Radeon Surrenders to GeForce 

Just wanna point out like some others have that the quote from Mr. Bergman at 8:20 is largely mistranslated. Gamer's Nexus reached out to AMD to get the original quote which is featured in their latest HW News video. The original quote was: "We're planning our next couple generations of products and we had to ask ourselves how high we can go. So obviously our competitor went to $1600 and 400 plus watts, so we could - graphics is scalabe - we could build a $2500 solution and be the highest performance at 600W or some other... At some junction, you just have to say we're going to draw a line on what we're targeting. So for this generation, we said $999." EDIT: As some people have pointed out, yes the point of the video still stands and the pricing target is nonsense since AMD seems to simply be following Nvidia, but the difference in phrasing here concerning the power target still matters.

Japanese here I want to address the translsion at 8:20 the correct translation would be such; "Technically, we could develop a GPU that can compete with theirs. But, when you introduce such a GPU, with 600W TDP and $1600, would the gaming market really welcome such a product-? We gave a hard thought into it and made a decision not to. " the part "was accepted by general PC gaming fans." from that translation got it completly wrong. the article was more that AMD thought the market wouldn't accept such a product.

Definitely no price fixing going on guys! Never investigate multinational corporations! Especially duopolies

While I agree their GPU-only sales are not profitable enough for the company to care, they still need RTG to keep the graphics part of their APUs up to speed. That's what gets them Sony and Microsoft for the consoles. That's also who develops the graphics portion of their new MI300 datacenter juggernaut. The accounting might not work in favor of the Radeon group, but they are still vital to the company's future.

The current state of affairs seems to boil down purely to a lack of wafer capacity at TSMC. As you mentioned; high-margin products are then preferred. It cannot be understated that the 7000-series graphics cards may be competing against the rest of AMD's portfolio, and that AMD is not even interested in actually selling them. Once wafer capacity becomes available, hopefully AMD will want to sell graphics cards again, and we will see more interesting offerings. I would rather bet on this than Intel shaking things up.

One thing that is overlooked in the analysis of AMD's largest customers is that they onboarded the Xilinx clients. This also includes MIcrosoft who has been pushing FPGA hardware in Azure. That does put who is that 16% customer into question. One other aspect of that is this is done in raw percentages but only to point out later in the video that AMD has doubled from 2020 to 2022. Again, part of that is from their recent acquisition of Xilinx. As for AMD going the route of a large die for a GPU, those times are effectively over. What we are likely to see is continuation and evolution of their chiplet strategy. Spin off PCIe + codec + display interfaces into their own chiplet as you only need one of them per package. A midrange die size (300 to 400 mm^2) as the base chiplet for compute and then scale upward in quantity to get the equivalent of a 600 mm^2 to 800+ mm^2 die. The memory controllers + cache have already been spin off into their own chiplets. It is possible to scale up GPU compute dies as demonstrated in Apple's M1 Ultra. The other side of chiplets is that you can develop fewer dies for chiplets and put them together for a wider range of products in a wider range of markets. I would be surprised if AMD continues separate development of RDNA and CDNA for much longer: they have already spun off much of the functional blocks used exclusively for graphics form the RDNA compute core. The consumer market is likely to get many of the chiplet dies first as a means to discover bugs/errata in them before the same chiplets are used in data center products. Worst case for AMD would be to delay the data center versions to release a new stepping. Lastly, AMD's graphics group isn't just about the PC GPU market. AMD is now in the business of licensing their IP and developing custom silicon for their customers (consoles). The PC market is the test bed of those technologies to demonstrate their capabilities. The PC board market is indeed getting smaller for them as a percentage of their overall business but it is an important stepping stone to retaining other businesses.

Imagine Radeon beating Nvidia flagship. People still will buy Nvidia anyways or I am mistaken?

GamersNExus reached out to AMD about that quote and it was a mistranslation. AMD provided GamersNexus with the original English "script" to that answer that then was translated to Japanese and then translated back to Egnlish for this article.

The 7900 XTX IMV was going to ge 1199, but then had to set it at 999 when they failed to hit their targets.

I had no idea there was such a major gulf in die size between 7900XTX and 4090! What a missed opportunity.

I thought this was a foregone conclusion! It's like they just don't want market share. They keep ignoring opportunities, not just flubbing them.

Great to see you posting more videos, Always enjoy your perspective on things.

I find this one a bit weird, as if it's either Radeon wins or Radeon is dead. With both RDNA 1 and RDNA 2 we saw the Radeon cards offering better value for raw performance, with Nvidia offering the top performers and an overall better feature set. This allowed me to get about 20% higher raw performance with Radeon than with Nvidia at my budget - with an RX 6600 over an RTX 3050. And this gen so far is weird pricing-wise, but at least the xtx continues that trend, and the xt is slowly falling in line by inching towards $800 retail. And yes, that is compared to Nvidia's 40's series which offers bad value compared to previous gen, and yes that means Radeon's 7000's series follows that trend in Nvidia's footsteps. But like you pointed out, they're playing second fiddle to Nvidia, content with just offering better raw performance with a worse feature set. And yes, they're not correcting Nvidia's pricing by pricing their own products more aggressively, but that doesn't mean that that relative value proposition isn't there. They're not the competition we hoped for as consumers, but at least it's there offering an alternative. I don't even know what 'giving up on Radeon' even is supposed to mean, I'll just upgrade when my system warrants an upgrade, and I hope I'll have more than just one brand to choose from. And I'll choose the card that I fancy most, regardless of which brand it is - though as I don't expect to get suddenly rich in the coming years, it'll probable be the better value Radeon card from whatever generation will be out by then.

