Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
ssp
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
43 ms
·
391.
▲
by
ssp
17y ago
Andy Rutledge's blog has lots of good stuff. His five-part introduction to gestalt principles is about the basics of graphic design: http://www.andyrutledge.com/gestalt-principles-1-figure-grou... http://www.andyrutledge.com/gestalt-prin
392.
▲
by
ssp
17y ago
The problem for Amazon is that they don't have much unique value to provide besides their brand and scale. The good experience, the Kindle etc., can all be replicated. If Amazon starts squeezing the publishers, some of them might leave and
393.
▲
by
ssp
17y ago
In heavy traffic most people would benefit from switching lanes, because most people are in the lane with the most cars.
394.
▲
by
ssp
17y ago
Note that I didn't say Walmart is bad. In fact I said: 'Is it good for consumers? Probably', and I mean that. I called their goods junk, just as I'm willing to call The Duke's Boardroom Affair junk, because that's how those products are s
395.
▲
by
ssp
17y ago
Amazon is behaving like Walmart here: squeezing suppliers in order to provide lower prices. Walmart is certainly not good for Walmart suppliers, because if you want to sell there, your profit margins will go down and down, and you will need
396.
▲
by
ssp
17y ago
His publications are here: http://www.mit.edu/~parrilo/pubs/index.html . This one in particular: http://arxiv.org/pdf/0909.0808v1 looks like it might be related.
397.
▲
The VP8 codec: High compression + low complexity
(dspdesignline.com)
2 points
by
ssp
17y ago
|
0 comments
398.
▲
by
ssp
17y ago
Well, UTF-8 is rather brilliant in many ways. It can represent all of Unicode while remaining backwards compatible with ASCII and C string representations. The drawback, that characters have variable encoding, is almost irrelevant, becaus
399.
▲
by
ssp
17y ago
Samsung would be an obvious guess at the fab, given Apple's history of working with them.
400.
▲
by
ssp
17y ago
So, what does this chip do that, say, a Snapdragon doesn't?
401.
▲
by
ssp
17y ago
Companies that are getting eaten from below have very few options. Do you really think Sun could have made a profit selling equipment at PC margins? That requires an entirely different cost structure. If they had tried, you could have writt
402.
▲
by
ssp
17y ago
Does the analogy work though? The fear in the rallying cry is that if musicians can't eat, they will stop making music, which means you can't have music anymore. But if accountants can't eat, it's because there is less need for accounting,
403.
▲
by
ssp
17y ago
Yes, it's a marketing piece, but the statement that AMD is never going to add threads to their CPUs is surprising. Given the trend in memory vs. CPU performance, adding threads to hide latency seems like a complete no-brainer. I'd really li
404.
▲
It's all about the cores. Why AMD doesn't use HyperThreading
(blogs.amd.com)
48 points
by
ssp
17y ago
|
44 comments
405.
▲
The Midlayer Mistake (Kernel Design Pattern)
(lwn.net)
25 points
by
ssp
17y ago
|
2 comments
406.
▲
by
ssp
17y ago
The main point of ext4 (and ext3) is that they are backwards compatible; you can use your existing on-disk layout. Google probably has a zillion disks that they can't just reformat.
407.
▲
by
ssp
17y ago
Hmm, I was actually about to edit my comment to say that you are right because usually you'll store the old stack pointer as the new base pointer. I guess the answer is that realigning at runtime costs a register, but most code has to pay
408.
▲
by
ssp
17y ago
You need to restore the stack to what it was before on return. Normally you can easily do that because you know the stack pointer delta at compile time. But when realigning at runtime, you don't, so you need to compute the difference and th
409.
▲
by
ssp
17y ago
Right, "just make your stack frame a multiple of 16 bytes" was wrong; I should have said "just make your stack frame 16 * k - 4 bytes long". This is basically free because the size of the stack frame is known at compile time. Realigning the
410.
▲
by
ssp
17y ago
Presumably, Apple is relying on SSE for floating point, like x86-64 does. That certainly is both faster and simpler than dealing with the bizarre x87. And you don't have to insert extra code at all call sites. You can assume that your s
411.
▲
by
ssp
17y ago
I always appreciate people giving me advice on my website, even if I don't end up taking it. Ok, here is what I would do. Dramatically simplify the whole experience: * The front page says "Make bingo cards on your computer" in large print
412.
▲
by
ssp
17y ago
"pi seconds in a nano-century" is how I remember it.
413.
▲
Some Actual Larrabee Information
(realtimerendering.com)
13 points
by
ssp
17y ago
|
0 comments
414.
▲
by
ssp
17y ago
job you just don't quit Did you really mean "just don't quit", and not "don't just quit"? If there really are jobs in Japan that you are obligated to stay at for life, that would seem to be a pretty large impediment to growth.
415.
▲
Larrabee talk at Stanford by Tom Forsyth
(stanford.edu)
1 points
by
ssp
17y ago
|
0 comments
416.
▲
by
ssp
17y ago
Eventually, only Trolls with a fair-case will fight Don't a lot of trolls actually have valid patents that read on whatever it is you do? If this worked, why wouldn't Microsoft just "self-insure" and never, ever settle a patent claim?
417.
▲
by
ssp
17y ago
Maybe now someone will actually die in a blogging accident ( http://xkcd.com/369/ ).
418.
▲
Qualcomm signs with Global Foundries
(semiaccurate.com)
1 points
by
ssp
17y ago
|
0 comments
419.
▲
A hack I'm disturbingly proud of, and its connection to some real math
(godplaysdice.blogspot.com)
16 points
by
ssp
17y ago
|
6 comments
420.
▲
by
ssp
17y ago
One thing worth considering is that parts of Intel may not so thrilled about the Atom. It's a really low-margin chip and it's being manufactured at TSMC, both of which are somewhat out-of-character for Intel. When Craig Barrett left as ch
More ›