The Intel Execution in [2024]

[raytracingfan]

The Intel-AMD cross-licensing agreement for x86 will be terminated if either company undergoes a merger or acquisition. The earliest AMD64 patents have already expired, but even so, making a performant x86_64 without infringing on AMD's IP is a potential legal minefield. I doubt any company will attempt a merger with Intel unless they know AMD will be willing to re-negotiate the cross-licensing agreement.

 

[pcchen]

I think the question is, does Intel need a "rescue" now?
Although Intel has a bad quarter, they still have quite some cash on hand (~$29B from last quarter's report), they are not in immediate danger for maybe a few years. So, they are not really in a hurry to accept some takeover bid.
If there's a deal which can strengthen both sides' positions I think it might have a slight chance.

 

[pcchen]

The Intel-AMD cross-licensing agreement for x86 will be terminated if either company undergoes a merger or acquisition. The earliest AMD64 patents have already expired, but even so, making a performant x86_64 without infringing on AMD's IP is a potential legal minefield. I doubt any company will attempt a merger with Intel unless they know AMD will be willing to re-negotiate the cross-licensing agreement.

Ironically such terms was there to prevent AMD from selling itself But I don't think that's a huge problem because AMD also needs a lot of Intel's patents so if such deal actually happens they'll be happy to negotiate a new agreement, although probably with better terms.

 

[tunafish]

I think the question is, does Intel need a "rescue" now?
Although Intel has a bad quarter, they still have quite some cash on hand (~$29B from last quarter's report), they are not in immediate danger for maybe a few years. So, they are not really in a hurry to accept some takeover bid.

If Intel wants to keep the fabs, they need to invest more than $100B in the next 3-4 years to keep them current. They absolutely cannot survive on that $29B, they need massive external capital investment.

 

[pcchen]

If Intel wants to keep the fabs, they need to invest more than $100B in the next 3-4 years to keep them current. They absolutely cannot survive on that $29B, they need massive external capital investment.

Of course, but that can't be solved by a merger. Qualcomm itself probably can't invest this much money (Qualcomm has less cash than Intel). On the other hand, if the capital market is willing to pay, they don't need to be merged with Qualcomm (or anyone else) to do so. They can easily issue their own corp bond to raise the money (just like what TSMC did).

 

[homerdog]

How did Intel fall so far behind in manufacturing? A decade ago it seemed they were as far ahead as they are behind these days.

 

[Kaotik]

How did Intel fall so far behind in manufacturing? A decade ago it seemed they were as far ahead as they are behind these days.

They failed at 10 nm and tried to fix it for too long

 

[homerdog]

They failed at 10 nm and tried to fix it for too long

Right but what was so messed up with their 10nm? Nobody else seemed to have this problem, or at least they moved on quickly enough.

Intel's manufacturing is really a hero to zero story and I'm interested in how they could fail so spectacularly when everything seemed to be going so right for them.

 

[DegustatoR]

Right but what was so messed up with their 10nm? Nobody else seemed to have this problem, or at least they moved on quickly enough.

Intel's manufacturing is really a hero to zero story and I'm interested in how they could fail so spectacularly when everything seemed to be going so right for them.

Well, they are a gen behind TSMC. Which is pretty much where everyone is in the semi world. I dunno if that means that they've "failed" exactly - or that TSMC just was better than everyone else.
It is also worth remembering that the whole "hero to zero" happened once AMD switched their CPUs to TSMC for production. Prior to that Intel were doing relatively fine even with all the issues they've had with the 10nm node.

 

[Kaotik]

Right but what was so messed up with their 10nm? Nobody else seemed to have this problem, or at least they moved on quickly enough.

Intel's manufacturing is really a hero to zero story and I'm interested in how they could fail so spectacularly when everything seemed to be going so right for them.

To my understanding main issues were
- Too aggressive schedule leads to all sorts of trouble
- Too ambitious targets (36nm metal pitch, while TSMC's 7nm hits 40nm)
- Contact Over Active Gate didn't work (and got eventually scrapped)
- Some issues with cobalt which were solved

 

[Seanspeed]

To my understanding main issues were
- Too aggressive schedule leads to all sorts of trouble
- Too ambitious targets (36nm metal pitch, while TSMC's 7nm hits 40nm)
- Contact Over Active Gate didn't work (and got eventually scrapped)
- Some issues with cobalt which were solved

Yea, the common narrative you'll see from many online is that Intel just got complacent, but it's really quite the opposite. Intel had gotten quite comfy with their essentially two node gen advantage over the competition, and with 14nm dropping them back a bit in schedule, they likely just pushed too hard to get 10nm back into extreme advantage territory, not happy with a mere 'good' advantage.

