Posts by mattaw2001 89 publicly visible posts • joined 8 Aug 2021
I'm not sure I agree. Linux was first released fractionally before the BSDs went open source. Maybe a couple of years after the gpl debuted.
It has been argued that Linux is the success it is due to the requirement to contribute code back.
Linus himself has credited gpl2 as being important: https://www.cio.com/article/238985/linus-torvalds-says-gpl-was-defining-factor-in-linuxs-success.html
I was deeply upset that windows on Linux is still gimped re GPU and some other hardware
Windows 11 for devs, what a joke. I literally invested two days to switch from vscode to neovim as MS make it work more and more how they want it to work and break my dev setup. Subscriptions coming soon!
Folks, I remember in the early internet multiple site would have a component (maybe and iframe? probably a table) which contained "you are on this site" and offered links to the "previous" and "next" one in the list and a link to the list itself. It was based on interest AFAIK and had nothing to do with search engines at the time.
What was it called‽ Driving me crazy.
Note, for the USA, legal monopolies are fine under the law as long as the monopoly comes about through providing an excellent product or service. It only becomes illegal when the monopoly is used to force competition out and interfere with other businesses or business relationships.
So I believe Apple would have to have used its purchasing power to bully TSMC (unlikely, NVidia tried and failed!) to to not sell to other companies, which is not happening here AFAIK. They also do this to various ICs like sensors or high capacity flash chips.
I'm kind of surprised this wasn't discussed in the article - a huge problem from any lawsuit is the enormous amount of time it sucks up from everyone involved and I don't see Microsoft making that good. Even if money could cure the damage, which it often can't.
Also I feel that unless you can enforce the guidelines through policy this kind of guarantee feels very wishy-washy. How does one prove that it was all within those guide rails, especially when you blend AI generated results into other material.
And surprisingly, performance. It's on a par or better then chrome on most benchmarks - a steady improvement over the last year or so.
That could also be due to chrome's constant API growth, eg. Webusb, because of course bidirectional access to and from the Internet to a wire protocol that has no trust model to speak of.
Professor left an expensive data capture card in it's brown box on his recycling bin and was furious when it was tossed.
For a week we're had to empty our own trash thanks to the cleaners, not unreasonably, refusing after they got shouted at.
Allegedly peer pressure forced the least genuine apology ever uttered and a new policy added that if you leave anything on a bin it's your fault - state the bleeding obvious!
If you had a simple DB with simple queries happening then you have already migrated, unless that means re-qualifying costs etc.
If you have taken "advantage" of any of the massive array of footguns sold by Oracle, followed by downing several glasses of their koolaid, then I downright disbelieve that a "wizard" will work. I just don't.
I wonder if they made the mistake of confusing access to the backups with access to the unencrypted content of the backups.
It's easy to assume that because your backup files are encrypted they're safe, but they're only "safe" against being read, not destroyed or in this case encrypted again!
'"Free Software" is a term chosen and used to describe a rather viral form of Open Source, '
"vaccine" has a lot more positivity around it, but virus does describe it better - the infectious nature it. I just can't think of a good clean positive metaphor for open source of the GPL variety!
I appreciate your perspective it's a good one. I would like to raise an objection on the subject of scale and authority though - I do think it really matters.
For example a newspaper is held to a different standard when it publishes a story versus anything you say. Having written textbooks, those textbooks are held to a higher standard than things I say.
I think that you probably look at your information source in deciding how much weight to give it so size and presentation does matter in these cases.
I gave you a down-doot as you may not realize that the uController / FPGA / whatever chip has hardware features supporting encrypted and signed firmware built in, just like the UEFI-TPM combination in many desktop PCs, and requires that to work at all. Take a quick look at android cell phone chipsets for example.
https://ww1.microchip.com/downloads/en/Appnotes/00002462B.pdf is the AES crypto secured version for relatively inexpensive AVR micro-controllers from 2018, the decryption keys are in read-protected memory in the micro, and it won't boot with an invalid program, nor will it allow itself to run an unsigned binary.
You can defeat it by disassembling the chip and using a bunch of EM probing on the die to get it, but that is typically outside what most folks can do. You can also sometimes exploit a badly designed firmware to read out its contents, however sometimes the firmware is tied to the specific chip via serial and is signed "on the fly" when you download it from a manufacturer's portal. While encrypted firmware is normal/standard for a consumer part these days, per-device keys and signing is not AFAIK.
The cloud is a self fulfilling prophecy. If it is not more expensive than on-prem, Microsoft will just make it more expensive until it is.
Its a great scheme: Microsoft control the pricing of both on-prem and cloud, so they can raise the prices of both almost at will to get the most profitable outcome for each.
Cloud providers are also in love with the fine-grained level of visibility they get into a customers operations. Meetings, calendars, emails, files, etc. - that will all prove very helpful when the cloud provider launches competing services. See "Amazon Basics" for a comparable strategy. Amazon is letting other companies take the risks and spend the money to create a market, and when it is established they launch a competing product which they advertise first.
