Intel Xeon "Sapphire Rapids" Volume Shipping Delayed Again: Company

[btarunr]

Intel's ambitious next-generation server processor, Xeon "Sapphire Rapids," hit its second shipping delay this year, according to the company. Speaking at the Bank of America Securities Global Technology Conference, Intel's Sandra Rivera stated that the volume ramp for "Sapphire Rapids," is not going as planned, indicating that its general availability could be delayed for the second time after original plans to do so in the first quarter of 2022.

Riviera was quick to defend the Intel 7 silicon fabrication node (10 nm Enhanced SuperFin) that the "Sapphire Rapids" chip is based on. "One thing I didn't mention on Sapphire, it sits on - it's on our 7-nanometer node and so the process is quite healthy. In fact Alder Lake, which is our client product ramped 15 million units. I think we announced at Q1 earnings, which makes that the fastest ramping, you know one of the fastest ramping client products in almost a decade."



The arrival of "Sapphire Rapids" is particularly important for Intel as demand for server processors are at an all time high, with growth in cloud data-centers. AMD is readying its 4th Generation EPYC "Genoa" Zen 4 processors for later this year. Both "Genoa" and "Sapphire Rapids" herald support for next-generation I/O, including DDR5 memory and PCI-Express Gen 5.

View at TechPowerUp Main Site | Source

 

[Tomorrow]

AMD is running away in the server market with performance and it will only get worse for Intel when AMD releases their 128c/256t Bergamo and 256c/512t Turin. Intel is still living on old marketshare and long contracts. Intel might still have a small edge in singe threaded and AVX-512 workloads. But TCO and multithreaded perf is better with AMD.

 

[usiname]

Now their intel 7 node is 7nm. How "unexpected"

 

[R0H1T]

Intel might still have a small edge in singe threaded and AVX-512 workloads.

They also have a slight(?) lead in AI acceleration but that'll vaporize once Xilinx IP makes their way into AMD server chips! The biggest lead for Intel is in 4s/8s systems with its higher density.

 

[KarymidoN]

AMD is running away in the server market with performance and it will only get worse for Intel when AMD releases their 128c/256t Bergamo and 256c/512t Turin. Intel is still living on old marketshare and long contracts. Intel might still have a small edge in singe threaded and AVX-512 workloads. But TCO and multithreaded perf is better with AMD.

yeah, but companies still slow to adapt to EPYC, a good example is 45Drives and their storinator, who still runs on XEON processors (no option to go AMD), even tho AMD Offers More Lanes for better storage servers (only the Hybryd storage server offers an option with AMD EPYC 7281, but the SSD servers still run on XEON). many other companies still slow to even offer an AMD option on their products.

 

[W1zzard]

It's almost like the bigger news is that they already produced 15 million Alder Lake CPUs?

 

[bonehead123]

I think we announced at Q1 earnings, which makes that the fastest ramping, you know one of the fastest ramping client products in almost a decade."

This chick does NOT seem to know for sure what the company is or is not doing, so why is she making public comments about their announcements, ramp-ups, or anything else for that matter. Sounds like they just needed someone, anyone, with limited knowledge of how to put a positive spin on yet ANUTHA epic failure, to say something so that when the media and stockholders start having a hissy fit over it, nobody with a real clue will be criticized.........

If I were a stockholder, this is not exactly the kind of statements I would want to hear....

Sooooo glad my main tech stocks are held in other companies....

 

[Shatun_Bear]

Intel are already a generation behind AMD in server CPU performance, this is a dire state of affairs as this should have been out last year.

 

[phanbuey]

They are way behind...

But then again:
China Must Seize Apple's Chipmaker in Taiwan If Sanctioned: Economist (businessinsider.com)

 

[eidairaman1]

AMD is running away in the server market with performance and it will only get worse for Intel when AMD releases their 128c/256t Bergamo and 256c/512t Turin. Intel is still living on old marketshare and long contracts. Intel might still have a small edge in singe threaded and AVX-512 workloads. But TCO and multithreaded perf is better with AMD.

