HNHacker News
TopNewBestAskShowJobs

cwizou

826 karma · joined May 6, 2020

Former hardware reviewer, I make the Aerial screensaver for mac https://aerialscreensaver.github.io
submissionscomments
cwizou··on Qualcomm’s M1-class laptop chips will be ready for PCs in “late 2023”
They did, but late 2023 is still a very short timeframe (less than 3 years) to get a custom arm64 CPU architecture, ready to ship.

Pre armv8, Qualcomm had strong custom cores, but Krait (their armv8 custom implementation) didn't exactly perform well compared to basic Cortexes from ARM (a problem that wasn't unique to Qualcomm, to be fair), and Qualcomm moved to using those generic ARM cores on most of their lineup.

Qualcomm reportedly did get some TMSC volume at 5/6nm that would correspond to that timeframe, but the timing would be extremely optimistic considering how long of a lead time you need, even if the Nuvia acquisition solved all their problems.

cwizou··on Qualcomm’s M1-class laptop chips will be ready for PCs in “late 2023”
I think the breakdown is with native/non-native apps and specific build systems.

Pretty much all the native apps that get updated for M1 are universal binaries. Xcode is (all things being relative) pretty good at making this work semi-transparently.

The only instances where I see separate builds for x86/M1 is with non native apps, a good example that I saw just yesterday being Postman. I would guess either their build system is the culprit, or some libraries that they rely on. It's not systematic though so this is particular to those vendors, it could also be that they chose to do separate builds for other reasons, though it's not what Apple suggests developers do.

Teams and Discord both have universal builds and are non-native, as counter examples to Postman. Another good example is that while VS Code [1] is universal, VS Codium [2] makes separate builds (those seem to have been posted in the last 12 hours, yay !), and this is mostly due to their build system [3].

This site is also a good resource that tells you what has native M1 support, and what still runs under Rosetta : https://isapplesiliconready.com/

[1]: https://code.visualstudio.com/Download

[2]: https://github.com/VSCodium/vscodium/releases

[3]: https://github.com/VSCodium/vscodium/issues/597#issuecomment...

cwizou··on Apple to Apple Comparison: M1 Max vs. Intel
Yep, I think it's interesting to note how slow x86 was in general with refcounting even compared to the PowerPC era.

It's something that showed early on quite a lot when you compared what was considered light usage at that point in time (say open Word and a couple of Safari tabs), a G4 wasn't ridiculous in those scenarios compared to the first Core 2 machines.

While I'm sure the designers paid particular attention to those instructions, it's really x86 being a terrible instruction set than the other way around that create that particular refcounting gap I think.

cwizou··on FTC sues to block Nvidia-Arm merger
For many different reasons too ! Some were a bit petty/misguided :

- Microsoft was mostly contractual on the Xbox, Nvidia didn't want to lower the price of the chips over the lifespan of the 1st Xbox which was a huge no-no. And some technical people also had a terrible experience (some people in the DirectX team had strong thoughts about working with Nvidia).

- Sony was mostly about terrible tech support and not wanting to share enough technical details which hampered AAA devs a whole lot, to the point most ended up using the SPUs in very creative (and technically fascinating) ways to compensate (the GTA V engine on PS3 was truly impressive in that regard).

- There were some minor (but acrimonious) back and forth with Apple over a failing gen of GPUs (G80s if memory serves), but the large fallout came for the same reason as for the rest of the mobile industry : the Kepler licensing initiative.

Long story short, Nvidia tried to assert patents on mobile GPU (ala Microsoft on Android, as a way to "sell" their exit of the mobile market to investors). They then sued Samsung and Qualcomm [1]. Samsung countersued Nvidia, which had to settle as Samsung was close to winning a ban on imports from some Nvidia products [2]. Many of the patents Nvidia tried to assert were thrown out and they had a terrible legal time.

The damage that ill thought strategy did to Nvidia is hard to measure but there's not a single mobile company that wanted that acquisition to go forward and many are probably quite relieved after this.

