When I saw your comment about X, my first thought was "Oh, what's wrong with X windows?", then a couple of seconds later my brain said "Oh, wrong decade, the writer probably means the artiste formerly known as Twitter".
Posts by Chappy
25 publicly visible posts • joined 4 Jan 2012
'Almost every Apple device' vulnerable to CocoaPods supply chain attack
Google pulls RISC-V support from generic Android kernel
AWS exec: 'Our understanding of open source has started to change'
Microsoft will upgrade Windows 10 21H2 users whether they like it or not
Samsung scores fresh Radeon deal with AMD for Exynos chip line ahead of profit crunch
Unix is dead. Long live Unix!
After less than half a year, Intel quietly kills RISC-V dev environment
Re: If it's not x86 it's not Intel
I don't think DEC should be in that list of companies supporting Itanic. DEC had its Alpha CPUs (and also designed and manufactured StrongARM). I remember the Alpha CPU team writing papers explaining why Alpha was able to be higher performance that Itanic, and their arguments made a lot of sense. DEC stuck with Alpha until they went bankrupt. Oops.
Windows Subsystem for Linux now packaged as a Microsoft Store app
SpaceX reportedly fires staffers behind open letter criticising Elon Musk
The letter asked SpaceX management to break the relationship between SpaceX and Elon Musk.
This is in effect asking "Whose continued contribution do SpaceX management value higher: us small group of grumbling employees or Elon Musk?"
The management answered their question by firing the party whose continued contribution was considered to have lower value.
If you ask a stupid question, don't be surprised if you don't like the answer.
Re: Does as I say, not as I do
I don't see too many articles in the Murdoch owned press criticizing Rupert Murdoch, his children, or his businesses (e.g. remember employees at one of his papers hacking the mobile phone of a murdered child? I don't think you'll find too many articles in the Murdoch owned papers and TV news berating the Murdoch empire for that one).
Microsoft unveils a Universal version of Office for Apple silicon
Apple have approx 9% of the PC & laptop market with their MacOS machines.
That 9% of the market is definitely moving to Arm (assuming that Apple's market share does not shrink).
In the Windows 10 market, I assume that Arm based machines is a tiny percentage of the market, probably < 1%.
So it makes sense for Microsoft to put more effort into Office for MacOS on Arm than into Office for Windows on Arm because there will be vastly more users of the Office for MacOS on Arm product.
What did they do – twist his Arm? Ex-Qualcomm senior veep joins SiFive as CEO, RISC-V PC for devs teased
Broadcom's Arm server chip lives – as Cavium's two-socket ThunderX2
Re: Is it just me?
As someone else posted, ARM started out as a desktop personal computer processor.
ARM was in the Archimedes 400/1 series computers released by Acorn in 1987.
So ARMs journey was Desktop --> embedded --> smartphones & tablets --> Chromebooks --> Windows laptops & servers
See https://en.wikipedia.org/wiki/Acorn_Archimedes
And https://www.youtube.com/watch?v=NF_9McTFR20
Acorn designed 3 generations of ARM processors before the technology and 12 engineers were spun out into a separate company Advanced RISC Machines in a joint venture with Apple in December 1990. Apple wanted to use ARM processors in the Apple Newton, which was released with an ARM610 processor in 1993.
ARM was originally an acronym for Acorn RISC Machine.
It was changed to Advanced RISC Machines when the joint venture company was formed.
Apple sues iPhone CPU design ace after he quits to run data-center chip upstart Nuvia
Re: Apple wouldn't do similar?
Apple gave ARM Money + A dream lead customer (i.e. lots of free marketing)
ARM was created as a joint venture in 1990 between UK computer company Acorn, Apple, and VLSI technology who at that time manufactured the ARM processor chips for Acorn, and got a small percentage holding (5% if I remember correctly) for providing VLSI design tools to ARM for a number of years.
Apple wanted to use ARM processors in the Apple Newton, but didn't want to buy them from a competitor in the desktop computer market, which Acorn was. Acorn wanted to spin out ARM because they couldn't afford to keep developing successive generations of CPUs and thought that spinning it out would also lead to value creation (they were right about that, SoftBank bought ARM for 32 Billon dollars).
So Acorn transferred the ARM technology + 12 people; Apple put in some money (a million or so pounds or dollars); VLSI put in the design tools. ARM was born, and the "A" in the ARM Acronym was changed from "Acorn" to "Advanced" and "Advanced RISC Machines Ltd" was born.
Having Apple as a customer launching a glamorous new category of hand held computer called the Newton with an ARM610 in it was pretty free good marketing too.
Intel's back. Can't keep it down. Back with 5G. Back in the game, back with modems... that have 'MediaTek' written on them
Nooooo
... except that it is not "Native Signal Processing". Native Signal Processing meant doing signal processing on the Intel CPU by incorporating SIMD instructions starting with MMX. This is adding a separate chip from MediaTek (which is probably full of ARM processors) so it is not Native Signal Processing.
Cause and Effect
I don't the article gives the correct order or causality of events.
I don't think that Intel gave up on 5G because Apple suddenly feel back in love with Qualcomm.
I think that Apple was forced to eat humble pie and go crawling back to Qualcomm offering several fist fulls of dollars to buy peace because Intel could not deliver a competitive 5G solution in the timeframe required to keep iPhones competitive.
This is based on my memory of articles I read at the time, and as in all things, I might be wrong.
Fresh fright of data-spilling Spectre CPU design flaws haunt Intel
Re: So Raspberry PI still in the clear?
ARM Cortex A7 is not affected by Spectre or Meltdown.
See https://developer.arm.com/support/arm-security-updates/speculative-processor-vulnerability
and note the statement above the table listing affected processors that
"Only affected cores are listed, all other Arm cores are NOT affected."