I have sort of expected the same thing for a while. The reason I think AMD is in the PC gaming space is mostly to support consoles, hear me out. 1)The console developers are kept much healthier by sales of PC ports keeping the high cost of console game development subsidized by a bit more market cushionsecondary market. 2) Radeon knows they can always compete on price with nVidia and likely win the low to mid end with chiplets making more ROI. 3) AMD can use the PC platform as a huge console tech testing platform that still creates some revenue. 4) With a larger presence in the market than if they were just on consoles they can shape the market to support and introduce the technolgies and standards good for them and consoles.

Amd showing everyone that you either dir a hero or live long enough to see yourself become the villian. But not in a cool Batman way. They just straight out said fuck it, I don't wanna be the hero anymore, I got kids to feed

The thing I hated the most was seeing that news article where it was stated AMD did not want to compete versus 4090. That they didn't want to make a 450 watt or more graphics card that ran hot. I find it absolutely appalling. Whoever at AMD keeps saying no to competition needs to be fired. We could right now drop me a 7970 XTX that is a 600 watt monster and I would buy it. This idea that we don't want massive fast cards is bullshit. 4090s are selling like hotcakes and AMD is nowhere to be found. The consumers are screaming for it but whoever's saying no at AMD is ruining it. They seriously need to be fired. It's Raja Koduri all over again. Raja didn't want to make large GPUs. He wanted to keep the core count small. And now look at them at Intel, while the chip itself is large the core count is low. He hates making fast GPUs and said so himself in an interview a long time ago. He wanted to push for max performance for low power. Gamers don't want that. Gamers want the fastest no matter what. If the graphics card is designed for gaming and meant to be sold to gamers then it needs to please the gamers. Playing this bullshit low power mediocre performance doesn't cut it. The selfishly choose what gamers want instead of giving gamers what they actually want. Granting the videos no better because they make one GPU and sell it to their entire product stack. Said GPU was sold for machine learning, AI, gaming, servers, etc. The video is no better. Honestly I'm disappointed with both brands. I miss the old Nvidia, GPUs made for gamers by gamers for gaming. Not this generic piece of shit that they pawn off on gamers while they market and profit on AI and servers.... It's sad that you actually think AMD or Nvidia sets their prices based on the competition. That's not how business works. Sale pricing is pre-planned long before the product ever gets told to the consumer that it exists. It's not like 1 hour before a press conference AMD picks a price. Pure insanity that people like you keep thinking that this is how business works. AMD 100% could have planned to make a GPU that was a thousand bucks. The funny thing is if you go on to their Reddit, which isn't even officially run by AMD, You will see tons of people saying that they will pay no more than $1,000 for a GPU. So it's obvious why AMD pushed for $1,000 GPU. It actually makes perfect logical sense unlike your idea of being 1000 only because the 4080 was 1200.... But I guess ignorance is bliss. You also ignorantly think that AMD lied about performance. The sad part is you were taking into account aftermarket cards which are overclocked and thinking that those are what AMD is using for their reference benchmarks. That isn't how business works. When AMD says 6950 XT versus 7900 XTX they are comparing their reference card versus reference card. The results are actually absolutely correct. The issue is the 6950 XT overclocked cards that you would typically buy from an AIB are already faster than the reference 6950 XT. Therefore the gap and performance to the new 7900 XTX reference is smaller. Which you pointed out. But you pointed it out ignorantly not realizing your mistake. When AMD compares anything to anything they aren't using aftermarket cards or overclocks. When AMD compares the 7800x3d to a 13900K they are not comparing it to an insanely overclocked 13900K that a typical Intel owner would do. They are comparing it to a bone stock 13900K as if it was slotted into a motherboard and tested. You'll even hear people say well my 13600K gets the same performance as a 13900K because I overclocked it. Well no shit you ever clocked it. The benchmarks aren't for people who overclock it's for people who buy a product and use it. Enthusiast will always have to wait for overclocking information from third party reviewers. To claim AMD being deceptive due to your own ignorance is actually hilarious. AMD not competing is the reason why they're not profiting. Why buy second best when you can buy the fastest. It's basic economy and basic psychology. A basic consumer mindset is I want the best for my money. AMD releasing a budget GPU for $1,000 that doesn't actually beat the competition won't sell. It's funny how all they have to do is compete. They just competed with the 6,900 XT last gen. It literally tied to 3090 when you look at a multi-game benchmark. Sure pick one game and it loses. But you can pick another game and it wins. They were there. And someone high up said no no no wait wait wait, we can't make a GPU that fast because no one will buy it. Whoever that person is needs to be fired.

for reasons just like this, are the main reason why I have zero brand loyalty. Yes I want AMD to do well but that mainly just so they keep everyone else in check. I fell the same way about intel. at the end of the day I dont buy anything until its out and had 3rd parties review it. only then will I get what I feel is good value for my money.

I had a 6900 XT at launch and now I have a 7900 XTX and I find it a big improvement over the 6900 XT at 4K. Also AMD made a great improvement to ray tracing as the 7900 XTX beats Nvidias 3090 Ti in ray tracing which is pretty good in my eyes.

It's especially sad because they are the only company atm who could work on synergizing their CPU+GPU units to achieve better results. :/

AMD exists so Nvidia can claim it doesn't hold a monopoly, when it in fact does. But this will keep the antitrust off their backs. Which makes sense considering AMD/Nvidia/Intel all have the same major shareholders. Lo and behold, the newcomer Intel also prices to match. Who'd have thought. Just like in politics, it's a collection of players that pretend to be engaging in competition/debate but are in reality alike and have the same owner. And that also means that you cannot "vote with your wallet", because you're still paying either of these two disingenous companies. Unless you forgo the sale completely.

