https://www.tomshardware.com/news/amd-ryzen-3000-six-core-processor,39436.html

https://wccftech.com/amd-zen-2-6-core-cpu-benchmark-leaked-faster-than-ryzen-...

Non Disclosure Agreement's rule the world right now, so please understand that any benchmarks that are posted by anybody who has a copy of the chip in question has to be taken with a grain of salt.

The official AMD specs for the chips in question are still evolving and improving right on up through the issuance and shipment of the first production lots.   The price is also not fixed until the first production lot actually ships.

AMD's entire line is seeing massive improvements in performance, and perhaps in price as well.   Glomming on to the end points of this entire range makes this point most clearly.  

The AMD top Consumer point is completely without Intel comparison, not unless you go halfway up the Xeon pantheon and run TWO $2000 chipsets on a LARGE Intel rackspace server board to make your comparison point.

A better weather mark comes lower down,  at the lowest end AMD consumer chip ---  the six core Ryzen 3000 which really is going to be a six core chipset and not a 4 core as was thought earlier.  

Tom's hardware is scrupulous in saying what it can say about this new Ryzen 3000 chipset, and about what is just being inferred by testing, testing which was done on a provided interim development board --- and done while intentionally not using the most current improved memory types that the chipset can actually utilize.   "Normal memory" was used for all Tom's Hardware testing.

The Geekbench test result outlines a six-core, 12-thread processor that comes with the Ryzen 3000 codename Matisse. AMD has discontinued using its standard processor identification strings, which are easily identifiable in public test databases and easily decoded. As such, this sample comes with the AMD 100-000000031-03 identifier.

The chip has a 3.20 GHz base frequency and a 3.99 GHz boost, though these likely aren't final clocks due to the Engineering Sample nature of the chipset, and the chipset comes with 16MB of on chip internal L3 cache apparently split into two shared 8MB slices (8MB x 2). AMD has also doubled the L1 instruction cache to 32KB x 6. The chip's TDP (thermal design power) is not listed as the standard speed and the overclock levels keep getting better and better up until launch.

A lot of information is being worked over here, but an Intel Fanboy reader down in the comments says the best summary statement ever.  

ANDREWJACKSONZA says:
It's taken four years, but this low end AMD CPU (if that's what it is ) is creaming my i7-6700, both in multi-threaded AND single-threaded workloads where AMD were traditionally quite weak. The summary says quite a bit, but it gets worse the more in-depth one looks

So, you can't compare the top end of AMD to anything reasonable from Intel because there simply isn't anything to compare it to.  

The bottom end of the new AMD line can be compared, and is ranked as midway up the Core i7 rank of Intel processors.   And on top of this if you take into account the fact that Intel loses 25% of its functional task related working speed right now due to 5 different security mitigations, then this whole situation simply gets a good bit worse for Intel.

Comparative cost is the killer item though, the lowest most humble AMD six core processor will cost $99 at list retail, and will actually sell regularly for as low as $66  on sale supposedly.    And if you match the Ryzen 3000 line up with some better class fast memory (and use enough memory to actually fully utilize its VERY GENEROUS throughput pipelines) it punches upwards even higher, well up towards the top half in Intel's Core i7 product rankings while costing a princely sum of $199.

https://wccftech.com/amd-zen-2-6-core-cpu-benchmark-leaked-faster-than-ryzen-...

That’s right, word on the street is that the Ryzen 3000 series will start out with 6 cores and 12 threads as a minimum. This specific chip has a base clock speed of 3.2 GHz and a Turbo of 4.0 GHz.  So it’s a six core, twelve thread part with a 3.2-4.0 GHz clock range, does that sound familiar to anybody?

Well, it should. Because these are the exact same specs of the leaked Ryzen 3 3300 CPU that we’ve reported on roughly six months back. Which based on everything we’ve heard to date, is going to be the cheapest Ryzen 3000 series CPU you can buy at $99 at the low end, trending up to $199 for the highest performer in the Ryzen 3000 series.

AMD’s New Entry Level Ryzen 3000 Series CPU Could Be Faster Than its Current AM4 Flagship, The Ryzen 7 2700X

So, how well does literally the cheapest low end next gen Ryzen CPU perform? Very well, as it turns out it performs better than the fastest AM4 compatible Ryzen 2000 series CPU you can buy today. The Ryzen 7 2700X. At least in Geekbench 4 that is.

