yvdriess

40 ms·

481.
by yvdriess 10y ago
Programming a modern Xeon x86 does count. Modernising your software for a haswell/skylake server Xeon (ISA was made public) also modernises it for the new Xeon Phi. You have a nearly identical ISA and programming model. In other words,
482.
by yvdriess 10y ago
Wait, Brussels was not even on the list?
483.
by yvdriess 10y ago
Mastering or thoroughly understand the system before moving on is implied :) It might also be important to play differently challenging games (#1), e.g. DotA2 + Europa Universalis 4 + CS:GO will give you a much wider spectrum of brain trai
484.
by yvdriess 10y ago
A similar issue is for the commodity hardware handed out to developers. The desktop setup gets standardized to the lowest common denominator, managers/IT do not get the argument that saving 1k on a developer's desk setup is not g
485.
by yvdriess 10y ago
It is interesting to point out that the last few nuclear tests were to test the supercomputer simulations.
486.
by yvdriess 10y ago
If you look at supercomputing as a national strategic asset, that is a small number. The K-computer was around the price of a battleship: 1b USD
487.
by yvdriess 11y ago
Yes, you are indeed right.
488.
by yvdriess 11y ago
Outside of its security model, E is an interesting language to study because of its actor/object system. Objects live in 'vats', an actor with a local stack and an event loop thread/process. Objects can sequentially&#x2F
489.
by yvdriess 11y ago
Intel's client computing chips plateaued, but server platforms have kept improving. The volumes might be lower, but the high margins more than make up for it.
490.
by yvdriess 11y ago
I would not be surprised, they have done both in the past.
491.
by yvdriess 11y ago
Please do, I would love to hear a piece with a pro-VLIW perspective. I always found Transmeta's approach to VLIW very interesting. In a way, it did for VLIW what current GPUs are doing for SIMD: take the dispatching away from the compi
492.
by yvdriess 11y ago
Well, they used to. http://www.xbitlabs.com/news/other/display/20121107223140_AM...
493.
by yvdriess 11y ago
All of them. There has been a steady march of applications gobbling up (or wasting) more cycles and memory as performance improved. I do not see how 'working fine right now' is an argument on how this march is suddenly stopping ri
494.
by yvdriess 11y ago
Intel's icc (and even gcc) got really good at vectorization these last few years. It's rare when I see some code that the vectorization-report says is not worth vectorizing that actually is. Last week, I got better results by lea
495.
by yvdriess 11y ago
GPUs are in an interesting place. They were not able to scale under 28nm due to the increased static leakage. Intel CPUs were able to scale because of dynamic power management, but GPUs cannot really use the same trick. It's only with
496.
by yvdriess 11y ago
(Here is my ELI5 attempt using the one-syllable G. Steele format, assuming you know what a transistor is ;) Transistors get warm when they work. We have so many of them now that we can not make all of them work at the same time. The CPU wo
497.
by yvdriess 11y ago
You can also look at it this way: Your old laptop is still good because subsequent generations are not increasing their general compute power at any great speed. When every new release increases performance dramatically, say >150%, you s
498.
by yvdriess 11y ago
Maybe if it would be sending AST in JSON, not recommended though :)
499.
by yvdriess 11y ago
A vastly lower lobby magnitude?
500.
by yvdriess 11y ago
The article points out that the majority of the Disney movies are based on public works. It may be interesting to note the controversy around the 'concidental similarities' between The Lion King and Kimba the White Lion. ( http:&#
501.
by yvdriess 11y ago
You are being a bit hard on sharks here, they provide a valuable service to the ecosystem.
502.
by yvdriess 11y ago
The evils of abstraction.
503.
by yvdriess 11y ago
The interesting question is if this was intentional neglect, unintentional neglect or incompetence.
504.
by yvdriess 11y ago
Yes, Transmeta was bought by Intel. x86, despite its quirks, does not seem to be holding back the hardware too much. The silicon for decode/ILP does a fantastic job.
505.
by yvdriess 11y ago
Average consumption at least. It's still the poorer layers of society who suffer the most obesity problems, so it would be interesting to see a graph of their average consumption. The highly-refined super-sugary foods are typically the
506.
by yvdriess 11y ago
sucrose is an 50/50 glucose/fructose combination: https://en.wikipedia.org/wiki/Sucrose But yeah, it's all sucrose except for 'dextro energy' (glucose==grape sugar) or baking.
507.
by yvdriess 11y ago
> I was wondering how radically this architecture differs from more conventional x86 cores The big difference besides memory and organisation is that the KNL cores are beefed-up Silvermont cores (the new Atom), which is an out of order c
508.
by yvdriess 11y ago
I was indeed more thinking of the fine-grained dataflow or event-driven tasks, in the style of VAL/SISAL or very recently OCR ( https://xstack.exascale-tech.com/wiki/index.php/Main_Page ). I heard there were so
509.
by yvdriess 11y ago
It works well if it code-generated down to efficient sequential code. It does not work well if you actually perform fine-grained dataflow execution: all individual operations can fire when their inputs become available. The former works we
510.
by yvdriess 11y ago
So does dynamic compilation, such as LuaJIT and PyPy ;)

More ›