Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
loser777
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
17 ms
·
121.
▲
by
loser777
8y ago
Isn't the most direct analogy: a stolen car refuses to start? But I would still download a car if I could. side note: What does flamethrower mean in this case? Some potentially totally legitimate car modifications (e.g., two-step lau
122.
▲
End-To-End Learning of Energy-Constrained Deep Neural Networks
(arxiv.org)
3 points
by
loser777
8y ago
|
0 comments
123.
▲
by
loser777
8y ago
I would argue that this is extremely shady by Intel. It being an overclocked part means that this could be the single best 28-core die Intel has ever produced, so good luck getting one yourself (even then it still required a 1700W+ water ch
124.
▲
We Got a Sneak Peek on Intel’s 28-Core: All You Need to Know
(anandtech.com)
2 points
by
loser777
8y ago
|
0 comments
125.
▲
by
loser777
8y ago
I think the IPC gap will become an interesting battle in the near future, especially as gains that came from aggressive speculation have proven to introduce security holes (take your pick of Meltdown, Spectre variant n ,...).
126.
▲
by
loser777
8y ago
Do you mean for liquid-cooled solutions only? We've been using this Noctua U14S TR4-SP3 ( https://noctua.at/en/nh-u14s-tr4-sp3 ) on a 1950X (no overclock though) without issues.
127.
▲
by
loser777
8y ago
Just put together a first gen Threadripper build for work a few months ago where out-of-the-box ECC support was too good to pass up for a box with >= 64GiB DRAM. If it’s possible to do a drop-in upgrade (cooling solution not withstanding
128.
▲
by
loser777
8y ago
I think the cooling could be a significant part of the picture here---you can nudge the power/frequency curve a bit with very aggressive cooling (phase-change is extremely aggressive) to drop the leakage power (e.g., as AMD did with th
129.
▲
by
loser777
8y ago
If we assume this design is based off of something similar to the 28-core 8180 which has a 2.5GHz base frequency and (generous, likely understated) 205W TDP, this chip could easily have a TDP north of 300W. Those boxes on stage sure don
130.
▲
Intel Announces the Core I7-8086K: Coffee Lake at 5 GHz
(anandtech.com)
3 points
by
loser777
8y ago
|
0 comments
131.
▲
by
loser777
8y ago
"Spam phone calls/telemarketers" are probably on the nicer side of typical voice traffic floating around (at least in the US) these days. The article doesn't really mention the vast number of spoofed police/(three l
132.
▲
by
loser777
8y ago
Perhaps one of the most interesting parts of this deal are the fact that the order seems to be on the scale where things appear to be ready off-the-shelf . Does an order of 62,000 imply that Waymo is close to having hardware that can be qu
133.
▲
by
loser777
8y ago
From the article: > That leaves Cruise, which got a $2.25 billion cash infusion from Softbank on Thursday, as seemingly Waymo's top rival. > Cruise has begun testing its cars at significant scale...
134.
▲
by
loser777
8y ago
This point doesn't seem to make sense. If you have 25 potential attackers, then for there to be an actual 51% attack, more than half of them (13/25) would have to band together to actually reach 51%. Unless the assumption is that
135.
▲
by
loser777
8y ago
Surprised that the TI-8X series and TRS-80 got mentioned as popular applications, but not the Gameboy! (well, it's not _exactly_ a Z80, but it's close enough that reasonable Gameboy emulators are feasible on a TI-83+SE: https:&#x
136.
▲
by
loser777
8y ago
>the first baseline was to freeze the tensorflow graph and just run it directly on the pi. this works without any problem, it's just the pi can only do 1 image / second :/ I wonder if this was done using full blown tensorf
137.
▲
by
loser777
8y ago
This type of noise is arguably worse than that produced by a well-managed fan setup. If you have fans that spin at a constant RPM, you can fairly easily tune out the noise. Coil whine will vary depending on the load of the system (e.g., whe
138.
▲
by
loser777
8y ago
That depends on what the source of the lurching is---if you live in an area where surface street road quality is rough (e.g., Seattle), then it won't help much. But electrification can mean the bus will use a fixed gear instead of a ro
139.
▲
by
loser777
8y ago
This anecdote is hard to believe (the slows the phone down part) because unless the app is constantly hammering the storage system with I/Os even as a background process, storage usage doesn't mean it's actively slowing the P
140.
▲
by
loser777
8y ago
Brood War AI may end up being a significantly easier problem (with this formulation) than DeepMind's StarCraft II research environment for several reasons: + As others have mentioned: Brood War is significantly cheaper to run (can be h
141.
▲
by
loser777
8y ago
0.25s for 200MB (in memory) seems pretty slow for a modern CPU (800MB/s) which is more than an order of magnitude below what you would expect from main memory. There can be a lot of performance left on the table even in standard libra
142.
▲
by
loser777
8y ago
A good time to bring up: https://newsroom.intel.com/editorials/moores-law-setting-the... >Second, in today’s world Moore’s Law can be delivered only by a few companies. Every new process node gets harder and therefo
143.
▲
by
loser777
8y ago
According to a quick search Keller is apparently moving to _Intel_. Wow, I really hope that doesn't worsen the current imbalance of the x86 and GPU markets. Given that Keller seemed to be to Ryzen at AMD what Raja was to Vega/Pola
144.
▲
by
loser777
8y ago
Not mentioned in the bloomberg article (at the time of reading): Keller is apparently moving to _Intel_ (according to a quick search). Wow, I really hope that doesn't worsen the current imbalance of the x86 and GPU markets. Given that
145.
▲
by
loser777
8y ago
Might want to ban any kind of aftermarket exhaust system and saws as well (in case people decide to just chop off their exhausts). /s
146.
▲
by
loser777
8y ago
I think you linked to a flip chip---you would expect a mirror finish from the bulk silicon side.
147.
▲
by
loser777
8y ago
Unfortunately it looks like the hardware acceleration used by the AI is _relatively_ modest in this case: 6660K, GTX 1060. I wonder what the strength improvement would be with just a beefier single node machine.
148.
▲
by
loser777
9y ago
The difference is that an interesting AI could likely be scaled back meaningfully for lower skill levels. Make MarineKing play with only the mouse against a lower skilled opponent? The results might be less stale. Rate limiting a brute-forc
149.
▲
by
loser777
9y ago
If this project uses the BWAPI (which it says it does in the paper), then I assume that this doesn't need to recognize units as that state is available from the game. On the other hand, the DeepMind "research environment" ( h
150.
▲
by
loser777
9y ago
My understanding is that a micromanagement-based AI is likely not to be an "interesting" difficult AI to beat. The DeepMind project is aimed more towards (with its apm caps) at building something that makes good decisions (e.g., u
More ›