Yes, Intel has floundered for a number of years, but at the same time they are a foundry themselves, second only to TSMC (for now), have incredible manufacturing capacity, and seem poised to start pushing interesting CPU designs as well as having their foundry half finally start executing correctly. Also their growth potential for their GPU business is huuuuuuge.
As someone who enjoys AMD for the competition they bring, I am worried what the next few years might look like for them for the following reasons:
1. AMD GPUs were actually very competitive with their RDNA2 architecture, but at the same time AMD enjoyed a huge silicon advantage as they were on TSMC while NVIDIA was on Samsung (for consumer). The next generation of NVIDIA consumer GPUs will be made by TSMC and if their professional GPUs (see A100 which are made by TSMC) are any indication we can see potentially up to 1.5x to 2x efficiency bump for NVIDIA.
2. Intel seems to be cleaning up shop and their latest generation of CPUs feel like a 1st Gen Ryzen moment. Also, Intel does not have to clearly surpass AMD, they just need to be competitive as they can simply out produce AMD given AMDs relatively small allocation at TSMC.
3. Intel GPUs pose a moderate threat on the horizon for AMD for both low end and portables and at the same time they are eating up available capacity at TSMC.
Actually, techwise AMD has outperformed Nvidia. Nvidia was only able to keep up because they used Samsung 8nm which is extremely cheap.
AMD silicon is doing much better when it comes to perf/area, Nvidia chips are massive, also Nvidia uses high end expensive GDDR6X memory and still can't completely win the throne.
Nvidia strength is in their software eco-system, things like CUDA and ML frameworks as well as it's popularity among gamers.
Intel is still far behind TSMC when it comes to manufacturing and retook the crown against an old year old architecture from AMD.
* to be clear, I'm not saying AMD is completely dominating Intel, I actually think things today are quite tight. close competition. And once you start looking at players like Apple or Qualcomm, things are looking very competitive.
Unforunately, the drivers of AMD for their consumer products havent kept up with NVIDIA's consumer drivers. For example, when I goto microcenter all of the AMD consumer GPUS are there and are available to order online. Nvidia GPUs have just recently been in stock.
Hmmmm... I don't share this experience, the AMD drivers have been rock solid for me. And on Linux there is simply no argumentation, AMD drivers are a million times better than NVIDIA there.
This is 100% true. I don't know why AMDGPU (the Linux driver) is somehow better than the Windows drivers, but for some reason AMD should focus on Windows driver development (or even a rewrite) since yesterday. Windows NVIDIA drivers may have a bad release every blue moon but for AMD's you need to test out if the update is worthwhile - or just curse when it's a security fix but it crashes on your favourite game.
> I don't know why AMDGPU (the Linux driver) is somehow better than the Windows drivers
Open source Linux drivers are developed by another team inside AMD. They can't talk with the other driver team either. When you remove the incentive to add game specific hacks, multi platform support and other stuff, and add community support at the back, you get a straightforward and superior driver at the end.
Also, the driver wasn't forked or derived from FGLRX or any driver for that matter. It's just written from scratch. I remember OpenSuSE had a dedicated AMDGPU driver team too.
Out of curiosity, are there any writeups on how the game specific hacks work? Like, how do they detect the game being run and what kind of optimization are they doing?
I don't have any links handy, but this is what I remember as I read these things over the years:
- Drivers match the binary name (and CRC32 of said binary sometimes if the name checks out) to trigger game, and game version specific "hacks" or optimizations in the driver.
- These can be small bug fixes for Z flicker or performance enhancements like fill dithering or other more advanced hacks. nVidia once "Optimized" 3DMark to errm... render the benchmark more fluidly I may say (They got caught, BTW).
- There are more advanced "behavioral changes" which alters internal behavior of the driver to improve game performance (a version of the Crysis engine called something expensive too often, and it had no visual impact, so nVidia's driver ignored most of them intelligently and game ran significantly faster for that, IIRC) or use a more optimal memory copy/access strategy than the game declares. Sometimes nVidia's driver takes its own initiative and does the things as it pleases, again if I recall this correctly. It's not impossible that a driver version uses optimized shaders or other hidden hacks to increase performance of a certain game (or contain optimized routines which work better for a particular game).
