"Overall, for 2021 AMD booked $16.4B in revenue, which was an increase of 68% over 2019" Looks like time is not moving since the last time AT's GPU review 😉
I purchased AMD stock when it was trading at $3 or so in 2016 and was learning about the upcoming Ryzen, it was a long-term play but unfortunately, I sold it all for around $50 when the COVID news came out in Q1 2020. I wish I kept it longer. I did make some money from Tesla too but also sold it a bit early (gains of 3x versus 20x).
* Dektops: ADL wins over Zen3 * Laptops: ADL/Raptor Lake/Meteor Lake will win because of big.Little efficiency advantage. No big.Little until 2024 Zen4. Apple leads this category by a wide margin. Qualcomm Nuvia incoming competitor. * GPUs: Nvidia leads * AI: Nvidia leads by a wide margin * Server: All hyperscalers are going inhouse ARM. In 2021, 50% of all new EC2 instances were Graviton2. 2022 should be blowout for ARM. * Government contracts: AMD is pretty good here.
This is probably peak AMD. I'd rather buy Intel stocks at this point.
I guess AMD can always say they're in Xbox and Playstation. But AMD refuses to break up semi-custom so we have no idea how much money they even make from them.
While most of this comment is factually correct, it is so selectively presented and ahistorical that its points are really problematic. A snapshot in time that selectively ignores history and developments is not something to make predictions on. Yes, ADL beats Zen3, but at high power (for K SKUs) and AMD still sells every chip they make, growing rapidly. And, notably, Intel is ahead for the first time in a decent while, which in and of itself is a rather unprecedented situation. No reason to suggest Zen4 won't hit back hard. Idk where you're getting 2024 Zen 4 from - APUs are typically later than desktop CPUs, but it's entirely feasible for 7000-series to be Zen4 still - "leaked" roadmaps suggest Phoenix/7000 APUs are Zen4, though of course those aren't necessarily accurate. They might not be big.little, but that's not necessarily needed given how well Zen scales (especially if that rumored 16c CCD 50/50 perf/efficiency tuned core split happens). Whether Nvidia _leads_ in GPUs and AI isn't really relevant, as long as there's sales and gains to be made - that's how competition works. In terms of performance, and especially perf/price, AMD is highly competitive. That the decades-long incumbent market leader hasn't been knocked off their perch just demonstrates that entrenched market power is a significant strength. This of course makes things more difficult for AMD, but again, your post just outright dismisses any opportunity for success, which is just ... rather odd.
It's really not that important, no, but fanbots need excuses to post FUD and they will take literally any feature Intel has that AMD don't and declare it to be game-changing.
big.LITTLE might bring real-life (and in select cases/benchmarks) improvements in battery life for mobile systems. My work laptop has 3-4 hours battery life (just normal usage). That's acceptable, but there is space for improvements.
Possible, depending on what is meant by "lead'. b.L. allows you to be more aggressive with your performance cores (ARM appears to have forgotten the script with their X-cores), so single thread perf should be fire, and for MT you optimize your cores for certain load traces (or something a bit more complex/inclusive). That's gonna be expensive, in absolute terms, and is likely to have catastrophic failures for some loads. So, if nothing else, and assuming fab parity, AMD should be, in general, better at multi-core than adl. M
who cares about big litle and and total core count is higher when your big cores require that amount of power to be competitive and this on there leading edge node.... time will tell. Its all hail and praising talk about ADL but everybody forgets that its comparing with a year old AMD architecture. 3D cache will alter already quite the difference for many desktop results let alone what this will do combined with Zen4...
3D cache is great, but is very situational though (big.LITTLE is also situational, but less so). Some loadouts will benefit very little out of it (at least in terms of performance)
Zen 3 becomes irrelevant by 2023 at the latest, ADL in notebooks isn't notably more power-efficient than Zen 3, ADL in notebooks will be competing with Zen 3+, Raptor Lake will compete with Zen 4 (late 2022 / early 2024, not 2024), Apple lead only in Apple devices, Nvidia /barely/ lead in GPU performance and AMD are increasing market share (and profit share, more importantly), AMD lead in datacentre GPU compute performance, hyperscalers aren't the entire market and still use x86 extensively (50% according to you).
The things you've cited as problems for AMD (ARM, Apple, hyperscalers) are much larger problems for Intel. Kinda gives the game away about your biases.
AMD will grow by 35% ~ 40% in 2022. Intel will not even dream to touch these percentages.
AMD will continue to gain notebook market share as Intel's 125W ADL "mobile" CPU is not a reference for the notebook market.