Great video - but I'm going to provide a steelman argument about AMD's current pricing: Global Economics. Economic projections of the global economy are terrible - spending on luxury goods are going to drop substantially - so the market for high end consumer goods are likely going to shrink. People aren't likely going to remember the really competitive products that were available at the point in time where they couldn't afford them. So AMD could be just staying in a holding pattern - don't rock the market and wait for things to recover before pushing for huge market share. Why show your best hand when the rewards are terrible? As I said - this is a steelman argument. I don't necessarily agree with it but there is an argument to be had.

I thought AMD really had a strong showing with the 6000 series, then they completely fumble the 7000 series.

It was proven that the "600 watt" quote was grossly mistranslated, i think steve from GN made it clear

AdoredTV upload notifications are always a pleasant surprise :)

Imho their business strategy is subpar because of pricing If the XTX sells for $900 now and their XT is around $650-700 They shud sell well even with less features

High end graphics cards used to be a market space where you could buy the future graphics technology. That's what SLI was all about, which happened to push power supplies to the next level.

I waited for an entire year before I finally built my first personal windows machine in over 15 years. I got and XFX Radeon 7900 XT, as the performance increase for the price wasn't worth it, IMO for the XTX. I paid $840 for my XFX 7900 XT (Early March 2023), the price has now drop another $10. But I had zero intensions of paying $900+ USD. All said, I built it exclusively for gaming. The performance paired with the 5900X has been amazing for me. Loving it.

Damn annoying for the customer but AMD are simply jumping onto Ngreedia's pricebar, 100% unwilling to repeat their 'aggressive pricing' Zen2 strategy. It's clear AMD has been collaborating with Nvidia on pricing since any fanboy argument over the 'benefits of RTX/DLSS/AI/etc' can be countered if AMD were only to make a serious attempt to price cut Nvidia. Nothing wrong with the 7900 apart from pricing. Of course, the same applies to Nvidia's cards, though they're price-set at even more sphincter-clenching levels.

16:43 I gotta throw up the red flag here. In my own experience with ChatGPT I have found that it will absolutely throw out WRONG information if what you are asking about is not something it was trained on. It also is incapable of doing math. I'll ask it for information on something like calculating the kinetic energy of a projectile, and not only will it pull a completely incorrect formula out of its ass, it will also give you an "answer" that wouldn't even be correct if you followed the formula it presented you with.

Jim, a pleasure to hear from you and see your point of view! It's really sad to see the RTG group lose all of the "goodwill" they would've earned with this launch of Radeon 7000, with issues like their vapor chamber problems, lack of proper drivers etc, however, I'll still support them as I don't find the offerings from Nvidia worth my $$$, I can't use Raytracing in any multi-player games, so I find that technology case "pointless", also after the cyptoboom has died customers like myself need to purchase these cards at these "inflated prices" is beyond nonsensical! I would like for the RTG group to really ensure that their cards have good use cases, such as proper support with the software developers like Adobe etc. If they can get more developers/development support i believe that this can help in getting more customer buy-in. Looking forward to the next video drop!

Great to hear you as always Jim! I have to say that as someone who built his first gaming PC with AMD XP2000+ Palomino and GF 4200Ti I am pretty into all this CPU/GPU saga for the last 20+ years. My take on this RDNA 3 story is quite simple - AMD quit on the GPU crowd because it is not worth their time. And I am 100% behind their decision! They have locked console market and there is no threat from nVidia or Intel in foreseeable future...or ever. Similar story is with mobile CPU division. They have the best price/performance ratio, excellent efficiency for the last 3 generation (5 years) of mobile Ryzen CPUs and still have 15% of laptop market...maybe not even that much. The latest intel mobile CPU draws more power than R9 5950X3D so right now for AMD management is like - WTF people...are you insane??? Mobile RDNA2 6800M/6600M are also great GPUs and I can bet that in the high end segment of gaming laptops they could not get 1% of market. I think that nVidia spent in last 5 years more on the marketing than AMD spent on R&D for both CPU and GPU and you just can not fight that. Next RTX 5090 will be 2500$ and marketing will make it great product and people will buy it. In the meantime AMD will use every square mm of TSMCs wafers to make the products where it is appreciated - datacenters. Epyc and MI cards will rule in performance, efficiency and design (chiplets all around) and make 70-100% margin on these products. For client segment they will make GPUs comparable with nVidias under 1000$ and will be happy. For nVidia i predict the same future as for German luxury car makers like BMW or Audi - they are running out of customers able to pay 100% premium over the competition for 20% of more performance. Bottom line - consumers quit on AMD in the mobile/desktop GPU segment so now AMD quit on them.

I wish you could have a private conversation with Lisa and carefully inform us what's going on behind the scenes at AMD -while also giving her your perspective. It's very frustrating b/c it feels like AMD is shooting itself in the foot on purpose with every decision.

My thought for AMD giving such good deals to M$ and Sony was to get devs to optimize for AMD architecture but there's still few games that do better on AMD than nvidia cards.

Great stuff once more Jim. I still have a gut feeling AMD is high on the APU sauce. If they can shift everything into one package and reduce the need for dedicated graphics. It would be something to see them trying to make nvidia obsolete, while not having to compete with dedicated graphics.

