video. I'm not even gonna post the images because you can go there and look those up yourself

Dr. DroI think it's more a matter of pride than anything else. It's better to have a great mid-range device than a steamy bunch of "flagship" that's only good at getting thrashed by its direct competition and making it hard for itself to show any advantage over the other brand's second class of products. Looking at the expansion numbers, the market doesn't currently hold any AMD cards for itself, although the 7900 XTX has sold pretty well – about half as well as the worst-selling NVIDIA card of this generation (the original 4080).

Well, these are competitors and pure facts. If there were more mature ones in this community of gamers faction. Could you also show the kiddies out there alternatives. Maybe the AMD Radeon cards would sell a little better.

But most of them are still very young. And therefore also very receptive to advertising. NVIDIA's marketing department knows this all too well. Constant sensory overload of ray tracing benefits on NVIDIA graphics cards.

They keep up the hype and keep putting new wood in the stove. This does not extinguish the flame of enthusiasm. These people don't even bother to look for alternatives.

And in the end, about 90% of the studios are also approached and paid by NVIDIA. This will optimize their products for NVIDIA cards. But at a few studios, games are also optimized for AMD hardware. The best example here is FarCry 6.

There are various performance tests on the TechPowerUp page. There I did a test where I make a comparison with XTX. In the game FarCry 6 you are ahead. Neither the RTX 4090 nor the Radeon card could be included. Very impressive. If only there were many more studios, AMD hardware would be better supported. Ryzen CPUs and Radeon GPUs. Maybe if the proportion was about 30 or 35%.

AMD cannot afford to pay many studios. But the green camp, yes.

SailorMan-PTWell, these are competitors, and pure facts. If there were older mentors in this community of NVIDIA disciples who would also show them alternatives. AMD Radeon cards would probably sell a little better.

But most of them are still very young. And therefore very receptive to advertising. NVIDIA's marketing department knows this only too well. Constant overstimulation of the benefits of ray tracing on NVIDIA graphics cards.

And in the end, around 90% of studios are also approached and paid by NVIDIA. This means that their products are optimized for NVIDIA cards. But a few studios also optimize games for AMD hardware. The best example of this is FarCry 6.

There are various performance tests on the TechPowerUp website. I did a test there in which I made a comparison with XTX. In the game FarCry 6 you are ahead. Neither the RTX 4090 nor the Radeon card could be included. Very impressive . If only there were a lot more studios, AMD hardware would be better supported. Ryzen CPUs and Radeon GPUs. Maybe if the share was around 30 or 35%.

AMD can't afford to pay a lot of studios. But the green camp can.

I believe the days of inserting developers from a graphics vendor on a game's development team are mostly past us - AMD, NVIDIA and Intel all work closely with Microsoft and engine vendors to better support their hardware and software configurations. However, AMD has a bad habit with poor documentation (at least that which is available to the public or even developers themselves), which cause incidents like the Anti-Lag+ VAC bans that Counter-Strike players got hit with some time ago. NVIDIA on the other hand, extensively writes documentation and makes these available to developers, writing, maintaining and updating APIs that give them control over said features. Not to mention that they also offer driver support for extended periods of time, the GTX 900 series are still being updated to this day and these cards are really, really old by now, to the tune of around a decade. Things like this reflect on the end result.

The 7900 series' redemption is their current market price. But they aren't cheap to manufacture, and AMD is selling them with much lower margins than NVIDIA can get on their cards. It's pretty obvious that Navi 31 was intended to compete with AD102, and that never worked out from day one. The current standoff between the 7900 XTX and the 4080 (and 4080 Super) results in a minor victory for AMD in raster performance, with a significant regression in ray tracing performance and power consumption, all while being a much larger and much more complex GPU in itself, that should, on paper, be better at practically everything. But it isn't, nor is the user experience that said card can provide.

Reducing the situation down to "NVIDIA cards sell better because their buyers are young and influenced by marketing" is not the right way to see this.

Dr. DroSigh. I shouldn't need to literally go on YT type in their names to give you the search results, but alas. Again, they aren't phony, their videos are entertaining and of high informational value, but you can't deny the thumbnail drama isn't there. There's nothing wrong with that, it's just a smart way to get people to click on the video. Things like "50% of all 4090 will end up in trash", "top reason not to buy AMD 7900 series GPU", "three melted 4090 melted connectors. This is toxic" (lol at the gas mask at the thumbnail of that one) and "4090 connectors are melting like chocolate", it's really just clickbait drama to draw in people to watch an actually good video. I'm not even gonna post the images because you can go there and look those up yourself

So where's the lie then, everything they said is in their videos. sigh.

Dr. DroRepeat that enough times and you may come to believe that yourself. People at large like money and performance per dollar far more than bragging rights, there's a reason the xx60-tier cards are historically the most popular amongst Steam users.

The 4090 is sold by marketing and reviews calling it the fastest card available. Its performance per dollar is nowhere to be seen compared to literally anything else. It's the worst value Nvidia has ever shoved down on people's throats. So yeah, bragging rights.



The x50, x60 and x70 series are historically popular because they've historically been on the top of this chart, not on the bottom, within a stone's throw to the 4090.