Intels underhanded sneeky bs is catching up with them.

Intel are already a generation behind AMD in server CPU performance, this is a dire state of affairs as this should have been out last year.

 

[mechtech]

I wasn’t aware Intel had 7nm node.

 

[defaultluser]

I wasn’t aware Intel had 7nm node.

They keep rebranding shit. And meanwhile, they swear up-and-down that their actual 7nm (AKA Intel 4) will be here next year...but as-of May, they still haven't worked-out who will actually provide the process node tech

Intel might ditch proprietary 7 nm process for full TSMC N5 on its 2023 Meteor Lake processors

Despite Intel's latest reports on the steady advancements with the proprietary 7 nm production nodes, industry sources close to DigiTimes claim that Intel might change the Meteor Lake blueprints to be produced exclusively on the TSMC N5 nodes in order to avoid delays and ensure competitive...

www.notebookcheck.net

The reason Sapphire Rapids is going to be delayed indefinitely is because Intel can't mass-produce anything much larger than Tiger Lake 8C. That's why They patched-up the under-performing-unless-factory-overclocked Alder Lake with 8 efficiency cores, and kept the die size under 220. Raptor Lake is performing the same trick, but just doubling the efficiency cores.

 

[P4-630]

I must say that I'm pretty satisfied with my Alder Lake CPU otherwise.

 

[ixi]

Nothing new, nothing terrible, nothing unexpected. Just intel being intel.

Intel moto: we are good with being late.

 

[Deleted member 24505]

Intel, we might be late to the party, but when we do get here, it ROCKS

 

[Minus Infinity]

God how embarrassing. Moore's Law is dead alluded to all the delays coming including Raptor Lake barely shipping at all this year. So much hype on new products and we still haven't seen Arc, SR delayed, Ponte Vecchio barely in the wild, and they are hyping it successor, BS timeframes for Meteor and Arrow Lake.

AMD's Milan and Milan-X were already going to crush SR, and Turin will be the coupe de grace with 256 cores with Zen 5. I wonder about the scumbag, IT departments still insisting on using Intel in servers, what possible reason could they have to keep using grossly inferior products other than kickbacks and/or coercion.

 

[Crackong]

Intel Delays again

Are we Surprised ?
It is like what, the forth delay?

 

[ghazi]

The process is healthy so they ship 15 million consumer CPUs while continually delaying the ramp on data center chips that have a far higher margin, which some customers are already using in limited quantities nonetheless. Sounds like they still have serious yield problems on these chips.

 

[defaultluser]

I must say that I'm pretty satisfied with my Alder Lake CPU otherwise.

Nothing wrong with it, if you like running Windows Eleventy; one unfortunately requires the other.

Its going to be several years before Windows 11 is anywhere near feature-complete, along with actually being stable + have the broken context-menus less insane..

 

[watzupken]

7nm or a refined 10nm? It will be quite odd for them to release Golden Cove on retail CPUs on 10nm, while Sapphire Rapids' Golden Cove on 7nm. At the end of the day, they can name whatever node they want, but if they cannot get the product out, it is just another phantom product. I do wonder how power hungry Sapphire Rapids will be since we have witness how power hungry those Golden Cove are. On retail CPUs, the good thing is that we have those supposed E-cores to pick up some tasks while the P-cores can take it easy. Under gaming load, I've observed my CPU don't go above 70 to 80W as the P-cores are not under heavy load. For DC and enterprise use, I would expect CPU utilization rate to be quite high, and may draw quite a fair bit of power.

The process is healthy so they ship 15 million consumer CPUs while continually delaying the ramp on data center chips that have a far higher margin, which some customers are already using in limited quantities nonetheless. Sounds like they still have serious yield problems on these chips.