This might be my eternal optimism talking but I actually read the current situation slightly differently. So with Nvidia clearly not interested in bringing GPU prices down or they they can't much because of how big their dies are and how expensive 4nm is but either way it means AMD can do their normal thing of price below Nvidia but this time actually make some profits for a change which will hopefully mean they don't give up. But also how I'm reading all the rumours at the moment is that AMD might actually surprise us and launch RDNA 4 much earlier than expected and take the fight to Nvidia that way. I think AMD are pretty far down the line with RDNA 4 development. N31 had issues and they had a choice of either spending a fair bit of money on a re-spin to try to fix these issues for a refresh release in Sept/Oct time or use that money and man-power to pull in the launch date of RDNA 4. And I believe they went with the latter! So even though it looks like Nvidia had already reacted to this by cancelling their own refresh too, it all depends how far along they are with Blackwell development and how much they can pull in the launch date. But if AMD can get 8900 XTX to complete with Nvidia's current 4000 series for maybe 4-6 months then that would result in massive sales for Radeon at decent margins for them.

I can kinda understand AMDs thinking. The sheep buy on brand, and the Nvidia brand is strong. Even if they came out and blew away Nvidia at every price point their sales wouldn't reflect it. It sucks massively but the consumer is as much a problem as AMD.

gezz that HD 4850 got me hooked on AMD ever since. I guess I am the kind if girl who thinks they can change him…

Really enjoyed your analysis, lets hope AMD decides to compete in the high-end again.

This is a classical economic problem, if AMD builds a faster chip, Nvidia will respond etc. Its similar to the prisoners dilemma.

Companies aren't our friends be it NVidia, AMD or Intel. The flip side to that is that we can't expect companies to give gamers something without there being some evidence that it will help their bottom line. AMD's got a ton of evidence that if they push price/performance gamers will largely ignore them and they'll just have a division making less money then it could. Until NVidia sales get punished be it either by shifting to ( the only viable alternative in discrete GPUs ) AMD or by a YoY decline in sales overall that's notably worse then general economic trends then there's no reason for AMD to believe that chasing market share with price is going to get them anywhere. A chance for this to change happened with the anemic launch sales NVidia managed that at least suggests that the stranglehold NVidia has on the gaming consumer may be loosening, what we've seen from AMD hasn't been dramatic certainly but steady whittling down of prices have it surpassing NVidia in mindfactory sales, a small sample to be sure . The reasoning is easy enough to understand, they want to test if the market actually reacts to a lower price with higher sales, not lock themselves into a lower MSRP that doesn't net them the sales needed to make up for the lost profit by cutting margins down. Not trying to make a top tier ~1600$ card is a shame absolutely for those in that market but I suspect they saw such a fight as a considerable gamble and as you noted their approach seems to be more focused on extracting cheaper costs out of chiplets instead of maximizing performance. Which along with less stingyness with memory then NVidia still allows them to have compelling offers at some price points, thus the mindfactory results, certainly that's a big part of why I went for an AMD card. A 4090 was just way more performance then I needed and my use of mods means I couldn't tolerate NVidia's VRAM stingyness lower down the product stack. If these positions were reversed I'd have to look at getting NVidia instead. I don't think AMD's position in the market is quite that hopeless but neither do I expect them to swoop in and 'fix' the GPU market any time soon, they have just as much to gain from high GPU prices as NVidia after all, only consistent rejection of those higher priced offerings by gamers will bring it back to earth and at best we're seeing the first hints of that this past year or so. Also as you pointed out Jim AMD's got a lot of other markets to focus on, dedicating a ton of resources to decisively push NVidia in their key market when they still only have a quarter the market cap NVidia has doesn't make a lot of sense compared to focusing on the markets Intel is struggling with holding onto and these days Intel is much closer to being a peer size wise then NVidia. This doesn't mean anyone should be happy or satisfied with somewhat backburner efforts from AMD but they're not going anywhere either and as a heathier company then they've been in years they at least have the option to tackle NVidia, even if they're a bit gun shy about it thusfar.

You are correct. AMD isn't really trying to out compete Nvidia. It's how and when they sell their GPU hardware that tell you they don't want the lead role for selling video cards. They are happy simply undercutting Nvidia but at the maximum price that they think the public will tolerate.

If they have truly given up, I can't even blame them. Whether they give a decent effort or do their best, what's the point if the public never rewards them when they're winning?

look at GN for the 600W statement, this was a wrong translation

Those graphs nicely show how good Polaris and first-gen Navi was. Well, I guess that they concluded that they make more money making cpu chiplets than rather large gpus. I guess that I'll upgrade my aging zen1 chips with zen 2 and 3 chips, and then keep my mix of Polaris and Navi2 GPUs until they die. There is not much time for gaming, my large library of games will last me for many years so there is no need for any further hw upgrades.

It's like radeon is aiming for the Mozilla Firefox business strategy get so little market share that Google gives them money to be in business but they want nvidia to do it. I would have bought the 7900 xtx if it was faster than the rtx 4090. Too bad it's more of a 4080 at best. It should have been named the 7800 none xt

Thank you for this video. I was not sure whether to get amd of nvidia gpu for my next build, but this convinced me that amd is the way to go. I couldn't care less about who has the best performance card, I care who has better price/performance product. That's what REALLY matters. Trust me, I work in HPC.

finally. NVDA doesnt even WANT to sell you a GPU.

Was really disappointed with 7000 series launch. As we speak, if I have to spend $1000 on a gpu, I don't really care adding extra $700 to get 4090.

I’m about halfway through, and I’m assuming I’m gonna mostly agree, but uh. But AMD did not say that. Gamersnexus video where they asked AMD: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-qlPSNOU2WEA.html Still bad, just less bad.