Based on the leaked performance figures, the company’s new entry level chip could very well dethrone its current flagship.  And that sounds just as insane to read as it was to write.

AFTER THE BIG PRESENTATIONS ARE ALL OVER ......

Both AMD and Intel had workshops where folks could get hands on with the prototypes.   Folks note that Intel is showing unknown processors in raw test beds, then making all sorts of  future claims about what they are planning to go do out sometimes in the future.  

Gist is a general 15-18% improvement in throughput is supposedly out on the table from Intel.   But not right now, a year from now, maybe.

AMD is being much more concrete about what they are selling, they say exactly what it is and exactly when it will be shipping (1-2 months from now).  Once again, a real 15-18% improvement in throughput is out on the table from AMD (with a lot more being held back in reserve).

Note please, AMD is showing a 15% upper on top of an earlier 15% improvement from earlier in the year.   AMD's improvements come on a six month cycle (2x a year).  Intel however hasn't made any real improvements in speed for a while (years and years) so you get no compound interest on the Intel side, instead you reality-lose a net 20-25% in real processing speed due to the required mitigations for the 5 Intel hyperthreading/predictive security illnesses .......

Cost is never discussed by Intel.  In contrast, AMD has a Lisa Su mantra promise going now,  "Twice the performance for half the cost" and that indeed sounds very intriguing to me.  So far nobody refutes that it is real, either.

AMD HAS SANDBAGGED US AGAIN like crazy on the top end, they won't even bring out their current big guns for us to see them until Intel actually shows them a real big gun for AMD to go blow out of the water.  

On the bottom end AMD has taken a large chunk out of the middle upper zone of the Intel Core i7 lineup with their very least, very cheapest new Ryzen 3000 line processor.   This is significant real item to me, the most significant thing seen this weekend actually.

This is my take away, that AMD has lapped Intel just about completely at this point in time.  

So, what is the big take away, what is lastingly good about this 2019 post show situation ????  


REAL COMPETITION IS HAPPENING AGAIN, LEADING TO REAL ADVANCEMENTS FROM BOTH AMD AND INTEL.



..... a Humorous side note from Qualcomm

The new 8cx Windows 10 laptop chipset from Qualcomm tests out at the midrange of the Intel Core i5 level, publicly testing on the very same weekend as the very lowest AMD Ryzen 3000 chipset laps up into the Intel Core i7 line performance wise.  

Once again, more competition is coming out for Intel, this time from the another side, the ARM processor side.  This ARM laptop chipset and its grandsons may hurt AMD in the Chromebook market once the prices come down considerably.   And they must come all the way down to the $99 level or else you would just go with a Ryzen 3000 core set.

https://techcrunch.com/2019/05/26/amd-unveils-the-12-core-ryzen-9-3900x-at-ha...


New "after the show" detailed analysis by various pundits looking for lies, exaggerations and data flaws in the AMD presentations have found several amazing discrepancies ......


AMD claimed a moderate 18% throughput improvement,  MSI (the motherboard maker) claimed only 15% throughput improvement on their motherboard spec sheet.   Neither of them claim the 20% increase in instructions completed per clock cycle that is inherent in the Ryzen gen 3 data pathway designs nor the 2x larger/faster data flow per clock cycle that is inherent in the AMD PCIe 4.0 data channel speeds (and those channels are also by count twice as many and twice as wide bit-wise compared to what is used in the current AMD or current Intel processors).  

This 7nm Ryzen 3 data channel advantage alone more than makes up for the slightly lower top clock speeds that AMD claims for their equipment, nor did AMD have to go try Intel's most recent dirty trick -- some bogus sticks of extra super fast memory to make their stuff seem to run synthetically better than it normally would run.

Nobody on the AMD side made any mention of the AMD advantage out of NOT HAVING TO DO THE 5 MITIGATIONS FOR THE 5 SPECTRE/MELTDOWN CLASS OF PREDICTIVE SECURITY HOLES that made Intel recommend shutting down its vaunted Hyperthreading.   Yep, AMD failed to claim any part of that 50% upper on that one, another serious oversight on AMD's side of the fence.