- Other than that, sometimes drivers allow some games to access more information than an ordinary application to allow higher performance via hacky or dangerous code paths, but they're almost considered as trade secrets. Sometimes game engines abuse the driver behavior too, but it's again "secret sauce" and nobody talks about them in the open much.
I remember reading such things for Need For Speed III, Quake III, A Crysis franchise game. Again, Doom III is criticized for being developed too nVidia specific (Carmack openly likes nVidia more IIRC).
This information is collated from my readings, past memory, some chats with a close friend who develops a game engine, and some other channels I had access in the past.
At least on the AMDGPU side, I've seen mentions on Phoronix of various game (detected based on binary name usually) specific compatibility hacks that force on/off various behaviors and other minor tweaks.
Given the size of the "driver" files on Windows these days, my GUESS is that they might ship literal patches that replace portions of the GPU code in some (popular) games.
There's also the issue of games developed with explicit support for the architecture of a given GPU (usually Nvidia), where algorithmic approaches bias towards the strengths of one type of product.
In addition to what other people have said, there's a lot of shared code between the various open source Linux drivers making things easier for everybody. Not just at the driver but I think stuff like Mesa[1] would be part of the driver on Windows?
Just speculation but it might be because AMD does focus a ton on the HPC GPU market which is huge cash cow for them so you drivers are likely to be taken care of.
I've used AMD graphics exclusively for my gaming PCs for decades and apart from the occasional game using an Nvidia specific library causing it to crash (e.g. Mirror's Edge), I have *never* encountered stability issues. That being said, I generally don't install their extra utilities like the streaming crap or console. I just stick with the bare drivers downloaded from AMD directly.
Similarly I have used Nvidia almost exclusively for work machines and I would say empirically that the drivers are less stable than AMD but not in an appreciable way. They do have unusual bugs that that I do not encounter with AMD. I can consistently cause the DWM to crash in Windows by using the Pen Tool in the Snip & Sketch app. After the latest driver update on a Quadro card, one of my monitors hooked up by an absurd series of dongles will no longer sleep.
Except, annoyingly, HDMI 2.1 doesn't work on Linux. So, I'm still using an Nvidia card on a mostly Linux desktop. But I wish I didn't - the Linux Nvidia drivers are absolutely horrible.
Sure - I'm using Gnome 40/Xorg, latest 5.10 kernel (on Nixos 21.11), nvidia 3080, LG CX48 4k/120hz screen with a certified ultra high speed HDMI cable.
I bought Radeon RX 5700 XT in 2019, roughly a month after it's release. It didn't work for me with the latest kernel at the time. I was so disappointed, I wanted to give my money to AMD, but had to go with Nvidia instead. So Nvidia's driver, hated as it is, does have advantages.
Until you try to work on GPGPU. AMD is a complete dog there. While you may argue this is irrelevant for consumer products, industrial know-how is built on students' personal computers.
If you're talking about the software and stability/performance for gaming, anecdotally I've been using Radeon GPUs for the past two years and they've been awesome.
I still agree that Nvidia is ahead with their software ecosystem in general, though. And AMD GPUs being in stock while Nvidia's are out of stock does not prove the quality of drivers. I suspect AMD is just doing better with their supply chain the past few months (and Nvidia has retained an edge on brand preference). On pure rasterization performance, the Radeons are currently a better value, but Geforce GPUs have some feature advantages.
I'm guessing that's true for Windows, but I've had the opposite experience on Linux (for graphics, not computation).
On Linux, I'm sometimes bitten by the fact that nVidia's closed-source drivers might not work well with a certain kernel version. And IIUC their closed-source nature has also slowed down nVidia-Wayland compatibility.
On the other hand, I fear that now that HDMI specs are closed, AMD's open-source drivers might never be capable of 4k@120Hz via HDMI. (IIUC, that is.)
This observation could easily be explained by brand loyalty and other market forces.
Why did Intel sell more of its objectively inferior 10th and 11th gen chips than AMD Zen 2? Because they have more brand loyalty, partnerships, sales channels, etc.
