https://i2.wp.com/static.techspot.com/articles-info/1613/images/Image_04.jpg?...



A new singing voice joins the Fat Lady Deathsong Auria, a worldwide rejoicing being sung at the demise of the Intel choke hold over the PC Industry

The Ryzen 7 2700 is marked down 25% on Amazon, making it possibly the best value CPU available right now. In games, AMD only a few percent slower than the Intel i7-9700K, which costs nearly double the AMD price at $410. The AMD Ryzen part looks especially good compared to the Intel i9-9900K, which costs 140% more.

Meanwhile, a 20% discount to the Ryzen 5 2600 means it annihilates the i5-8400, which performs roughly the same for $35 more. Thumbs up to AMD.

   Ryzen 5      Ryzen 5X      Ryzen 7      Ryzen 7X
MSRP          $200      $230      $300      $330
Amazon      $165      $190      $220      $290
Newegg      $165      $185      $250      $295

Zen 2 will offer up to 16 cores, faster speeds and PCIe 4.0, but it’s also expected to arrive at a bit of an initial price premium. If you already have a first gen Ryzen part and are looking for a high-end CPU upgrade for gaming, waiting is not a bad idea because the next generation will introduce performance levels Ryzen has never reached before. Professionals should also wait, as higher core counts will push productivity thresholds, too.

But for most people, buying these AMAZON discounted AMD processors could be a good idea.

Of course, that begs the question: how long do we have to wait?  Not very long, if new leaks are anything to go by.   Some Ryzen new product announcements will likely be made before Computex in May, but it is now clear that AMD has already told the Ryzen distribution channel to participate in some large inventory clearing events ASAP to move out their existing inventory as the new chips are already built and sitting in inventory awaiting the Computex release announcements.   And once the new AMD parts are available, anybody's old stock on hand at their warehouse isn't going to be so comparatively desirable.   Double ditto for the old Intel stock they may be carrying in warehouse stocks, as it will be 1) very very overpriced and 2) technically overlapped by AMD's newest generation at 7nm with wider faster internal data transfer pathways and multiple extra 7nm processor cores using much less power to run the AMD faster  more secure throughput .......

Intel is still acting somewhat arrogantly un-reactive expecting their Intel fanboys to pay over a 50% premium in addition to their old historically high Intel pricing just to have "Intel Inside" ---- this in an era where Intel is being run by a finance bean picker instead of a real CEO ---- Intel is being slowly eaten up alive with synthetically inflated overly high prices, some very serious production shortfalls and having to continuously run 3 required mitigation slowdowns due to being security risk contaminated by all the bad choices made by Intel's past management groups.    

Intel is not really even making a serious effort to reference off of AMD's much cheaper current pricing and higher relative performance levels, instead Intel simply says their processors are "better" when they demonstratively ARE NOT EVEN EQUIVALENT to AMD's best stuff any longer.  

Intel simply lies, in other words ......

This is all weakly excused on the plus side as acceptable by the excuse that "Intel is better for gaming" which is simply no longer true in a world that now has streamed gaming available that can run on a Chomebook just as well as it can on a powerhouse PC.

Remember, Intel does always require all three sequential slowdown mitigations for all three 3 major security issue groups (Spectre, Meltdown and Foreshadow).  Three (3) mitigations that depending on the task at hand can easily peel off 10% to 30% of the claimed Intel processing speeds when the processors are run in the real world.

And please remember, that the 3 Intel sequential mitigations ALWAYS takes at least a solid 10% off on all Intel processing speeds since you always have to run all of the 3 mitigations continuously all the time just to have "somewhat safe" computing on your Intel inside machine.



So, AMD picks up a 10% to 30% speed boost relative to Intel because the Foreshadow exploit is not active against AMD at all and the mitigations for Spectre and Meltdown are both a lot less processor clock cycle consuming on AMD processors than they are on Intel processors.

 Are there larger timeframes of that data available somewhere?

 The largest span of change is only 30 days, has this happened before?

https://www.oregonlive.com/silicon-forest/2019/03/intel-lays-off-hundreds-of-...

"If Intel doesn't show some clear steps in the right directions soon, they may indeed go down the toilet bowl whirlpool swirl yet even deeper ....."

