Now they only have to do the same on the GPU side...
Looks like it won't happen honestly. As I pointed out in the other thread it's arguably the lowest it's been in ages. Intel is putting pressure on them from below and Nvidia from above. I think they had a shot with RDNA3 but the pricing is terrible on it. I almost wonder if the MCM design helped the price or if the packaging costs more than the savings through yield.
Looks like it won't happen honestly. As I pointed out in the other thread it's arguably the lowest it's been in ages. Intel is putting pressure on them from below and Nvidia from above. I think they had a shot with RDNA3 but the pricing is terrible on it. I almost wonder if the MCM design helped the price or if the packaging costs more than the savings through yield.
I think the MCM route is more expensive than a single chip on the short run but on the long term is going to be the opposite.
As for AMD being able to turn around their fortunes on the long run, i think there`s a 50/50 chance of that happening...
Looks like it won't happen honestly. As I pointed out in the other thread it's arguably the lowest it's been in ages. Intel is putting pressure on them from below and Nvidia from above. I think they had a shot with RDNA3 but the pricing is terrible on it. I almost wonder if the MCM design helped the price or if the packaging costs more than the savings through yield.
H83:
I think the MCM route is more expensive than a single chip on the short run but on the long term is going to be the opposite.
As for AMD being able to turn around their fortunes on the long run, i think there`s a 50/50 chance of that happening...
I think the bigger problem with RDNA3 was the heatsink and driver situation. Since they've only released their top-end GPUs, the price for those is "fine". Not good by any means, but many are still selling well above MSRP so whatever they set the price to would be irrelevant anyway.
MCM likely had a hefty R&D cost but it should reduce production costs, and that's what matters more in the long run.
If the RDNA3 mainstream GPUs are plentiful (to keep pricing at/below MSRP) and have compelling performance-per- (the 7900s are cheaper but the value proposition is not compelling) then I think they have a good chance of success. By the time those GPUs are released the drivers should be in better shape. They won't be power hungry enough to have thermal issues. So long as they're priced based on their raytracing performance, I think they have a good chance of turning around their fortunes.
This is a pretty specific situation though, and considering Su openly admitted to manipulating S&D for the sake of profits, I'm not feeling so confident RDNA3 is going to be what it should be.
I think the bigger problem with RDNA3 was the heatsink and driver situation. Since they've only released their top-end GPUs, the price for those is "fine". Not good by any means, but many are still selling well above MSRP so whatever they set the price to would be irrelevant anyway.
AMD couldn't present a card that would leave Nvidia's best to eat the dust. AMD also couldn't present cards that would be unquestionably better than Nvidia's equivalents below the flagship. AMD's cards aren't even energy savers compared to Nvidia's. Although we haven't yet seen more than two cards from AMD and three from Nvidia. AMD's 10% market share tells something about lacking prestige as well, so if a random person is choosing between AMD and Nvidia, and the offerings look pretty much the same price-performance wise, 9 out of 10 would choose Nvidia, statistically. In such a situation price is the only thing AMD could use as a weapon, but it's not doing it, at least so far.
There was a time when the Radeon brand had a bigger market share than MX/FX but it was when Radeon had ATI in front..now it has AMD and that's it...
Do you remember? https://www.gamesindustry.biz/ati-edges-past-nvidia-in-graphics-market-share
As for increase in marketshare for CPU's....probably this is one of the reasons why they started to pull Intel moves recently(space heaters CPU's, expensive platforms, etc)....
AMD couldn't present a card that would leave Nvidia's best to eat the dust.
They didn't need to in the past. What they did years ago was release GPUs that had near the perf of the NVidia tier they were targetting, but were much cheaper. Now apparently Intel is doing that, except people just don't trust Intel GPUs right now (with good reason, the perf is all over the place.)
It doesn't look like AMD is focusing on winning GPU market share. They're focusing on maximizing profit with the market share they already have.