And so far nobody is taking any exception to Lisa Su and her "twice the throughput at half the cost" AMD mantra as on the face of things even when using these slack AMD conservatively reduced metrics Lisa's mantra seems easy to prove on both the throughput and on the lower price.

Intel thinks that putting their way way overpriced existing stuff on sale, taking it all the way back down to close to the old Suggested Retail Pricing from 2 years ago is going to move a bunch of their huge warehouse stocks of already finished products.     right, who wants it now that Ryzen is coming in 30 days with twice the throughput at half the cost .....

The pundits are also criticizing AMD for sandbagging and for sand trapping (a new computer term that means a pretty white pool of raked fluffy sand in front of, behind and lying on both sides next to the green, very pretty pools of raked fluffy white sand that buries your opponent's ball underground when it flies  up in the air and lands into it).

AMD is waiting for Intel's new 10nm chipset products to fly on up into the air, knowing full well that they will eventually land hard (burying themselves deeply into the soft raked fluffy sand) right next to the very least of the much faster and much cheaper AMD 3rd gen Ryzen 3000 chipsets.

AMD chipsets which are going to bebop along at $99 to $199 (or even less if they are on sale) happily swinging their 6, count them, yes 6 faster Ryzen processor cores and 50% larger data pathways & 15-18% greater throughput per clock cycle and eventually using all the raw 2x data bus speed increase that PCIe 4.0 provides .....  

..... Hey, and please remember those 6 Ryzen cores are also not needing the 5 Intel mitigations for Meltdown, Spectre and all the rest of those 5 ugly software security vulnerabilities (and that BTW also means no hyperthreading will be available for poor poor Intel, so Intel is now down like around 8-10 thread counts compared to the very lowest end AMD Ryzen processor which CAN BE safely hyperthreaded using open source Simultanious Multi Threading (which is the open source precursor of Intel's Hyperthreading).   Yep, FOSS based SMT which lacks the 5 new security vunerabilities that are eating up Intel's Hyperthreading performance so very very badly which by itself is worth well over a 50% hit for Intel thread performance-wise) ......

...... AND REMEMBER, Lisa Su has stated bluntly she will always win on price by a factor of at least two as she is currently getting over 75% first pass yields at the full 8 core counts per chiplet, with the sorted out and cut down 6 core Ryzen chiplets being seen basically seen as "save them up until you get enough of them to mess with" freebie processors.

Them rabid Intel fanboys are currently jest a whining nonstop like a bunch of bad waterpump bearings and jest keep on lodging all them repeated protests about this thoroughly stacked AMD deck, saying AMD is jest being "too too harsh" in how AMD is taking such bad advantage of poor poor Intel's production problems and Intel's problem ridden lackluster processor designs at the moment.


                  



....... what do you guys think?    Is AMD jest being mean to the Intel fanboys again ???



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



https://www.extremetech.com/computing/292400-intel-amd-7nm-epyc-vs-xeon-computex

https://www.tomshardware.com/news/intel-amd-computex-benchmark-epyc-xeon,3954...

Intel isn't happy at all with AMD's "Twice the performance at half the cost claims", and to "make Intel real again" Intel wants to throw forth a 350-450 watt power draw $10,000-$17,000 specialty dual-processor motherboard with built-in processors (must be bought with required Optane memory as an assembly at over $25,000) with a list of special optimizations using a special non-published benchmark, none of which apply in the consumer market in any fashion at all.

But Intel doesn't want to just be more competitive. It wants to prove that it will continue to lead even after AMD's 7nm Rome processors come to market. To that effect, Intel also included test results with its Xeon Platinum 9000-series (Cascade Lake-AP) that come armed with as many as 56 cores, 112 threads, and 12 memory channels crammed into a package that dissipates up to 400W. These new behemoths, which are essentially two Skylake-SP CPUs in a single socket, only come in OEM servers, so they aren't available on their own like AMD's Rome chips will be.

When Intel brings its high-powered -AP battleships to bear, it takes a 1.5% lead with the 96-core 9242 server, while the 112-core 9282 server takes a 23% lead over AMD's 128-core Epyc server.