Uncle Sam's treatment of Huawei is world-class hypocrisy – consumers will pay the price
Re: Politicised journalism over political theatre
Arm started out as British (Original CPU designed in Cambridge by Acorn staff).
Arm added a CPU design centre in Austin, TX in the 1990s; a design centre in Silicon Valley that came from the acquisition of Artisan Components in ~2004, and has had a CPU design centre in Sophia Antipolis for many years. There is a GPU design centre in Norway, other design offices in Sweden, Israel, Grenoble in France, and more recently in Taiwan and China. There are probably others that I have forgotten. So in terms of design centres, Arm has been multi-national for more than 20 years. The Cambridge UK campus is growing significantly with the continuous construction of new buildings for the rapidly growing Cambridge workforce.
In terms of ownership, Arm listed on the stock market in 1998. Since then anyone in any country could buy its shares. Softbank bought all of the shares in Arm in 2017.
It seems to me clear that Arm has long been an international company, The main remaining claim to Britishness could be that its corporate headquarters is in Cambridge, UK; even though its British CEO Simon Segars very sensibly lives in Silicon Valley (nowhere else on earth is comparable in terms of proximity to other industry players). Mike Muller who is CTO and has been a director of Arm for 20+ years is also British and lives in Cambridge, UK. So perhaps some British culture remains, but by all other sensible measures it is an international company and has been for a long time.
Re: Politicised journalism over political theatre
"If you want defect-free chips you need to change the design process to use a mathematically-verifiable method, not the country where they're designed or fabricated."
Using a mathematically-verifiable method is not going to help you protect against a type of side-channel attack that no one has ever thought of at the time of doing the design. The Meltdown and Spectre security flaws are exactly that: Multiple companies (Intel, AMD, IBM, Arm, and probably others) have been designing processors with speculative execution for years, some of them for a decade, and no one ever thought of these side channel attack.
One of the side channel attacks works by training the branch predictor in User code, so that speculative execution in kernel space will do a specific branch prediction so that it speculatively executes code that will use a secret in the calculation of a branch address or of a load address, so that the cache line that is loaded is dependent on the secret; and then user code can work out the secret by using timing analysis to work out whether the targeted cache line was loaded or not. It is seriously non-obvious, which is why no one found it for a decade.
Other side channel attacks have the same property - e.g. think of differential power analysis attacks. Before these type of attacks were thought of and demonstrated, you could design a perfectly secure chip with all of the mathematical proof of correctness that you want. Then someone comes along with a completely new tool for looking inside your design. Mathematics and formal verification won't help you, because you didn't know that that new tool was going to be invented.
Introducing mathematical techniques will not help if you don't know what you are protecting yourself from.
Leaked screenshots show next Windows kernel to be a perfect 10
Re: the 64 bit address space is "huge"
@Tom13:
You can build computers with 128-bit data buses without increasing the width of the address bus, and people have been doing that for a long time.
Most modern computer architectures - including intel and ARM - have a set of wide registers used for floating-point and SIMD aka "Media" processing. Register widths of 128-bits and 256-bits are common.The contents of the registers can be treated as multiple 8-bit, 16-bit, 32-bit, or 64-bit data items, etc. to allow parallel operations (e.g. treat a 128-bit register as four 32-bit floating-point values, and do four parallel adds to the four values in another 128-bit register and write four results into a third 128-bit register).
So the industry is already using wider buses and wider registers to get more performance.
Note that there is no need for the register widths and the bus widths to be the same. Different price performance points can be implemented that can run the same code at different performance levels. Also in the example above, different price performance points can implement different numbers of floating-point adders. Four operating in parallel gives maximum performance. One doing each of the four adds sequentially gives lowest cost.
These machines with wide buses and wide media registers are still considered 64-bit machines if they hold addresses (pointers) in 64-bit registers and have 64-bit scalar integer math and logical operations.
Finally - a solution to let people make money online WITHOUT ads?

Re: He said reduce "transaction cost" to pennies.
> There are no transaction costs when a photographer sells a picture to a newspaper.
Of course there are transaction costs.
e.g. The cost of paying someone at the newspaper to find a relevant available photo, talk to you, look at your photos, negotiate the price with you, send you the licensing contract, check that you signed it, check that you are credibly the owner of the copyright and so have the right to sign the contract, taking the contract to someone internal who has signing authority for the newspaper, sending you a copy of the signed contract, entering the photo in a system that tracks its future use against the contract that authorises it, setting up the payment to be made to you, etc.
The more this is standardised and automated, the lower the transaction costs.
That does not have to mean that the price paid is standardised. e.g. ebay provides an electronic auction platform that allows market forces to determine the price paid for a disparate range of items, while lowering the transaction cost. In the case of ebay, it particularly lowers the cost of the buyer and seller finding each other, as well as providing standard ways for them to interact, and a standard legal agreement behind the transaction.
Imagine an ebay designed specifically for licensing photos rapidly, with standard watermarking requirements, etc.
Smartmobe Wi-Fi blabs FAR TOO MUCH about us, warn experts
Re: Clearing some things up.
A strong motivation for turning on WiFi in some locations (e.g. city centres) is to improve the accuracy of positioning information where tall buildings hide GPS satellite transmissions.
So is there a simple way to tell my iphone or android phone to forget all of the WiFi networks it has previously connected to, so that these probe requests are not sent?
Or any other way to leave WiFi on (for improved location) while disabling the probe requests?
Guy
Virtual sanity: How to get a grip on your home PCs

Tutorial please, or a pointer
If I want to create a setup like that, what do I need to do to create the virtual server?
What software do I need to run on the virtual server box?
Do I need special license to allow multiple instances of XP (or Windows 7) to run?
Can anyone point me to a good tutorial on that?
Thanks