SoylentNews is people
SoylentNews
Intel may be planning to sue Microsoft for its plans to include x86 emulation in Windows 10 for ARM machines:
In celebrating the x86 architecture's 39th birthday yesterday—the 8086 processor first came to market on June 8, 1978—Intel took the rather uncelebratory step of threatening any company working on x86 emulator technology.
[...] The post doesn't name any names, but it's not too hard to figure out who it's likely to be aimed at: Microsoft, perhaps with a hint of Qualcomm. Later in the year, companies including Asus, HP, and Lenovo will be releasing Windows laptops using Qualcomm's Snapdragon 835 processor. This is not the first time that Windows has been released on ARM processors—Microsoft's first attempt to bring Windows to ARM was the ill-fated Windows 8-era Windows RT in 2012—but this time around there's a key difference. Windows RT systems could not run any x86 applications. Windows 10 for ARM machines, however, will include a software-based x86 emulator that will provide compatibility with most or all 32-bit x86 applications.
This compatibility makes these ARM-based machines a threat to Intel in a way that Windows RT never was; if WinARM can run Wintel software but still offer lower prices, better battery life, lower weight, or similar, Intel's dominance of the laptop space is no longer assured. The implication of Intel's post is that the chip giant isn't just going to be relying on technology to secure its position in this space, but the legal system, too.
(Score: 1, Insightful) by Anonymous Coward on Saturday June 10 2017, @04:40PM (3 children)
More innovation stifling patent wars. So they going to sue everyone else too? ( like QEMU? )
I haven't been fond of Intel since they killed the ix432. Growing to hate them over the years, much like microsoft and oracle.
(Score: -1, Offtopic) by Anonymous Coward on Saturday June 10 2017, @05:09PM
I haven't been fond of Intel since they put in backdoors into their products so that they could make sure that mmmaaallleeesss don't have wrongthought about cute young girls vs THICCCC ADULT PROUD WOMYN
(Score: 2) by frojack on Saturday June 10 2017, @09:26PM (1 child)
Intel already licensed this tech and its instruction set to AMD, and IBM So it hasn't been exclusive for a long time, and the instruction set is pretty much in the public domain, except for recent additions. But even those could emulated in RISK environments for a slight speed loss. Every compiler in the world generates this code.
I think Intel would lose this case.
No, you are mistaken. I've always had this sig.
(Score: 2) by BK on Sunday June 11 2017, @02:17AM
Well, patents run 20 years around here... 1997 puts us back to the 'Pentium' type x86 instructions +/-. I'm guessing there's been a few new developments since then. M$ may have included some of those.
Presumably, M$ hasn't purchased such a license. If Intel sells licenses on this tech, that means that they've retained exclusive control. Of course, with patents, (unlike trademarks,) that doesn't actually matter.
That's copyright. It isn't about the code a compiler might generate, it's about the machine on which that compiled code can run.
...but you HAVE heard of me.
(Score: 2) by bradley13 on Saturday June 10 2017, @05:12PM (7 children)
I suppose it depends on which patents, and what is being emulated, but: Don't patents expire after 17 years? 17 years ago was already the first Xeons. Anything up to that point ought to be out of patent protection. That certainly covers the general architecture; the only thing not covered would be instructions developed since then.
Even then: I would submit that an instruction set is actually an interface, specifically, an interface to allow software to interact with the hardware. The implementation of the instruction set would be patentable, but the simple definition of instructions? Not so much. An emulator will have a different underlying implementation, but providing the same interface should be fine.
Everyone is somebody else's weirdo.
(Score: 0) by Anonymous Coward on Saturday June 10 2017, @05:51PM (4 children)
They don't fall under 10+10 patents?
If so you're looking at 2020 before 2000 era patents qualify, which is pre Pentium 4.
That means no SSE2+, NX, AVX, VMX, x86_64, no 180nm process tech, no IPC improvements from the past 20 years, etc.
Obviously *SOME* of them might actually be unpatentable/overturnable thanks to prior art, but the majority of these there is at least one patent they can swat you with. Having said that, Intel's relevance in the market is fading, even if it isn't obvious to the business community/stock market yet. SoftBank is looking more and more like the heir apparent to the technological throne. Owner of the ARM IP, Owner of Boston Dynamics and other robot/manufacturing technology. Internet/Telecommunications master. Financial supporter of many anime (and even a few MMOs?) They have their hands in almost everything tech nowadays, but unlike Intel they don't seem to be flailling about with their purchases. They are systematically buying up specific companies in specific markets that are both undervalued for their potential and capable of much more given the proper corporate support and guidance (IE taking boston dynamics' designs and actually producing some field prototypes... rad-hardened for use in Fukashima Daiichi, for instance.)
