Tech "intel's f-----"

lol AMD’s server market share isn’t even in the double digits.

it might actually be double digits as of now-ish, but yeah. the greater point is how hard intel's been pushing xeon and still losing share to amd - and amd's been seemingly selling all the epycs that they get. with greater capacity from tsm, amd's expected to increase their server share faster than previously expected*. but this also pushes into rumour/unofficial and based off recent news/tidbits (ie: tsm/huawei).

iirc, su said they were aiming for mid-teens by q4 during the last earning's call. which seemed to startle/disappoint some analysts who expected a larger figure. after the recent news, ~20%ish seems to be expected again, which might be good or bad, depending on results/etc.
 
  • Like
Reactions: PEB
lol AMD’s server market share isn’t even in the double digits.

This is not about currently this is about the future and I have friends who work on DELL storage servers and DELL/EMC have already moved to AMD. The message is as you move to massive multi-core server scale in the 1000's of cores AMD not just eating Intel's lunch it's taking out the dinner and the breakfast in the morning. Articles from 2019 hardly go into the details how far the landscape has shifted. Go ahead say I am an AMD fan boy and I will point out my current computer is Intel and Nvidia based machines. I just think it's amazing how much the landscape has changed even as Intel continues to make huge money.

"
Accelerate performance
Deliver better performance for traditional and emerging workloads with core or frequency updates, including:

  • 100%1 more processing cores and faster data transfers
  • 20%1 faster memory speed to reduce latency and deliver faster response
  • 2X1 PCIe performance with Gen4 at 16GT/s to overcome bottlenecks
"
https://www.delltechnologies.com/en...-sQAvD_BwE&gclsrc=aw.ds#accordion0&accordion1
 
This is not about currently this is about the future and I have friends who work on DELL storage servers and DELL/EMC have already moved to AMD. The message is as you move to massive multi-core server scale in the 1000's of cores AMD not just eating Intel's lunch it's taking out the dinner and the breakfast in the morning. Articles from 2019 hardly go into the details how far the landscape has shifted. Go ahead say I am an AMD fan boy and I will point out my current computer is Intel and Nvidia based machines. I just think it's amazing how much the landscape has changed even as Intel continues to make huge money.

"
Accelerate performance
Deliver better performance for traditional and emerging workloads with core or frequency updates, including:

  • 100%1 more processing cores and faster data transfers
  • 20%1 faster memory speed to reduce latency and deliver faster response
  • 2X1 PCIe performance with Gen4 at 16GT/s to overcome bottlenecks
"
https://www.delltechnologies.com/en...-sQAvD_BwE&gclsrc=aw.ds#accordion0&accordion1
Articles from 2020 show Intel making record profits in their Server division in the most recently reported Quarter.

I think you're just talking out of your ass, and you had no concept of the market.
 
as Intel continues to make huge money.

i mean... they definitely make a lot of money, but they're making less and less. the red flags are that their sales were higher than ever, but their profit was... flat/down. additionally, they recently took on a strange amount of debt for their position. it's customary for companies to take on cheap loans during covid/etc, but they were sitting on cash - which brings into question what the hell they're doing.
 
Articles from 2020 show Intel making record profits in their Server division in the most recently reported Quarter.

I think you're just talking out of your ass, and you had no concept of the market.

like i already said, they changed that category entirely, so it includes more than servers/etc. it's why "data center" became "data centric," (2 ERs ago, i think?).

ie: their numbers for that are inflated because they merged it with another department.
 
They didn't merge any departments, and it has nothing to do with internal reorganization. MobileEye was already part of the DCG in Q1 2019, so those YoY and QoQ earnings surges were completely apples to apples.

This shit is so basic.
 
dcg (data center group) restructured and there is now dpg (platforms).
 
this thread make my braen big
 
I only skimmed all those updates. Don’t really feel like trying to figure that all out especially since I’m sure we can’t see the whole story. Seems like both companies are doing well this year, so that is great. I’m happy for them. Hopefully the space keeps getting more competitive. I miss stressing over which parts I should buy
 
<Grimes01>

https://www.tomshardware.com/news/i...m-processors-now-one-year-behind-expectations

https://finance.yahoo.com/news/intel-considers-once-heresy-not-013625863.html

“You didn’t need to read any more,” Sanford C. Bernstein analyst Stacy Rasgon said. “Whatever little credibility they had is out of the window.”


and old, but funny:
91e70ivgqca51.png



thread tl;dr - when this started, intel had the largest market cap of all the semis and they claimed to have the best fab.

their fab sucks, they're still stuck on 14nm (++++++++++++++++++++++++++), they've taken on a ton of debt, their buybacks ended (it's like focusing on boosting the share price instead of products wasn't a good idea or something?!), and they're not even the 2nd largest market cap (nvidia and tsm overtook them... even before today's collapse). they'll be lucky to have the 3rd best fab behind tsm and samsung.

lolzturtle.gif
lolzturtle.gif
lolzturtle.gif
 
Last edited:
I wonder how close 7nm is to the limit for how small they can go before they can’t go any smaller.

For anyone that’s interested this is an interesting video from some university professor from 2013 when 22nm was the present transistor size and he talks about how probably by 2025 transistors won’t be able to get any smaller because of quantum tunneling, and Moore’s law will come to an end.



If 7nm is small enough for quantum tunneling to be a problem maybe that’s part of Intel’s difficulties.
 
I wonder how close 7nm is to the limit for how small they can go before they can’t go any smaller.

For anyone that’s interested this is an interesting video from some university professor from 2013 when 22nm was the present transistor size and he talks about how probably by 2025 transistors won’t be able to get any smaller because of quantum tunneling, and Moore’s law will come to an end.



If 7nm is small enough for quantum tunneling to be a problem maybe that’s part of Intel’s difficulties.


tsm is already developing finfet down to 2-3nm and samsung is looking at ~2nm via gaafet, iirc.
 
Wow. So what’s going on with Intel’s problems at 7nm?

tl;dr - their fab's fucked and the yields are unworkable.

samsung's also had problems with low yields on their advanced nodes, but intel has compounded their problems. tsm is vastly ahead.
 
Damn. WTF is going on at Intel? What a fall from grace.
 
I did see that AMD's price share is finally worth more than Intel's, first time ever.
 
Wow. So what’s going on with Intel’s problems at 7nm?
It's due to the higher sophistication of their fabrication design. It's more difficult to shrink a busier architecture.

Nevertheless, Samsung isn't a peer (nor yet a competitor) to Intel in CPU design, and they're own CPUs shat the bed this year. Their latest Exynos was outperformed by Qualcomm so badly that for the first time in the Galaxy line's history they put the Qualcomm in the flagship Galaxy models released in South Korea itself. Historically, only the USA versions have had Qualcomm, and for a select few special variants for some of the international phones, but never for South Korea.
 
It's due to the higher sophistication of their fabrication design. It's more difficult to shrink a busier architecture.

Nevertheless, Samsung isn't a peer (nor yet a competitor) to Intel in CPU design, and they're own CPUs shat the bed this year. Their latest Exynos was outperformed by Qualcomm so badly that for the first time in the Galaxy line's history they put the Qualcomm in the flagship Galaxy models released in South Korea itself. Historically, only the USA versions have had Qualcomm, and for a select few special variants for some of the international phones, but never for South Korea.

Are you saying Samsung reserves its best products for Korea and exports its B team products? What an odd business strategy.
 
Back
Top