Sales volume != product quality
On top of that, it's easy to imagine two different cards from fabricated by two different factories having much different levels of supply constraints. If Tesla runs out of chips but Toyota doesn't, the lack of Teslas can't on its own imply that Toyota is a making an inferior product.
Their coding for AGESA doesn't seem to be any better. My Ryzen 5000 series CPU was repeatedly faulting when the system was under load until recently I updated to a beta BIOS whose only change was a newer AGESA version. (I actually thought it was my NVIDIA GPU for a while because it only happened when the GPU was being driven hard).
Intel CPUs (at stock clock speeds/multipliers) never gave me stability issues.
Pretty much false. Having the worst GPU compute stack in the whole industry is not "outperforming", and that's very much part of the tech aspect.
> Nvidia was only able to keep up because they used Samsung 8nm which is extremely cheap.
They got cheaper to produce chips at pretty much a perf/W parity with the competition's GPUs.
NVIDIA were able to avoid using the latest nodes because of their architecture advantages, keeping their 7nm allocation where it's needed the most: high-margin datacenter products.
I have to agree with this. Intel has floundered, but if a lot of AMD's resurgence has been on the back of TSMC, that's a very tenuous resurgence. We've seen reports that Intel has ordered a lot of 3nm capacity from TSMC (https://optocrypto.com/intel-to-be-tsmcs-1st-customer-for-3n...). We also saw a report about Qualcomm moving their production from Samsung to TSMC for the Snapdragon 8 Gen 1 due to bad yields from Samsung (https://www.techspot.com/news/93520-low-yield-samsung-4nm-pr..., something like this was on the HN front-page in the past few days).
As you note, AMD is supply constrained by what they can get out of TSMC. As I noted, Intel is looking to improve their offering by using TSMC's 3nm process. I'd take this a step further: if Intel can buy up TSMC capacity, they prevent AMD from being able to offer new products on those processes and buy time for Intel to get their fabs in order.
Think about it this way: let's say Intel bribed TSMC to refuse orders from AMD for anything better than 7nm for the next 2-3 years. That would given Intel a lot of time to catch up on their fabrication. Of course, regulators would come down extremely hard on such an anti-competitive move. Instead, Intel knows that TSMC has limited capacity and can legitimately buy some of that capacity for some of their volume over the next few years. They'll put out some great 3nm laptop chips (and maybe some others) that will get them good press and AMD will lose some of its luster. If Intel 4 and Intel 3 are delivered on-time and they're able to use TSMC's 3nm capacity in the interim, that kinda takes the wind out of AMD's sails.
AMD's $16B in revenue compared to Intel's $79B in revenue seems to indicate that AMD is doing a lot less volume than Intel. AMD has great growth, but given that TSMC's capacity might be a bottleneck to that growth and given that Intel seems to have locked down a lot of TSMC's next-gen supply, Intel may have thrown up a huge roadblock to AMD's continued advancement. If Intel 4 is supposed to have 17-46% more transistor density than TSMC 5nm, that will put Intel in a good place in 2023. Intel 3 is supposed to improve efficiency by 18% over Intel 4 and be available in late 2023. So it seems like Intel should catch up by mid/late 2023 on the fab side of things - or at least be close enough. As you noted, Intel doesn't need to clearly surpass AMD, they just need to be competitive since they can out-produce on volume.
AMD has been doing great things. A lot of that is coming from AMD and some is a result of TSMC's fabs (and Intel's fabs languishing by comparison). Intel is looking to close that gap and cut AMD off from their supply in the interim. That's not to say that AMD is doomed or anything. Samsung's foundry exists, TSMC might be able to increase their capacity, and likely other things I haven't thought of. AMD has done great work and I hope they continue to provide strong competition for Intel. It just seems premature to value AMD higher than Intel given some of their limitations - and some of Intel's plans.
Intel will not get more favorable treatment from TSMC than AMD.
- Intel compete with TSMC for foundry services
- Intel try to get government subsidy by attacking TSMC's location, national security concerns etc, which caused TSMC original founder publicly respond with not too kind words.
- Let one customer buy capacity to starve other customers will hurt the basic business model and trust of TSMC as a foundry service, but people keep suggesting this from their imagination.