[1] : https://www.zdnet.com/article/nvidia-launches-patent-suit-ag... [2] : https://www.reuters.com/article/us-nvidia-samsung-elec-idUSK...

cwizou··on Private keys used to sign EU Digital Covid Certificate might have been leaked
So I haven't had an issue yet and having the QR code is definitely sufficient in France (and always attempted to check when appropriate in my personal experience, no ID check though as that was deemed illegal).

There is however a prevalent issue with QR code scanning that I noticed twice personally (and friends reported the same experiences at other places), which is that most restaurants use a super cheap phone/tablet with a terrible camera that usually is unable to focus on a smartphone screen correctly.

Once, the person was honest about it and explained they kept having issues with it, and I tried to help him focus correctly his device (QR code size + luminosity seems to be the issue from what I saw).

The second one tried a couple of time, didn't manage to focus and told me "all ok" despite it's device not having scanned me (I could see the screen).

If I had to travel though, I'd definitely take the paper certificate on top of it, as, at least here in France, that's considered for better of for worse the source of truth.

cwizou··on Private keys used to sign EU Digital Covid Certificate might have been leaked
It's definitely a solution when you have an app (assuming you have a 3rd party source of identification of users), but since it's a French key, a sizeable portion of people, mostly elderly, are not using the "official" app but a paper printed QR code that was given to them at the time of the second injection (or a picture of said code).

Since the French app is pretty terrible (it used to be the barely functional contact tracing app that didn't want to use Google/Apple's API, so there's a pretty strong popular stigma against it), even a sizeable portion of smartphone users use an alternative too (I personally have my QR code in Apple's Wallet, but there are other 3rd party solutions).

Depending on when that key was in use, this may be an epic bureaucratic mess to solve here in France if they have to reissue many.

Edit : I'll quickly add that while there's a database of vaccinated people handled by CNAM (the single payer social security administration), it doesn't seem extremely accurate : my best friend and one of his collegues were vaccinated, issued a QR code, but they weren't added to said database and had a terrible time getting the situation fixed. In the end, a pharmacist reissued a "vaccination" for them to "get them in the system". So that database is probably at best incomplete (for the process of reissuing).

cwizou··on AMD’s Lisa Su
As part of my former job I used to follow AMD very closely and I have to say that while maybe misplaced in some ways, she deserves all the praise she gets and then some.

Some context, when she took over, not only was the CPU division in shambles and the product line not that great, but the internal infighting was terrible between the GPU and the CPU division. Most of it was fuelled by the marketing department taking the GPU side.

Not only did she had to handle this, but soon enough she had part of the company actively working against her to get her removed and replaced. I can't tell you how many times some very senior AMD people would go to the press and bad mouth her, talking about her incompetence (and I'm putting things mildly), and seeding rumors about her near and ineluctable ousting.

Political infighting in a large org is always a thing, but I never noticed anything that publicly overt (at least to the press), prolonged, and acrimonious.

She focused a lot on righting the CPU division and getting the trust of key senior people there and in the end, it paid. During that same period, same people from above were seeding rumors that Ryzen was going to be a bust, and the character attacks went up to 11.

I remember the first Ryzen press event, and you could see a mix of relief and confidence (I would say that prior to that event, she never appeared confident on stage) that came with it during her presentation. At that time she knew that not only she had a good product line for the next few years, but that the coup attempt would die down.

It took some time but those behind the attempts left the company and AMD ended up in a better place.

So yep, her resilience always impressed me and she will never get enough credit for it. AMD's turnaround is obviously not just hers, it's shared with many key people (eg. on Zen, Mike Clark, and Keller who - check for the Anandtech interview in another comment [1] - made the battered and bruised teams believe they actually could).

But she definitely played a large part in it. A bit before Ryzen was released, I remember casually asking one very senior person on the CPU division what he thought about Lisa. He started talking about the fact that sometimes she sets impossible deadlines (which boss doesn't?), but then he paused and reflected a bit and said "but she listens and understands". Coming from that person, that was about as much praise anyone in management could get.

