2,896 karma · joined April 8, 2021
Could you qualify what you mean regarding double precision, though? nvidia consumer GPUs have pretty terrible double precision (usually in the range of 1/64th single precision). And FWIW, the normal cores in the M1 (Max|Pro) have fantastic double precision performance, and comprise the bulk of the SPECfp dominance.
The benchmark you linked is grossly overclocked, as are many others people have linked throughout this thread. Which is fundamentally the problem with the Geekbench browser and people searching around for something to prove a point, when the numbers on there are completely unverified (and sometimes wholly fictitious), unproven, and often under ridiculous scenarios.
Use actual reviews from credible reviewers and extract the GB results from that. e.g. https://www.anandtech.com/show/16214/amd-zen-3-ryzen-deep-di...
And FWIW, I don't even put any credibility on these M1 results. I'll just wait until Anandtech or someone similar has a real review with competently created, reproducible benchmarks.
Technical and business limitations drove the design of MapQuest. That client computers had little computing power, bandwidth (on both sides of the pipe) was limited and expensive, and processing power on the server for a very low value user (per interaction) had to be meted. No one who used MapQuest didn't wish they could just smoothly scroll the map, but we were fine given many of us were on a terrible connection, ran on systems with limited graphics power (where even smoothly blitting a high resolution raster graphic was taxing), etc.
Every improvement (more computing power, memory, storage, bandwidth, or even business model, etc) in the industry invariably leads to many people all independently seeing the same obvious next step, many groups building the same eventual thing, and then the losers (from a market perspective) claiming that their ideas were stolen. It is the story of this industry.
A high end graphics card from nvidia these days has 1000GB/s all to itself, not in competition with the CPUs. If these GPUs are really as high of performance as claimed, there may be situations where one subsystem or the other is starved.
It's an Apple product launch. Like every product launch ever in the history of ever, they point out the features of the thing they launch.
There is no mistake. Your first paragraph is unfortunately founded on some misunderstandings of optics, however I calculated the hyperfocal length for an equivalent ASP-C 35mm system and an iPhone at the same crop (which anyone with an SLR and an iPhone can replicate in moments). The iPhone has a dramatically higher DoF. There are no mistakes in that calculation. This is the reason why you need computational bokeh. It's why it's so easy for everything to always be in focus. Could someone contrive ridiculous focal length / f-ratio / CoC parameters? Of course they can -- it's just a function with parameters that you punch in, and they can offset. In actual reality, however, short focal length is the primary input into why small cameras feature larger depths of field. Why we talk about the equivalent aperture in the way that we talk about equivalent focal length.
sudosysgen's argument in the end seems to distill down to "yes, but compare it via the equivalent DoF f-stop on the larger camera" which is a short circuit of the entire argument. It is basically saying that AMC is worth the same as Apple if AMC shares were each worth $4636.
Okay.
I calculated the hyperfocal length for an iPhone and an equivalent zoom SLR, at the same aperture. These yield effectively identical degrees of focus from 1/2 the HF to infinity. The iPhone is from 3.2ft to infinity, the SLR is from 27 feet to infinity.
Nothing else matters if you can't tell me why that's wrong. Because it isn't wrong. It's absolutely right. The same zoom level and cropping. MASSIVELY larger focus zone.
If we doubled both dimensions of the sensor, thus doubling the CoC, it would halve the HF. If we instead doubled the focal length it QUADRUPLES the HF. The focal length is a squared factor and outweighs any other component. For a reason.
Further my 70mm lens has a smaller aperture than my 35mm f1.4 lens. Yet it has a much smaller depth of field for a given distance. Weird! Lens makers must not know your remarkable "slit lens" trick.
At this point I'm convinced you are either trolling, or have dug so far into the depths of wrongness that you're dedicated to sticking with it. So good luck with that. I'm out of this conversation.
At this point I feel like you are just posting things hoping some future visitor will think that your commitment must demonstrate that you are right. I guess.
It yields a practically perfect comparison of focus. This isn't a trick or handicapping, and the degree of focus/defocus is identical whether that SLR had 10x the resolution. There is utterly nothing arbitrary chosen here, and the amount a tree 10 feet outside the focus is out of focus will be identical on a 12MP SLR or a 24, 48, or 96MP version with the same focal length / f / sensor size.
No, you can't, because you are painfully ignorant on this topic.
Literally, spend 30 minutes with an iPhone and an SLR and you'd be illuminated. Instead you seriously argue that I need to look at the "physics" (which is farcical when you ignore the most important part of a camera, which is the focusing from the lens to the sensor. Dismissing that betrays a complete misunderstanding of optics).