I think it all boils down to opportunity cost, especially when it comes to wafer availability. If AMD had the resources and wafers to do so then they would compete in the high-end but they decided to focus on data-center and the consoles instead. Regarding RTG not making much profit it could be partly explained if the profit is made by other parts of AMD. If 3 divisions are involved in a product then it is not uncommon for 1 division to book the profit for the product and the others get break-even.

I think this video comes one generation too early. This gen is in my eyes more or less a beta test for the chiplets which AMD plays safe. If the RX8000 series dont have an ultra high end card that would be the perfect time for this video.

08:48 the citation is wrong and is a double translation: English -> Japan -> English. (at least GN said so)

In India the price of the RTX 4080 and 7900xtx are the same so I got the 4080. CPU wise AMD are killing it, got the 7600x but the mobo is too expensive, got the AS ROCK Pro RS x670e for the gen 5 GPU and SSD support.

Corteks has been suggesting price fixing for ages, I tend to agree. AMD have pulled an Intel/Nvidia, screw the customer for as much as possible

I wouldn't think that AMD could double the amount of CUs just like that. The command frontend does have to run increasingly faster than the shaders in RDNA3, the higher you push their clocks, just in order to keep them fed. Heck, in OC scenarios, the frontend is touching 3.5Ghz+ when the shaders haven't even reached 3.0. I'd argue that a completely redesigned frontend will be needed as CU counts increase further, and we'll probably see that in a future generation of RDNA. Doing that extra design, validation and driver work for a low volume, halo RDNA3 product wouldn't make any business sense and probably result in a net loss of money; thus I can understand why they didn't approach it just now.

Why should they Adored? Let's be honest here, take for example the RTX 3050 and RX 6600XT situation, the 6600XT is more than 50% faster than the RTX 3050 for less or about same money, what consumers buy you ask? Yes, they buy the RTX 3050, the RX 6600XT is faster than the RTX 3060 too, do you see the RX 6600XT in the leading spot on Steam Hardware Survey? No it is the RTX 3060 which costed and still cost way more than the RX 6600XT, why should AMD heavily undercut NVIDIA or even launch an card for let's say $1800 that beats the RTX 4090 if people won't buy it? This is also consumers fault at the end of the day, people want AMD to beat NVIDIA so they can buy NVIDIA cards cheaper, in the end AMD decided it would be better to try and at least maximize margins with MCM if they can't reach market share since consumers don't even consider their cards.

I would say that when Zen came out it was rather expensive, had some great new design ideas, but was lacklustre, and did not blast past competitors. I can see the same coming from their GPU path. Multi chip GPUs are the way forward, and in two generational development cycles could have a much more impressive outcome. Personally I expected AMD to go for lower prices (£800 max for the RX7900XTX) and keep the mid/high gaming segment happy to try AMD, sadly they fell in with the 'grab the cash' nVidia scam. There is still time for AMD to turn this around, and that profit margin will grow if they win over a good 20-30% of the market. Only an opinion, let's face it Companies run on profit and if they so wish can tell the consumers to do one if they do not like the price.

Först och främst 😌 I always appreciate the content you make available for us mere plebeian content consumers to watch/listen to so many thanks for that m8. You truly have a talent to highlight certain/unique aspects/circumstances regarding a topic/issue/question which often provides more context, nuance, insight and/or clarity from a different/unique angle/perspective. This in my opinion almost always contributes genuine informative value delivered in a way that’s incredibly interesting and unique. I mean from the first video I happened upon by you..jeez some 5 years ago I just realized, to the latest ones I sincerely thank you for the work you do for us to enjoy. 🙏🙌 Sist men inte minst, ta inte skiten som alla dönickar till fanboys ger dig när de inbillar sig att du attackerar märket de har en sån bisarr relation till att de börjar lipa, gnälla och kasta ur sig massa skit mot/om dig. Fuck’em 😎 Du är bäst! Fredens 👌✌️🖖

8:27 That translation is wrong, as confirmed by AMD

So now Intel is our only hope? Oof.

I do agree that AMD's strategy is not quite there. They need to get very aggressive with performance targets and pricing to do to Nvidia what they did to Intel with Zen. However I do want to give them some credit as the better aftermarket 7900 XTX cards are actually pretty good. And now they are starting to show better raster performance than a 4080 in a lot of titles for at least $100 cheaper if not more. You can sometimes grab one for less than $1,000 USB open box at MicroCenter.

Regarding the other comments pointing out the apparent mistranslation as covered by GamersNexus, the overall spirit of what they say doesn't change It is still just another excuse. Radeon does NOT have any will to compete. They are not incapable, they are unwilling. And it shows, as for the past 3-5 gens, they have always launched AFTER. They DON'T want to set the stage, they just WAIT for Nvidia and aferwards slot their cards in between the pricing and performance gaps. In my opinion, the last time they really tried to compete was in the R9 290x and the Fury generations. Even with the 6900XT, it didn't feel like they were serious.

If you're arguing that Radeon has low profits, I don't think releasing the 7900XTX for $200 less than it was would help that out.

I was actually seriously considering to buy an AMD GPU this time around, but then saw they priced there XTX card at 1000€ and went immediately fuck that, i aint gonna lay down 1000 eddies for a GPU no way no how. So AMD shot themselfs in the foot with there dumb price following of GreedVidia. Definately a hugely missed chance, dont know what i'll do now, probably buy a 4070 -_- and maintain my current workflow on Daz3D and iRay.