But we have to keep thermal design power (TDP) in mind. AMD hasn't released pricing for its Rome chips, but TDP serves as a decent indicator of competitive price ranges. We're told by our sources here at Computex that AMD's Rome has a maximum 240W TDP (to be clear, that’s not officially confirmed by AMD), slotting between Intel's Xeon 8280 and the low-end -AP models (if there is such a thing). Meanwhile, Intel's 9282 and 9242 weigh in at 400W and 350W, respectively.

TDP is a decent litmus test of system pricing, as more heat generation and power consumption require more expensive components and equate to higher operating costs. That means Intel's behemoths require exotic cooling, and because they aren't socketed processors like we see in most servers, unique system designs that impact price heavily. Meanwhile, AMD's chips are definitely designed for the general-purpose market, whereas Intel's -AP models are pricey pieces of silicon that are only available as OEM systems and command premiums so high that pricing isn't public.

Intel's 8280 processors have a recommended price range of $10,000 to $17,000 apiece, and two of these processors are required depending upon the options you choose, and while AMD hasn't announced pricing for its Rome models, it's fair to assume it picked the 8280's as a comparison point based on pricing. While Intel's 9282 and 9242 may be the fastest on the market, they're likely priced significantly higher than AMD's Rome parts. It's all about the price-to-performance ratio, so unless Intel is going to significantly reduce its pricing, the -AP's performance advantage is a hollow win.

Well, with an Intel price range of $10,000 to $17,000 for each one of the pair of processors that Intel wants to use so they can win their proposed comparison, then Lisa Su is easily already 4x to 6x less expensive than Intel already, and unless you want to drop another $20,000+ for the yet undisclosed super duper motherboard and all the specialty Optane memory that the slots require, well then you don't really need to go pay the extra for the Intel specialty benchmarking program that Intel has to use to say their stuff is faster (by only 1.5% to 23% depending on the components and tests used).

And, killer point here ---- at the Intel price you can easily buy two or three of the far cheaper AMD units complete and Really Really Really bury Intel for keeps.

Reminder Again, Intel is still ignoring their Hyperthreading issue and the various mitigation performance hits again, and these "ignore" items by themselves more than wipes out the Intel claimed advantage of 23% that the $20,000+ worth of hardware was needed to buy.

I don't think most normal people really have much use for a pair of 350 - 450 watt processors, nor for all the other custom motherboard & stuff that costs well over $25.000 for the full kit, so for us more normal folks Intel's whole objection is really becoming just some more stinky brown vapor oozing slowly out of an puffy swollen Intel PR orifice  ......

A win and a lose for Intel

Apple has announced their by far most powerful desktop ever, using a mainframe version of a Xeon motherboard and dual Xeon processors and tons & tons of Optane memory.

Apple feels their Apple fanboys can afford a $10,000 to $12,000 iMac Pro machine (@ the very lowest price listed), and Apple management apparently feels their real top crust Apple users will always want Intel Inside rather than attempt to use the rather plebeian AMD Threadripper or the commoner's AMD Epyc processor motherboards.

I think Apple's fanboys can get a whole lot better machine from somebody else using AMD Epyc processors for way way less than half the price, and quite frankly Apple's yuppy snob appeal isn't really all that mind controlling for their fan base any more.  

We shall see, shan't we ????      





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



Intel sez ..... "Whupsie ...... sorry for your misunderstanding of what we actually said at the big Computex show presentation".

https://liliputing.com/2019/06/intels-14nm-comet-lake-u-chips-are-also-coming...



Look at the pictures --- when we said "next generation processor", YOU ASSUMED we meant the 10nm Ice Lake on the left because Ice Lake was what was presented at the big show presentation,  but hey,  now you say you didn't even know about the 14nm Comet Lake processor line that is being set up as the drop in 10nm replacement for whenever we eventually fail to ship at 10nm in quantity yet again ..........   (good 'ol bait and switch trick for the 4th time no less)

Does any of this feel like Deja Vu to you, again?

That's understandable, although we just finalized it it all yesterday IT IS THE EXACT SAME STICK WE DID 2 YEARS AGO --- but please remember, we were actually talking about the thick fat one at the big presentation too, the fat old flaming hot running 14nm Comet Lake with the noisy fan not just the little skinny mini cool running 10nm Ice Lake.    

