frankchn

9 ms·

61.
by frankchn 6y ago
It doesn’t matter to the device OEM when the customer demands a refund because the device stopped working, even though the root cause is the ISP hijacking the DNS request and the device OEM has nothing to do with it. Easier to work around t
62.
by frankchn 6y ago
I don't think the "developers who want a high performance ARM laptop" niche is large enough for any serious hardware manufacturer to address.
63.
by frankchn 6y ago
The A14 is significantly faster than the Kirin 9000 (Cortex A77 on TSMC 5nm) on SPECint and SPECfp: https://www.anandtech.com/show/16226/apple-silicon-m1-a14-de...
64.
by frankchn 6y ago
> Apple has already laid the architectural groundwork, a chip with similar performance numbers can probably be commissioned with ~50 million. I think you might be two orders of magnitude off here. A class leading general purpose compute
65.
by frankchn 6y ago
> If the notifications are frustrating, users will turn off the permissions or remove the application entirely. Crappy snooping applications are gone. Mission accomplished. Will they? That didn’t happen with the UAC dialogs in Windows.
66.
by frankchn 6y ago
The correction would be about 3% if we reclassify those as unemployed: https://www.bls.gov/news.release/pdf/empsit.pdf > If the workers who were recorded as employed but absent from work due to “other reasons”
67.
by frankchn 6y ago
They did disclose survey response anomalies in the report: https://www.bls.gov/news.release/pdf/empsit.pdf > If the workers who were recorded as employed but absent from work due to “other reasons” (over and ab
68.
by frankchn 6y ago
Actually it looks like the double precision number in general GPU usage only went up 25% (Volta did 7.8 TFLOPS). To get the 2.5x number, you need to use FP64 in conjunction with TensorCores, which then gets you 19.5 TFLOPS. Considering how
69.
by frankchn 6y ago
The double-precision number probably best represents the generational improvement. The 20x 32-bit floating point improvement is probably achieved by comparing doing full FP32 calculations on the previous generation vs doing TF32 calculation
70.
by frankchn 6y ago
It seems like the TF32 format is similar to BF16 but with 3 more precision bits (in other words, it is FP32 with 13 low-order bits dropped instead of 16). Full adders in FP units scale with the square of the mantissa bits, so if the number
71.
by frankchn 6y ago
That’s true but not really relevant here. if traders are holding on to any open contracts for May past tomorrow, then I believe they will have to take physical delivery. Thus, traders who still have contracts open but cannot take physical d
72.
by frankchn 6y ago
The CME Crude Oil Futures Contract is settled through delivery at Cushing, OK: https://www.cmegroup.com/trading/energy/crude-oil/light-swee...
73.
by frankchn 6y ago
Note that this is for the May contract, which closes tomorrow, and anyone left with a contract then will have to actually take delivery of the physical product. Since we are in supercontango (oil storage is full, causing spot prices to be s
74.
by frankchn 7y ago
According to https://academic.oup.com/cid/article/49/9/1405/301441 , the aspirin doses that were used in 1918 (8.0 - 31.2 g/day) were much higher than what people normally take today (usually a
75.
by frankchn 7y ago
True, but an F1 car is on another level with 900+ hp (including ERS) at 740 kg with driver but not fuel.
76.
by frankchn 7y ago
Adding friction will more likely resulting in most users blindly giving permissions to anything that asks for it. We’ve seen this before with the endless UAC dialogs in Windows Vista.
77.
by frankchn 7y ago
Chiplet designs means that you still have to route signals either onto an interposer or onto a PCB. If you have a silicon interposer you have the same issue of making a really large silicon die. If you route into the PCB, then you may need
78.
by frankchn 7y ago
Does the Neoverse core have hyperthreading? I know that vCPU threads on the major cloud providers are hyperthreads instead of real hardware cores.
79.
by frankchn 7y ago
A big difference is that AirBnB (and Uber, Lyft, etc...) don’t own or lease any of the underlying assets (apartments, cars) but are purely middlemen.
80.
by frankchn 7y ago
Relevant xkcd: https://xkcd.com/1105/
81.
by frankchn 7y ago
Amazon already has 1st generation EPYC servers available: https://aws.amazon.com/ec2/amd/ so I assume they will follow up as well.
82.
by frankchn 7y ago
That’s the beauty of their business model — you can only opt out by opting out of most of banking altogether.
83.
by frankchn 7y ago
Your new employer has to file a H1B petition de novo, with the only difference that you are not subject to the lottery any more. USCIS may well take the view that your new job, even if it is doing the same thing, is not a specialty occupati
84.
by frankchn 7y ago
I think hardware acquisition costs can dominate GPU prices, so perf/watt might not be a realistic measure for the cost of running such large scale/high performance experiments. For instance, the DGX-2 has an MSRP of $399,000 and c
85.
by frankchn 7y ago
Vehicles getting bigger is a trend in general across all manufacturers and all vehicle categories, from the Toyota Corolla to the Lincoln Navigator. I think the only car I can think of which hasn’t gotten bigger and heavier over the years i
86.
by frankchn 7y ago
BMW ultimately is in the business of selling cars, not preserving a brand legacy. All the German brands are doing the same — AMG and S aren’t what they once were. Even Porsche GT cars are more pliant and daily-able than ever.
87.
by frankchn 7y ago
Two tapeouts, one for the IO die and one for the 8-core chiplet. For lower core count versions they can simply leave chiplet slots blank: https://img.purch.com/o/aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNv...
88.
by frankchn 7y ago
I think a key difference here is that the McMarksFace building is given to the university, while the bribery scandal involves giving money to individuals. An analogy would be "buying" a car by giving $5k to the salesperson to get
89.
by frankchn 7y ago
Xeons, even at wholesale prices, are probably much more profitable than laptop chips though.
90.
by frankchn 7y ago
The fact that they will stay competitive is already priced in, given that Tesla is by market cap about the size of GM and Ford. However, I find some of the loftier predictions of analysts ($1000+/share) somewhat unjustifiable at presen

More ›