Take 80 watts way from ADL's TDP and suddenly Ryzen 6000 in pole position.
AMD will continue Data Center growth at a pace significantly higher than Intel.
On the GPU side ... it remains to be seen how much can AMD grow and if nVidia doesn't come up with a significant competing set of products. The market is leaning more towards nVidia , despite AMD's higher efficiency.
In the past 2-3 years, it seems Intel and nVidia managed to brainwash everybody to not care about efficiency. The Intel Prescott days and nVidia Rankine seem to be back :)
Also, AMD's valuation will soon have to be adjusted for the Xillinix acquisition, so my stock target price is over 200 USD for this year, just based on the market's enthusiasm with the company.
Also, we expect some interesting surprise from GlobalFoundries this year.
TMSC is still the bread & butter, the bleeding edge, but hey .. when you can sell every chip you can possibly make, the stuff GF will announce will be a nice to have.
AMD has enormous space to grow and as long as they have the room to grow, they are and they will, at a much higher pace than Intel.
Sure, if you want your dividends, then Intel is the way to go.
* Desktop - Zen4 will win over ADL/Raptor lake with 20+% IPC and optimised TSMC 5nm, zen4 is even more than 1800x to 3800x performance increase.
* Laptops - currently AMD is leading in efficiency even against ADL big.little, ADL H youtube stream - 6 hrs , Rembrandt H youtube stream - 12hrs. Rembrandt U - 24hrs . Zen4 based pheonix in CES 2023 5nm will also beat Raptor lake Intel7
*GPU - 6900xt is head to head with rtx3090 but loses in Ray tracing and upscaling. RDNA3 MCM will include machine learning chip according to greymon (leaker) But can't say now about performance, maybe 4090 wins.
*HPC - AMD INSTINCT 1200 CDNA3 leaves nvidia A100 in dust. Milanx already rocking servers, upcoming Genoa with 96 Zen 4 cores.
Further - AMD supplying tesla, playstation, xbox, handhelds, samsung Exynos.
AMD leads in the DC/HPC segment where SPR is getting delay after delay. SPR will now be going head to head with Genoa, and will be trounced. Hopefully Intel gives a good enough discount on SPR to make it worth while, they have already had to do that with Aurora and its ongoing delays.
ADL beat Zen3 for sure! Zen3 is a year old now, I should hope Intel can catch a year old product. Zen 4 launching this year against raptor lake will be the true test.
GPU's is Nvidias market to lose, and AMD is eating into that as will Intel. AMD RDNA3 (7000 series) MCM designs are set to beat nvidias monolithic 4000 series this year. So AMD will take the performance crown there.
AI Nvidia owns this for sure, but with all of the R&D that has come into AMD you can be sure this is a focal point for them.
AMD is a juggernaut that has just started to gain speed, the next few years are going to be amazing for technology and AMD will be driving that competition. So stop with the FUD, and be happy we have 2-3 big players all with competitive products. AMD is in a awesome position for the future now to compete head on with both Nvidia and Intel.
Software is pretty difficult to scale, just dumping in money often doesn't work. But more resources/money/focus on ROCm/OpenCL seems appropriate now - afaik Cuda is def one of the reasons why Instinct isn't as successful as could be.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
26 Comments
Back to Article
0iron - Tuesday, February 1, 2022 - link
"Overall, for 2021 AMD booked $16.4B in revenue, which was an increase of 68% over 2019"Looks like time is not moving since the last time AT's GPU review 😉
Ryan Smith - Wednesday, February 2, 2022 - link
Time is moving. Apparently I just can't count!Calin - Wednesday, February 2, 2022 - link
Meanwhile, for 2021 Intel had about $80 billions of revenue compared to the $16 billion of AMD.AMD has a lot of space to grow ;)
krumme - Wednesday, February 2, 2022 - link
AMD market cap is 141B vs Intel market cap of 199B.AMD market cap in 2015 was 1.3B ! - From 1.3 to 141 in 6 years or so. Thats a stunning change.
Shlong - Wednesday, February 2, 2022 - link
I purchased AMD stock when it was trading at $3 or so in 2016 and was learning about the upcoming Ryzen, it was a long-term play but unfortunately, I sold it all for around $50 when the COVID news came out in Q1 2020. I wish I kept it longer. I did make some money from Tesla too but also sold it a bit early (gains of 3x versus 20x).Spunjji - Wednesday, February 2, 2022 - link
That's a big pie for AMD, Nvidia and ARM to take increasingly large slices of.lemurbutton - Wednesday, February 2, 2022 - link
Where does AMD actually lead in the next 2 years?* Dektops: ADL wins over Zen3
* Laptops: ADL/Raptor Lake/Meteor Lake will win because of big.Little efficiency advantage. No big.Little until 2024 Zen4. Apple leads this category by a wide margin. Qualcomm Nuvia incoming competitor.
