Search results

I am fortunate enough to have a hands-on job (mix of bench science and onsite engineering). We were allowed back into lab mid-June and my mental health/happiness shot through the roof, even though I had been pretty successful about doing meaningful work during the mandatory WFH period. WFH has...

The joke/adage I've known is it's always 25 years in the future. But, yeah, I really *really* want to be wrong, but we've heard this before. At least ITER is building now!

Fortunately this one is a MSI Super Ventus. :) I am definitely loving the lack of noise and heat! Not quite the deal /dev/null/ found but I'm very happy about my purchase nonetheless.

You can see my thread a few down, which is very similar in scope. Tl;dr: just picked up a returned-to-Amazon 1650 Super for <$150, so keep an eye on things. You have to make your own value assessment on new vs used. I needed a card asap, so didn't have time to hunt used, but I'm sure I could...

It's all good; thanks guys. Did manage to score a returned 1650 super for $<150 and Amazon already told me it's in the mail. I didn't necessarily have the luxury of waiting too long to get the best bargain since the machine is down until I get a replacement (I have a habit of running video cards...

Thanks all -- you guys did seem to reinforce that that 1650 super is probably the best bet new in the price range I'm looking and the RX580 is the go-to used card. Sadly, a used GTX 1080 isn't in the budget, nor the 1660 super (which does seem like another good performer!). Appreciate the help...

Had been getting some BSOD's lately and this morning the graphics card was throwning funky lines all over the place. Bought the sucker used in 2016, so I'm not too heartbroken. I have been living under a rock in terms of computer hardware so could use some help. Truth be told, I was pretty...

I'm an engineering postdoc doing research at an academic hospital. I do not have access to many of my usual and relevant journals, but if I need anything clinically-related, we have a contract to supply it. The institution pays a huge amount for access to those journals and it makes sense for...

There's a lot of private grant organizations that are now mandating publications funded through them have open access--HHMI being a big one in this realm. We're in the midst of a sea change in this regard. Similarly, you have things like this happening...

5 nm process is not a 5 nm fin pitch: https://en.wikichip.org/wiki/5_nm_lithography_process

You make sure you don't need long metal lines for these ultra small feature levels, so you use heavier, higher resistance metals, eg. Ta, W, Cr that do not suffer from electromigration anywhere near as bad, then switch over to Cu BEOL. At least that's what I remember, been out of processing for...

You're correct I flipped yield and defect density, my apologies. Although I haven't seen an official statement as to what TSMC is calling a defect; https://en.wikichip.org/wiki/defect_density leaves some ambiguity. A core disabling defect is different to a defect that renders an entire chip...

Not likely that elegant of a "solution" to suggest it's all artificial fragmentation -- how they are reporting defects makes a difference. And with these dimensions, a yielded core will have variable performance, so there will be qualifications on that as well (e.g 2 dumpster fire, but working...

I was more going off of Mathworks lazy programming of blindly using the Intel libraries and not even doing a proper check for AMD processors to provide an alternate math library. I believe this part of Matlab code isn't exactly new, so it was done ages ago and never re-examined. Like it or not...

This sounds like lazy coding more than anything nefarious, but, hey, I'm a big fan of Hanlon's Razor. Hopefully Mathworks will fix their code base.

Thanks for this detailed post -- how much of this scheduling should be happening at the CPU/driver level? (or, more appropriately, given recent scale-up in physical cores, is this something that can/should leverage that capability?) My hair brain wonders if this kind of scheduling is something...

I mean this as a complement to your post: If only to allude to my post earlier in this thread -- it's worth looking at the specific demands of the software you need when in the prosumer/professional space. Some applications are memory starved and you'd be better with a 2000 series with its quad...

Thanks for asking, maybe someone will be able to answer this, as I don't know! I have a 1600 at home that's treating me well and I haven't done this form of heavy lifting on it. For FEA, it'd be a work computer and if I cannot configure it through a major vendor, no soup for me (womp womp). We...

Yeahhhhhh, I'm one of those three: our FEA work benefits from crazy fast cores with monster memory bandwidth, but doesn't scale cores all that hot, so 8-16 is more than enough. So if I could buy the absolute fastest cores at a lower total count and get the benefits of the platform, I'd be a...