           

What’s a bit strange is that the company will indeed offer two different U-series chips this year: Ice Lake-U and Comet Lake-U.

Anandtech speculates that this could be Intel’s way of ensuring it can produce enough chips to meet demand after years of struggling with its 10nm manufacturing process (and an associated shortage of 14nm chips).

It’ll be interesting to see how Comet Lake-U and Ice Lake-U chips compare in terms of performance and efficiency. Since battery life isn’t an issue for desktop processors, I wouldn’t be surprised if Comet Lake chips support higher clock speeds and run a littler hotter than their mobile counterparts in order to try to offer similar or better performance.

https://wccftech.com/amd-ryzen-9-3950x-16-core-cpu-5-ghz-overclock-ddr4-5100-...

No more Intel fanboy based grossly exaggerated (silicon bulked) super-duper tit for tat liquid nitrogen supercooled cheating to the max contests -- what can the Ryzen 9 3950x do with plain ol' standard air cooling (a stock fan assembly) ???


AMD Ryzen 9 3950X Overclocked To 5 GHz Across All 16 Cores On LN2, Destroys The Intel Core i9-9960X – MSI MEG X570 Motherboard Pushes DDR4 To 5100 MHz On Air

I guess my synopsis is this ---- do same same exact things to both sides equally, then Intel always loses.   Big time.  

And the AMD unit cost really is half as much as the Intel unit costs ......



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


I used to think that Intel cheated until they won and was ungracious whenever they did finally win.   Now I extend that rotten attitude to losing, because Intel doesn't lose very well either.

What Intel does do well is to price gouge the heck out of their fan boys for LOTS and LOTS of money.

Shows like this last one should educate the old radical Intel fanboys to go buy AMD next gaming rig they need.   Seeing AMD beat Intel and live post it and do the calculations all in real time AT THE SAME TIME points out something else AMD can do.  Get somebody to make a really neat game up that has to use that sort of extra Ryzen 9 capability just to play it.

Simply stated, do a modern version of the old "Can it play Crysis?" trick, except do it to Intel this time ---- turn about is fair play after all .......  this one action would move ALL the radical gamer boys off of Intel faster than any other reason you could possibly come up with ........  yep, give them a game they simply ain't got the chops to play when using their old Intel rigs.

https://www.google.com/search?q=too+many+cores&oq=too+many+cores&aqs=chrome.....


Intel supporters out in computer land are beginning to whine about how that they cannot keep paying for all these extra cores --- too much money, too much heat, too much power consumption.   They are talking about Intel products in all these comments, because Intel is all they will ever talk about.

Considering that these commentators really are Intel's paid talking heads (they are given new equipment for free by Intel), what we are really hearing is that Intel can't afford to go any further up into the core count wars without losing out even worse on yield/price/heat/power.

AMD at 105 watts can pull together a 16 core Ryzen 9 that costs less than half what the proposed Intel 18 core units cost and the AMD unit beats the Intel 18 core unit on any front you wish to name including COST, power used, heat, data throughput, gaming and overclockability.



Let's hold back for a bit and intentionally not talk about Epyc or Xeon rackspace level units from either side, let's agree to stop at High Performance Computing level which is the upper end of what Consumers do.   So lets talk the "shortly due for update" Threadripper level, the very top end of the AMD consumer spectrum.

Intel is in a bind there at the top end of Consumer because they have in essence just re-labeled some of their older huge Xeon rackspace units from a few years ago to be their top end HPC units.

These ex-Xeon HPC Intel units could not compete against the old Threadripper units from AMD for the last two years running, not for energy use, cores counts or computing throughput.    On raw power, yes Intel could reach up high enough into their Xeon line up to pull down something to match AMD Threadripper on computing raw power, but at a 2-3x price bump and a 2x power usage penalty.  

Lisa Su intends to fix this inbalance this time around .......   She can easily build whatever she needs out of chiplets to do the full complete hit job on Intel Consumer HPC products very very completely, then she can declare that as the start point of the pumped up Epyc line that is going to get refreshed next year.   Remember, the exaggerated bogus 10nm Intel claims were considered to be real back when the 7nm AMD chiplets were designed, and the AMD 7nm chiplets were designed to beat Intel's completely bogus claims for Intel's 10nm products, products which never came about (and still haven't).