Looks like it won't happen honestly. As I pointed out in the other thread it's arguably the lowest it's been in ages. Intel is putting pressure on them from below and Nvidia from above. I think they had a shot with RDNA3 but the pricing is terrible on it. I almost wonder if the MCM design helped the price or if the packaging costs more than the savings through yield.
The MCM definitely improves their cost significantly, not only does is reduce wafer cost since 6nm is much cheaper than 5nm, they can reuse their cache chiplets for a range of products, from CDNA compute cards all the way down to the budget products. it allows them some extra flexibility with silicon allocation, and if they want to jump onto a newer high risk node, they can do so more quickly. with less cost and higher yield, Its pretty much as good as it sounds. With the drawback being a small performance penalty.
its so good , that I'd even wager despite the botched launch, they are probably still in the green so to speak.
Just a comparison (about what happened since AMD bought the Radeon brand):
ATI Radeon: 24.9 percent market share vs Nvidia 24.7
AMD Radeon: 10 percent vs Nvidia 86 percent
AMD never cared too much about their GPU division - they seem to suffer from an inferiority complex, they lack ideas, their only notable architectural innovation is the MCM design but only because it is something new not necessarily better, they don't exist on the GPU data processing market, they don't invest in programs with universities and research institutes like Nvidia / Intel does - they care only to get as little profit as possible milking the brand.
The Radeon brand would be much better in other hands - AMD was toxic for them. And to think that when AMD bought them everybody thought that Nvidia will be history in a couple of years...
AMD couldn't present a card that would leave Nvidia's best to eat the dust. AMD also couldn't present cards that would be unquestionably better than Nvidia's equivalents below the flagship. AMD's cards aren't even energy savers compared to Nvidia's. Although we haven't yet seen more than two cards from AMD and three from Nvidia. AMD's 10% market share tells something about lacking prestige as well, so if a random person is choosing between AMD and Nvidia, and the offerings look pretty much the same price-performance wise, 9 out of 10 would choose Nvidia, statistically. In such a situation price is the only thing AMD could use as a weapon, but it's not doing it, at least so far.
I don't think anyone [without biases] expected AMD to best Nvidia. I think most of us were surprised Nvidia ended up overestimating their power consumption while AMD underestimated, but RDNA3 on the charts is adequate so far, even with the sub-par drivers. Depending on your definition of "prestige", I don't know if I'd say that's the problem. Here's what I think made AMD's marketshare suffer:
First, Nvidia released the RTX 2000 series. It was overpriced, DLSS was underwhelming, and raytracing accomplished nothing that traditional rastering couldn't do, but, it was an interesting step forward compared to Vega, which is really all AMD had at the time. Nvidia gained marketshare because it was the only interesting product released in years.
AMD then released RDNA, which honestly felt like an incomplete product, but they needed to get something out the door; it was blindsided by RT. They didn't even attempt to release a high-end model and you were better off buying older generations than the low-end models. So, while the 5700s were solid performers in raster against Nvidia, that's really all they had. With no relevant GPU since the R9 580, AMD's marketshare continued to shrink.
Next, Nvidia released the RTX 3000 series, which fixed pretty much every issue with the 2000 series, and then some. It was popular.
Shortly after, GPU mining became profitable. Key word there is "shortly", because there was actually enough time for many enthusiasts to get the RTX 3000 series, helping Nvidia's marketshare further when AMD at the time had no response.
AMD then released RDNA2, which was basically the completed version of RDNA1. However, it was too late - Nvidia had a better product that hit the market before the mining craze. AMD couldn't get production numbers high enough to outpace the demand, and as a result, lost more marketshare.
Nvidia, for a little while, made it easier for gamers by crippling hash rates. This didn't last long, but it made their GPUs just a little more accessible to gamers. AMD didn't attempt to do the same.
Once the mining bubble burst, AMD released RDNA3, which despite the initial issues is a pretty good architecture. The problem is... if you're going to spend 4 figures on a GPU, you might as well get an RTX 4090 or 4080Ti, which are better overall. Perhaps with enough "fine wine" RDNA3 will prove to be the better GPU in the end, but that's probably going to be at least 2 years away.