Gosh, no sooner than I type it, it takes place -- is it possibly magical or what, I ask you.

“Changes in our workforce are driven by the needs and priorities of our business, which we continually evaluate. We are committed to treating all impacted employees with professionalism and respect,”  Intel said in a brief statement acknowledging the cuts to The Oregonian/OregonLive.

The white collar layoffs numbered in the multiple hundreds, according to people with direct knowledge of the cuts who asked not to be identified because they were not authorized to speak about their employer.

Intel confirmed the layoffs but declined to specify how many people lost their jobs, what the jobs were or to describe the rationale for the cutbacks.

The cuts took place at sites across the company, including Oregon, Intel’s largest site with 20,000 workers. A person with direct knowledge of the cuts said the Oregon layoffs were in proportion to those elsewhere.   A grand total is not known at this time.

Cuts also took place at other Intel facilities in the United States and at a large administrative facility in Costa Rica, according to people familiar with the layoffs.

Though Intel forecasts flat sales in 2019, people inside the company said this week’s layoffs don’t appear to be strictly a cost-cutting move. Rather, they said the cuts appeared to reflect a broad change in the way Intel is approaching its internal technical systems.

Aa an example of this sort of consolidation/cutback,  Intel has previously used several information technology contractors. An internal memo obtained by The Oregonian/OregonLive said Intel will now consolidate operations under a single contractor, the Indian technology giant Infosys.

The Big Change as known stated in a nutshell ---- Intel is offloading "technical management" as a complete management function primarily to outsourcing folks like the Indian technology giant Infosys.  

Why?  Intel has no strong technical visionary at the helm any longer, just a promoted finance bean picker.  This CEO is somewhat contemptuous of the current crop of technological people that he manages.  They can't do what he asks them to do so who needs to pay them those fat salaries any more?  Get rid of them.  

Hire younger "currently skilled" less costly replacements from IBM and AMD.  This is seen as a simple non-visonary non-technical bean picker type of move as Intel battens the hatches down tighter for the next year's worth of market share losses.

If you wanted to see this as a "fall back and regroup" by Intel's bean picker CEO in the face of ever increasing competition, you could.   You could also see this as an attempt to co-opt outside company talent and strengths that Intel certainly no longer has internally.   If you were charitable, you could see it as the start of a "cut and build plan" (with the build part still pending).

It is hard to see it as any part of a roadmap that is actually leading out of the existing Intel tar pit that Chipzilla finds itself so firmly mired into ......    

It is a bean picker knee jerk reaction type move, nothing more.

Wacky Prediction time .....

Intel really doesn't have a plan for expanding their 14mn production.

They realize that they are going to lose so much market share this year they actually have more than enough 14nm production capacity already, when all the furor and all the blood and dust finally settles.

Intel's 10nm plan is the same as it has always been -- it isn't a real plan, it never was and it  will never really happen -- new excuses and new delays will be made up as needed.  11-12nm is as good as Intel's current equipment can do AFTER it has been upgraded.

This is the Intel bean picker CEO's REAL executive plan for the next 2 years.   Protect the stock price by any means necessary.  Gouge the hell out of his installed base for every penny he can by heavy price gouging and lying.  Use delay and smoke screen freely as much as is needed.



When this actually changes, we will actually see some Intel 5nm-7nm ASML production equipment orders hitting the ASML waiting list.   Watch this ASML waiting list -- it says plainly who is doing what for the next 2 years .....

You will see the very first 11-12nm Intel big-little cores and chiplet designs come out from Intel later this year as copying ARM Holdings and AMD's ideas is simply what Intel does now-a-days.


Some of these Intel runs may be at a real 11-12nm size, some will claim to be 10nm but take that with a grain of salt as current production Intel 10nm really isn't 10nm as Intel's installed lithography isn't really up to 10nm (as proven by the many many many failed 10nm trial lots).



===================================================



IBM and Google are actually showing some real progress on Quantum Computing, but nobody in the customer real world really seems to understand what Quantum Computing is or what "advantageous things" it will do for you.    