Personally, I've never feel that Intel are out of the woods with their node woes. 10nm has been a disaster for them, so much so that it cost them their fab advantage. New Intel CEO can only do so much by the time he joined, to try and rectify it so they can ship out 10nm products in good enough numbers instead of backporting to 14nm. As competition heats up, you start to see these "cracks" getting worst. Fortunately for Intel, this overall "shortage" problem have a roadblock for AMD to capture a more significant chunk of market share from Intel because (1)they have limited resources to secure allocation from TSMC, and, (2) TSMC is barely keeping up with supply for all their customers, especially when all the big customers are all stuck on 5 if not 7nm. So I suspect Intel deliberately use TSMC as a strategy to keep AMD at bay while they try and fix their own node issues. Intel may have deep pockets, but for a none strategic partner + competitor to TSMC, I don't think they are getting any preferential treatment from TSMC. Which means it probably cost them a lot to forcefully secure sizeable allocation for TSMC's cutting edge node. If their 7/10nm is that good, there is no reason for Intel to go to such extremes.

 

[Deleted member 24505]

Nothing wrong with it, if you like running Windows Eleventy; one unfortunately requires the other.

Its going to be several years before Windows 11 is anywhere near feature-complete, along with actually being stable + have the broken context-menus less insane..

How is win 11 not stable? I have been running it since beta and never had a single bsod.

Do you actually need win 11 to run ADL, I have run win 10 on my 12700k with zero problems. Even if the E cores do not work on 10 it is still a very fast 8/16 core CPU. I am sure there are plenty of people running Win 10 on ADL CPU's with zero problems.

ADL does not specifically require Win 11 does it.

 

[P4-630]

How is win 11 not stable? I have been running it since beta and never had a single bsod.

Also no issues here running windows 11.

Its going to be several years before Windows 11 is anywhere near feature-complete, along with actually being stable + have the broken context-menus less insane..

If you have crashes/bsods you might have faulty hardware.

 

[AnarchoPrimitiv]

7nm or a refined 10nm? It will be quite odd for them to release Golden Cove on retail CPUs on 10nm, while Sapphire Rapids' Golden Cove on 7nm. At the end of the day, they can name whatever node they want, but if they cannot get the product out, it is just another phantom product. I do wonder how power hungry Sapphire Rapids will be since we have witness how power hungry those Golden Cove are. On retail CPUs, the good thing is that we have those supposed E-cores to pick up some tasks while the P-cores can take it easy. Under gaming load, I've observed my CPU don't go above 70 to 80W as the P-cores are not under heavy load. For DC and enterprise use, I would expect CPU utilization rate to be quite high, and may draw quite a fair bit of power.


Personally, I've never feel that Intel are out of the woods with their node woes. 10nm has been a disaster for them, so much so that it cost them their fab advantage. New Intel CEO can only do so much by the time he joined, to try and rectify it so they can ship out 10nm products in good enough numbers instead of backporting to 14nm. As competition heats up, you start to see these "cracks" getting worst. Fortunately for Intel, this overall "shortage" problem have a roadblock for AMD to capture a more significant chunk of market share from Intel because (1)they have limited resources to secure allocation from TSMC, and, (2) TSMC is barely keeping up with supply for all their customers, especially when all the big customers are all stuck on 5 if not 7nm. So I suspect Intel deliberately use TSMC as a strategy to keep AMD at bay while they try and fix their own node issues. Intel may have deep pockets, but for a none strategic partner + competitor to TSMC, I don't think they are getting any preferential treatment from TSMC. Which means it probably cost them a lot to forcefully secure sizeable allocation for TSMC's cutting edge node. If their 7/10nm is that good, there is no reason for Intel to go to such extremes.

Exactly, Intel trying to buy up TSMC 3nm is just their underhanded way of spiking AMD because they can't out innovate them (despite spending over 7x more on R&D than AMD

 

[Deleted member 24505]

Exactly, Intel trying to buy up TSMC 3nm is just their underhanded way of spiking AMD because they can't out innovate them (despite spending over 7x more on R&D than AMD

Same crap and always from AMD users