So, AMD's small marketshare to me is a series of missed opportunities, underfunded development, and a little bit of bad luck.
The MCM definitely improves their cost significantly, not only does is reduce wafer cost since 6nm is much cheaper than 5nm, they can reuse their cache chiplets for a range of products, from CDNA compute cards all the way down to the budget products. it allows them some extra flexibility with silicon allocation, and if they want to jump onto a newer high risk node, they can do so more quickly. with less cost and higher yield, Its pretty much as good as it sounds. With the drawback being a small performance penalty.
its so good , that I'd even wager despite the botched launch, they are probably still in the green so to speak.
Eh - I don't buy it. I understand what MCM allows them to do but you're wrong about there only being one drawback. The second drawback is the increased complexity in packaging which does have a cost penalty. We don't know what that penalty is but I can only imagine it wipes out some or all of the other MCM savings at the moment.. because the rest of your post doesn't really make sense given the context of their price and waning marketshare, right?
If you're a company that's going from ~20% GPU marketshare down to ~10% quarter over quarter. (which is what the latest John Peddy numbers show and is honestly really sad given that RDNA3 is arguably AMD's most competitive product in years) and you just launched a product that in theory is massively increasing your margins, then why the hell would you price your product where AMD is pricing the RDNA 3 cards? It makes no sense. No one is buying them. Their market is literally crumbling and all they would have to do is lower prices.. which based on MCM and your post, they should easily be able to do.. but they aren't. They are just letting it all crater.
And I get there's other factors, like @schmidtbag said with drivers & cooling issues.. but if you price those cards even just $100-150 cheaper, no one is going to care about those issues.. everyone is going to buy them - hell I'd buy one. There's constantly posts here on reddit, etc of people complaining about being priced out of the market and AMD with MCM should in theory have the opportunity to do this but they haven't.. why? I think because it's not saving them as much as they thought or other people think it is.
...So, AMD's small marketshare to me is a series of missed opportunities, underfunded development, and a little bit of bad luck.
You forgot incompetence and lack of interest - I don't feel that they are trying....
When I was young everybody wanted a red ATI card - if you had an Nvidia card people looked at you like meh...poor guy. Crazy how things turned around and it can't be only because of missed opportunities and bad luck....
Eh - I don't buy it. I understand what MCM allows them to do but you're wrong about there only being one drawback. The second drawback is the increased complexity in packaging which does have a cost penalty. We don't know what that penalty is but I can only imagine it wipes out some or all of the other MCM savings at the moment.. because the rest of your post doesn't really make sense given the context of their price and waning marketshare, right?
If you're a company that's going from ~15% GPU marketshare down to 8% (which is what the latest John Peddy numbers show) and you just launched a product that in theory is massively increasing your margins, then why the hell would you price your product where AMD is pricing the RDNA 3 cards? It makes no sense. No one is buying them. Their market is literally crumbling and all they would have to do is lower prices.. which based on MCM and your post, they should easily be able to do.. but they aren't. They are just letting it all crater.
And I get there's other factors, like @schmidtbag said with drivers & cooling issues.. but if you price those cards even just $100-150 cheaper, no one is going to care about those issues.. everyone is going to buy them. There's constantly posts here on reddit, etc of people complaining about being priced out of the market and AMD with MCM should in theory have the opportunity to do this but they haven't.. why? I think because it's not saving them as much as they thought or other people think it is.
the packaging they are using is similar to what intel uses for their laptops(EMIB) ,(*correction, its actually something completely as seen here , basically an enchanced version of what they use for ryzen and epyc) its really not that expensive, there is no interposer like fiji and vega.
As for why amd is pricing it where it is, basically everyone is anticipating a market crash, so why is amd pricing it there? because they don't have supply, intentionally, and self admittedly. Hypothetically If you restrict supply pre-emptively and jack up the price before a crash, It might be better for you. since you will have made bigger margin on what you have sold and because you won't be holding inventory that won't sell when it does crash.