Edit: First, love thay you're back! I hope life is going great! Thats the thing aint it, the wafers. Why would they want to sell big GPU chips for cheap when they can sell tiny CPU chips for big $$$? The Radeon Group is stuck between a rock and a hard place, the AMD data centers and Nvidia. The time of cheapish, big chip pure gaming GPUs in PCs was over the second they got useful for things that actually makes moneys.

So PC Gaming segment / dGPUs is just marketing for them, we are like the bottom of the barrel guys demanding the most, for the least, the annoying customer base. It's the businesses which make it profitable, both Sony & MS and HPC & Datacenter sectors. So a lot of Radeon revenue is actually on the embedded and datacenter segments. It would be curious to see a breakdown of how much is GPUs and how much CPUs in their revenue.

Given the state of PC games derived from cross platform, a high end GPU is a waste of money, you’d be better served with a console. PC gaming is in rude health, but in the Indy space where it’s competing with Switch and also legacy support. That 4090 is a very good deal for a professional workstation card, but the rest of the line up below it is crippled by VRAM and the Radeons can’t even run general purpose AI work due to the stranglehold CUDA has. CDNA makes sense for certain subsets of computation where you are buying 10000 units and throwing them at a FP64 problem, modelling weather etc.

In regards to 4090 and AMD competing against it. You have to take into account that AD102 is used in RTX 6000 Ada (and other similar products) at 7000$ if you can find it at that low price. If you can't afford it you can use a 4090 in many cases. AMD dose not have this dual use / dual market for a large die. In this generation i bet that close to 50% of all dies of AD102 are not bought by gamers.

I think AMD and nVidia have completely different philosiphies. AMD wants their graphics to be pervasive. They have APUs, they go into Teslas, console SOCs, handhelds and they have recently renewed their licensing agreenment with Samsung. nVidia has chosen AI as a target and wants to be the best at that while at the same time holding the PC graphics market. Make no mistake though, they are making less revenue in PC gaming as well.

Thanks for this Jim. I'm more than a bit depressed by what I'm seeing. As far as the current start of the AI "inflection point" (its real, but GPT is starting at LOL levels) GPT is crypto now for AMD/Nvidia/etc. Where this will really take hold is in the related merging of various coding mindsets around GPT/GAN/Diffusion/etc. Large Language models are just a start, the compute requirements are going to crater over the next few years (LOL for AMD/Nvidia if they think data crunching will be the cash-cow) But it is interesting. Having a capable agent that can interact with you and act upon your personal needs advocating for your benefit? I think that's where the real end-game of this generation of learning systems end up. People who trust corporate systems and their biased agents are going to get hard-burned by the rise. People who learn how this stuff works and build their own or work on community projects to build agents for community consumption? That's where we will see the benefits. At least, I think that's where the evolution plateau is going to stick after the logarithmic utility increase we're going to see. I'm not sure its going to work out for big businesses the way they hope - too many people with the skills to use what exists to build what's next in the wild. But this is me being a wishful thinker - but I can definitely see the large tech companies biting early and finding out the AI they consumed was poison and not worth the cost.

I wish AMD would plan a GPU that has 4-6x the RT performance and completely disregards raster performance favoring RT for chip real estate. When AMD first mentioned they were moving to chiplets I was freakin stoked because I was thinking likely what everyone else was, dual GPUs would pwn and the perf would double but that's not what they did... When I saw the single GCD and 6x MCD design, I was actually really confused and sad. They've gotten a lot out of architecture in this generation though and it should afford some advantages in future iterations and that's good to see. Its just not what I wanted to see.

Nvidia didn’t screw over the GPU market. Nvidia has been doing its job exceptionally well as a cooperation maximizing its profits. The only job of a company is to make money, and maximize value for its shareholders. AMD screwed the market over by not providing proper competition by offering a superior and cheaper product to challenge Nvidia’s dominance. Competition exists to keep companies in check naturally. And AMD has failed. Corporations aren’t your friend. And Nvidia never pretended to be your friend.

Been using radeons and Ryzen chips for years trying to support the underdog but I'm tired of sandbagging. Going to jump ship again like I did when the core 2 duo was released. I'm waiting to see what zen5 and rdna4 bring. I have a 5950x and a 6900XT that I purchased at launch and I'm tired of the games they seem to be playing

As a Linux user I'm basically stuck to AMD or maybe Intel if they have a high end offering down the line. It's OK for my personal demands so far but hope in the future I can still buy something ok

I'm curious what's going on with Navi 32, the expected 7800XT chip. I would hope to have heard something about it by now, but maybe they're trying to figure out what didn't go quite right with Navi 31, first.

It used to be in business that "the customer are always right". Now it's "the shareholders are always right".

Hiya Jim, been awhile I know, but as usual I hope this finds you well. I'm watching this and about half through, but what's gnawing at me my good friend, is let's suppose AMD can always get close to NVidia but with way better power specs and efficiency....and a better price. Is it that important that they "win" against NV's top card? Is it for bragging rights or is there something bigger at play that I'm missing? I'm just curious what your take is, I hope you don't think I'm trying to abrasive, I'm truly curious. My feeling is that, if they get close to NV but at a great price and efficiency, I'm still a happy camper. :) What say you my good man?

23:24 I never purchased AMD hardware because of some weird duty to "save the company." That's just bizarre. Radeon has consistently offered better value at my shopping price-point for my last three gpu purchases (R7 260X, RX 470, RX 5700XT). That's why I purchased Radeon gpus.

