HN Simulatornew | past | comments | lists | submit | legitronics's commentslogin

When I actually went to the source for the 240hz claim, I didn’t reach the same conclusion as the testers?

Maybe they just let their rigol scope spit out 240hz, but 480hz worst case and I would actually say 960hz. I think what’s happening with the tandem oled is they both pwm at 480hz and 180 degrees out of phase giving effective 960hz when both are in use. Which appears to be most of the usable brightness.

Edit: I’m so confused.

“The PWM behavior is interesting, because we can see DC-dimming for all brightness levels. Usually, OLED panels use DC-Dimming only at higher brightness levels. The frequency is very low at 240 Hz, however, and this might be an issue for sensitive users, despite the DC-dimming. ” https://www.notebookcheck.net/Amazing-OLED-panel-Apple-MacBo...


You’re right - looking at the scope pictures, the horizontal period of the grid is 2ms, and the period of the wave is just over that.

A 240Hz wave would have a period of 4.17ms. 480Hz is 2.08ms. Much closer to the oscilloscope waveforms. And even then the waveforms are periodic, with smaller troughs at every second dip.

I wish I could see their test setup. It would be ironic if the panel was fully DC dimmed and the troughs were from another light source or something…

I’m not sure why the scope is saying 240 though. It makes me doubt my understanding :-/.

[Edit: Ah - got it! It’s measuring the frequency of the very lowest points - the ‘every second dip’]


Just adding a note to future readers: frequency is correctly measured dip to dip or top to top. Or every second time a particular value is reached.

> Or every second time a particular value is reached.

The value is waving, so a complete cycle is when it got back to where it started: that is, not just its value but its _direction of change_ is also the same.

(A value waving in 3D could coincide with its start value multiple times before it completes a whole cycle.)


The difficulty here is that it's 480Hz top-to-top, but there are two definitions of dip-to-dip depending on if you count every 'dip' from 100% (480Hz), or just the ones that reach the lowest voltage ⃰ (240Hz).

⃰ ⃰(brightness - presumably the test setup is an LDR or something?)


Here's my hypothesis: I'm guessing that the reviewer is using some kind of photodetector that is sensitive, to different extents, to different portions of the screen. And that the actual display modulates at 240Hz but that some portions of the screen are 180 degrees out of phase with other portions. Or that one of the subpixel colors is out of phase with the others. This could explain the weird alternating-depth signal at ~2ms intervals.

Here's notebookcheck's review methodology:

https://www.notebookcheck.net/How-does-Notebookcheck-test-la...

And the sensor they use might be this one:

https://www.thorlabs.com/item/PDA100A2

That doesn't look like it's even close to being able to focus on a single subpixel or even a single row without additional optics.

In any case, the 0% and 25% waveforms look pretty bad. In my experience, light with sharp, short pulses (like "driverless" LED lights produce) are unpleasant, but the unpleasantness is related to how bright the source is compared to everything else in the field of view. (The 0% and 25% waveforms are solidly at twice the frequency of driverless LEDs, which will make the effect considerably less bad.)


I like this hypothesis! It sounds very plausible.

Surely someone on HN has the equipment and curiosity to double-check all this as a Sunday afternoon project?…


Older samsung galaxy phones (till ~S21) had HW 960fps mode where it recorded 0.2s on sensor before switching back to normal fps. It is not macro, but maybe it could show the flicker rate of the whole screen or halves out of phase.

It would be entertaining to bring a serious high-speed camera with a macro lens to the Apple Store :)

  > [Edit: Ah - got it! It’s measuring the frequency of the very lowest points - the ‘every second dip’]
So that's the danger of [the authors] just reading the number off the screen without understanding what the scope is showing.

To measure frequency, typically the scope will measure the max and min of a waveform, then select a point halfway between them as the trigger point, then measure the time between successive triggers. Also, the trigger usually has an intentional hysteresis, so that noise near the midpoint does not cause multiple triggers on a slowly-rising waveform.