* GPUs: Nvidia leads
* AI: Nvidia leads by a wide margin
* Server: All hyperscalers are going inhouse ARM. In 2021, 50% of all new EC2 instances were Graviton2. 2022 should be blowout for ARM.
* Government contracts: AMD is pretty good here.
This is probably peak AMD. I'd rather buy Intel stocks at this point.
lemurbutton - Wednesday, February 2, 2022 - link
I guess AMD can always say they're in Xbox and Playstation. But AMD refuses to break up semi-custom so we have no idea how much money they even make from them.Spunjji - Wednesday, February 2, 2022 - link
Big on revenue, small on margin. Not really a problem though as your post is misleading FUD.Valantar - Wednesday, February 2, 2022 - link
While most of this comment is factually correct, it is so selectively presented and ahistorical that its points are really problematic. A snapshot in time that selectively ignores history and developments is not something to make predictions on. Yes, ADL beats Zen3, but at high power (for K SKUs) and AMD still sells every chip they make, growing rapidly. And, notably, Intel is ahead for the first time in a decent while, which in and of itself is a rather unprecedented situation. No reason to suggest Zen4 won't hit back hard. Idk where you're getting 2024 Zen 4 from - APUs are typically later than desktop CPUs, but it's entirely feasible for 7000-series to be Zen4 still - "leaked" roadmaps suggest Phoenix/7000 APUs are Zen4, though of course those aren't necessarily accurate. They might not be big.little, but that's not necessarily needed given how well Zen scales (especially if that rumored 16c CCD 50/50 perf/efficiency tuned core split happens). Whether Nvidia _leads_ in GPUs and AI isn't really relevant, as long as there's sales and gains to be made - that's how competition works. In terms of performance, and especially perf/price, AMD is highly competitive. That the decades-long incumbent market leader hasn't been knocked off their perch just demonstrates that entrenched market power is a significant strength. This of course makes things more difficult for AMD, but again, your post just outright dismisses any opportunity for success, which is just ... rather odd.lemurbutton - Wednesday, February 2, 2022 - link
Correction: AMD won't have big.Little until Zen5, which is about 3 years away. In 3 years, it'll be Intel's 3rd or 4th big.Little iteration.patel21 - Wednesday, February 2, 2022 - link
Is big little really that important for taking the lead?Spunjji - Wednesday, February 2, 2022 - link
It's really not that important, no, but fanbots need excuses to post FUD and they will take literally any feature Intel has that AMD don't and declare it to be game-changing.Calin - Thursday, February 3, 2022 - link
big.LITTLE might bring real-life (and in select cases/benchmarks) improvements in battery life for mobile systems.My work laptop has 3-4 hours battery life (just normal usage). That's acceptable, but there is space for improvements.
tuxRoller - Tuesday, February 8, 2022 - link
Possible, depending on what is meant by "lead'.b.L. allows you to be more aggressive with your performance cores (ARM appears to have forgotten the script with their X-cores), so single thread perf should be fire, and for MT you optimize your cores for certain load traces (or something a bit more complex/inclusive). That's gonna be expensive, in absolute terms, and is likely to have catastrophic failures for some loads.
So, if nothing else, and assuming fab parity, AMD should be, in general, better at multi-core than adl.
M
duploxxx - Wednesday, February 2, 2022 - link
who cares about big litle and and total core count is higher when your big cores require that amount of power to be competitive and this on there leading edge node.... time will tell.Its all hail and praising talk about ADL but everybody forgets that its comparing with a year old AMD architecture. 3D cache will alter already quite the difference for many desktop results let alone what this will do combined with Zen4...
Calin - Thursday, February 3, 2022 - link
3D cache is great, but is very situational though (big.LITTLE is also situational, but less so). Some loadouts will benefit very little out of it (at least in terms of performance)SunMaster - Thursday, February 3, 2022 - link
You seem to assume the performance crown will be held by a big+little cpu.Spunjji - Wednesday, February 2, 2022 - link
Boring anti-AMD FUD post is boring.Zen 3 becomes irrelevant by 2023 at the latest, ADL in notebooks isn't notably more power-efficient than Zen 3, ADL in notebooks will be competing with Zen 3+, Raptor Lake will compete with Zen 4 (late 2022 / early 2024, not 2024), Apple lead only in Apple devices, Nvidia /barely/ lead in GPU performance and AMD are increasing market share (and profit share, more importantly), AMD lead in datacentre GPU compute performance, hyperscalers aren't the entire market and still use x86 extensively (50% according to you).