[1] : https://www.anandtech.com/show/16762/an-anandtech-interview-...

cwizou··on AMD’s Lisa Su
That's the first confirmation I see of it too, and I followed that pretty closely a while back. We knew they had both at x86-64 and a ARMv8 project running at the same time (K12 was the ARMv8) but it never was really spelled out how they implemented it (best I got back then was that they shared a lot of commonalities).

They cancelled it though as they saw : 1) the ARM server market wasn't happenning, at least not at the timeframe the product was to be ready for 2) the x86-64 could really be competitive with Intel

Then Intel had some issues and here we are.

cwizou··on iOS 15
> If they ever get the port to work on a newer version of Safari, it could work on iOS.

I looked at it at the time but sadly the Content Blocker API that Apple released is incredibly crippling compared to the feature set of uBlock Origin.

Basically, you can onlly generate lists of domains that you want to block, package that in (multiple !) batches of 50k (at the time I think) and let Safari do the blocking for you.

This removes the ability to do anything non-static basically, which is where uBlock Origin really shines, so I don't expect anyone to make a meaningful port anytime soon, sadly.

cwizou··on Why the American shoe disappeared and is hard to bring back (2019)
TL;DR: C-States improve performance.

I'm not sure calling C-States throttling is completely accurate, at least not in the way that you imply.

Modern CPUs have multiple mechanisms that can be called throttling in some way, but C-States are the least offensive and in modern CPUs actually increase performance.

There's a good basic explanation here [1] but in short :

- C-States is about idle cores, and powering them down. You can actually configure how deep you can power them down. Waking up does have a tiny cost in latency, yes, but it's mostly irrelevant (your core spends most of it's time waiting for data).

- P-States are what I would call "ok" throttling, those happen when non-idle, modern cores can scale their frequency up and down based on how much workload they have. Again, in theory purely about power saving and close to 0 impact on performance.

This was simple, and used to be just about power saving, disabling was ok. But things changed massively about 10+ years ago with the introductions of various Turbo mechanisms.

Turbo is about looking at your CPU as a whole (and not per core), and maximising performance of the loaded cores by pushing their frequency, using the power budget that used to be allocated to the idle cores (that are now powered off).

The way its done is by adding a few extra P-states that push the frequency a few hundred MHz. The more "C-State-down" cores you have, the higher the frequency of the working cores is pushed.

Extreme example : you have a 28 cores CPUs running a single thread workload on one core only, with 27 idle cores. With C-States disabled, you lose Turbo and are stuck with the single core running at the "non-Turbo" frequency. With C-States enabled, because you idled the non used cores, your working core gets higher clock speeds and higher performances.

(Things get more complex with AVX2 because those units have a large power consumption, and this is where you get into the very bad throttling)

[1] : https://software.intel.com/content/www/us/en/develop/blogs/c...

cwizou··on Social media influencer/model created from AI lands 100 sponsorships
I think that this is also part of the craft/craftmanship, it's not just the tools but the knowledge and inspiration to use them, to express something. It's that entire package that's more valuable than the end resulting physical object.

Do we get part of that vision/window of the world from a high resolution digital reproduction of the painting? Could we get most of it from a (very good) replica too? Would most people know the difference if we swapped the original for a replica in Le Louvre? Or let's say the painting gets damaged and restored by artists. Did we lose that vision?

cwizou··on ‘Havana syndrome’ and the mystery of the microwaves
That article was much more informative than I thought it would be, there's some interesting background on previous public research on the topic of microwaves, but what was completely new to me was the fact that such cases go back decades.

Slight off topic, but I do see a bit of a parallel between medical teams claiming that those are psychogenic illnesses and the way "long covid" has been considered by some practicians. There's a bit of "if we can't measure it, you must be (unconsciously) making it up". It's rationally convenient.