[I also believe companies use the cloud to reduce/eliminate technical competency and knowledge in a customer. Then they *have* to buy cloud and other management services as they have no-one who can manage local systems.]
While I am sure a few folks may get away with it, I know that many judges will take a dim view of such a case. Default judgements are a real thing though. Your plan seems pretty solid but we need to beef it up :)
1. Legality - no problem, purchasing a PS5 for someone is totally legal. Hopefully we can supply some legal good or service on our side.
2. Consideration - while twitter supplying the PS5 is their value, what are we bringing to the table? Maybe commit to advertise the whole story on social media, news sights, etc (keep any royalties etc. of course). Seems legit.
3. Meeting of the minds - This one is tougher, how would you show that twitter knew what they were committing to? Maybe we should let this one slide, or double down on Elon's use of emoji's to conduct business? Actually due to Musk's behavior this might be easier than it looks....
Thoughts before we go ahead with this? (grin)
I really don't know any government that is pro this, unless it is deliberate. I have heard rumors (from several decades ago) that someone lost a lot of classified docs, which were subsequently found, forcing a government backpedal. The individual concerned was reprimanded, but promoted the year afterwards, no doubt having done their "duty" by their branch.
So the belief that Elon is deliberately changing how Twitter shares tweets etc. is fairly based on evidence. Of course, Twitter already did this, but as Musk is directly involving himself now and owns the company it is "his fault".
The Twitter code release may have spurred this particular suspicion re LGBTIQ+, however Musk has been cited as pushing himself. There was a period mid-February when lots of folks complained they were being shown tons of Musk's tweets. Articles were published stating that Elon was agitated his tweet during the Superbowl received less engagement than President Biden's. Apparently the twitter core team of about 80 employees was summoned by Musk to "fix this" which resulted in the flood of Elon tweets everyone saw. He also reportedly fired one of Twitter's principle engineers for suggesting his tweets were less interesting as interest in him was declining. https://www.platformer.news/p/yes-elon-musk-created-a-special-system?r
A colleague of mine was visiting the Synopsys User Group conference, SNUG, recently and discovered a massive downside: you do not get the model stored with the design. Chip design is iterative, and you store the pseudo-random number seed with the design. If you need to recreate it, you download the version of the EDA tool, fab design kit, and run the scripts - bingo identical design.
I cannot count the number of times things have to be ripped up and replaced - its constant as the design evolves towards tapeout.
Synopsys' new tool does not save the model history or state, it learns from itself and your design over time - bottom line Synopsys' engineers said its designs are not reproducible.
Yay, a tool with millions of dollars riding on it is now unable to recreate a design / find that local minima again. God I love the black-box nature of ML.
I predict they will fix this and you will be able to snapshot the model state with your design in the very near future.
I wonder if we are looking at a domain problem, where folks not skilled in ML are trying to apply ML techniques?
Two things I would like to highlight in your comment though. Firstly, apparently the Nature article does not describe the actual flow Google used, namely that the ML flow optimized an initial layout performed by Synopsys' EDA tools - this seems a serious omission that requires Google to respond and or amend their paper in Nature. Secondly, there are no modern large opensource IC designs, especially resembling a commercial flow on a modern fab for Kahng et al to test on.
Overall while I appreciated your comment and how it highlights shortcomings in Kahng's approach, I feel this is proper science and publication should go ahead. If Google publishes a paper claiming something they have to defend it against criticism. Kahng tries to reproduce it and is struggling, documenting his approach in a reproducible way and cannot achieve the same results. You read both, spot flaws in Kahng et al's approach and highlight them.
TLDR I do not agree on your take of not publishing, if nothing else the flow omission, but shortcomings of Kahng et al. will come out, and the onus is on Google to put up or shut up - published papers should not be waved through on a "trust me bro" basis
I have never understood why companies find it so hard to comply with a simple license? The GPL has almost no boilerplate or pages of rubbish it's really clear.
In actual fact I feel open source's plain English has helped it stand up in court more than almost anything else. The more complicated the license the easier it is to screw something up or to demonstrate there was no meeting is the minds - that you didn't really understand it.
The only two things I found really attractive from immigrating from the UK to the USA financially was that *I* own my mention fund and 30-year fixed mortgages are normal.
The UK military and the members of Parliament pension are the only two final salary schemes I would ever trust.
I'm not sure how to answer this argument that modern C++ solves memory problems. The argument normally comes with a trivial example as proof. And the example is typically already easy to identify & fix with linting or static tools even before modern C++.
Those are not the actual kind of coding errors that leak memory that the approach rust takes is intended to address.
From my experience the real problems lie in thread / hardware module interactions (often not fully defined even on existing x86 hardware), or the use of an API that requires a guarantee that the coder failed to meet in 1% of cases. Or where someone just simply screwed up and the tools never ran on that part of code due to error.
In my experience C and C++ requires the coder to have an ungodly understanding of the entire code base and operating system and sometimes even hardware specifics to be sure the code they are writing is correct.
In some ways this is why explicit ownership enforcement is the best paradigm to address this problem.
Unless you tried to use one of its gimped implementations forced by the carriers of course.