Having said all this, from a consumer perspective the greatest benefit to come from this would be a non-profit organization devoted to producing RISC-V chips with user friendly features, like wholly device owner controlled TPM/TrustZone/ME implementation that allows the user to input keys (whether into efuses inside the CPU, efuses outside the cpu in a seperate package, or simply loaded from flash during bios initialization, with physical write protection to ensure it stays disabled... Which BTW is dangerous on SPI flash, where the write disable pin only works if the flash has write-disable *ENABLED* in software after poweron! Meaning glitching could exploit and allow a write to flash even when you believe yourself protected by a read-only behavior! Unlikely to happen in practice, but impossible to ensure without architectural changes!)
(Score: 1) by Ethanol-fueled on Saturday June 10 2017, @06:17PM (3 children)
As an employee of Boston Dynamics who was working there throughout the selloff to SoftBank, I can state that there was a good reason why Google sold off Boston Dynamics -- because nobody in their right mind would buy any of Boston Dynamics' inefficient rube-Goldberg contraptions and all the worth is in the IP.
We are understandably demoralized, as the new management SoftBank put in love to torment us. For example, since a few employees suffered repetitive-motion injuries, group stretches are now mandatory twice a day and I've already been reprimanded for saying, "This is bullshit," and walking out of a group stretch session.
Then during our all-hands meetings, the new management started enforcing "team-building exercises" -- and again, we are forced to perform such humiliating acts as playing pin the tail on the Donkey under threat of reprimand and poor performance reviews, although I managed to weasel out of that idiocy by showing up a few minutes late and eating my dish of catered food really slow.
I'm usually the kind of guy who likes to wait around until I get laid off, so I can collect, but Boston Dynamics has become so terrible that I've already lined up some extra work and am bailing. Had some good times there, though, like that time I caught Sergei Brin and Eric Schmidt stealing our inventories of expensive gold-plated pins and trying to melt them into ingots using one of our blowtorches. They might be smart guys, but they need a crash-course in metallurgy and inventory management.
(Score: 2) by takyon on Saturday June 10 2017, @07:13PM
Didn't hear about the sale so I'm submitting it now. I'm expecting big things on that one.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by kaszz on Saturday June 10 2017, @11:00PM (1 child)
The military has no use for those robotic burros? Nor Fukushima Daiichi as a tool to get into contaminated areas?
Seems Sergei Brin and Eric Schmidt forgot the ask-first-do-later thing. Especially in hi-tech where health dangers and economic loss lies present in unexpected ways. Doing such things in a nuclear lab can end quickly.
(Score: 2) by takyon on Sunday June 11 2017, @12:54AM
The military found them to be too noisy [soylentnews.org].
They found Fukushima nuclear robots to clean up after that mess, but it's been messy:
https://www.theverge.com/2017/2/17/14652274/fukushima-nuclear-robot-power-plant-radiation-decomission-tepco [theverge.com]
http://www.independent.co.uk/news/world/robots-fukushima-nuclear-disaster-dying-probe-clean-up-tepco-toshiba-reactor-nuclear-radiation-a7612396.html [independent.co.uk]
https://www.theguardian.com/world/2017/mar/09/fukushima-nuclear-cleanup-falters-six-years-after-tsunami [theguardian.com]
https://techcrunch.com/2017/03/25/japanese-authorities-decry-ongoing-robot-failures-at-fukushima/ [techcrunch.com]
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 0) by Anonymous Coward on Saturday June 10 2017, @05:53PM (1 child)
Not to mention most of Intel's x86 code is actually licensed from AMD and their amd64 instruction set. Every intel chip running 64 and 32 bit code is using this instruction set because it was better than their in house version called IA64.
It's saber rattling with no real threat imho. If they do take it to court it will be opening a can of worms as qualcomm and microsoft will devote legal resources to tracking down the origin and original patent/licenser of each instruction set on a modern intel cpu. Intel would have a good chance of being found to violate someone else's licenses/patents and costing them major money in turn.
They won't risk it, that is unless intel is not only too badly hurt from the hit by ryzen. And too scared of how top end phones can pretty much be drop in replacements for low end and corporate desk jockey systems. Ie i3's and celerons.