In the "Havana Syndrome" case, the article claims that some blood markers have been found recently that correlate with brain damage (as long as measurements are done immediately), which certainly helps. But the mix of secrecy inherent to diplomacy and classical governmental bureaucracy makes considering it psychogenic a very satisfying solution to push the problem aside for years.

cwizou··on Matrix trailer dynamically uses current time
15k is getting very blue in terms of temperature. Maybe bitrates? Those would also fit.
cwizou··on Incident: Airbus A330 at Taipei, primary computers failed on touchdown (2020)
> The deceleration performance of the occurrence flight between 6,600 feet and 7,300 feet from the threshold of runway 10 deteriorated. It may be due to paint marking and rubber deposit on the touchdown zone of runway 28

Just to clarify for others, this is the same runway but taken in two different directions (west to east and east to west, multiply the number by 10 and you get [rounded] angle in degree, so 100ish vs 280ish). It's 8500 ft. long and it looks like there was a loss of performance around the "touchdown" point of going the other way on that runway, where you had accumulated rubber from those landings (which is pretty normal and monitored, as pointed by leecb below), if that makes any sense.

cwizou··on Incident: Airbus A330 at Taipei, primary computers failed on touchdown (2020)
> Perhaps the root cause is an issue in the weight-on-wheel sensor inputs that the computers use to transition. The computers need to use redundant sensors so they don't all rely on the same faulty sensor, but maybe the sensors are not appropriately cross-fed or the sensor input combination logic is too different between the computers.

This would indeed explain the discrepancies between COM/MON which is what they called asynchronicity (which is a bit different from where my mind went when I read that word the first time). There was no particular fault found in the computers.

Maybe wet runway factored in a bit (or some other issue with the runway itself), but this is a common occurence around the world and this was a first on the whole 330/340 family.

> The problem with not disconnecting a computer in this situation is that bad inputs from that computer may also cause a crash. But maybe if no other computer is available to take its place the logic should be different, at least during takeoff/landing.

Yes, only thing I would add is that they failed in cascade within one second of each other, 3 seconds after main gear touching (and before the nose gear touched down):

> Three seconds after the main gear touched down, autobrake system fault was recorded on FDR. One second later, PRIM1/PRIM2/PRIM3 faults were recorded at the same time and the spoilers retracted, as the ground spoiler function was lost.

Whether all three should be allowed to be failed within one second then default back to direct law (manual) is a hard question to solve, in this precise case it seems like try again would have maybe worked? Would it always be safe to do a retry? I don't know. This is a super critical phase of flight so retrying doesn't sound safe.

Apparently they would have needed 2 functional for full autobrake:

> Ground spoilers function requires at least one functional FCPC, arming autobrake requires at least two functional FCPCs, deployment of thrust reversers require unlock signal from either FCPC1 or FCPC3

cwizou··on Apple Exploring RISC-V, Hiring RISC-V ‘High Performance’ Programmers
Apple kicked Nvidia out of its product line a long while ago, after Nvidia tried to collect licensing fees on every iPhone sales for their GPU IP. They launched their legal actions with Samsung first, and it backfired spectacularly. [1]

I believe there are some higher ups who would not welcome at all a Nvidia takeover of ARM, and that's putting aside potential increases of licensing fees, or other actions.

I'm sure they have a plan B in place should that happen. Whether those hiring are it or not, I don't know.

[1] : https://www.reuters.com/article/us-nvidia-samsung-elec-idUSK...

cwizou··on Apple Delays Rollout of Child Safety Features
But that's the thing, they have reached a compromise regarding iMessage being "E2EE" by keeping all those avenues (iCloud backups and iMessage on iCloud) that essentially trade E2EE for the convenience.

Most users opt-in to the convenience for iMessage making it, for all purposes, no longer E2EE.

I don't see them removing that precisely because of iMessage.

cwizou··on Apple Delays Rollout of Child Safety Features
Yep, Apple currently only scans iCloud Mail attachements (I believe they confirmed this, don't have the source at hand though), which explains the discrepancy.