I still remember the eternal feud between the carriers & Nokia, e.g. debating whether being able to view .jpeg pictures should be possible on a business handset, as that was an entertainment feature.
Honestly this is why, word for word, ARM does not often sell architecture licenses and prohibits custom instructions.
The reason ARM standardized the CPU + interrupt controller and forced custom hardware out of the CPU into accelerators was to create an ecosystem of good compilers, transferable software and transferable employee skills vs. the absolute horror of the old ARM7 & ARM11 days.
ARM was getting stuck with the support burden of customized IP blocks, custom instructions, different interrupt controllers & the inability of software to be reused especially to design a compiler properly to support lots of this stuff.
They are OK, up until you use C/C++ libraries to interface with the DB.
Then you are faced with either:
1. Custom code paths to return the NULLs out of band with the actual data buffers
2. Return a struct for every data value containing a bool representing Null & the data buffer
3. Use an actual binary code that is normally mapped to data to map to Null (e.g. -1, "\0"), etc.
All three options stink, can introduce hard-to-find bugs, and waste a bunch of performance as you can no longer have simple, consistent, buffers of ints/floats/text info but have to search for Null-ness or holes, etc.
Note, even higher level languages like Python interface with DBs through C/C++ libraries so suffer in various ways from the 1-3 choice above.
Particularly impressive and beneficial is how much code Rust can save you writing in the first place, especially in drivers with complex memory interactions (nearly all of them). For complex memory interactions in C / C++ you either have to personally identify the memory interactions as safe, or write code to do memory isolation around them. In the case of Rust you often don't have to write any of that, as the compiler will guarantee correctness. It also reduces the maintenance burden by automatically accommodating memory management for new patches over time, and it can bring speed boosts as the original locking scheme can be too conservative and fail to keep up with patches and updates.
Since semiconductor manufacturing is still one of the toughest technical jobs I think the determining factor is access to a workforce, laws applying to same, and the costs of paying them. I guess Arizona, for whatever combination of laws, living costs, education system, etc. is their sweet spot.
Of course it could simply be the best state at incentivizing **cough**bribe**cough** them to invest
I believe you deserve more recognition for your thoughtful post. A knowledge of history is essential to understanding the present, and the idea there ever was a time of uncensored, private access to information or communications is essentially a fantasy.
Complete privacy and uncensored access may not be what the majority of people actually want, and we live in a democracy. Balancing rights and freedoms is what society is all about - there are no absolutes.
For example, right now I am controlling my children's access to the Internet, while being undermined by the tech companies I do not want uncensored access or private communications for my seven and eleven year olds. I also believe most people, if asked, would be not only be OK with their online activities being looked over for criminal activities or terrorism but surprised it wasn't already happening.
I am pretty much OK with it, although paging through millions of lines of my eBay searches for a cheap multi-meter might be considered a health hazard...
Honestly, this is one of the more nuanced discussion of the technology I have seen. I was expecting the end of the article to claim that we were within five years of quantum computing which will replace everything we know, not a statement that "we have to try it out to see, really".
I just don't think humanity as a whole can afford to have such an amazingly expensive and energy intensive technical device be replaced at the rate of cell phones.
I hate to recommend it but I think we're going to have to have regulation to insist on replaceable batteries and software updates for say 5 years.
In a way you are right and completely wrong at the same time. Commercially, 180nm is a legacy tech that is long dead, perhaps surviving in some exotic niches like MEMs, FRAM, etc.
For defense and aerospace however the 180nm production is very helpful for certain "radiation rich environments". Having a local wholly owned and operated fab is the best way to prevent certain critical designs from being shared outside of their users.
Honestly, Moore's law as he spoke it is dead, but "more than Moore" (MtM) is alive and well with some creating thinking.
I do think folks are overlooking some advantages that chiplets bring to AMD
1. AMD can simply vary the number of chiplets in a package to deliver multiple product families, both server and desktop
2. AMD can direct flows of chiplets to the highest margin sector cheaper and quicker than monolithic dies
3. AMD can engineer/optimize entire product families by improving one chiplet design!
4. Smaller dies lead to better binning with these new variable fab processes, for example low power chiplets go to servers, high performance & power chiplets to workstations, slow and hot to desktops
Disadvantages are more wafer lost to sawing, more expensive packaging, more dies to handle and test.
The higher power draw of chiplets on first glance should lead to a more power-hungry CPU, but in practice this is not true as a large monolithic die has to be powered to the lowest performing transistors on the die, while can be binned to create CPUs which more than recover that energy cost.
QC is terribly important, and these new hybrid cores are new to the x86 space, so testing on the actual hardware is essential.
However, I am not overly fond of sharing our nearly completed project(s) with Intel first - what are the guarantees for privacy etc.? I assume their cloud system will record everything, and possibly be swept up into backups as well. I also guess that Intel will require us to indemnify them from any responsibility for a leak, and we are not big enough to get a custom arrangement in place.
It is still a big step up for smaller shops who don't have lots of different chips and chipsets to test on though.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2023
