Dash Lambda

You guys want an actual solution to GPU sag? Like, a real, systemic, non-patchwork solution? It's remarkably simple: Horizontal motherboard trays. I have no idea why they aren't more common. It just makes so much more sense.

Honestly, that bugged me back when they started it, but seeing how marvelously absurd the Gx100 chips have gotten I totally understand it now.

Ever since I got a second 980ti, I've been periodically screwing around with SLI for curiosity's sake. In games that support it, it works great. In games that don't support it, I see one of two things: Some games only load a single GPU, but most put half the load on both. The former is what I expected, but the latter is by far the most common, and also seems really weird. My guess is that it's actually doing AFR, but the developer's job is to tell it how to plan future frames, and since they didn't do that it's still just rendering one frame at a time. But, then, why do some games just ignore the second GPU entirely? Is it perhaps easier to make custom SLI profiles for games that already have AFR running?

I can personally vouch for the build quality of Ducky keyboards. My personal recommendation is the Ducky Shine, any iteration. All Ducky keyboards are well built, but the Shines are the ones that really feel well built. They've got a thick, solid plate of metal that the board and case are built around, and everything is really well put together -they even come with double-shot transparent-legend keycaps from 5 onward. Other than that they're really no-nonsense keyboards. The lighting is handled onboard, so you don't need any software running at all times to keep it from going into rave mode -just set and forget. Even has a detachable USB cable, mine is micro-USB but apparently the 7 switched over to type-C. The 7 also has a zinc alloy top panel. That... Would make me really want a 7 if my 6 wasn't capable of outliving me. I did originally have a 4, but it had a disagreement with a glass of orange soda that I didn't notice for several hours, and it's now really unsure whether or not it wants to work at any given time. So, basically, you can murder someone with a Ducky, but be careful where the blood goes.

There's a full-fat, honest-to-god text adventure game written in BrainFuck called Lost Kingdom. I used it as a benchmarking tool for my BrainFuck interpreters. If you have a good optimizing interpreter (like mine!) it should run fine, but I've never gotten a naive interpreter to run it fast enough to even print the first character before I give up and compiling interpreters can take a very, very long time to compile depending on the intermediate language.

Honestly, the Steam Controller has made every other gamepad I've touched feel like banging rocks together in a cave. I absolutely cannot use an XBox, Playstation, or any other twin-stick controller now without constantly thinking about what an awful input method analog sticks are for most of the games I play nowadays. Of course, Steam discontinued it because so few people even gave it a chance... So I guess Xbox One or Elite controller.

A friend of mine asked my advice on what board to get a few months ago, before anyone I know even heard of COVID-19. Prices were about the same. I just checked Wikipedia and I do see the $999 launch price, but this is the first time I've ever seen that number. I said I stopped paying close attention to the market for a while, but I can't imagine I was paying that little attention... Is it one of those MSRPs that's purely theoretical because nobody follows it? I'm genuinely confused. Of course, there will always be cheaper and more expensive boards. My hybrid 980ti was $710 vs ~$600-680 normal cards, I remember the 1080ti being $700 with good boards going for around $750, hybrids at $800, and the Kingpin was $1000 (not counting the FE nonsense). Then I quite distinctly remember balking at a $1200 price tag for the 2080ti. Then again, it did launch a difficult relationship, a huge internship, and a first year of college ago, so maybe I'm just remembering wrong?

Nvidia's price is $1199, most are around $1300, hybrids and higher-end boards are $1400-1500. Where are you seeing $999?

