Intel is reportedly going to kill the CPU socket


Recommended Posts

Apart from that, I still think Phouchg's post is excellent, as it is a nice take on how some people would be totally careless about it happening and the message is true with other things going on in the tech industry where many people downplay the importance of repairability, maintaining your gear yourself and so forth.

I often feel alienated with my sentiments around some people, so yeah, nice read!

Glassed Silver:ios

Well, this is basically the situation with laptops. Yes, it might be possible to replace things, but the average consumer won't, and the trend it make the cases near impossible to open / tamper with. If this happened to consumer desktops, well, it wouldn't surprise me.

The thing you fail to note is that if, say, your GPU blows up, there's no fixing it... you send it off to get RMA'd, or toss it in the trash. We're well beyond the days of breaking out the soldering iron and attaching some new capacitors, so it's really just a matter of throw it out in pieces, or throw it out all at once. From a quality control standpoint, something soldered on in house in a cleanroom is obviously going to be more reliable than something pieced together in someone's office workshop.

Beyond all this is the fact that if you plug a component in and it works, the odds of it failing in its useable lifespan is pretty slim. I think PC enthusiasts still worry about this because we're often dealing with components that are intentionally overworked or relatively poorly built (specifically graphics cards, which are the only things I've had die on me in my years of PC building) and will fail randomly. The failure rate on a properly designed CPU / board / memory has to be incredibly small.

Well, this is basically the situation with laptops. Yes, it might be possible to replace things, but the average consumer won't, and the trend it make the cases near impossible to open / tamper with. If this happened to consumer desktops, well, it wouldn't surprise me.

The thing you fail to note is that if, say, your GPU blows up, there's no fixing it... you send it off to get RMA'd, or toss it in the trash. We're well beyond the days of breaking out the soldering iron and attaching some new capacitors, so it's really just a matter of throw it out in pieces, or throw it out all at once. From a quality control standpoint, something soldered on in house in a cleanroom is obviously going to be more reliable than something pieced together in someone's office workshop.

Beyond all this is the fact that if you plug a component in and it works, the odds of it failing in its useable lifespan is pretty slim. I think PC enthusiasts still worry about this because we're often dealing with components that are intentionally overworked or relatively poorly built (specifically graphics cards, which are the only things I've had die on me in my years of PC building) and will fail randomly. The failure rate on a properly designed CPU / board / memory has to be incredibly small.

That was the point of the post GS quoted really...

The failure rate may be low (no data on if it is low or not), but when you fall on the unfavorable side of the statistic you can recover quickly. Otherwise, you're at the mercy of someone else to get you up and running and they won't care half as much as you do about your ability to get back up and running quickly.

Additionally, even with super high tech clean rooms and etc. you still have parts that fail prematurely or arrive dead. So soldering it all in one unit and building it all in house won't eliminate this. It may lower it, but again you're hoping not to fall on the wrong side of the statistic as you're at the mercy of someone else when you do.

  • Like 1

It's inevitable for some basic engineering purposes that every EE has to deal with: Cost, thermal, parasitics, size.

These sockets are EXPENSIVE.

They offer decreased thermal performance. You can use thermal vias with a BGA package, and suck heat out the back of the board.

Electrically, sockets tend to suck, a lot. You're sticking a little LC network on every pin. That's bad, and hurts switching. You can get lower voltages and better performance off a BGA.

And size. Ah size. LGA2011 has a pitch of ~1mm in a hex array. 0.3mm square pitches exist in BGA form. The pin density available is insanely better. You can't achieve that with a socket.

i could see Intel leaving the Enthusiast market where it is.... say, the current 2011 socket is only available to Enthusiasts with swappable CPUs, while the 1155 socket/cpus are the ones that are soldered.

It's inevitable for some basic engineering purposes that every EE has to deal with: Cost, thermal, parasitics, size.

These sockets are EXPENSIVE.

They offer decreased thermal performance. You can use thermal vias with a BGA package, and suck heat out the back of the board.

Electrically, sockets tend to suck, a lot. You're sticking a little LC network on every pin. That's bad, and hurts switching. You can get lower voltages and better performance off a BGA.