Here is a readable, easy to understand catch up on what Quantum is and what it can do for you .....  hint, it isn't going into your earbuds any time soon as it still fills up whole small rooms and uses liquid nitrogen for cooling (still primitive).

https://www.itweb.co.za/office/in2ittech/content/kLgB17eJPYzM59N4

Best explanation yet, very NON-technical and fairly easy to read and understand.

https://www.tomshardware.com/reviews/intel-cascade-lake-xeon-optane,6061-4.html



Intel trots out their new bean picker CEO to try to explain their proposed new dual socket Xenon-Optane motherboard plus co-processors plus other stuff package that also has to have three different types of separate special Optane memory boards and a separate FPGA processor and a HUGE Ethernet communications card involved in the mix -- all supposedly working together to compete against the new single chip AMD EPYC processors that are 1) faster and 2) take far less than half the energy that Intel needs (400 watts is just for the dual Intel CPUs and Intel is holding mum about how much energy the rest of the stuff requires).  The AMD EPYC single socket CPU has more and faster cores and COSTS WAY WAY WAY LESS THAN HALF OF THE TOTAL INTEL PRICE TAG.






Yep, that is a book sized Optane SSD in that lexan block and below it is a heat sinked Optane buffer card that is the length of a standard laptop just to get it all to go fast ......

You also have to fill up the RAM slots in your motherboard completely with lots of expensive Optane RAM cards.  These Optane RAM cards by themselves costs $850 --- you can expect the other Intel items to run up to similar amounts of money.

OK, how do you get this Intel mishmash to try to work as a unit -- you need a HUGE separately programmable Ethernet communications card of course with super high speed Ethernet running in between the various plug in cards and memory buses.   This separate motherboard mounted communications card by itself dumps more heat and more wasted watts than the AMD EPYC processor takes period.  Ditto for the heat loving always watt sucking three separate types of Optane memory that are also required by the composite Intel system.



Intel's huge programmable Ethernet adapters don't draw as many flashbulbs normally, but in this case they serve as a key strategic asset for the company by acting as the glue that binds all the disparate Intel stuff together. Intel's new 800 Series adapters offer up to 100Gbps port speeds and support the new ADQ (Application Device Queue) feature that partitions network traffic into dedicated swim-lanes to boost performance significantly. The adapters also support Dynamic Device Personalization (DPP) profiles that enable smart packet sorting to route data intelligently. These adapters also support RDMA via iWARP and RoCE v2.

56 Cores, 112 Threads, 6 major components (some on separate boards) and a Whole Lotta current draw for the whole entire "you gotta buy it all on a new motherboard" Intel package.    

The AMD alternative is comparatively small, much lower cost, can run on your existing EYPIC motherboard using your old memory while being much more power efficient and simpler to implement.  AMD has already made some market share headway with its existing EPYC data center processors, but the company’s forthcoming 7m Rome processors pose an even bigger threat to Intel with up to 64 cores and 128 threads packed into a single socket and chip, wielding a massive 128 cores and 256 threads in a single server die instead of an entire rack full of Intel stuff.

The increased performance, and reduced power consumption of AMD EPYC purportedly outweighs Intel’s existing lineup of Xeon processors, so Intel has turned to a completely new line of "entire new system is required" Cascade Lake-AP co-processors to shore up its defenses in the high core-count space. These new co-processor heavy motherboard packages slot in as a new upper tier of Intel's Xeon Platinum lineup.

These new two 9000-series (with 2 sockets and 4 co-processing items required)  come packing up to a total of 56 cores and 112 threads in a dual-CPU die dual socket MCM (Multi-Chip Module) design ---- meaning that the two processor sockets and processor dies come together with four other major components in order to form a "single chip".  

Intel claims the disparate co-processor system offer the highest-performance available for HPC, AI, and IAAS workloads. The co-processors also offer the most memory channels, and thus access to the highest memory bandwidth, of any data center processor. Performance density, high memory capacity, and blistering memory throughput are the goal here, which plays well to the HPC crowd. This approach signifies Intel's embrace of a disparate multi-card multi-chip design, similar to AMD's EPYC processor's single die, for its highest core-count models.   Yeah, it takes all that huh?