Net improvement. The 10 series was indeed great, but net improvement is then vs now, not in-between. I'm aware that in other product tiers it was improved. I'm not talking about those. Equivalent to 980ti SLI is not double the performance, getting 100% higher numbers is double the performance. The 780ti was actually a Titan with less VRAM and no native double-precision compute. So that's 3 generations of $650-700 cards within sensual whispering distance of their respective Titans over 7 years, then they say to hell with it and double the price. The model structure wasn't really comparable nor consistent before that. I'm not denying that there have been improvements in some brackets, but that's not what I'm looking at. I prefer to buy the top of the stack regular GPU, so I buy a 980ti. I stop following the GPU market so closely until I run into a game where my 980ti just isn't enough (Half-Life: Alyx), but now this generation's equivalent of the 980ti is more expensive than the Titan was last time I bought a GPU. So, basically, I'm now given the option to spend twice as much for double the performance or drop down a tier or two. I don't expect pricing of the top non-halo card to literally double from one generation to the next.

Then... Why not get a 2080 Super?

A 2080ti is about twice as fast as a 980ti, and literally twice as expensive. There has been absolutely no net improvement in price to performance in the past 5 years on the high end. The 2080ti is more expensive than the preceding Titan that itself had a controversial price jump. That's truly absurd. They try to justify it by saying it's not just an increase in raw horsepower, but that doesn't mean much when you don't have the option to not get the extra features. I really, really hope the next generation resumes reasonable pricing.

Actually, I did not intend aesthetics to be a part of that argument. It is purely a convenience argument. I do prefer the aesthetics of an AIO, and that is part of the reason I prefer them, but my point there was only that they are so much more convenient compared to air coolers that it justifies the price if you frequently open up your PC.

That's a little misleading. It's not that AIOs can't match the performance of air coolers, it's that they're usually much more expensive for the same performance. On top of that, the performance floor for air coolers is way below that of AIOs; most AIOs will cool just fine, but air coolers range from super high performance to indistinguishable from no cooling at all. I used to use air coolers, but then I switched to an AIO and have no intention of going back -I would happily pay quadruple the price of a comparable air cooler just so I don't have to deal with a giant tower. A friend of mine built a PC a couple years ago and chose an NH-D15, and despite its great cooling performance, he recently said he needs to get an AIO because the D15 is a pain in the ass. Really, no form of cooler has an overall performance advantage over the other. Air cooling, AIOs, and custom loops can all have poor or amazing performance, the difference is cost, reliability, and ease of use.

Sort of. It's Schrodinger's Cat. Before it's measured, the qubit is literally in both states -this is called a superposition. When you measure it, something called wavefunction collapse occurs, which means it more or less 'decides' what state to be in, and after that, the state you measured it in is set in stone for that measurement. The probability is influenced by its interaction with other particles, through entanglement and interference. A quantum algorithm is a way to set up interactions between the qubits available to the computer in such a way that, of all the possible answers that you could measure at the end of the algorithm, the wrong ones are rendered virtually impossible to get due to destructive interference. You know how, with the double slit experiment, the particles hit the detector in a pattern with bands and gaps? Designing a quantum algorithm is like setting up the system so that the correct answers are in the bands and the wrong answers are in the gaps. Here's a MinutePhysics video explaining Shor's Algorithm, still a little hard to follow but a better explanation than I think I'll give.

A qubit can exist in a superposition of states, which means that it has a certain probability of being in either state when measured. When it is measured, though, it will be in one state or the other. The problem with room-temperature supercomputing is suspending and controlling the particles. The higher the temperature, the more noise you get and the harder it is to carefully isolate and manage individual electrons (or sometimes photons). Superconductors are also commonly used in a couple critical components (I'm not entirely certain for what role), which thus far require cryogenic temperatures.

Oh damn, I didn't even notice. I see a topic on the first page and I just sort of assume it's relatively recent.