So now Intel is in a real quandary, AMD is now using the exact SAME throughput improved, more energy efficient compute chiplets in all of their newest products.  Chiplets which test yield at over 75% yield rates at full speed, test all good on all 8 cores, making up very power efficient 7nm chiplets that are very easy to make by the wafer full on current TSMC processes.   AMD then slices the wafers up into chiplets and automatically sorts them for grading/binning once the chiplet wafers arrive at AMD.  

And the throughput grading results are simply getting a lot better over time as the lot after lot after lot of a steady production pace tends to make the TSMC production process more closely tuned and the chiplets tend to simply perform better and better and better.    

So, AMD keeps on raising their chipset performance numbers accordingly.

By pre-testing all the chiplets and putting like-speed chiplets into the same warehouse storage bins AMD has put together a streamlined production system to build an exactly controlled range of known good CPUs at a very minimal manufacturing & scrap cost.

This leads Lisa Su to be able to do the "Twice the throughput at HALF the cost" mantra thing and to actually mean it.

Intel has a very basic problem here.  The much larger, more complex the "built as one piece" Intel products becomes, the lower the resulting production yield numbers will be --- so by increasing their core counts any further Intel is actually cutting their own throats monetarily --- and so now Intel is orchestrating their paid yak yak tech press taking heads to all tell us that Intel wants to quit the core count race.


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


Guess what Intel, Lisa says she is just now getting started good.   Keeping to 25% less current draw (that's 75% of the power that the current Intel products require) Lisa Su is again giving you half again more to twice as many cores and twice or more on the throughput, and she is really actually doing it at half the retail cost of the old Intel products.

Intel is now beginning to panic.  

AMD just announced the Threadripper Refresh is coming ---- going up from 32 cores up to 64 of these same Gen 3 Ryzen / Epyc chiplet cores in a configuration that smells similar to a smaller version of full sized Epyc.   This is the proposed upper end of Consumer for AMD, and yet this is going past what you get in the lower part of the current Epyc server line (Epyc which is likely to get stronger yet again during their next refresh in 2020).   AMD's Ryzen and Threadripper competition is just about totally going to lap Intel's existing Xeon rackspace line of processors, in other words.   AMD's Epyc line will simply kill Intel completely.

AMD is refreshing and improving on a 2x per year frequency, rolling through their product lines one at a time in an organised fashion.  Intel has no timely response to this as Intel's natural improvement pace has never been better than a 1-2 year tick-tock improvement cycle.   Remember, Intel is already 3-4 real lithography generations behind AMD at this point in time, and Intel is falling further and further behind as AMD keeps rolling out the lithography improvements.

AMD is beginning to talk about 4-5nm as their next chiplet lithography level as Samsung is taking over the next AMD lithography improvement step, doing the 4-5nm lithography bump now not 2 years from now as TSMC was originally talking about doing.  

TSMC is responding by setting themselves up to leapfrog all the way down to 2-3nm in 2021-22 as their next big competitive lithography move.   This leapfrog action on lithography is a repeated pattern between Samsung and TSMC and it has occurred multiple times in the past decades.

Intel is beginning to see the handwriting up on the wall.

Intel is losing 20-25% of their real compute power to mitigations for their 5 current predictive security illnesses, Intel has had to shut off their hyperthreading completely due to the same security illnesses, Intel is on the wrong end of the production yield curve on having to make more and more cores into more and more complex chipsets and Intel is currently suffering total line competitive price/cost losses to AMD on all fronts.

AND NOW their Intel fanboys and Intel press supporters are beginning to actually talk out loud about these issues ........


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


Samsung and AMD and ARM collaboration side effects.  

ARM is British based and ARM is used to being at the forefront of their cell phone based technology.   They have had customers "jump ship" for better graphics before, it caused them to regroup on graphics and equal and better what folks had jumped ship to go get.

AMD has VERY GOOD graphics and ARM will have to raise their graphics bar a LOT to equal AMD graphics.   Samsung is very good at tuning ASML lithography and at chipset packaging and Samsung does co-development with folks like Samsung, IBM, AMD, Global Foundries, etc. very well.  