The little dips don't go down far enough below the midpoint to activate the hysteresis window, which is why it only counts every second dip.

Edit to add: Also, while it's nice that they show all the waveforms with the same vertical scale, they really need to adjust the vertical attenuation for the 0% and 25% brightness steps if they want to measure frequency by reading a number off the screen. Auto measurements won't work in these cases, when the waveform fills less than half a vertical division.

And the waveform trigger is also set incorrectly (0V, with DC coupling on the input...) because none of the waveforms have a rising edge on the center graticule line, where the trigger annunciator is.


In this case that's not necessarily incorrect, though. With the pulses alternating in amplitude by that much, there will be a component of the brightness variation which is at 240Hz and still visible (if the pulses were more closely matched, then it would be fainter). You don't need the trigger to measure this frequency in a single shot, either.

(The waveform looks more than a little odd, to me. Like they say, the DC value is changing as well, so the circuitry should be able to just output a flat value. The pulses don't really look like typical PWM, either. So I don't really know what exactly is going on there)


The authors are correct. The signal repeats every 4.2 ms, so the fundamental frequency is unambiguously 240 Hz. There is an even stronger second-harmonic component at 480 Hz, but this doesn't erase the 240 Hz flicker.

240 Hz looks correct to me. Of course it also has 480 Hz and other even higher components but the lowest significant component is 240 Hz. You can't disregard that half of the dips are much deeper. That will create a lower frequency flickering.

Their method is a photo detector driving the oscilloscope, right?

How can they assume the rise/fall slope in those plots is characterizing the driving electrical signal? Wouldn't a PWM driver be able to have much faster transitions and look more square than that at these time scales?

It makes me wonder if this is the response of the organic phosphors and the difference in apparent "depth" of the dip is really just a difference of duration, since the dips are very V-shaped, rather than bottoming out.

It is slightly odd that it has this paired rhythm, but I don't think you can assume it is "DC-dimming" rather than just PWM quanta that are being anti-alias filtered by the phosphors and measurement methodology.


Isn't it the other way around? They're claiming it's PWM, not DC dimming - DC dimming would be smooth, this isn't.

And tbh I don't think the PWM driver is particularly interesting, nor are they claiming to be measuring it. The display itself is flickering, which is what they're measuring and what the problem (for some) is.


The quoted text claims "dc-dimming" because some of the pulse "dips" do not reach the same detected photon level as the "off" screen state.

Hey, Rigol is dependable ;) - What do you mean in tandem? I thought oled doesn't use a backlight (I have little to no knowledge about displays). If it did, you would probably see a difference in amplitude depending on the diode you are using.

At least they probably measured with some form of photo diode instead of poking their osci into their mac trying to find the correct signal. This would measure the effective frequency for the eyes, no? So it would be 240Hz.


I have a grey market a100 that had hbm errors but the card used some of its spare rows to remap. I believe they introduced this with ampere, and I was a little annoyed that my “new” card had these but I got it so cheap I didn’t really care. It was 7k usd (vs 9k and 10k) and I remember being unsure if it was worth it… None of them needed to remap any more in they years I’ve had them but I think I got that one so cheap because of it. The most expensive one came from the best rated seller and appears to be an engineering sample but has passed every test I’ve thrown at it, the only thing “wrong” with it is that I can set the power limit to whatever I want? This was back when b100 was about to be formally announced and before the pricing so eBay sellers were trying to unload them.

Well I’m not even sure they did that. Not really.

They took signals security goals/aim/target and used models to generate a mathematical model. Which is a bit “hey trust us bro” in 3 ways but they did have human experts fully review 1 path, partially review 1 path, and the last I’m not really sure.

The fully reviewed path was that the aims goals were turned into positive mathematical statements “we will achieve X Y Z” so if we trust the humans that part should be as strong as we achieved X Y Z.