If this doesn't read like it is actual or real or even really very technically feasible, remember it is being put out there by a brand new Intel bean picker CEO who is just reading the teleprompter slides -- he is contemptuous of his own people with the required technical expertise as he only believes in PR and stock price protection activities.   And that is what he is doing -- using an improbable wad of BS PR announcements over several disparate Intel products to protect his faltering stock price.   He has no intention of ever really making any of these conglomerated monstrosities, it is all for show so he can preserve that all important stock price.

Needless to say, it is doubtful that Intel will sell any of these mash ups at the price Bob Swan would want for one, even if he could get it to really work at all.


===================================================




Remember, it takes all of that 6 pieces of disparate (and desperate) Intel stuff hooked up together inside a single rack case sucking in lots and lots and lots of power to not quite equal the throughput and speed of what Lisa Su is holding up in her dainty little hand.    Lisa's stuff is real and has been shipping test programming samples to key AMD vendors for months now (the public shipping announcement for the completed vendor units and updated vendor motherboards is inside 26 days to take place at the May 27th AMD keynote address at Computex 2019)  ---  Intel's homegrown BS fictional stuff is a vague, sometimes in "late 2019, early 2020" and is unlikely to ever be really sold to anyone possessing both a brain and a price - benefit analysis speadsheet.

Bob, what are you gonna do when AMD drops down to 5nm chiplets and increases the core count before you can even start shipping this somewhat smelly Frankenstein mess ????

Worse yet, suppose the existing already built and warehoused EPYC Rome 7nm Server Processor actually already has 162 PCIe Gen 4 Lanes of data throughput, more than twice as much as Intel's Xenon 9200 will have (if and when they get produced and shipped as promised -- mebbe inside the next 2 years that is ......)           

https://www.cnbc.com/2019/04/05/wells-fargo-downgrades-intel-on-weak-2019-dem...

https://www.marketwatch.com/story/intel-stock-falls-after-wells-fargo-downgra...

https://www.thestreet.com/markets/intel-slips-after-wells-fargo-downgrades-ch...


Oh, how embarrassing for Intel ........ the official FAT LADY for Intel is Wells Fargo ?!!?

Wells Fargo  (Wells Fargo, of all people, I mean really, Wells Fargo.   Talk about pots calling kettles black .....)  has just downgraded Intel due to weak 2019 demand and increased AMD competition.

"Weak semi demand data points thus far in 2019, coupled with an expectation of cautious outlooks into [the second quarter of 2019], leave us to consider some downside risk" to estimates for the first half of the year for Intel shares, Wells Fargo analyst Aaron Rakers wrote in a note to investors.


Stock price for Intel immediately dropped OVER a percentage point, and increased to over a 2% drop inside hours and the Intel stock price is still down but moving back up very slowly.  

Expect even more unannounced illegal point stock repurchases by Intel shills to continue and to increase sharply simply to stop this market dip and stage something of a visible mini-recovery before close of trading today.

This is also being seen by some as a vote of "no market confidence" on Intel CEO Bob Swan after his EMBARASSINGLY BAD teleprompter reading presentation yesterday on "the new innovative Intel rack space processor set".  Anyone watching that stumbling slightly technically ridiculous presentation would realize you had somebody behind the mic that really didn't understand what he was talking about (or worse, he didn't think what he was saying was true, which is actually even worse when you think about it).

Intel is now FOUR lithography levels behind the market leaders, perhaps that fact also plays in this Wells Fargo decision to downgrade Intel's stock price a bit ???

 
German Mindshare.de First Quarter sales data for 2019 is in.   AMD is still outselling Intel 2-1 in individual chip sales, verging on a 5-1 ratio for some of the more poplar AMD models.

The Big Box finished unit sales still tend to be the more expensive gaming rigs, and there sales still tilt more towards the Intel boxes.  Sales of large expensive gaming rigs are light at the moment, with the main flow of gaming sales going towards the Xbox/PlayStation sort of device.

The gaming stations are almost 100% AMD based now.   This has been consistent for over a year now.

New Finished PC Units announced are predominately AMD based and tend towards the middle to lower total power range.   This is also about the fact that AMD's middle to lower power level is as strong as the very best Intel ever was a couple of years ago.  Smart shoppers know this and buy accordingly.

New Unit Sales is different from the Total Installed Units comparisons.   Watch this point when folks throw numbers around a lot, it is VERY misleading when they get them mixed up.   Total Installed Units total data is old enough to say that Win 7 is still the dominant OS installed on the machines comprising this data ......  

Some Stuff to remember as we head into the big AMD ZEN 3000 announcements for 7nm production .......



IT IS ALL BASED ON THIS STUFF .....
The new AMD tech is this server tech, with some of the EPYC chiplets actually being the exact same as the RYZEN chiplets in some cases.  

The AMD Ryzen 3000 lineup is based on the new Zen 2 core architecture which is made possible with TSMC’s bleeding edge 7nm process node. AMD has reaffirmed that their Zen 2 based Ryzen 3000 series processors for the AM4 desktop platform will be available in mid of 2019. We are now hearing multiple reports of a possible launch in early July and that might be it as far as the launch day is concerned for the new desktop processors.

AMD has made significant changes to their CPU architecture which help deliver twice the throughput of their first generation Zen architecture. The major points include an entirely redesigned execution pipeline, major floating point advances with doubled the floating point to 256-bit and double bandwidth for load/store units. One of the key upgrades for Zen 2 is the doubling of the core density which means we are now looking at 2x the core count for each core complex (CCX).

Improved Execution Pipeline
Doubled Floating Point (256-bit) and Load/Store (Doubled Bandwidth)
Doubled Core Density
Half the Energy Per Operation
Improved Branch Prediction
Better Instruction Pre-Fetching
Re-Optimized Instruction Cache
Larger Op Cache
Increased Dispatch / Retire Bandwidth
Maintaining High Throughput for All Modes

So, Lisa Su actually sandbagged very strongly at her last big show presentation because Intel had totally failed in their 10nm execution and all the advanced items AMD had prepped for use against Intel 10nm were not needed at all, any whatsoever in the last six months.  

So Lisa Su totally sandbagged on all that advanced 10nm Intel beating stuff, very very strongly sandbagging all the new features she could have shown in her previous show presentations.

She did this SPECIFICALLY to allow more time for more of AMD's old 12nm inventory to sell through at full price, and indeed AMD is even now hosting a MASSIVE Retail Discount Push to try to move all the rest of the old AMD 12nm inventory all the way out of the way before the new AMD 7nm stuff arrives in a few months.

AMD knows that the new 7nm stuff will sell much much better than the old 12nm stuff did to the point if AMD doesn't help their vendors to move out all their old 12nm inventory then their vendor base might take some damage when the new 7nm stuff makes the old 12nm stuff relatively undesirable and "obsolete".    

The vendors have got to move the 12nm products before it becomes un-moveable.

Since some of the most recent 7nm AMD advances are all server sized pipeline widths, greatly increased pipeline through puts and various other basic structural changes, these changes have got to come out now anyway.   Ditto for the 7nm chiplet structure being like the server/rackspace structure and all the advancements based on that.   If you think some of the consumer RYZEN data handling, floating point and branch prediction is going to be more like mainframe/server/rackspace type stuff, you are correct in that particular thought.

Look to see Intel Consumer processors drop way behind the leader at this point in time, almost shamefully far behind.   Intel consumer can't compete on a wide range of features any more.

Intel is gonna be playing catch up from now on --- extending out for around 5 years to a decade or thereabouts.


===================================================


Understand what comes after 7nm and why it will arrive so very quickly.

5nm is what it is called.   It it can run on some of the most current ASML equipment that was purchased for 7nm production, so all players can take advantage of the 14-20% speed and efficiency improvements.   You can also get twice as many wafers off your silicon wafer when compared to 10nm, so that is another good reason to use 5nm.

The final reason is one Intel cannot touch or understand ..... unless they want to start stacking their 14nm chips up into towering infernos again.  5nm direct burn EUV can burn stacked layers directly into a solid thick silicone wafer when carefully calibrated and focused and controlled.  

Like ..... well ..... nothing we have ever seen done before.   Current 7nm EUV can lay down 4 relatively crude stacked layers into a simple thick memory wafer --- generally used for simple stuff like memory chips, but the first generation 7nm EUV machines had to be customized out the wazoo to do that stacked laydown trick at all.   But both TSMC and Samsung worked it all out and got it to yield enough good chips to make it work well.

