HN user
Isn't the real reason to use char8_t over char that it that char8_t* are subject to the same strict aliasing rules as all other non-char primitive types? (i.e., the compiler doesn't have to worry that a char8_t* could point to any random piece of memory like it would for char*?).
Elpida was purchased by Micron after the financial crisis (they bought it out of bankdruptcy for a swan song in 2013). Much of the Micron DRAM you might buy is made at the former Elipida fab in Hiroshima.
They're also wrong in suggesting that J-1 visas are intended to just be used for "cultural exchange". They are quite explicitly intended for use by trainees[1]. I assume these folks are intended to be trained in New Mexico so they can ramp the advanced packaging fab in Malaysia when they need to.
Thank you for the link to Robin Sloan’s blog and temporarily answering my question about whether I should upgrade iOS. I love his books and for some reason never thought to look to see if he wrote anything else online.
This just doesn't make technical sense. I completely agree that backdooring encryption standards is a bad thing. But Dual EC DRBG is a clear example of a NOBUS backdoor actually being that. The backdoor is equivalent to "knowing" a private key. The weakness is not some sort of computational reduction. Using this logic, you would say that no encryption method is possibly secure because you can't rely on its security once the key is exposed.
If the refund is paid in less than 45 days after the tax filing deadline or the day you file your taxes, they will pay you interest. See here: https://www.irs.gov/payments/interest#pay
Plenty of actual research costs count as overhead to avoid the need to hire an army of accountants to allocate every single bit of spend.
For example, the electricity costs of the lab in which the research is run would typically be paid for by the university and would be considered overhead. It's not "administrative bloat". Most of the particularly gross administrative bloat is on the undergraduate side of things where higher tuition costs have paid for more "activities".
I think the "X" looks like a pair of scissors. No good explanation for ZUndo though.
I like how you edited this to add "at work" after folks provided examples of it happening outside of work. If you'd like a slightly more work adjacent example, see the rampant increase in IEPs for students in the bay area. (I'm sure that the increased time for tests provided in many of these cases is not being abused at all...)
How did you choose the process noise covariance in your `grad` function? It doesn't seem like a single process noise covariance structure should be globally applicable across all possible functions.
This headline kind of misses the point. The NTSB is very explicitly structured to not be a regulator to enhance cooperation between all parties, which is part of the reason that confidentiality is so important.
The tiny yellow blobs are the bonded lasers, so the locations of the bonded lasers in the subsequent figures are incorrect but still get the point across. This page from IMEC has a video actually showing some of the process. (The image at the top is from a press release from them about successfully doing flip chop bonding of lasers.) https://www.imec-int.com/en/silicon-photonic-ICs-prototyping
To be clear, you want to decrease the permittivity that adds capacitance where the accumulation of charge serves no purpose (i.e., in the interconnects) and increase permittivity where we want to have a lot of charge (i.e., on the actual transistor gates so we can produce a large enough electric field to switch the transistor), which is why gate dielectrics use high-k materials like HfO2 and interconnect layers use low-k dielectrics like carbon-doped oxides.)
A red LED (or really any LED due to the physics of the type of energy transitions used to produce light with an LED) has a much shorter wavelength (significantly higher energy photons) than any type of EM wave sued for cellular connectivity (including the mm-wave stuff that isn't being deployed very widely).
The more important reason this is not true is that there is technically feasible DIMM form factor for LPDDR4/4x/5/5x RAM. And there are actual power savings (~20% reduction in J/bit transferred) to be had for using LPDDR RAM as opposed to similar generation DDR RAM. Laptops that support replaceable memory have to use DDR4/DDR5 and therefore cannot take advantage of the power savings you get from LPDDR4/4x memory.
The document layer seems to be unmaintained since the end of 2019 and has quite a few bugs. The 1.8.x releases fixed a bunch of bugs but also dropped transaction support. It honestly seems like a great idea, but I'm guessing no one at Apple could justify the time spent working on it.
That’s just not true? They have significant amounts of production in Arizona, some in Oregon (along with TD) and then the rest in Ireland and Israel (which probably aren’t much lower cost than Arizona)
This doesn't particularly make sense. You can't even use non Intel chipsets with Intel CPUs and integrated GPUs are inherently tied to the CPU. (In the case of the discrete GPU segment, Intel didn't even have an offering.) The only place where alternatives are even technically possible are in WiFi where there is no impact on AMD.
It could only solve the two language problem if the "users" were writing their program in Julia themselves. Otherwise your ideal solution is still using two languages. And if they are writing their program in Julia, there's no reason to compile your code into a shared library; you'd just share a Julia package with them.
A typical new car practically lasts way longer than a car built 30-50 years ago, early 2-3x the value there. Account for significant reduction in chance of death, MPG improvements, and these numbers make complete sense.
This is most definitely not true. Trumpf makes CO2 lasers which are used by Cymer to produce EUV radiation via excitation of liquid tin droplets.