It's frustrating to see AMD not attempt to compete at the high end. But consider that AMD put considerable engineering effort into chipletizing the GPU. Next year they could conceivably double the 7900XTX with almost no effort. If they don't beat the 7900XTX by at least 50% in the next 12-18 months we should give up on them. If AMD is smart, they'll recognize that their hold on consoles isn't as strong as it was 3 years ago. Nvidia has very compelling SoCs. Qualcomm has the IP to produce such an SoC, but they don't currently have it. Intel almost has it. A dying company like Imagination might just bet the farm and win a console design bid.

300mm2 is the gpu size, but that already pulls +400w when uncapped.

I do agree with what you say but if we have to give up AMD and the Future of Intel is so uncertain and their current products are not that usable without accepting a lot of issues, what is left? im for sure NOT going to buy an nVidia card. I banned them after GPP that was the final straw for me. With prices going up i dont like the PC hobby as much as i used to. Maybe i should just stick to buying second hand hardware 2 years down the road to keep things affordable and hopefully the AMD cards will be good enough to run the games i want to play. So far i notice im sticking to older titles anyway. But more than 50% of the PC hobby was the hardware the new architectures tweaking the stuff. I feel that is slowly dying out. I really dont get why AMD is not trying at least in the more low end segment. Dont they have enough wafers to do that or what is up with that? The 7600 series seems to perform well in the laptop space. So if they can make a lot of those an price them well they can have their margins and gain some market share. Most people dont care about 400+ dollar GPU's anyway they are to expensive. Polaris did really well the refreshes where to mediocre it needed a real followup that never came. I understand that these high end N31 GPU's are not that interesting for them but selling a product like N33 at mass might be? They are way cheaper to design and manufacture dont need stupid big coolers to keep cool and address a way bigger market. I still hope they would do something like that but like you said they just not seem that willing to do so.

To be honest, RDNA3 had some teething issues, so I am not surprised they picked some less aggressive path. Imagine if they would invest in that big die, only to see rdna3 have hardware clog they cannot solve. AMD don't have a budget for that. They might have sat out this one gen, to get the next one, when drivers and architecture will get more stable, to strike harder then. I really hope they will do more in next gen, (or update to this one). About price and marketing, INTEL did layoff all the lying bastards, I fell like some of them attached themselves to AMD now. IMHO it's a yellow card for Radeon, but its not as bad as the video makes it. Might be a mistake they never make again. its a shame they did that, but not all hope is lost yet, until they repeat it.

The last 3 months after watching your video of "does AMD want to win?" and Coreteks video about a possible price fixing, which I personally think it is not exactly that case but close, I repeated this theory that AMD does not want to compete with Nvidia as a matter of FACT. My theory that I already explained is more like economic/organization theory, corporations work in a way that they avoid any possible competition, usually this is done with government state law making help, but also, like in this case, corporations can make a deal: "Jensen, I buy ATI, we play competition, then I make Radeon brand trash and leave the market to you" "What do you get in return AMD?" "Jensen, I get the console market, IBM no longer will produce the console CPUs and you leave the GPUs to me, you win the PC gpu market, I get the consoles". This is what happened, corporations work this way because this is the BEST outcome for both, AMD and Nvidia. If you think that any of them could just "eat the other, why not do it" the reason is that in fact, that is really really hard to do, look at Intel now and AMD on cpu market, and the risks of competing are very high, AMD almost disappeared, now Intel is in trouble, why? Because they are actually competing. What Intel could do is move to another market where there is no AMD nor Nvidia to fight, and if that happens.... you will have 3 de facto monopolies. This is a way for these corporations to be actual monopolies and look as if they are not, as if they are competing. This is possible in this highly, extremely regulated market. Also AMD sells are related to government, not sure if just American government or others too, I read that data somewhere. We don't have a free world, this is the fundamental issue. Intellectual property for example, sounds nice and needed, I have bad news, corporations exists since these laws exist, since patents were created, you might find this something needed, but it makes things worse. Anyway, unpopular opinion.

I don't believe AMD will be completely uncompetitive forever. I think as Intel gets better and better, they will steal AMD's marketshare and AMD will start competing again thus NVIDIA will a bit as well.

8:45 that's a miss-translation AFAIK. They said, it'd have been 600W (4090 competitor). EDIT: It might be me, who's wrong edit2: 16:00 i think that amd chose this strategy, 'cos even if they sell GPUS that are similar perf, for significantly cheaper, people still wont even consider it. I have sysadmins and and user support in my company who just fucking hate AMD for their past failures, and 'cos of their views it's company policy to not buy amd stuff. nor do they ever recommend AMD to family and friends. We're strategic partners with Lenovo and all the current computers are intel 11 gen cpus. Total clown word. Because of driver issues 10 years ago, people that are in IT are still bitter and won't even touch AMD. And that's mindset of the average consumer. That AMD is the bad, cheap, budget option, therefor they wont even consider it. If you want to know how bad it is, watch a few content creators on RU-vid. They call Ray tracing RTX. That's really all you need to know.

I just purchased a 4090 last week and I also put a 4090 video on my channel. 4090 is clearly more than gaming. With AMD's severely lacking support on the software side, assuming there could be a 4090 competitor from AMD with some pricing discount, no one with a right mind would purchase such product. For example, for machine learning, almost all the people working in this field choose Nvidia's cards. For me, I learned this the hard way with Vega FE. Back then, they promised a lot with software support for Tensorflow. But what I found was that it was just not usable. In the end, I purchased 1080 Ti instead.

This is why I keep telling people AMD isnt going to go hard on graphics cards, they just make way more money by making chiplets to sell everywhere.