The things you've cited as problems for AMD (ARM, Apple, hyperscalers) are much larger problems for Intel. Kinda gives the game away about your biases.
IGTrading - Wednesday, February 2, 2022 - link
AMD will grow by 35% ~ 40% in 2022. Intel will not even dream to touch these percentages.AMD will continue to gain notebook market share as Intel's 125W ADL "mobile" CPU is not a reference for the notebook market.
Take 80 watts way from ADL's TDP and suddenly Ryzen 6000 in pole position.
AMD will continue Data Center growth at a pace significantly higher than Intel.
On the GPU side ... it remains to be seen how much can AMD grow and if nVidia doesn't come up with a significant competing set of products. The market is leaning more towards nVidia , despite AMD's higher efficiency.
In the past 2-3 years, it seems Intel and nVidia managed to brainwash everybody to not care about efficiency. The Intel Prescott days and nVidia Rankine seem to be back :)
Also, AMD's valuation will soon have to be adjusted for the Xillinix acquisition, so my stock target price is over 200 USD for this year, just based on the market's enthusiasm with the company.
Also, we expect some interesting surprise from GlobalFoundries this year.
TMSC is still the bread & butter, the bleeding edge, but hey .. when you can sell every chip you can possibly make, the stuff GF will announce will be a nice to have.
AMD has enormous space to grow and as long as they have the room to grow, they are and they will, at a much higher pace than Intel.
Sure, if you want your dividends, then Intel is the way to go.
The Futuristic - Thursday, February 3, 2022 - link
* Desktop - Zen4 will win over ADL/Raptor lake with 20+% IPC and optimised TSMC 5nm, zen4 is even more than 1800x to 3800x performance increase.* Laptops - currently AMD is leading in efficiency even against ADL big.little, ADL H youtube stream - 6 hrs , Rembrandt H youtube stream - 12hrs. Rembrandt U - 24hrs .
Zen4 based pheonix in CES 2023 5nm will also beat Raptor lake Intel7
*GPU - 6900xt is head to head with rtx3090 but loses in Ray tracing and upscaling.
RDNA3 MCM will include machine learning chip according to greymon (leaker)
But can't say now about performance, maybe 4090 wins.
*HPC - AMD INSTINCT 1200 CDNA3 leaves nvidia A100 in dust. Milanx already rocking servers, upcoming Genoa with 96 Zen 4 cores.
Further - AMD supplying tesla, playstation, xbox, handhelds, samsung Exynos.
Isn't this leadership?
psyclist80 - Friday, February 4, 2022 - link
AMD leads in the DC/HPC segment where SPR is getting delay after delay. SPR will now be going head to head with Genoa, and will be trounced. Hopefully Intel gives a good enough discount on SPR to make it worth while, they have already had to do that with Aurora and its ongoing delays.ADL beat Zen3 for sure! Zen3 is a year old now, I should hope Intel can catch a year old product. Zen 4 launching this year against raptor lake will be the true test.
GPU's is Nvidias market to lose, and AMD is eating into that as will Intel. AMD RDNA3 (7000 series) MCM designs are set to beat nvidias monolithic 4000 series this year. So AMD will take the performance crown there.
AI Nvidia owns this for sure, but with all of the R&D that has come into AMD you can be sure this is a focal point for them.
AMD is a juggernaut that has just started to gain speed, the next few years are going to be amazing for technology and AMD will be driving that competition. So stop with the FUD, and be happy we have 2-3 big players all with competitive products. AMD is in a awesome position for the future now to compete head on with both Nvidia and Intel.
Tams80 - Saturday, February 5, 2022 - link
They sell every processor that they can make.If they're products were bad, people wouldn't buy them, even if there was no other choice. But they aren't and they do what customers want them to do.
Maxime Chambonnet - Wednesday, February 2, 2022 - link
Now with this money if ROCm could become a thing that'd be great!Mil0 - Friday, February 4, 2022 - link
Software is pretty difficult to scale, just dumping in money often doesn't work. But more resources/money/focus on ROCm/OpenCL seems appropriate now - afaik Cuda is def one of the reasons why Instinct isn't as successful as could be.Oxford Guy - Tuesday, February 8, 2022 - link
Turning OEM scalper pricing into...