And then having designed all their new architectures post-Skylake specifically for 10nm, they were genuinely completely stuck until it got properly resolved, and the bar was high to get 10nm in shape not only because of competition shooting up, but also because in terms of performance, they had pushed 14nm to such a degree that it was even harder to get 10nm to a level where it didn't have a large clock disadvantage, which was significant in the desktop market.

 

[Sega_Model_4]

Intel missed the smartphone moment and was comfortable in the PC world.
The immense competition and evolution of smartphones made TSMC evolve its manufacturing method very quickly.

 

[Kaotik]

Intel missed the smartphone moment and was comfortable in the PC world.
The immense competition and evolution of smartphones made TSMC evolve its manufacturing method very quickly.

It doesn't matter what chips you're trying to make if your process is broken

 

[arandomguy]

Intel didn't exactly miss the smartphone moment. They did have a large concious part in passing on manufacturing for Apple and the iphone but I feel that's a bit revisionist in terms of the actual impact. Apple afterall didn't start with TSMC and stick through them even though you associate Apple with TSMC today. So it's far from guaranteed that had Intel aquisced to Apple's requirements for the first iphone that Apple would've stuck with them until today.

In terms of smartphones in general Intel wasn't setup as a third party fab. They were always going to be limited in terms of directly penetrating the smart phone market without a modem and modem IP (which was corned by Qualcomm). Nvidia faced the same problem as well.

 

[DavidGraham]

Intel is claiming to have found the root cause of 13th and 14th gen problems.

https://twitter.com/x/status/1839125650596393290

https://twitter.com/x/status/1839126397543133565

 

[DavidGraham]

Sadly it seems even if setting at idle, the 13th and 14th gen CPUs also degrades. Intel is releasing a new microcode update to fix that.

Microcode and BIOS code requesting elevated core voltages which can cause Vmin shift especially during periods of idle and/or light activity.

Mitigation: Intel® is releasing microcode 0x12B, which encompasses 0x125 and 0x129 microcode updates, and addresses elevated voltage requests by the processor during idle and/or light activity periods.

Intel Core 13th and 14th Gen Desktop Instability Root Cause Update

Following extensive investigation of the Intel® Core™ 13th and 14th Gen desktop processor Vmin Shift Instability issue, Intel can now confirm the root cause diagnosis for the issue. This post will cover Intel’s understanding of the root cause, as well as additional mitigations and next steps for...

community.intel.com

 

[Albuquerque]

Good news is they found it.

Bad news is, if they're fighting vMin at idle and low loads, then basically everyone's 13th and 14th gen ended up taking some sort of hit. Is it enough to cause them all to fail? Now we're back in a race against entropy, so there's probably no way to truly know without literally destroying the chip after it's already dead in a sort of silicon autopsy. Bleh.

 

[Shifty Geezer]

What and how this 'clock tree circuit' and vMin? Is the silicon of this circuit the one that's degrading, or is the whole chip degrading?

 

[pharma]

Intel Customizing Granite Rapids Server Chips for Nvidia GPUs

Intel is now customizing its latest Xeon 6 server chips for use with Nvidia’s GPUs that dominate the AI landscape. The chipmaker’s new Xeon 6 chips, also called Granite Rapids, […]

www.hpcwire.com

Intel is now customizing its latest Xeon 6 server chips for use with Nvidia’s GPUs that dominate the AI landscape. The chipmaker’s new Xeon 6 chips, also called Granite Rapids, have been customized and validated specifically for server boxes with Nvidia’s latest and upcoming GPUs.

“Nvidia is the leader on the GPU side…so we’re partnering closely with them to make sure that people deploying MGX or HGX-based systems, we have a full suite of CPUs that have been qualified together with Nvidia for those systems,” said Ronak Singhal, senior fellow at Intel.

 

[DavidGraham]

Intel Customizing Granite Rapids Server Chips for Nvidia GPUs

Intel is now customizing its latest Xeon 6 server chips for use with Nvidia’s GPUs that dominate the AI landscape. The chipmaker’s new Xeon 6 chips, also called Granite Rapids, […]

www.hpcwire.com

It’s a shocking reversal of roles. There was a time when Intel’s server chips ruled the roost while Nvidia’s GPUs relied heavily on CPU sales. Intel is now playing second fiddle to Nvidia.

I very well remember when the rumor mill was saying that Intel is threatening NVIDIA to cut them off from their PCI-Express interconnect. Now the situation is reversed!

I also remember, back in 2011 when Charlie Demerjian (of SemiAccurate) was shouting from rooftops that NVIDIA is living on a borrowed time, as Intel and AMD are going to corner NVIDIA with their invincible APUs and iGPUs, leaving NVIDIA with a very small niche market, on their way to extinction.

How the times have changed!