There's a big difference though between scanning a library (which is closer to what DropBox/OneDrive are doing) and when it's messaging platform whose ultimate goal is communication between people. It does make sense that those catch way more.

If Apple had roll those scans on iMessage attachements (which they didn't planned to, though the feature looked to be designed for it way more than it was for a library scan), you would probably see comparable numbers (modulo those platforms relative marketshares).

One big problem though is are those large numbers actually actionable ? I think I remember seeing a quote from Swiss (?) law enforcement that complained that the reports they got from NCMEC were nearly all of the time unactionable.

There are possibly many reasons from this, it could be differences in local laws, algorithms that wrongly report things, or other factors (maybe not enough to identify the user of the acount, etc).

This is where one feature of Apple's design was a bit better: they didn't plan to report every single occurence, but would only if 30 such images were detected. Part of the reason they did was because of the imperfections of the NeuralHash mechanism, sure, but in the end it doesn't matter.

One can argue doing this would at least generate more actionnable reports. It shouldn't be a numbers game.

cwizou··on Apple Delays Rollout of Child Safety Features
Interesting thought but Apple never gives choice on much (well except Safari UI now!), and I don't see them giving an explicit choice on this. If they really had for a goal to implement E2EE, I could maybe see this now that you are mentioning it.

However if I had to make a guess, they'll go server-side scanning only, and E2EE backups/Photos will never happen.

To me it's a pipe dream to believe they would offer E2EE, there's too much international pressure from governments around the world to keep iCloud backups open and going E2EE would trigger unpalatable retaliations.

cwizou··on PhotoDNA and Limitations
> They also use a second, undisclosed system on the visual derivative before human review. While it’s possible they have invented two separate perceptual hashing systems, it’s also quite possible that they would use PhotoDNA for this.

Yep I probably didn't phrase my question well enough but that was exactly part of what I was wondering.

They claim it's a second independent hash as you pointed out, so I would assume this is PhotoDNA server side, unless they made two separate hash systems (one which was not initially disclosed, if not implemented, at announcement time) ?

I'm also wondering exactly how they trained their NeuralHash, but may have missed some parts of the discussion. Supposedly NCMEC is not sharing the database of the "original" content that one would use to train such a system.

Edit : Guess that's a moot point now, they postponed it!

cwizou··on PhotoDNA and Limitations
Very interesting read.

Side question : my initial assumption on Apple's announcement was that they (like everyone else) were using PhotoDNA.

Apple's own perceptual hash has been found in current iOS and macOS releases, do we actually know if it is matching the results of that hash to a PhotoDNA database server side or not ? If not, did they train their model on the raw database with NCMEC's help ?

There was so much hand waving on their "tech" papers (and the recent announcement that they have a server side check anyway as a response to collisions being found), that I have a hard time reconciling everything.

cwizou··on Apple Watch production delayed as engineers wrestle with quality issues
> What's wrong with digitimes?

They publish pretty much any rumor that comes to them with close to 0 verification. Hard to trust anything when they are the root source.

And not just on Apple btw, on anything IT.

cwizou··on Show HN: HN Badges – Displays game-like badges for your HN profile
Also, I'm not sure if it's a Safari 15 beta thing or just a Safari thing, but the hovers don't work on Safari 15 beta from macOS Monterey either.

They work fine on Firefox though.

cwizou··on M1 Icestorm cores can still perform well
Completely agree on everything.

That's the reason we didn't review laptop CPUs when I reviewed CPUs. You can get exact CPU power draw on a desktop motherboard (by using an amp clamp on the P8 connector) but it's hard (or not possible) to do that across multiple laptop chassis.

Removing battery (when possible) is not a solution either as what you get may differ a lot from classic "plugged in" usage (see the references to the MacBook Pro and Dell that used an i9 that still drained the battery when plugged in, because they can use more power than the power adapter brings).