So, the stacked laydown tricks have been refined twice now and are now built right into the new 5nm ASML production machines as a basic function.  

Standard 5nm ASML EUV direct burn machines can lay down stacked layers 14 layers deep right into the same silicon wafer and that is with the standard 5nm production equipment.

Apple is going 5nm ASAP (A-14 production) because they can do just about the entire device inside the layers of the CPU itself, including a goodly bit of memory.

This saves a lot of money and a lot of post burn assembly processing time.   The chips run a lot faster and the energy they use to run is a lot less.


So, watch the Apple A-14 chipset when it comes out for radical new capabilities due to 5nm 14 layer direct burn EUV.

https://www.theverge.com/2019/4/16/18411332/intel-exiting-5g-smartphone-modem...

https://www.theverge.com/2019/4/16/18410985/apple-qualcomm-settle-royalty-dis...

https://semiaccurate.com/2018/02/22/intel-talks-5g-connected-pcs-mwc/



Intel this evening says it has decided to leave the 5G mobile modem market to focus its efforts more on 4G and 5G modems for PCs and smart home devices, as well as its broader 5G infrastructure business. The announcement comes within hours of Apple and Qualcomm striking a surprise settlement in the two companies’ ongoing patent infringement and royalties dispute related to Apple’s use of Qualcomm modems in the iPhone.

It’s likely Intel’s decision here was what prompted Apple and Qualcomm’s decision to settle — which all came as quite a surprise since it happened just as lawyers were presenting opening arguments at the latest courtroom trial that began yesterday in Southern California.

But it’s unclear when Intel came to this decision, or when it informed Apple, and Intel has declined to comment. Either way, phone manufacturers like Apple will need to look elsewhere for their 5G radios now, and that means Intel just ceded that 5G cellular phone business to Qualcomm.  And to Huawei.  And to Mediatek.  And to Samsung.

It is clear that Intel has TOTALLY FAILED yet again to meet their Apple commitments to do the production design and to make fully functional energy efficient mobile 5G modem production runs that they contracted to do for Apple.

“IT HAS BECOME APPARENT THAT THERE IS NO CLEAR PATH TO PROFITABILITY AND POSITIVE RETURNS” Intel bean picker CEO Bob Swan said in a statement today, perhaps misleading some folks to think this is perhaps an Apple "failure to pay enough" to support all the development costs and/or an Apple caused "modem design" failure of some sort.  

Not so, Intel designs the 5G modems, not Apple.   Apple has LOTS of development money, enough to fund entire new levels of lithography whenever they want a new one.   Apple isn't the failure side on this one at all.  

Intel took Apple's money last year for 5G cellular modem development, fumbled it away and now has failed repeatedly to produce a successful run of 5G cellular modems for Apple, failing again and again and again until “IT HAS BECOME APPARENT (to Intel) THAT THERE IS NO CLEAR PATH TO PROFITABILITY AND POSITIVE RETURNS”.

How bad is Intel's failure this time?   Bad enough to make Apple do a 180^o ass-swapping reversal during the heat of a major court battle with Qualcomm and then to immediately sign a six year contract with Qualcomm after only 24 hours of negotiations.

Intel has failed Apple multiple times now, and Apple isn't all that forgiving for those who fail and cost Apple lots of money.   Expect Apple to put their own ARM based Mac processor chips for Apple laptops back into high gear again as they really can't trust Intel to do ANYTHING right for them any more.


===================================================


LOOK AT THE PICTURE, notice the clustered crowd of "support personnel" who are now clustered around Bob Swan, currently the head "technically challenged beanpicker" CEO of Intel.  This is the crowd that has surrounded him ever since he FLAT OUT FIRED all his old Intel Tech people and then hired Infosys out of India to be his major technological resource (and mebbe to provide his personal bodyguard security force as well ????   ).

Folks, I think Intel is fundamentally broken at the top, and things are not going to do very well under this new CEO's regime.   Bob Swan is systematically firing and replacing anyone who could unseat him as CEO --- by doing so he is chopping Intel off right about at the armpits.