That is why I keep repeating AMD is still not doing enough on Severs and GPU market, both on Anandtech and HN. But most AMD fans tell me they are fine. And this was before Pat Gelsinger took back Intel. Now judging from Intel's roadmap and their current speed and execution, things aren't looking too good.
But I disagree with GPU, Intel ( or more precisely Raja koduri ) still has a lot to prove about their GPU. It is coming close to 5 years since he joined Intel, on a project that started before his arrival, and they still have yet to ship a single thing. And Drivers Quality. I am still waiting for actual reviews.
And that is speaking as someone who bought AMD shares at under $3.
You're bringing in a lot of company strategies and achievements and stuff, but... they are meaningless in this case, because this is the stock market. A company's stock value and market cap are not necessarily related to what they do, and the stock market is not rationally linked to a company and its strategy.
I mean Gamestop's stock and market cap exploded because of investing hype, not because it changed anything in its strategy.
Any stock is priced on multiple factors, one of which is the fundamentals, or the real value of assets and the company's ability to generate revenue. Another factor includes sentiment or guesses about the company's ability to accumulate assets and continue to generate revenue. And the third would be the general state of the investment market, and how equities are valued against other asset classes.
So sure, sentiment can absolutely be bonkers in relation to fundamentals, which is what the parent comment is saying.
Company strategy and achievement are very much relevant to assessing the fundamentals, and attempting to identify the underlying value of a stock, while seeing past the bonkers, irrational valuations.
While it's only "in theory", over a longer timeline, in general, any individual stock will "revert to the mean", that is to say, end up closer to the fundamentals, rather than remain bonkers indefinitely.
Of course you don't have to invest merely on fundamentals. You can use bonkers speculative valuations to your advantage. But to do so, you kind of have to evaluate all the components of an individual stock's pricing.
Unfortunately, the stock market also has no real distinction between what a long or short run is :\.
"The market can remain irrational longer than your can remain solvent" is the relevant platitude.
I worked in prop trading for half a decade and the number of really catchy, pithy sayings they have NEVER ceased to amuse me. My firm was very much on the "voting machine" time horizon so at that time granularity all the notions of company value mostly end up being 0 effect and everything is totally dominated by statistics, but that never stopped us from all joking about these catch-phrases and nodding sagely (as the computers do all the work of actually trading the stocks).
> I mean Gamestop's stock and market cap exploded because of investing hype, not because it changed anything in its strategy.
Some intellectual honesty would go a long way here.
From Gamestop's latest 10-Q filing on December 8, 2021, page 15 [1]:
> BUSINESS PRIORITIES
> GameStop has two long-term goals: delighting customers and delivering value for stockholders. We are evolving from a video game retailer to a technology company that connects customers with games, entertainment and a wide assortment of products. We are focused on offering vast product selection, competitive pricing and fast shipping – supported by high-touch customer service and a frictionless e-commerce and in-store experience. Accordingly, we prioritize long-term revenue growth and market leadership over short-term margins. We are taking steps that include: 1) Increasing the size of our addressable market by growing our product catalog across consumer electronics, PC gaming, collectibles, toys and other categories that represent natural extensions of our business; 2) Expanding fulfillment operations to improve speed of delivery and service to our customers; 3) Building a superior customer experience, including by establishing a U.S.-based customer care operation, and; 4) Strengthening technology capabilities, including by investing in new systems, modernized e-commerce assets and an expanded, experienced talent base. The Company will continue to invest in growth initiatives, while continuing to prioritize maintaining a strong balance sheet.
Pages 15 and 16:
Net sales and gross profits are up YoY.
From Page 17:
> 2021 Debt Payments
> On March 15, 2021, we repaid at maturity $73.2 million outstanding principal amount of our 2021 Senior Notes. On April 30, 2021, we completed the voluntary early redemption of $216.4 million outstanding principal amount of our 10.00% Senior Notes due 2023 ("2023 Senior Notes"). This voluntary early redemption covered the entire amount of then outstanding 2023 Senior Notes, which represented all of our long-term debt. In connection with the voluntary early redemption of our 2023 Senior Notes, we paid approximately $219.1 million in aggregate consideration, including accrued and unpaid interest. In connection with the voluntary early redemption of our 2023 Senior Notes, we paid a $17.8 million make-whole premium which is recognized in interest expense in our Consolidated Statements of Operations. Additionally, we accelerated amortization of $0.4 million deferred financing costs associated with our 2023 Senior Notes.