Tons of lean code generated to prove X Y Z correct, a lot of it is entirely self consistent. But then because it’s too much to look at we had the AI judge all of it for problem areas and we had humans look at that 1% of it… okay? That’s great if your magic box can pick out the problem areas but that wasn’t proved.

My last issue is, and please tell me if I’m wrong or missed something, but what says or proves the security goals/models are necessary and sufficient? We aim to do X Y Z. Great. We aim to not do A B C. not sure if I’m quibbling or this is strictly needed but it seems missing (unless it can be captured in entirely in the positive statement) let’s assume it’s included. We did X Y and Z, we did not do A B or C. Okay what’s my prize? Unbreakable crypto?

“Um well no… we know we can have X Y and Z while not having A B and C”

And so if we have X Y Z none of A B C we know the model is good.

“Yeah, probably”

And that the code at some point in time is no better than that?

“Yeah, certainly”


We may be able to say that X Y and Z are necessary since I can show you a concrete attack if you omit one. But we can never be absolutely certain that they're sufficient. Still, we've had something like 30-40 years of research by now looking at how to specify security properties. We certainly didn't get many things right the first time around. For the "basic" stuff, like confidentiality and authenticity we likely have a reasonably good grip on it under some, let's say "reasonable" assumptions. But a lot of research also still goes into developing notions of security with weaker assumptions or more exotic goals.

Very cool will check this out more

The color configurable bulbs or the high CRI bulbs typically use a 400nm led source and some do use a 380nm but those are much rarer and expensive. Also almost any amount of plastic or glass will significantly absorb that wavelength of UV.

I have a 405nm laser pointer and it makes certain things (any% titanium dioxide) absurdly bright from fluorescence. So you probably aren’t getting UV from those bulbs, but certainly some IR.


"The color configurable bulbs or the high CRI bulbs typically use a 400nm led source and some do use a 380nm but those are much rarer and expensive."

Cree, Nichia, et.al. use ~460nm LED with a cerium-doped aluminum-yttrium garnet cover to convert to white light and have done so for at least half a decade. Just a bare look at the spectral output chart from any company will tell you the wavelength of the base emitter, with a MASSIVE output peak centered around 460nm.

Nobody has used UV as a white LED base for well over a decade or more.


I really don’t think UV is playing much of a role here. You could also sort their chart by power usage, or IR produced. The paper conclusion paragraph even states UV and IR as candidates but the article just says UV.


Ever used a blacklight outdoors? They are not bright, but will be covered in bugs.


Don’t those bug collectors (the big white sheets) biologists set up at night usually use UV lights?


And the insect killers in kitchens


I have no idea what you’re trying to say.


They are trying to say you cannot blindly swap for noctua if you don't know what specs you need for a fan.


Intel ai gpu division is so aggressively anti “small” consumer. I think it has something to do with low volume and a small sales/distribution network. The data center cards I have from amd and nvidia are grey market at best. I don’t think intel moved enough units to establish alternate channels.


Serious question, would they bother?

You can work with 10 players that will buy all of them with minimal effort.


What I don't understand is: when you only plan for your product to have 10 customers, why even make a marketing page like this for it? At that point it's essentially the hardware equivalent of consultingware, something you sell to your existing contacts through outbound calls and by shoving ads through the B2B equivalent of a post-consumer sales channel (i.e. proactive outreach by your org's internal champion for that customer, to said customer.)


Because they want the hype to make the stock go up.


In fairness, they spent 5 cents generating it with Claude.


How so? You can walk into Micro Center and buy a B70 right now. Intel is the only one to offer license-free SR-IOV too.


Thats not an AI GPU its a consumer product.


It's a workstation card targeted specifically at inference. It's also their only late gen "AI GPU" arch available, baring the release of the D270 this thread is about.


Guidelines | FAQ | Lists | API | Security | DMCA | Apply to YC | Contact

Search: