neilmovva

8 ms·

31.
The Seven Deadly Sins of Predicting the Future of AI(rodneybrooks.com)
15 points by neilmovva 9y ago|1 comments
32.
by neilmovva 9y ago
GPUs are generally not latency optimized anyway - if an app is sensitive to tens of microseconds then it belongs on the CPU, broadly speaking. AMD's advantage here is getting drop-in throughput, by switching the SSD directly to the GPU
33.
by neilmovva 9y ago
Vega is already drawing near 300W, and is so high up on the voltage/frequency curve that even a measly 5-10% gain in core clock can easily cost over 100W more. Here, AMD is once again (see last year's Polaris) a victim of the infe
34.
by neilmovva 9y ago
"graphics memory" is probably not the right term here. While the Pro SSG does have 2TB of storage on board, that comes in the form of a NAND flash, PCIe SSD (from Samsung IIRC). While fast, this SSD provides single digit GB/s
35.
by neilmovva 9y ago
Taking a more skeptical look at the issue, we might have Intel to blame for the stagnant system interconnect. As it gets harder each year to improve CPU performance, the HPC community has shifted to more workload-specific accelerators, most
36.
by neilmovva 10y ago
fp32 is almost never "gimped," since that's the basis of performance for most video games and 3D applications. perhaps you meant fp64? That sees more use in industrial apps (e.g. oil and gas exploration) and so is often slow
37.
by neilmovva 10y ago
minor correction - the 15" retina MacBook pro is 2880x1800. The 13" retina MacBook pro is 2560x1600.
38.
by neilmovva 10y ago
> "only worth 28B" I feel like this statement is telling in and of itself - since when is a $28B startup with Uber's (lack of) revenue and lasting IP considered normal? Perhaps these numbers have yet a ways to fall.
39.
by neilmovva 10y ago
The hurdle is heat - we can do something like this in mobile SoCs, which commonly place DRAM on top of the CPU (package-on-package). But for a TDP > 10 watts, the memory layer effectively insulates the main die from whatever thermal mana
40.
by neilmovva 11y ago
Great work. I wonder if the author had heard of the Eyewriter project [1]? Similar system, fully open source, but I'm not sure how active development is nowadays. It's been up since ~2011, though, so quite old by software standard
41.
by neilmovva 11y ago
While I agree that perhaps we could use some more diversity in science research competition, I think you underestimate the scale of work and dedication required - look at Sara Volz, the 2013 winner, who notably did much of her work in an im
42.
by neilmovva 11y ago
That is pretty disappointing to hear - Intel's execution and involvement with the fair has been a great model for all STEM competition. The other major high school research event, the Siemens Competition, seems much less rigorously con
43.
by neilmovva 11y ago
However, some manufacturers have been backpedaling on NAND process size recently, since larger cells have better endurance. Capacity still scales well thanks to tech like Triple Level Cells (TLC) and 3D-stacked dies.
44.
Windows Insiders to Receive Full Windows 10(anandtech.com)
1 points by neilmovva 11y ago|0 comments
45.
by neilmovva 11y ago
Console hardware is always high volume and low margin; MS and Sony usually breakeven on costs or even eat losses at the start of a new console generation. For AMD, it was the volume they needed just to keep orders going to the fab and a cas
46.
by neilmovva 11y ago
The issue with TSMC/GF "16/14 nm" is that not all important features have been brought down to that level, so the actual shrink is less pronounced than usual. Furthermore, Intel's FinFET's are likely to be more
47.
by neilmovva 11y ago
Intel's ever-widening process advantage makes that unlikely, processor design entirely notwithstanding. Unfortunate, but it's the result of being stuck on 28nm for 4 years and counting now.
48.
Intel Desktop Broadwell – High Performance 14nm CPU and 128MB EDRAM L4$(anandtech.com)
2 points by neilmovva 11y ago|0 comments
49.
by neilmovva 12y ago
Man, even at 20nm, the die size must be a bear. Especially the L3$... I know TSMC has a ~600mm^2 max (for remotely reasonable yields), even on the highly mature 28nm process. Thus, I doubt this would be a viable chip for deployment today, r
50.
by neilmovva 12y ago
To be pedantic - you might be looking for "price discrimination" instead of perfectly elastic.
51.
Intel Announces Curie: Tiny Module for Wearables(anandtech.com)
5 points by neilmovva 12y ago|0 comments
52.
by neilmovva 12y ago
It seems that "Veti" just refers to the veterinary application. I've played with similar would healing biomaterials before, and from the (scarce) available materials, it looks like they're using a sodium-alginate based p