And size. Ah size. LGA2011 has a pitch of ~1mm in a hex array. 0.3mm square pitches exist in BGA form. The pin density available is insanely better. You can't achieve that with a socket.

sounds like you just finished some related college course :p

great, so now to upgrade a CPU I have to buy a new motherboard...... Motherboard issues, but a new CPU too..... and vise versa....

people wonder why the desktop market is going down, it's stuff like this that takes it down for the builder market.....

If they are going BGA instead of LGA couldn't we still have sockets though? just because it's BGA doesn't mean it has to be soldered to a board, they do have BGA sockets out there

heck maybe this chip line will be laptop only, which would make sense with a BGA package

The name of the source is Semiaccurate.com. That should leave some room for error.

That's what i thought at first too,... then i stumbled upon this other one over the weekend

http://www.legitreviews.com/news/14561/

AM3+ Here i come :woot:

That was the point of the post GS quoted really...

The failure rate may be low (no data on if it is low or not), but when you fall on the unfavorable side of the statistic you can recover quickly. Otherwise, you're at the mercy of someone else to get you up and running and they won't care half as much as you do about your ability to get back up and running quickly.

Additionally, even with super high tech clean rooms and etc. you still have parts that fail prematurely or arrive dead. So soldering it all in one unit and building it all in house won't eliminate this. It may lower it, but again you're hoping not to fall on the wrong side of the statistic as you're at the mercy of someone else when you do.

You're always at the mercy of someone else. As I said before, it's not like if my CPU dies I pop it out and rewire the transistors. It goes in the bin and I get a new one. If the whole thing goes, it's just a matter of cost, and if soldering the chips on makes the process cheaper, then that's a non issue as well.

To put it another way, if your CPU or Mobo dies, your SOL in either scenario. I imagine the case where you happen to have one or the other with a compatible chip lying around is pretty slim.

sounds like you just finished some related college course :p

Not really. I design industrial power supplies, for the most part. Haven't been in college for several years. The problems I listed are pretty basic, and every design decision to fix any problem always comes with a price.

It's inevitable for some basic engineering purposes that every EE has to deal with: Cost, thermal, parasitics, size.

These sockets are EXPENSIVE.

They offer decreased thermal performance. You can use thermal vias with a BGA package, and suck heat out the back of the board.

Electrically, sockets tend to suck, a lot. You're sticking a little LC network on every pin. That's bad, and hurts switching. You can get lower voltages and better performance off a BGA.

And size. Ah size. LGA2011 has a pitch of ~1mm in a hex array. 0.3mm square pitches exist in BGA form. The pin density available is insanely better. You can't achieve that with a socket.

Actually the downside or benefit (downside for consumers, benefit for production company) of BGA is that any BGA that gets hot causes the lead-free solder to melt and over time causes joints to come into contact with each other or not connect the bottom of the BGA to the board.

I don't see how sockets suck much if any power. If you're talking about nano-volts they yes, maybe, the resistance in the pins would be nil unless you could get an incredibly accurate meter that could measure to thousands of a nano-ohm. Again, solder on BGA means they need to run cooler and to run cooler then need to run with less performance.

Intel will realize once again why the enthusiast market should not be over looked.

We may not be the demographic they are targeting, but we influence the demographic they do target very heavily. If they go through with this, I personally (and my company) will push AMD.

Don't Intel force you to buy a new socket board for each new CPU release anyway?

That's not really the point. Even if you always buy a new motherboard with a new CPU, if they were integrated you would have far less choices. Right now ASUS makes 9 socket 1155 motherboards. Intel has ~30 1155 processors. You can mix and match between ~270 combinations to maximize performance for your budget. If Mobo+CPU were integrated, you can be damn sure than ASUS wouldn't manufacture 270 different motherboard+cpu combos.

Actually the downside or benefit (downside for consumers, benefit for production company) of BGA is that any BGA that gets hot causes the lead-free solder to melt and over time causes joints to come into contact with each other or not connect the bottom of the BGA to the board.