On top of that, way too much depends on the OEM design and the performance of a given CPU will greatly vary from one chassis to another, because of the various throttling mechanism and the various configurable things that OEM can do (it's not just the cTDP, you can as an OEM play with various turbo times, another person mentionned P2 states, which is one of those).

So a given mobile CPU performance means nothing at the end of the day, only the laptop "as a whole" can be measured, which is why you don't see good quality benchmarks of mobile CPUs.

Anyway, just a small complement :

> laptops / cores have an "amp-counter" on board somewhere

Intel (and AMD to some measure) CPUs all have various sensors on chip that gives you the power consumption in watts (or amps, depending). They can be read with software such as hwinfo [1].

Those are usually not incredibly reliable though, they are not calibrated per CPU and it's very much a guestimate that could easily be in some cases +/- 5W off.

So sadly, not usable either (especially on mobile).

[1] : https://www.hwinfo.com/

cwizou··on M1 Icestorm cores can still perform well
You'd be surprised, this is how most modern "high end" Intel laptop actually works. Even some Intel Macbook Pro exhibited this.

This is 100% an Intel thing.

cwizou··on Documents ≠ Programs
That's a very fair point, I guess I skipped over that part, my bad.

My core issue was around this and I should have sticked to it :

> The original purpose of the Web was to be a set of protocols and formats used for the purpose of publishing documents.

I still think this is a bad argument, I disagree that HTML was conceived as a way to make "documents", but it shouldn't matter, things evolve on their own and people may use your creation in ways you disagree with.

> The problem is, if the core content is actually a document, but behaves like an app (and you cannot access simple text because you disabled JS).

I 100% agree with you, but the distinction is the "core content" part.

Part of the page is that core content, and what we actually want to see/read indeed (and why various web browser brought a "reader mode" that extracts that part). But the page itself is not a document that only contains this core content, and hasn't been for a couple of decades.

Most blogs/publication, even personal ones have been backed by a publishing system with the core content in a database. At this point these are applications, and many of these system can still output "pages" that 100% fit the description of being easily readable without JS, while still adding a lot of utility. Hence my disagreement around the "Documents ≠ Programs".

That's the distinction I was trying to make.

cwizou··on Documents ≠ Programs
I thought this was gonna be about how we've been tying data or documents in native mobile apps to their program of origin, making the use of several apps on a same document very complicated (or importing data from other apps). iOS went all hog on this and it's been a pain for "pro" workflows.

Instead it's complaining about bad JavaScript usage and needless "apps" for "personal websites", which by the definition of the author are just documents to be shared. I love badmouthing SPAs and needless JS like anyone else, but what if I want a bit of interactivity such as comments? Am I not allowed to do that on my personal website by the author's definition?

I think most importantly it misses the point that the web never was about defining documents. HTML was about linking pages and resources together. And even early on, CGI was a thing, and interactivity quickly became the "killer app", starting with search.

I mean, Mosaic added forms in 1993. [1]

[1] : https://www.w3.org/People/Raggett/book4/ch02.html

cwizou··on M1 Icestorm cores can still perform well
No, the TDP is configurable up to 25W (and down to 10). Again check the link of the score distribution I put in another comment, it's all over the place, and the passmark results are not a reliable comparison at that point.
cwizou··on M1 Icestorm cores can still perform well
Check the sibling comments, your CPU (like most mobile Intel/AMd) can be configured for different TDPs by the OEMs, so you can't compare anything, especially not PassMark results.
cwizou··on M1 Icestorm cores can still perform well
Yep, you can see the distribution of results (there aren't a lot) here and it's all over the place for the Ryzen vs M1 :

Ryzen : https://www.cpubenchmark.net/cpu.php?cpu=AMD+Ryzen+7+PRO+585...

M1 : https://www.cpubenchmark.net/cpu.php?cpu=Apple+M1+8+Core+320...

TDPs don't mean anything even within a vendor lineup, and doing cross comparison is futile (without properly measuring yourself the power consumption at load, then you can start doing real efficiency comparison or apple to apple).

← PreviousPage 3 of 6Next →