"Snapdragon" is a core designed by Qualcomm (based on ARM IP), which is a fabless semiconductor company, and has no relation to Huawei. TSMC is a semiconductor foundry (fab) company that does not design any chips but manufacturers them for other companies. Kirin is currently being manufactured by TSMC for Huawei, but will not be in 2 months.
I've played around in Tea For God (https://sidequestvr.com/app/65/tea-for-god-under-development) which does this on the quest, and it works really well, and I didn't even use the full 6.5'x6.5' (in my case ~6'x4') possible. Within a minute I lost all sense of where I was in the real world.
You don't seem to understand the California Jungle/Blanket primary system... All candidates from any/all parties run against each other in the primary, and the top two are the contenders for the general election. Recently this has led to two Democrats always being facing each other in the general election.
Title is wrong; Kitty Hawk is not shutting down... they are winding down their "Flyer" project (for sub 250lb flight) and focusing on their "Heaviside" plane that is much more capable.
Me too; sadly we had silicon that ran great (We had better performance per watt for FP32 and FP64 on 28nm process vs the A100's 7nm GFLOPs/watt) but targeting a market that had 3 customers that didn't want to work with a startup, plus investors that were opposed to us going into the "risky" and "unproven" AI space. Still have 183 of the chips in my closet waiting to see the light of day :/
Not the case... AFAIK (based on talking to people who have played with Habana test systems), any real performant applications still need to be hand optimized with their assembly.
Since it says his favorite band of all time is Pink Floyd, it is fair he is using the name as Storm and Hipgnosis designed all their major album covers. Hopefully he paid for the name...
Except datacenter is not defined by NVIDIA in their EULA at all, and plenty of large and small datacenters continue to use "consumer cards" regardless of NVIDIA's fear mongering. I know that Tesla, OpenAI, Microsoft, Apple, and many others all continue to primarily buy primarily 2080Ti's, RTX Titans, and Titan V's since the EULA change.
> The difference is 8 more GB of RAM that comes at a steep premium
This is incorrect. The RTX 6000 has 24GB of VRAM and is $4000, and the RTX 8000 has 48GB of VRAM (double the amount) and is $5500. Is it worth the price increase? For a lot of people I know it is.
Also, the RTX Titan is $2500 and is identical to the RTX 6000 (at the chip level) and also with 24GB of VRAM, with the only difference being software enabling of additional H.264/5 encoding features on the Quadro. Definitely not worth the cost increase, especially for anyone doing ML.
Well Apollo 10 was the last time NASA let the astronauts name the CM and LM, precisely because they named it Snoopy. I think it is well known that in general the early astronauts had a pretty good sense of humor (along with balls of steel), but ground control, not so much.
Except this is exacerbated by a very dumb software oversight that is corrected for by virtually everyone in the Tesla software modding community, and in many cases Tesla has not covered the board replacement after the initial warranty expires (which happens just before the estimated failure time).
Virtually every government office I have been in the past 5 years dealing with any classified information has disallowed the use of flash drives being connected to government machines, and it is pretty strictly enforced with reminder posters all over.
Except Bitmain has announced and will be shipping to preorders the first Equihash ASIC (Antminer z9) in July... only 3000 z9's would be needed to equal 30 Megahash, which would be $6M at the z9's $2k price tag. Bitmain obviously has been mining themselves with the z9's for a while, so I would bet that they noticed this and jumped on it before shipping their "lightly used" units out.
I know a lot of people using LVT transistors in 28 and 16/14nm processes, including relatively low power (mobile and embedded) designs. I personally have used LVT variant SRAM blocks for both our 28nm and 16nm designs, and ULVT cells manually placed for critical path for Neo's FPU for our 28nm chip.
The fact that private lenders are incentivized to give out more and more leads to universities charging more and more. Even with all the money being spent, the same (or in some cases less) level of education is being provided. On top of that, how do you repossess someones education in the case of default? You can't, so the government can do one of three things:
1.Pay off the loan holders (by printing money, huge inflation)
2.Allowing indentured servitude
3.Debt forgiveness (now all the lenders are out of money)
Sadly, I think the 3rd option is the best one, but is by no means actually a good one.
In all of those cases there was legislation passed after the crises... in this case, the US government has already guaranteed all of the loans and has no way of getting out of the without the US government defaulting. That only leaves room for printing money, which hurts all holders of US dollars.
Most of the student loan debt is owed to private creditors and it is the federal government that insures these loans. If a student doesn't make payments, the lender gets a guaranteed bailout by the government (no need for congress to step in like they did for the banks and auto companies). Basically, all the lenders are incentivized to give out as much money as possible, which incentivizes universities to raise tuitions, all because the US government decided to give lenders free money to play with at the cost of those who take on the debt and the ones footing the bill, which are usually the same people.
I have been skeptical of the real "value" of cryptocurrencies (outside the technical ability to allow for value transfer, which does not necessitate the coins having value inthemselves), but assuming this is true I have a lot less faith in dollar bills and the fact that they are backed by the "full faith and credit" of the US government. If the student debt bubble really does pop, the government (and thus all of us tax payers) are going to be the ones holding the bag.
The Titan and Tesla NVIDIA cards are still built with 64 bit/double precision in mind with "only" a 50% performance reduction compared to FP32 in most of those devices. It is the GeForce gaming cards and the ones being targeted now for ML that are gimping FP64, where it is 1/4th or even 1/16th the speed of FP32.
The vast majority of "real" HPC still requires FP64 (only large industry categorized under HPC that I know is perfectly fine with FP32 is geospatial oil & gas stuff)... a valid LINPACK has a hard FP64 requirement.
There has been a lot of talk about lower precision and approximate computing over the past couple of years, which I hope will be recognized as a fad with John's (and to a lesser extent mine/REX Computing's) work... Why sacrifice precision/accuracy if you don't have to? Posit's are a great solution for this for a vast number of problems, and the Stanford lecture is great.
The thing about Posits in particular is, depending on the environment (e.g. es=3 for 32 bit Posits as referenced in the Beating Floating Point paper), you can actually have greater accuracy/precision and dynamic range than 64 bit IEEE floats while using fewer bits. The OP article here about CERN is doing something significantly worse, which is just using lower precision and taking the accuracy loss. While this may be acceptable in some applications, I would not want to losing any accuracy in the original data from expensive scientific experiments and using potentially inaccurate results for future calculations and assumptions.
Founder of REX Computing here, we taped out in July of 2016 and got our silicon brought up and working back in February. I gave a talk at Stanford showing our hardware and a tiny bit of software: https://www.youtube.com/watch?v=ki6jVXZM2XU
Type 2 unums are pretty much entirely deprecated by type 3 unums (now given the name 'posits' as referred to in the OP's linked paper)... they are basically superior in every way, and I recommend watching John's February 2nd talk at Stanford.
Just wanted to second close to the top that Augie is not only a great lawyer, but a fantastic human being and committed to helping the startup community.