From Page 18:
> Our principal sources of liquidity are cash from operations, cash on hand, and borrowings from the capital markets, which include our revolving credit facilities. As of October 30, 2021, we had total unrestricted cash on hand of $1.4 billion and an additional $202.4 million of available borrowing capacity under our revolving credit facilities. On March 15, 2021, we repaid our outstanding borrowings of $25.0 million under our asset-based revolving credit facility due November 2022 ("Revolver").
Nothing has changed in Gamestop's strategy? Really?
Did you perhaps get taken in by "We are evolving from a video game retailer to a technology company that connects customers with games, entertainment and a wide assortment of products"?
Everybody and their grandmother is claiming that they are evolving into a technology company if they aren't already one. Doesn't actually mean they are. "Connecting customers with games, entertainment and a wide assortment of products" would have been a reasonable description of gamestop's business since I first went to one in the 90s. I can't read the following 4 points after that statement, which go slightly into specifics, any way other than "we're going to keep doing what we've always done, but even harder".
So yes, their strategy seems effectively unchanged to me. Not sure how all the pablum about their financial juggling is relevant to their change in strategy, or lack thereof.
What significant change to their business model did they announce + demonstrate the value of around the time of the stock price explosion to justify the increased price? I see absolutely nothing. GP seems obviously correct in the assessment that it's hype. Everything you quoted seems like irrelevant corporate speak malarkey to me.
Key word in the GP is because. Yes, Gamestop changed their strategy, maybe--though what you quoted sounds like generic marketing/business-speak. But that press release isn't what made its stock shoot up from $18 to $325 in a month.
I agree, this is a bit bonkers. Tells me the Street thinks the new CEO is all hot air.
I will say for point #3: Intel has mystifyingly dropped the ball on GPUs for going on a decade (and really, why not four decades given how long they've been in PCs and you needed graphics acceleration for displays).
The manufacturing capacity is only as good as the current process, so this may mean the Street thinks they still are several generations behind the current process leaders. Intel's process technology would normally be what is used for mobo chipsets, not frontline CPUs. Intel's manufacturing assets are only useful if they get a modern process going.
The CEO change may have come too late, if the corporation has already chased off all the talent, or all the remaining talent is old and just waiting for retirement. I distinctly get the impression the noxious management has kept anyone good from joining and staying at Intel for a while.
Although Lisa denies it, there is a correlation between AMD's surge and the rise of crypto mining with ASICs. Watch the stock tumble when the crypto ponzi scheme crumbles.
AMD, as far as I know is a GPU/CPU maker so how is mining crypto with ASICs even relevant? Ethereum is mined with GPUs but I think that market doesn’t have a lot of slack with or without crypto.
Yes, Intel has floundered for a number of years, but at the same time they are a foundry themselves, second only to TSMC (for now), have incredible manufacturing capacity, and seem poised to start pushing interesting CPU designs as well as having their foundry half finally start executing correctly. Also their growth potential for their GPU business is huuuuuuge.
As someone who enjoys AMD for the competition they bring, I am worried what the next few years might look like for them for the following reasons:
1. AMD GPUs were actually very competitive with their RDNA2 architecture, but at the same time AMD enjoyed a huge silicon advantage as they were on TSMC while NVIDIA was on Samsung (for consumer). The next generation of NVIDIA consumer GPUs will be made by TSMC and if their professional GPUs (see A100 which are made by TSMC) are any indication we can see potentially up to 1.5x to 2x efficiency bump for NVIDIA.
2. Intel seems to be cleaning up shop and their latest generation of CPUs feel like a 1st Gen Ryzen moment. Also, Intel does not have to clearly surpass AMD, they just need to be competitive as they can simply out produce AMD given AMDs relatively small allocation at TSMC.
3. Intel GPUs pose a moderate threat on the horizon for AMD for both low end and portables and at the same time they are eating up available capacity at TSMC.