The implication of working coronavirus vaccine by end of October was that it would be available for use (you can find quotes from the Trump administration saying they were working on distribution of hundreds of millions of doses by the end of the month), not that there would be a preliminary report from a still ongoing trial indicating that the vaccine works.
The Steele dossier includes information from Russian sources. It was not "paid for" by Russia.
No one has argued that Hunter Biden was not being paid because he was a Biden. The question was whether it led to improper behavior on the part of Joe Biden. That email proves nothing. It's entirely plausible that Hunter Biden got this guy an invite to some social function where shook hands with the Vice President. This is also entirely ignoring the fact that the firing of the prosecutor wasn't some sort of one off idea that Biden came up with on his own. There are independent speeches by all sorts of people from 6 years ago that complain about the corruption of the prosecutor in question, which makes this whole point moot. The prosecutor was fired for not investigating the company that hired Hunter Biden.
Like most liars, Trump isn't inventing everything out of thin air, but he misrepresents so often and so much that you can't draw any useful conclusions from anything he says.
AMD and Intel don't actually own it all, certainly not from a software perspective. That's the difference here. The upshot of this article is that Nvidia basically wants to "CUDAify" the entire datacenter software stack. Intel and AMD absolutely do not have that kind of lock-in. You don't need to use their proprietary language to write programs that run on their systems.
That's probably 38 degrees Fahrenheit (~3 degrees Celsius).
That is not an Icelake chip. IceLake and Comet Lake are both "10th Generation" Core. Icelake was limited to more premium offerings and was only 4 cores and lower.
Mainly because what you're saying isn't true for many workloads? (Also this already works for most existing integrated GPUs.)
The types of workloads run on GPUs typically like very high memory bandwidth and are usually willing to live with higher memory latency to get it. Onboard GPU memory is usually built with this in mind (trade off capacity and latency for increases in bandwidth). This is generally speaking the opposite of what you want in a CPU where people often want very high memory capacity and lower latencies, but may not limited by memory bandwidth, so simply sticking a GPU on die and giving it access to a memory subsystem that was not designed to feed a GPU is not going to make anything better.
No...the whole point is that the single instruction induced halt for downclocking isn't a real issue. Even in the pathological case where you insert a single instruction spaced 760 us apart in order to induce the maximum number of clock shifts, the total performance degradation due to the clock halts is only 3% (the frequency drop that is induced by the use of these instructions has a much larger impact.) Furthermore, on Icelake-SP, the halted time due to frequency transitions is supposed to go to 0, which makes this aspect of the problem entirely irrelevant.
Yes, if you insert a single 512-bit FMA that runs every so often in your code you will get a 15% performance hit from the lower frequency, but that's much less likely than the old case where people who were trying to use AVX-512 for memcpy and the like would slow down scalar code.
This isn't the real problem though...the problem with the existing AVX-512 implementations is the "relaxation time" causes subsequent scalar code to be slow.
From later in the same post:
Here, we have the worst case scenario of transitions packed as closely as possible, but we lose only ~20 μs (for 2 transitions) out of 760 μs, less than a 3% impact. The impact of running at the lower frequency is much higher: 2.8 vs 3.2 GHz: a 12.5% impact in the case that the lowered frequency was not useful (i.e., because the wide SIMD payload represents a vanishingly small part of the total work).
Interestingly enough, this is another feature that is supposed to have been improved on server Icelake. The frequency transition halt time is now pretty much negligible. The "core frequency transition block time" goes from ~12 us on CLX (similar to the number quoted above) to ~0 us on ICX.
(Slide with frequency transition info: https://images.anandtech.com/doci/15984/202008171754441.jpg)
To reiterate, the problem with the existing license based downclocking is that a few AVX-512 operations can drop your frequency for subsequent scalar loads on the same core so you need to carefully analyze the overall application to make sure that you have enough AVX-512 work over which you can amortize the loss in performance in the rest of your code that is affected by the frequency drop.
If the only issue with AVX-512 is thermal downclocking because you end up using more power, it's almost definitely because you are getting more work done per time. A few AVX-512 instructions in a mostly scalar workload is not going to significantly increase power dissipation and therefore should not induce thermal downclocking, while a heavily utilized AVX-512 kernel will burn power, but should also be doing work twice as fast per instruction.
Hopefully this behavior change will help improve AVX-512 uptake and end the somewhat ridiculous conception people have that the instructions are entirely useless. Intel's HotChips presentation on Icelake-SP also indicated that the behavior will be significantly better on server chips, but the behavior is more instruction dependent with 3 license levels and only 512-bit instructions that utilize the FMA unit being subject to downclocking (by ~15-17% as opposed to ~27-29% on SKX derived chips).
(Intel Hotchips slide: https://images.anandtech.com/doci/15984/202008171757161.jpg)