David Patterson Says It’s Time for New Computer Architectures and Languages • Hacker News Board

Re: David Patterson Says It’s Time for New Computer Architectures and Languages

#51

Post by austincummings » Mon, Sep 17, 2018, 10:54 PM UTC

so_tired wrote: post #43

ericand wrote: post #4

I'm a VC and I've seen lots of pitches lately for various tech related to new architectures, be it GPU, FPGA or RISC. I'd add cryptocurrency to list workloads with insatiable computing demands.

Do u see more pitches for source-code level solutions ? i.e. analyzing and modifying the huge legacy of existing code base ? This is a very difficult area of course (both financially and technologically). But the pay off can be very large. Imagine a vmware-like transformation at the source code level.

> Imagine a vmware-like transformation at the source code level.

I'm not sure what you mean by this, but I'm interested. Could you explain?

Re: David Patterson Says It’s Time for New Computer Architectures and Languages

#52

Post by marcosdumay » Mon, Sep 17, 2018, 11:00 PM UTC

scott_s wrote: post #35

marcosdumay wrote: post #9

Earlier quoted context omitted.

Well, past experiences are not that useful right now. I don't think the Itanium or the Cell were great architectures, but it's clear that their most obvious failure mode (mainstream architectures improving faster than them) isn't a showstopper anymore. For some guess on the next big thing, I would imagine that stuff that avoid the need of memory coherence have nice odds.

The most obvious failure for the Cell architecture was that it was horrendous to program. That was partly caused by the lack of memory coherence between the PPE (main core that ran the OS) and the SPEs (the faster cores meant for offloading computations to), but also caused by a lack of developer tools. The main lesson I would take away from Cell is that you want to design an architecture that allows gradual performa…

Developer tools come with time, and time is the one thing non-standard architectures completely lacked until recently.

I still don't think the Cell had much future, but we will never be sure. (Also, the Cell had complete memory coherence.)

Re: David Patterson Says It’s Time for New Computer Architectures and Languages

#53

Post by stochastic_monk » Mon, Sep 17, 2018, 11:01 PM UTC

the8472 wrote: post #45

philipkglass wrote: post #7

Are there languages that have first-class support for representing/optimizing memory hierarchy characteristics? Optimizing C compilers, for example, may have extensions to force specific alignments: https://software.intel.com/en-us/articles/coding-for-perform... But I'm not aware of languages where e.g. declaring alignments is part of the base language. Awareness of L1, L2, L3 cache characteristics, plus NUMA nodes,…

Rust has alignment annotations. https://github.com/rust-lang/rfcs/blob/master/text/1358-repr...

Edit: RcouF1 posted a more thorough response covering cache coherence as well earlier, which I hadn’t seen. Read that instead. https://news.ycombinator.com/item?id=18010256

Original post:

As does C++ as of 11. [0] For heap allocations, one used the POSIX (not stdlib) posix_memalign until C++17/C11, when std::aligned_alloc became standard. [1] The other option was manually aligning heap memory; it’s so much more convenient to use a transparent aligned allocator.

[0] https://en.cppreference.com/w/cpp/language/alignas

[1] https://en.cppreference.com/w/cpp/memory/c/aligned_alloc

Re: David Patterson Says It’s Time for New Computer Architectures and Languages

#54

Post by saltcured » Mon, Sep 17, 2018, 11:08 PM UTC

gone35 wrote: post #39

chubot wrote: post #32

Earlier quoted context omitted.

Yes, a pithier way of saying this is: Software is a gas; it expands to fill its container - Nathan Myhrvold Bad performance isn't really a software design problem; it's an economics problem (ditto for security). You could speed up the browser by 2x now, or reduce it's memory usage by 2x, and websites would be exactly as slow after a few months. I'm not trying to be negative here, just stating an unfortunate systems d…

Bad performance isn't really a software design problem; it's an economics problem (ditto for security). Why do you say so?

Product developers make cost-benefit decisions all the time and much of the cost of slow or insecure software is externalized. A knee in the developer-visible cost-benefit curve represents the MVP everyone is so keen to release to market.

Re: David Patterson Says It’s Time for New Computer Architectures and Languages

#55

Post by riazrizvi » Mon, Sep 17, 2018, 11:12 PM UTC

philipkglass wrote: post #7

Are there languages that have first-class support for representing/optimizing memory hierarchy characteristics? Optimizing C compilers, for example, may have extensions to force specific alignments: https://software.intel.com/en-us/articles/coding-for-perform... But I'm not aware of languages where e.g. declaring alignments is part of the base language. Awareness of L1, L2, L3 cache characteristics, plus NUMA nodes,…

It seems to me that it is generally better to keep architecture-specific memory commands as part of a platform-based library that is imported into the code, depending on where it is being compiled (Separation-of-concerns). Imagine the amount of inappropriate/unnecessary commands that one would create in a language if you moved the superset of all possible memory commands on all possible architectures, into the language itself. There would be enough to make it ugly.

Re: David Patterson Says It’s Time for New Computer Architectures and Languages

#56

Post by Ologn » Mon, Sep 17, 2018, 11:13 PM UTC

Razengan wrote: post #44

I fear, the more we advance, the more we are going to become permanently entrenched in The Way Things Are. To really shake things up at a fundamental level will probably require running into an alien species that does basic things differently than us. Even creative people with lots and lots of spare time and resources at their disposal will still be too "colored" by existing human knowledge and established practices…

qwertyuiop

Re: David Patterson Says It’s Time for New Computer Architectures and Languages

#57

Post by luma » Mon, Sep 17, 2018, 11:18 PM UTC

Only tangentially related to the article but... I'm left to wonder about the photograph selected. His left hand appears to be on a stack of 20-year old HP 9000s (maybe rp3440, pre-grey box?), and standing across from a rack full of ancient 4U systems mounted into 2-post relay racks. The white blanking panels are early 2000's-era HP kit, and in the background are racks full of tower systems stacked vertically.

Is this a recent photo? Do they have a computer history museum @ UCB or is that datacenter actually running production workloads on that kit?

Re: David Patterson Says It’s Time for New Computer Architectures and Languages

#58

Post by mjfl » Mon, Sep 17, 2018, 11:24 PM UTC

It would be nice if low volume foundry costs went down an order of magnitude. I'm in a bio lab right now, but I did CS in college and really enjoyed VLSI. If I want to turn a design into a prototype in my bio lab, it's going to cost hundreds of dollars, but if I want to turn a chip I designed into reality, it's going to cost thousands to tens of thousands of dollars even with specialized low volume foundries like MOSIS. I feel like it should be cheaper.

Re: David Patterson Says It’s Time for New Computer Architectures and Languages

#59

Post by Varcht » Mon, Sep 17, 2018, 11:25 PM UTC

luma wrote: post #57

Only tangentially related to the article but... I'm left to wonder about the photograph selected. His left hand appears to be on a stack of 20-year old HP 9000s (maybe rp3440, pre-grey box?), and standing across from a rack full of ancient 4U systems mounted into 2-post relay racks. The white blanking panels are early 2000's-era HP kit, and in the background are racks full of tower systems stacked vertically. Is this…

Per the exif data embeded in the image "Date Time Digitized: May 1, 2006, 9:10:59 PM, Creator: Peg Skorpinski photo".

Re: David Patterson Says It’s Time for New Computer Architectures and Languages

#60

Post by nuguy » Mon, Sep 17, 2018, 11:29 PM UTC

This is a question of computer science but more a question of economics. Computer systems will drop their own compassing solutions for more specific solutions as time goes forward.