Ha! Okay, you're on the inside -- sometimes it's hard to tell ignorance from humor. :) We looked at getting some legacy 150mm stuff for our very basic research fab (and I argued our dollar would go further on 100 mm...), but TCO was too high for unmaintained machines where we don't have a shop...

Dunno about the latter half of your message but EVERYTHING that can be used in a fab line is being used in a fab line, I was told that our lead times for 150mm wafer sputter tool is long because they cannot make enough 300 and 200mm equipment. Could be running a lower precision node for low cost...

I don't understand the controversy? Games = human i/o = there's going to be a critical path somewhere = that path is going to be single threaded. Ergo, we absolutely want fast cores to deal with that problem. I have not seen any trend that suggests that 8 really fast cores *for games* (no but...

Scaling for a huge number of computational problems is, especially for a single user, is generally not kind. So for a single user, big, flexible, superscalar general purprose cores are definitely moving into marginal gains territory. That's even with demanding software like CAD/FEA, which have a...

Funny I did read the article -- they own magenta as much as the legal system lets them. And will continue to do so (as they have). Given they haven't gone after Pantone's magenta colors and a bevy of other players in the color world says something about whether TMobile owns "magenta".

Clickbait titles are clickbaity. Yawn. How is this tech news?

Mobile or OEM contracts? Retail isn't the only way these chips/cards are getting used, nor do we know if they'll effectively kill off the other two 1660 SKU's even if they remain ostensibly around. And we have the 1650 super coming down the line which looks to truly obviate the vanilla 1660 (and...

I guess process improvements/maturity and DDR6 prices allowed them a population of the TU116 chips that were worthwhile to bin accordingly? As long as the cost of greater fractionalization in SKU's doesn't hurt, it works out well, especially if it allows Nvidia provide OEM desktop assemblers...

You've already made the case that this isn't a rational argument for you, as like the grand majority of the US (and nuclear-capable countries), -- blind eye to the ugliness (often hidden to developing world/Middle East) to oil and its *enormous* externalities and hyping the problems with nuclear...

Yeah yeah yeah, I know it's a complete shocker, but a wildly boring alternative to a new carrier being built using superhypersectret-shamwow fusion reactor...

Evidently no one knows that the governments of the world can issue secret patents? This one might be an old design that is now too close to public domain ideas and it was no longer effectively serving its purpose...

Ummmmm, this whole thread took a totally wide turn to a very specific subset of demands that are "needed" by a handful of folks where wider SMT may benefit. Not exactly translatable to most common *consumer* use cases. Server workflows, on the other hand, I can see the utility!

^Agreed, and sorry if I made it sound like the yields were terrible (the huge huge benefit of getting the die size down) when I was trying for a play on words that the fab capacity is the likely limiting factor. Some combination of top-frequency binning or power-to-hit-X_frequency might be the...

And quite possibly TSMC isn't spitting out enough dies period, much less viable ones. But, yes, consumer space isn't going to pull in the same coin a server part will. Zero surprises here.

Best part of the talk is the optimism and the fact that he tries hard to prime the imagination of these engineers without BS'ing them. I know myself that I get bogged down in the "that's too hard" mentality and try for marginal gains that appear more tractable, and Jim's mentality is just the...

There's a much more plausible explanation that their respective technologies overlap closely and there's an interlocking web of patent claims that goes back and forth, especially with broad claims. For all we know (and there's a LOT of precedent) that TSMC will make respective infringement...

The first Falcon heavy simu-land of the 2 outer boosters was actually a mistake, albeit one that was incredibly cool. They're *supposed* to land staggered, for some reason or another. (Safety?) Edit -- RTFP: me, Meeho wrote the exact same 2 above, whoopsies. :)

The irony is that the Merlin engines have evolved/upgraded since the contract for this satellite, such that the "standard" Falcon 9 has the lift capabilities to put this payload into proper orbit. Still, it's space stuff so I'm excited, and will give SpaceX an "easier" flight path for their...

Sure, good luck getting them to pay!

Exactly. Vapor phase heat transfer is extremely good and we need as much help getting heat out of the small chip area as possible. The heat pipes in a 212 probably aren't optimized for running 44C to 10C, but 65C to 55C or so, so there's (relatively) low hanging fruit there. Even with...

Pity he didn't feed the cold water from the bottom and exhaust out the top(which would be the sides when installed in a tower), probably can get relatively decent pumping purely off buoyancy forces. That is, above 4 °C, water's density decreases with temperature, so it will naturally flow...