I don't see how sockets suck much if any power. If you're talking about nano-volts they yes, maybe, the resistance in the pins would be nil unless you could get an incredibly accurate meter that could measure to thousands of a nano-ohm. Again, solder on BGA means they need to run cooler and to run cooler then need to run with less performance.

If the BGA joint fails then CTE matching and cooling was done incorrectly. (nVidia knows the CTE mismatch problem quite well. Remember all those laptop failures?)

nano-ohm

Who is talking ohms?

I said they act like a very tiny LC network. Even a few nanohenries of inductance / a few picofarads of capacitance is non-trivial at these speeds. Why do you think they put zigzags/squiggles in traces on a board for these signals? Gotta keep all things equal for good signal integrity.

That's not really the point. Even if you always buy a new motherboard with a new CPU, if they were integrated you would have far less choices. Right now ASUS makes 9 socket 1155 motherboards. Intel has ~30 1155 processors. You can mix and match between ~270 combinations to maximize performance for your budget. If Mobo+CPU were integrated, you can be damn sure than ASUS wouldn't manufacture 270 different motherboard+cpu combos.

Yea true, probably a couple low spec, med spec and high end, much the same choice we get with Windows versions

AMD isn't going anywhere. I don't know how much longer they'll be in the CPU-only business for, but they'll be around.

well that's what I meant, their CPU division goes under

A lot of consumer PCs have soldered chips already.

which? outside of thin systems, and laptops, and ultra low wattage systems, virtually all desktops are LGA type systems

This topic is now closed to further replies.
  • Posts

    • They keep your data encrypted too. Including the calendar.
    • Microsoft makes billions redirecting people to use Bing in Microsoft Edge and selling OneDrive space. All thanks to Windows.
    • Teams is cancer, ebola and aids combined.
    • Claude on Windows is eating up massive amounts of RAM, with no way to stop it by Usama Jawad Anthropic has been in the headlines a lot lately, primarily due to its latest revenue and valuation figures, along with its release of its state-of-the-art (SOTA) Fable model. While its flagship product, Claude, may be very popular among millions of users, a lot of them are now reporting memory issues when using the tool. Over on Claude Code's GitHub repository, an issue raised in February has been gaining traction once again. Basically, Claude Desktop on Windows spins up a 1.8GB Hyper-V virtual machine if you use Claude Cowork or agent mode even once. This happens on each launch of Claude Code even if you plan to use the tool in chat mode only. Several users have upvoted this bug and stated that it's happening on their machine as well. However, it seemingly affects only Claude desktop users on Windows, not customers of the CLI or any other platform. Once the bug is triggered, it also shows a Vmmem process in Task Manager, indicating CPU usage of 0% and RAM utilization of a whopping ~1.8GB. Claude users complain that this process should only spin up when you explicitly launch agent mode or Cowork in Claude, with session files efficiently cleaned up after use. Additionally, they are calling for Claude to gracefully handle the absence of virtual machine-based infrastructure, without compromising on chat performance. It's unclear when this issue originated or what the root cause behind it is, but people are once again actively engaging in the GitHub thread as well as Hacker News. You can also find other technical details and log events over on GitHub. It's unclear if Anthropic will look into this issue, especially since it's already been reported for a few months. However, the bug is also causing major annoyance for users, with many claiming that it has led them to uninstall Claude Code on desktop, as a concrete workaround is not yet known.
    • "The US innovates, China replicates, Europe regulates" -- let's see who makes the cut
  • Recent Achievements

    • One Month Later
      Sopa flores earned a badge
      One Month Later
    • First Post
      StaticMatrix earned a badge
      First Post
    • Week One Done
      StaticMatrix earned a badge
      Week One Done
    • Rookie
      lamborghiniv10 went up a rank
      Rookie
    • One Month Later
      pinnclepd earned a badge
      One Month Later
  • Popular Contributors

    1. 1
      +primortal
      534
    2. 2
      PsYcHoKiLLa
      209
    3. 3
      +Edouard
      154
    4. 4
      Steven P.
      100
    5. 5
      ATLien_0
      84
  • Tell a friend

    Love Neowin? Tell a friend!