(Score: 0) by Anonymous Coward on Sunday June 11 2017, @02:13AM
No high end smartphone is anywhere near the speed of any i3. You need to come back to reality.
(Score: 0) by Anonymous Coward on Saturday June 10 2017, @05:12PM
Patents are done, and the "innovation" is obvious to anyone in the field as simple obvious extensions.
They are threatening having to force the other Co's to pay legal fees over a non existent frivolous claim. This is extortion. An illegal method. Perhaps their targets should use some other methods aswell in turn?
(Score: 0) by Anonymous Coward on Saturday June 10 2017, @06:22PM (5 children)
Any idea how fast the emulation will be? If it's sort of like a 386, than my old MS-DOS software should run just fine? Remember, this is MS, with a long history of bloat and terrible performance. Just say'n...
(Score: 3, Informative) by takyon on Saturday June 10 2017, @07:26PM (4 children)
https://www.thurrott.com/windows/windows-10/85365/thinking-x86-emulation-arm [thurrott.com]
https://arstechnica.com/information-technology/2016/11/x86-emulation-rumored-to-be-coming-to-windows-for-arm-in-late-2017/ [arstechnica.com]
Speed/efficiency is unknown AFAICT. The emulation isn't for Win16, so fire up DosBox. We'll have to see what benchmarks say about the speed, but many Windows applications probably don't need full speed.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 1) by oakgrove on Sunday June 11 2017, @02:24AM (3 children)
No matter how many canned demo marketing videos Microsoft puts out, the speed in real day to day work is going to be garbage for one simple reason. Even if it was perfect, the Snapdragon 835 is literally about half the speed of the lowest end Core m3. This shit isn't magic. If you fancy trading in your laptop for one that runs at half speed on a cellphone CPU then go for it. I doubt the market will follow you. This may get some good battery life and light weights but laptops already go for more than ten hours and weigh less than two pounds. This is going to be the modern EeePC. A fad that will drive the fanboys and geeks crazy but won't even make a blip anywhere else. At least let's hope not. Nothing will put a blackeye to the idea of non-Core laptops like one that is so frustratingly slow the average consumer wants to chunk it out the window.
Now tell me how fast and smooth the web browser is on your smartphone is as if that is remotely relevant so I can laugh.
(Score: 2) by takyon on Sunday June 11 2017, @03:00AM (1 child)
There are already ARM laptops/Chromebooks. Anything like what you described would at least beat Windows RT. Compatibility with x86 makes the systems more useful than RT was.
Not sure if Core m3 is the right comparison when people regularly buy Intel CPUs that are dramatically slower than the m3. You know, dual-core Celerons like the Intel Celeron N2840 (or the newer Intel Celeron N3060). I'm looking at Passmark and seeing numbers that are substantially lower for the N3060 [cpubenchmark.net] than Intel Core M-5Y31 [cpubenchmark.net] or even the Intel Core M-5Y10 [cpubenchmark.net]. I believe that's one of the slowest Core M chips Intel made.
Core M was not so cheap when it debuted [anandtech.com]:
Snapdragon 835 devices may be somewhat cheaper [pcadvisor.co.uk]:
Anything that launches at $400 will be at sale for $250-300 before too long.
The point is that even if Snapdragon 835 turns out to be only half as fast as today's Core m3 chips, something you have no evidence of since Microsoft hasn't released any of their own benchmarks and the Windows on ARM products are not on the market, it might be priced appropriately. A lot of these Core m3 products started [slickdeals.net] at [bestbuy.com] $900 [wikipedia.org] or more.
I wouldn't worry about the emulation speed right now. We'll know how good or bad it really is when it gets benchmarked. I doubt anyone is saving up in the piggy bank to specifically buy a Windows 10 for ARM laptop. We can have another story about this when the first Windows on ARM machines have been benchmarked.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 1) by oakgrove on Sunday June 11 2017, @05:53AM
None of these run win32 applications nor do they bear the user expectation burden of being performant while running said applications so they aren't really comparable. The closest thing to what this is was Windows RT which was a dismal failure so if that's what you're going with then.. maybe rethink something.
Most Windows laptops run Core processors so that's the right processor to compare to. Besides, the most common complaint of Atom powered stuff is "it's slow" which supports my point. I chose the m3 because it's the lowest of the low end. I'm actually giving SD 835 a sporting chance in that regard. If I was being completely neutral, I'd compare it to a typical i3 or i5-7200U which would completely slaughter it.
I'm not arguing they won't be cheap. I'm arguing they will be too slow to meet consumer expectation at basically any realistic price. Just like the cheap Android tablets that while cheap, have the highest return rates because they're just too slow.