This conversation is clearly futile, but again - spend 30 minutes and actually test your theories. Or, you know, read any single article on the tubes.
Or how about simply ask yourself "why does the iPhone need to do computational bokeh"? 65mm equivalent lens, f/2.2...should be the easiest thing in the world. In SLR world that is bokeh gold.
You don't, which is why higher depth of field is the golden standard. See: The entirety of this discussion.
"so confidently to take such a superior tone"
To be clear, you dismissed my post by claiming that it's a "fixed" issue, then posting proof that doesn't show it to fixed. I don't believe I'm the one who attempted a superior tone.
Depth of field is *THE* issue in macro photography. Small focal length cameras are at an advantage in that regard. It's pretty simple.
The iPhone is widely assumed to have a CoC of 0.004mm (this actually increases on the most recent iPhone, though it's tough to get precise numbers). A Nikon D5000 (going with an equivalent resolution -- larger pixels -- on an ASP-C camera) has a CoC of 0.020.
So let's calculate hyperfocal distance of the two systems for the same effective focal length (but obviously very different real focal lengths)-
iPhone 12 telephoto lens - 65mm (7.5mm real) equivalent, f2.2.
Nikon D5000 equivalent lens - 65mm (43mm real) equivalent, f2.2.
For the iPhone, the HF is 6.4m. For the Nikon, it is 54.3m. For those who don't know, hyperfocus is the point where everything from 1/2 of that distance to infinity is in focus if you set the focus to that magical point. It's a proxy for the other depth of field calculations, and is the simplest to demonstrate.
Anyone who owns an iPhone w a "telephoto" and an ASP-C SLR w/ a 50mm lens needs to try to replicate bokeh at various distances without the computational bokeh. Focus on a subject at 1m, 2m, 4m, etc at the same aperture. Close down the aperture on the SLR even.
I don't think that is the demonstration you think it is. Most macro photographers would not rack that up as a successful photo.
And again, focus stacking is what everyone does to compensate for the DoF weakness.
Go to the wikipedia page on depth of field and see how it is calculated.
"Cheap macro lenses in 2021 are typically around 24mm"
You claimed they were the norm. Now it's that they simply exist.
"Focus stacking is needed"
Focus stacking is needed when the depth of field is so small that the resulting photo would be unpleasant. This is the case for almost all macro photographs shot on SLRs. It's interesting that someone else claimed this is a fixed issue and posted a photo that looks like it was taken with one of those terrible lens adapter kits. If that is one's standard for "fixed", then sure, but most of us have higher standards.
Humorously years back I had authored a giant depth of field essay with online calculators specifically because so many people just couldn't understand why their iPhone couldn't get bokeh. Yes, f/512 would be the impossible equivalent. This is easily calculated.
Regardless, the lens Apple uses for macro mode has a 1.54mm focal length. The 4mm example was just demonstrating how fundamentally small cameras win on depth of field, at least if you want maximal depth of field. Conversely they lose when you want to limit depth of field, which is why we have computational bokeh.
"Macro lenses are typically 24mm or so."
The smallest from most makers is 35mm, but the majority are 50mm+.
This conversation has turned weird. As someone who has had many SLRs, and many lenses, and has taken thousands of macro photos, I know that in the real world macro photography is a massive pain. That DoF is by far the number one obstacle (which is why focus stacking is simply necessary, often with ten or more varied focuses). Physics benefits small camera systems for that specific scenario.
This is basic math. It's interesting that someone else claimed it's "physics" as a retort, when yes indeed it IS physics. It's why you can make a tiny lens fixed focus camera that seems to have everything in focus, from near to far, because the DoF becomes enormous.
https://www.photopills.com/calculators/dof
(That site has different camera selections purely because the circle of confusion differs based upon the sensor size / resolution, so it's the relative values that you should pay attention to)
Take a gander at the DoF variations for a 4mm versus a 35mm FL system.
This is the reason why iPhones have to implement fake bokeh -- because the depth of field on tiny cameras is so much larger, even with a wide open aperture. But the inverse is that where you want a wide depth of field it is a feature of the size. It's also why people seem to be much more successful taking photos on smartphones, because the focus is much more forgiving.
A wide DoF is a function of those physics.
And FWIW, the curse of SLR macro lenses is minuscule depth of field, so much so that many take many photos (presuming a perfectly stationary subject), bracket the focus, and stack by sharpness. It's a hugely involved process. A macro lens with a tiny focal length instantly has a big advantage, though depth of field is still going to be a problem given the fixed aperture.
It most real world scenarios I think the average person will have a much better chance of successful results.
