shaklee3

40 ms·

631.
by shaklee3 6y ago
For real time games, yes, for the vast majority of traffic, no. Streaming and web browsing are plenty interactive enough for this to matter. Even VoIP does perfectly fine if it's tuned properly.
632.
by shaklee3 6y ago
Agreed! That's why I was bummed when the second-gen was a host system. Didn't fit well to my use case.
633.
by shaklee3 6y ago
Now I'm remembering... They had the phi as a coprocessor in a PCI slot, effectively making it just as issue as a GPU. But the second gen (knights landing) made the phi the host processor, but removed almost all ability for external dev
634.
by shaklee3 6y ago
That looks like 28 cores, and I think Phi went to 72 cores (or 144 with HT). Of course, the Phi was clocked much lower. The AMD is definitely more comparable.
635.
by shaklee3 6y ago
That's right -- I always wished they made a Phi with PCIe connections out to other peripherals. Imagine a Phi host that could connect to a GPU to offload things it was better at.
636.
by shaklee3 6y ago
I was using CUDA heavily in 2015, and I also looked at the first/second gen of the Xeon Phi at the time. I thought it was much harder to program for than cuda was at the time (and certainly that gap has widened). I recall things like a
637.
by shaklee3 6y ago
Good point, but from what I recall the phi had two different memory types, and you had to specify which you were targeting. This didn't necessarily translate to the server CPUs.
638.
by shaklee3 6y ago
There's almost zero reason to run a phi in general.
639.
by shaklee3 6y ago
Wrong. A beta program where they don't want things leaked is very different from the real system. They self-selected a group of people capable of doing this. Do you really think that millions of real customers are capable installing on
640.
by shaklee3 6y ago
I would strongly disagree. Etcd, for what it does, is extremely simple. What is so complex about it? The configuration?
641.
by shaklee3 6y ago
Kube-vip is another good alternative to metallb.
642.
by shaklee3 6y ago
Or search for multi-stage builds
643.
by shaklee3 6y ago
> These are both things that are unlikely to ever be able to run on anything other than nvidia hardware. I think it's very clear that the framework was important here. Because you were pointing out that they don't open source m
644.
by shaklee3 6y ago
Well, of course. They're written in cuda. Why would Nvidia try to write it in a language like opencl that even AMD is moving away from, and instead making fft/blas library calls? At least the source code is there if you wanted to
645.
Walmart is converting its parking lots into free drive-in theaters(thewalmartdrivein.com)
1 points by shaklee3 6y ago|0 comments
646.
by shaklee3 6y ago
Please reread this thread. This is the comment I was replying to: > SpaceX isn't doing the install work work on these Starlink kits. They just sell you some hardware. If you're a terrestrial-based internet provider, you need a
647.
by shaklee3 6y ago
This is increasingly becoming not true. See cutlass and the new release of cufftdx.
648.
by shaklee3 6y ago
Just to add some color to my comment: https://mobile.twitter.com/Megaconstellati/status/1294628802... So SpaceX only has 2 PoP locations, and it happens to be the areas where they are doing trials. As soon as the
649.
by shaklee3 6y ago
I'm very aware of MIMO and the other physical layer technologies. But the original topic was about how wireless could somehow progress faster in these metrics than physical mediums. Mediums like fiber have progress much more in the las
650.
by shaklee3 6y ago
> Musk stated that Starlink's 20ms latency is designed to run "real-time, competitive video games." He also stated that Version 2, at a lower altitude, "could be as lows 8ms latency." https://www.shack
651.
by shaklee3 6y ago
No satellite companies do that. They are independent contractors.
652.
by shaklee3 6y ago
I expect for them to make money it's going to need to cost $150 starting for the low end. Although, if they get government RDOF money, all bets are off since they're burning through government money.
653.
by shaklee3 6y ago
They were supposed to do a large rollout with sprint, and it never happened. The technology itself was vaporware; it only worked in their lab setting.
654.
by shaklee3 6y ago
As pointed out in many places, the launch costs are still being paid for by their lack of funding from other commercial launches. The bulk of the cost is not in launches; it's running an ISP and equipment.
655.
by shaklee3 6y ago
I think the latency will go up after launch. They don't have much of the complexity in the system they'll soon need.
656.
by shaklee3 6y ago
I'm maintaining these are best case numbers and that they'll go up when the service is live. I would bet you'll never see 20ms when paying customers are on it.
657.
by shaklee3 6y ago
Everything you just said applies to starlink as well. Why the double standard? Still need line of sight, they still won't provide advertised speed guarantees ( no ISP does). And most people don't care about latency. Bandwidth is f
658.
by shaklee3 6y ago
Let's revisit this. There's no point in arguing, but a single data point during a beta is clearly not what he was referring to.
659.
by shaklee3 6y ago
Nobody has ever claimed 20ms to go around the world, because that's physically impossible. He did claim ping times would be 20ms, which is patently false, even within CONUS. There is no "second generation" to date. They have
660.
by shaklee3 6y ago
Yes, they will.

More ›