Just to put my two cents in: There are two types of math you'll use as an engineer: There's the tool math, which consists of plug'n'chug software tools and/or formulas, which would be silly to figure out every time and so are prepared for your usage such that you don't need to think about it. This is the equivalent of pulling out a calculator because you have more important things to think about than each step of 638*799.32. With minimal training, anyone can be taught to use these tools proficiently. Then there's problem-solving math. This is a way of thinking, and is what you get paid for as an engineer. It isn't the ability to re-derive one of your tools for shits and giggles, it's the ability to think about a solution to your problem in terms of the actual mechanics at play and understand what different factors and features in the design do. For an example: Let's say you're a mechanical engineer working on an engine. - Anyone can answer the question "is this engine balanced?" with the correct tools. It's simply a matter of observation. - A mathematician should be able to answer "why isn't this engine balanced?", because such an answer requires delving into the actual geometry of the system and understanding how the parts interact. - An engineer should answer the question "how do I balance this engine?", using a great enough understanding of how the system works at the most fundamental levels to not only explain the problem, but solve it. In essence, an engineer (a good one, at least) uses mathematics as a way of thinking about the world rigorously. It is not merely a tool, it is the most fundamental component of how they solve problems. You can't do that with only a cursory understanding of the math involved. Oh god yes. Modern math goes into shit like analytic continuation, abstract algebra, tensor calculus, and tons of other way more specific and advanced topics. To my knowledge, every mathematician who's contributed significantly to the characterization and measure of infinities has ended up going insane. Fluid dynamics has a similar problem.

This makes more sense, though I do wonder what benefit hybrid electrical/optical chips will have. Signal propagation is one of the most important factors at play in chip design nowadays, but adding extra steps to sending a signal is usually more detrimental to latency than whatever your transmission medium is. Maybe this'll eliminate a couple of those steps? I guess we'll see, some day.

Granted. People your age like you so much that they never leave you alone, no matter the hour of day or location, for the rest of your life. I wish my PC could run Alyx at max settings and 144Hz.

That article doesn't say exactly what the alloy does. It says it's "light-emitting," but then it describes it as allowing 'free flow of photons,' which is transmissive. So does it produce photons or just let them through? Either way, how is this a breakthrough? I thought the difficulty with optical computing had more to do with electron-photon-electron conversion efficiency and logic representation than miniaturization. Or is this purely for data transmission? If so, I guess it would be a breakthrough to have nano-scale light sources, but wouldn't converting to and from light add a ton of latency? We already use fiber-optics in large scale stuff where the conversion latency is negligible, but at the chip scale we're talking less than a mm. This is too small for RAM or PCIe, right? If it is just for data transmission, how does that reduce the number of transistors needed? You're not replacing transistors with optical gates, you're just replacing buses with optical wires -if anything, that will mean more transistors for signal conversion. I'm just so confused.

What? Define "viable."

It's a $175 cooler according to Newegg. To my knowledge, there aren't very many AIOs that cost more than that -most good ones are in the $100-$150 range. So they spent more to go with that cooler over an AIO. Now, if they had just went with a normal NH-D15 then sure it would cost less, but an extra $20-30 for a great AIO kind of pales in comparison to thirty two terabytes of storage. This video was filled with strange decisions.

Just to let you know, it's a good idea to quote people you're replying to. That's the only way to reliably give them notifications. As for relatively rare hardware fault: The most recent one I had was due to a second RAM kit I got. I originally had a 16GB kit, got another 16GBs, it's the same series but different supplier for the chips (no way to tell before putting it in your system), and Windows would occasionally tell me that half of it was suddenly 'reserved'. I swapped the modules around probably 4-5 times before it stopped doing that, it would have been hell trying to work around a D15, but with an AIO I have immediate direct access. There was also when my 780 burnt up and I had to take the computer to bits to make sure nothing else was fried. I have a friend who does use a D15, and my best guess is that his computer is possessed. Every time I work on it I have to deal with that big friggin' tower. I have to use my longest spudger just to unlock the GPU.

I'm in my computer much more often than that, even if just to clean it. And actually, of every component in my machine, the AIOs are pretty much second only to the case in terms of reliability. Every single component has had multiple issues at some point, whether I caused it or not. But, of my two AIOs (CPU and primary GPU), I've only had one problem I can recall. And I think it was just because I didn't plug it back in after working on something else. The GPU's AIO is 4 years old, the CPU's is 5.