Also they are probably allocating their decent chunk of 5nm to something else, the El Capitan super computer which is going to be a load of big chip compute cards, expected to be completed this year. supposed to be 2 exaflops , so its a big boy. ( for comparison, "frontier" also an amd based super computer , finished last year, is has 37,888 MI250x accelerators , which have 729mm^2 dies, and 9,248 3rd gen 64 core epyc chips. El Capitan will likely be similar or greater in number except it will be using 5nm cpus and gpus.) edit: some corrections and formatting changes
Checks out. 2 of my last 3 CPU purchases were AMD. Prior to that, I think my history was something like only 1 out of 10 being AMD (from Core2Duo until Ryzen).
Both companies are putting out solid CPUs now. It's hard to go wrong.
Depending on your definition of "prestige", I don't know if I'd say that's the problem.
No, it's definitely one of the problems. You can see it when you consider the basic setting: People don't need any reason to choose Nvidia, but you suddenly need to start listing reasons if you say you are considering an AMD card. It might also depend on your country of residence, but at least over here it's easy to see Nvidia is the default choice for stores, whereas AMD is an option. For years now shop clerks have recommended Nvidia, but of course they will also get an AMD card for you if you separately ask for one. On shop shelves the selection naturally reflects the market shares, with Nvidia claiming most of the spots.
5700 XT actually was an interesting card: A couple of traditional Nvidia customers among my friends got it because of the kickass pricing and because RT wasn't such a big deal back then.
AMD couldn't present a card that would leave Nvidia's best to eat the dust. AMD also couldn't present cards that would be unquestionably better than Nvidia's equivalents below the flagship. AMD's cards aren't even energy savers compared to Nvidia's. Although we haven't yet seen more than two cards from AMD and three from Nvidia. AMD's 10% market share tells something about lacking prestige as well, so if a random person is choosing between AMD and Nvidia, and the offerings look pretty much the same price-performance wise, 9 out of 10 would choose Nvidia, statistically. In such a situation price is the only thing AMD could use as a weapon, but it's not doing it, at least so far.
I don`t understand why people continue to use and abuse the 10% market share if that`s relative to just one quarter!
There might even exist good reasons for that measly market share but unless that 10% are repeated in the next 3/4 quarters, please refrain from using that figure.
This is not a personall critic against you or any other else, just a quick reminder that those 10% mean very little.
I don`t understand why people continue to use and abuse the 10% market share if that`s relative to just one quarter!
There might even exist good reasons for that measly market share but unless that 10% are repeated in the next 3/4 quarters, please refrain from using that figure.
This is not a personall critic against you or any other else, just a quick reminder that those 10% mean very little.
Wait till someone mention steam survey. Nothing triggers me more. 😀
Wait till someone mention steam survey. Nothing triggers me more. 😀
funny you should say that. I keep getting the Steam survey on my laptop with Intel i58300h UHD/Nvidia 1050ti but I never get it on my primary rig of a 5700x and a 3090ti
is similar to what intel uses for their laptops(EMIB),(*correction, its actually something completely as seen here , basically an enchanced version of what they use for ryzen and epyc) its really not that expensive, there is no interposer like fiji and vega. As for why amd is pricing it where it is, basically everyone is anticipating a market crash, so why is amd pricing it there? because they don't have supply, intentionally, and self admittedly. Hypothetically If you restrict supply pre-emptively and jack up the price before a crash, It might be better for you. since you will have made bigger margin on what you have sold and because you won't be holding inventory that won't sell when it does crash. Also they are probably allocating their decent chunk of 5nm to something else, the El Capitan super computer which is going to be a load of big chip compute cards, expected to be completed this year. supposed to be 2 exaflops , so its a big boy. ( for comparison, "frontier" also an amd based super computer , finished last year, is has 37,888 MI250x accelerators , which have 729mm^2 dies, and 9,248 3rd gen 64 core epyc chips. El Capitan will likely be similar or greater in number except it will be using 5nm cpus and gpus.) edit: some corrections and formatting changes