TSMC just likes to copy the advancements made by folks like Samsung and IBM.

AMD and Samsung and IBM and Global all have existing full design licenses with ARM, so they all have access to all the guts and they all have the ability to tweek the guts.   ARM technology is already inside the AMD chiplets and I look for AMD chiplet tech to get better after being shrunk to mate up with 3-4nm Samsung lithography.   I also look for AMD to start supplying a right killer small laptop / Chromebook chipset out of this collaboration.

ARM will improve their processor designs to meet or beat what Samsung/AMD puts together --- it will take a year or so but it will wind up happening the same way it has done multiple times before when Samsung/Qualcomm took the tech lead for a few years and then gave the lead up to TSMC who had spent that time in leapfrog lithography activities.   ARM tech catches up and takes a leap each time this happens, incorporating the advancements.

IBM is pretty much pure research at this point in time, but remember IBM tech advancements are co-licensed by Samsung, AMD and some of the others.   IBM quantum processors and AI processor tech keep the overall advancement pace bumped up to snuff pretty much.

Shunted off to the side now by legal issues and/or a general inability to keep up with the leapfrog pace of change -- Qualcomm and Intel.

COMPLETELY Sidelined by US gov. sanctions -- Huawei

Trying to make a come back into some major league play and to take up some of Huawei's lost positions -- Mediatek


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


Question for you to think about ......  AMD is going down to 4-5nm lithography soon but the existing chiplet size and connection pattern will remain pretty much the same as it is all automated at that size and the traces are all known and programmed into the automated assembly equipment already.   Core count for the existing 7nm chiplets was 8 cores per chiplet, with room left to squeeze in mebbe 2 more cores or some extra memory if AMD had really really felt it needed to go do that badly enough to change their chiplet layouts in between lithography levels.

So, If AMD uses up all the free standard chiplet layout space very efficiently with tightly packed 4-5nm cores and with adding in some more on chip memory (remember this next level of production equipment can lay down 14 layers of stuff into the same solid piece of silicon (3x more layers than before) so how many cores and how much extra memory will that wind up being inside each new generation chiplet?  12 to 16 cores per chiplet is my wild arsed guess ...... and remember these same sized chiplets will drop right into the existing overall automated assembly chiplet layout carrying that same proportionate increase in core counts, and in memory, and in throughput, and in higher efficiency & lower current draw.

Poor Intel.    

Your future sure does look bleak, boys.

AMD is improving strongly 2x per year, far far faster than you can even try to respond.

https://www.google.com/search?q=nsa+issues+warning+on+ms+windows+10&oq=nsa+is...

This is a CRITICAL WARNING issued by your Government compute experts at the NSA

This is a CRITICAL WARNING issued by Microsoft, the maker of the OS versions affected.   Bluekeep affects ALL OLD VERSIONS OF Windows as well as Win 10.   It is pervasive.

This is a CRITICAL WARNING reported by FORBES as an urgent matter that you must take care of ----- or potentially lose everything of value even mentioned in your machine.

The issue is BLUEKEEP, a form of worm attack that HOMELAND SECURITY lists as a most urgent critical attack vector for Windows 10.

https://www.zdnet.com/article/homeland-security-weve-tested-windows-bluekeep-...

This is one computer illness that I would be very careful to TAKE ACTION about ---- false patches are out on the web already so you also need to be very careful about patching your Windows machine.    

YOU need to go find where your real Windows patches are located and YOU need to go install them ASAP.   Ironically, Linux has already patched the Linux kernel and the new Linux kernel has already been pushed out to all the distros, so I already have protection against Bluekeep on my Linux box that stops me from passing any of the trash along to any of you Windows users.  

By Linux's very nature and structure, Bluekeep isn't really primarily a Linux exposure ----- unless you as the Linux root user get tricked into doing something really VERY STUPID and click on something "directly authorizing" it .....  

Watch out for stuff you don't know where it came from as this may be incorporated as part of an organized attack on American computing.

You Windows people are still all very exposed and will stay exposed until MS gets off their corporate kister and does something constructive about it and then pushes that out to all versions of Windows (yes, all the way back to Win 95).

If you can't trust MS to patch your machine, then you will have to do it manually.