But that's wrong though. I have plenty of evidence. The internet is full of benchmarks of the 835 and the m3 running general purpose workloads which almost always show the 835 to be about half as fast. The term you're looking for is "proof" and, yes, we won't know for absolute sure until these laptops come out but that's where critical thinking comes into play. If every benchmark shows the 835 to be half as fast as an m3 and the laptops in this case are not just running ARM compiled Windows but actually doing dynamic translation of x86 programs on top in the process, what is the likelihood that the end result is going to be anything but at best the equivalent of the m3 running at half speed. If full Windows laptops with a processor that slow were viable beyond the tepid sales of Atom powered models, we'd know it by now. To put an even finer point on it, I have a Lenovo T420s with the Sandybridge i5-2520m which also benchmarks twice as fast as the SD835. That is a 6 year old laptop with a low to midrange processor. Now imagine everything processor bound on it taking twice as long. That doesn't even get into things like comparative memory I/O etc.
To reiterate the crux of my argument, I believe that SD835 powered laptops will be too slow to meet consumer expectations of a full Windows laptop experience. Benchmarks back that up by showing the chip to be half as fast as the current low-end (Core m3) with acceptable performance. And anything lower than this "low-end" like Atom powered stuff just doesn't sell very well for that exact reason, i.e., it's slow. I also argue that price floor is not as important as people believe based on how cheap current laptops can be and there being a minimum of acceptability of performance at any price, that last part being the fundamental issue. I also argue that battery life and weight is an overblown concern as we have laptops now that weigh less than two pounds and will go all day. It can be argued that, yes, there are Core powered laptops that are fast, have long battery life, and are thin and light but you aren't going to get all that for cheap. True but that isn't an argument in favor of the SD835 stuff since, again, there is a floor of acceptable performance which I assert this will not reach. That's my position but ultimately we'll all see. Just remember that I told you so.
(Score: 2) by urza9814 on Monday June 12 2017, @05:16PM
I know organizations that still deal with Visual Basic 6 apps on a daily basis. I tried to migrate them to something newer, but we couldn't find anything they agreed was cheap/easy enough, so the VB6 crap lives on...and now I've started supporting OTHER groups across the country still using this crap because the original devs are long gone. And a few years ago I ran into a company still running Windows 3.1. On a laptop. For their accounting software. I'm sure there's a large number of "business critical" wintel apps that would run just fine on an ARM system because they ran just fine on single core systems a decade or more back and they haven't changed since.
This isn't for gaming. Most office apps aren't loading a 3GHz quad core to 100% utilization. Plenty of software will run just fine.
(Score: 2) by tekk on Saturday June 10 2017, @09:41PM (1 child)
Why is MS working with Qualcomm, a hardware company, to make use of software emulation? I wouldn't be surprised if this ends up being some sort of weird hybrid.
(Score: 0) by Anonymous Coward on Sunday June 11 2017, @06:01AM
QCOM is a *HUGE* portion of the phone market. MS wants back in. They actually have been working together for years all the way back to the WinCE days.
Intel will have a pretty large problem with emulation and patents. They let they sit for to long at this point. They basically may have just let it slide too long. QCOM will not let that go down without a fight. QCOM has some pretty good lawyers and their own set of patents. My bet is Intel is wanting to make their costs cheaper. QCOM charges a premium for their stuff. Hence Intel's saber rattling. MS is also not without expertise and emulation in the court. They bought the company that beat Sony and the same company that is at the core of their emulation. Watch for a settlement and Intel gets a better price.
(Score: 2) by KritonK on Sunday June 11 2017, @10:21AM
Isn't it a bit too late to try defending their patents against emulators? There were software x86 emulators for the Amiga [wikipedia.org], decades ago, which were capable of running MS-DOS. These days there is DOSBox [dosbox.com], QEMU [qemu.org], and a whole slew [wikipedia.org] of other x86 emulators. Now they remember that they don't like them?
(Score: 0) by Anonymous Coward on Sunday June 11 2017, @12:59PM
Great, just what we need. Another technology giant using patents as a lock-in. How does this move "promote the sciences and useful arts"? I thought that interoperability was specifically excluded from the scope of patent protection?
Besides, patents are supposed to protect the implementation, not the entire concept, and hardware-assisted ISA virtualization is hardly new -- IIRC Sparc already had Java acceleration some 15 years ago, at there were POCs for that published in CE literature.