They will need better cooling and better materials to handle any significant spike in power. It doesn't come with snapping fingers, more R&D = more money and a lot of it, if they can do it It was always like this Nvidia release something then AMD release months or a year later and says "look we have nearly the same performance as the second best Nvidia card" what they don't want to say is we are fucking a whole cycle behind and we can't even put the fastest card in the market Those VPs talks is a wishful thinkin begging fans to believe " yeah we can do it but our super amazing card which doesn't exist will cost a lot and we are a poor man video card company" The day they loose the console market or even one of Sony or Microsoft, the GPU department is going to close.

So you think AMD is giving up because they *aren't* launching a $1600 GPU?

Thank you for saying what I started seeing when Vega launched. it always felt weird how they where hyped up but would always aim for the "cool/good guy" right under NVIDIA. at the end of the day it reminds me of where the female wolf will act scared and hide under the alpha during a fight but in reality is just protecting their neck/lowers from attacks. end of the day these are both companies, they will only go for profit, they are not friends or family, they are an entity, a corporate face and that's all they will ever be. unless something forces them to aim higher and truly innovate and push the limits then we will see the same thing we have seen since the RX480 series with it just working good enough to keep people interested. I genuinely hope that the chiplet design will age like Ryzen did but if it does, you said it right, they will stay just under and never go over, or another words, not bite the hand that feeds them. overall i will still buy AMD because of the partner brand sapphire, I would have gone with NVIDIA but their practices disgusted EVGA go much they left so i see no reason yet in my eyes for them End of the day though all i want to see is path tracing take over and for us to properly have good rays, just have no clue when these brands will be pushed to go for kill on that one thanks for putting into words what needed to be said, but one thing i would like for you to do and continue on, for both NVIDIA and AMD, is breakdowns of architecture. The way you present the inner workings of gpus and cpus are fantastic and provide great incite into what is normally a box of black magic for many people including myself. keep up the good work Jim, who knows what we may find technology wise that has yet to surprise us, there is always something right around the corner!

Way way off my man. This is exactly the problem that AMD struggles with, which is that being no where near the market leader, they 1) are probably going to follow the leader's pricing (and as you mentioned, PC gaming isn't generating them much profit if any at all, so nVidia being greedy has helped them actually turn a profit this time around [will have to wait and see 2023 H1 financials to see how much exactly], ESPECIALLY since nVidia came out several months before RTX 4xxx saying to expect price increases, 2) don't know how outlandish nVidia will go on their "top-end" gaming GPU's, which the 4090 is in old Titan territory previously, i.e. a niche market 3) AMD has a rep (deserved or not, doesn't matter) for having hot, loud, and inefficient cards -- we can see how they've really tried to change the narrative as a claimed performance per watt leader, etc. I still think a 600W TGP/TBP GPU is absolutely lunacrious, even if real-world useage is much lower. Again, with nVidia is a firmly-established market leader, they set the standard on what is "normal." 3) Why does it matter "if" AMD could have made a better GPU than the 4090? Most of the gaming world slots in well below that. The value perspective is great on Radeons, and they're wise enough to know that's how they have to hold being 2nd in market. Beating nVidia both in Rasterization and Ray Tracing is an expensive endeavor that will probably only result in lower margins or even going in the red. It's a fool's game. Still, if they can chip into market share each generation, and I'd argue they did with the 7900's, that's about all they can. I appreciate your hyper-deep analysis and speculation, but it's still only from the lens of one person and someone that might not appreciate all the complexities of the business aspects. We forget that these are corporations with a fine focus on making money, not necessarily being "the best" or doing everyone that a certain class of customer wants (gamer, general PC enthusiast, mainstream, and so on). You mentioned the video being sad/disheartening. It's a pretty classic case of expectations vs. perceptions, IMO. Not that you are entirely wrong, but you went on a lot of circumstantial evidence and predicates.

It'd certainly be very ironic if Intel's next Arc GPU generation manages to be competitive with nVidia, and that gets them and AMD to drop prices, just like Ryzen "magically" got Intel to halve their CPU prices overnight

Thanks for this, I will buy last gen gpus for this year. The prices are much better and they perform just fine for my gaming setup.

Awesome video, I have the same feelings since the last time AMD presented the cards.

Maybe intel will save us. LMAO. Even when their Margins blow Nvidia away, they STILL refuse to compete for market share.

Well actually u have to attribute some money from APUs to the Radeon department as they use the graphics core.

Thank you for a great analysis.

With the caveat of 'for Radeon', I believe the 7900 cards have sold pretty well and if you are right that the original plan was to sell them at around $699 and $549 then surely that means Radeon profits will get a much needed boost this year, possibly giving the Radeon department the budget/justification to do something they maybe wouldn't have if it wasn't profitable anymore???

If I had a nickel for every time someone suggested AMD should shut Radeon down, I would have enough money to buy RTG

I understand what you're saying, but it feels you're contradicting yourself. Revenue is up, profits are stagnant. That means they are making less profits, which means they are pricing their latest products with lower margins so it seems they are trying to compete in price. Selling 7900xtx at $699 would have generated a loss it seems. Another huge problem for RTG/AMD is that the mindset of GPU consumers is to always buy nvidia no matter what. Look at the 1650 cards, they sold in droves even though there were AMD alternatives with better price and performance. It was the same in the CPU space, but AMD got lucky with Intel not being able to execute for a decade, forcing consumers to re-evaluate. Nvidia keeps executing so the mindset wont change. I honestly don't think a $799 7900 XTX would have changed sales all that much, it would only have impacted AMDs bottom line.