Blender is obviously a huge component of the software stack.
Israel was an extreme outlier. It is not a counterpoint whatsoever. Israel got priority first shipping before Canada, and even the European countries that hosted the factories.
Many of these comments are weird in that they seem to hold Canada's vaccination effort as some sort of failure. In reality it was one of the most rapid on the planet, dramatically faster than just about every other OECD country outside of the US (courtesy of nationalizing production) and the UK (courtesy of protectionism), and then testbed Israel.
I'm not sure what you don't buy -- the fact that Canada signed up early and for massive quantities is very well accepted fact. The reality that Pfizer and Moderna's European factories became massively oversubscribed (while US factories were basically off limits to everyone else) is also accepted fact.
[1] - Even if Canada was suitable as a testbed (e.g. it's too large and distributed), healthcare is run at the provincial level. There are 13 sovereign healthcare agencies, and this is constitutionally decreed, so it was unlikely to have been possible even if the country agreed to do it.
At the same time Canada had placed orders for literally hundreds of millions of vaccines from the major pharmas, doing so before almost anyone (your claim about shipment dates is just as factually deficient, as an aside).
Canada was ill prepared for a pandemic because it relied upon free trade and allowed for the hollowing out of medical manufacturing. Hopefully a lesson has been learned.
The US nationalized production in that government decree made it illegal to ship to anyone else unless US contracts had been satisfied in full (NOTE: I am not judging this at all. The merits of it are arguable, but it does point out a weakness in a reliance on free trade). The change of government didn't change this an iota.
The defense production act dictated that 100% of production was for the US until the government said otherwise. Moderna and Pfizer were both forced to supply Canada from the European factories as a result, even long after US vaccine manufacturing far exceeded consumption.
Canada was one of the first countries to pre-order vaccines, signing massive agreements with just about every candidate maker. I'm aware of literally a single deal happening with another country prior to Canada's orders.
What thwarted Canada's efforts was that the US defacto nationalized the medical establishment in that country, while Canada had basically become beholden to larger centers courtesy of an over reliance on free trade. With the US production nationalized, Canada, and truly the rest of the world, had to rely upon manufacturing in Europe. The EU at times lurched towards protectionism but never did, and Canada is in its debt.
Though in the wake of it Canada has committed to no less than four vaccine manufacturing plants, including mRNA vaccines. Lesson learned.
`"Bold” strategy? Canada had no option if they wanted to vaccinate quickly.'
Yes, it was a strategy. The manufacturers were dictating a timeline for shots that would have meant that the number of people fully vaccinated would be quicker, but the number of people partly vaccinated would be much slower. People much smarter than you or I did an analysis and decided on another approach, including mixing shots. And yes, of course it was because of a supply constraint (for reasons mentioned elsewhere, but having positively nothing to do with Canada "dragging it's heels"), otherwise such approaches wouldn't be necessary, no?
Companies are building the same sorts of solutions in Go today that they built in Ruby in 2009. Loads of web applications are being built in Go.
Like I said, around 2010 even payment (e.g. the fundamental opposite of where a dynamic type system makes any sense) systems were being built in Ruby because it was the hotness. Though no one has wed themselves to an ephemeral choice quite like Shopify has, and most other examples from that era now have a rich, diverse technology platform.
"Need to prototype a CRUD website super quickly? Ruby, Python, PHP, JS."
That's like the first 0.05% of a product. Like building your GUI in Adobe XD.
Because there are loads. I mean, almost every major technology company on the planet.
Indeed, there are lists of the "biggest" YC companies and it's remarkable how stratified the core technologies are with the launch year. Around 2008-2011 and it was Ruby all over the place. A number of recent companies list Go as the foundation of their solution. Recently Rust has started to appear in the lists.
Shopify is successful because of a business model at the right time and right place. That they use Ruby should be ancillary, but the company seemed to have hired the "big fish in a small pond" from the Ruby world so now they're cargo culting hardcore with Ruby there, making Ruby their core technology identity, to the point that they're spending enormous engineering effort to gain the benefits that everyone else is enjoying for free with alternative platforms. Which...they can afford it. But it doesn't make it right.
I posted a couple of weeks back that it is remarkable that almost all Ruby advocacy/development comes from Shopify, and here is yet another example. Similarly, recent claims about Ruby's growth among the "top N websites" was more than 100% accounted for purely by a quirk in the way Shopify implements their storefront (as a subdomain which allowed Shopify's choice of technology to trojan horse in the metrics onto hundreds of thousands of domains).
Clearly people make solutions with it, in the same way that people made remarkable solutions with PHP.