mgaunard

34 ms·

691.
by mgaunard 2y ago
That's different from what you hinted in your previous message. "Making" has specific meaning as well, and I don't think it's what you're trying to say.
692.
by mgaunard 2y ago
Doing anything other than the dumb trigger-release logic on FPGA is counter-productive IMHO. You're already heavily constrained by placement in order to achieve the lowest latency; you can't afford to have logic that's too co
693.
by mgaunard 2y ago
It's not that simple. It could just be that exchange is the first one to drop to 99 but all others will as well.
694.
by mgaunard 2y ago
The idea for aggressive orders in HFT is to buy before the market goes up, and sell before the market goes down. HFT signals are about detecting those patterns right before they occur, but as early as the information is available globally.
695.
by mgaunard 2y ago
Even if the low-latency logic moves to the FPGA, you still need your slow path to be reasonably fast, say about 100us, and absolutely never more than 1ms. To my knowledge Python is not suitable, though I know some players embed scripting la
696.
by mgaunard 2y ago
Latency isn't even as important in HFT as people claim. What's most important is deterministically staying into a reasonable enveloppe (even at the 99.9 percentile) to satisfy real-time requirements and not fall behind. When it&#x
697.
by mgaunard 2y ago
Which don't have the same level of control as C++, it only provides a subset of sequentially consistent primitives, which are slow by design.
698.
by mgaunard 2y ago
ACCU London is not very active anymore, I think CppLondon pretty much took over (it was the same guy as the CppOnSea conference, but I think he started asking other people to arrange it now). I don't know if it's that great these
699.
by mgaunard 2y ago
A good language just gives you the necessary tooling to do whatever you want, it doesn't magically fix problems. Only languages like C++ have a memory model that allows you to do lock-free programming for example (C and Rust copied the
700.
by mgaunard 2y ago
The mistake was using golang apparently. If you need high-performance and fine control of synchronization, just use a low-level systems programming language.
701.
by mgaunard 2y ago
There are regular C++ meetups, which I expect would cover systems programming. It seems however cloud/web people have a different idea of what systems programming is.
702.
by mgaunard 2y ago
Most of my web apps are built around tabular data.
703.
by mgaunard 2y ago
DPDK polls the hardware directly from userland. XDP reads the data in the normal NAPI kernel way, integrating with the IRQ system etc., which might or might not be desirable depending on your use case. Then if you want to forward it to user
704.
by mgaunard 2y ago
Yes for XDP there is a dedicated API, but for any of the other hooks like tracepoints, it's all designed to give you read-only access. The whole CO-RE thing is about having a kernel-version-agnostic way of reading fields from kernel da
705.
by mgaunard 2y ago
This is just normal asynchronous I/O with callbacks instead of coroutines.
706.
by mgaunard 2y ago
Myself I've only built simple things, like tracing sched switch events for certain threads, and killing the process if they happen (specifically designed as a safety for pinned threads).
707.
by mgaunard 2y ago
It lets you hook into various points in the kernel; ultimately you need to learn how the Linux kernel is structured to make the most of it. Unlike a module, it can only really read data, not modify data structures, so it's nice for thi
708.
by mgaunard 2y ago
To me the duff's device is just a mechanism to unroll a loop without having to duplicate the code for the trailing case. While you can't use SIMD you can still benefit from instruction-level parallelism. It's potentially bett
709.
by mgaunard 2y ago
Well but I might have a hundred files, only one of which changed. The 99 other ones are still in the browser cache and don't need to be re-(down)loaded. If I bundle everything, then I have to scrap and reload everything, which is proba
710.
by mgaunard 2y ago
I'm not a web developer, though I still develop web apps regularly. What exactly is the point of a bundler in the rapid development cycle? If you want your web app to load up fast, it's better if you only need to redownload the pa
711.
by mgaunard 2y ago
sounds like a step backward to be honest
712.
by mgaunard 2y ago
aren't the switch shenanigans important to the duff's device?
713.
by mgaunard 2y ago
It doesn't really work, the libc is widely non-conforming. I'd rather have a proper interface designed for portability rather than a hacked-together POSIX-but-not-really.
714.
by mgaunard 2y ago
How does it compare to dumping a core, or regarding what the process is doing for reverse debugging?
715.
by mgaunard 2y ago
The real fix is to not use -ffast-math. It breaks all sorts of things by design. If you really want to relax the correctness of your code to get some potential speedup, either do the optimizations yourself instead of letting the compiler do
716.
by mgaunard 2y ago
That sounds like a borderline insult, but I'm not familiar with American literature so I can't really feel targeted by it. If you're a leader with a strategic role, time is your most important and limited resource. You just c
717.
by mgaunard 2y ago
Is there?
718.
by mgaunard 2y ago
Why would you invest time and effort in solving someone else's problems? As a manager your responsibilities are to the shareholders. A business is not a charity.
719.
by mgaunard 2y ago
You're speaking under the assumption that "you" can't be a team lead or manager able to make the bad performer redundant. Moreover as an IC there is a lot you can do; most importantly you can quit if you feel the team yo
720.
by mgaunard 2y ago
Will you be kind as well to the people who repeatedly fuck up and don't care about delivering good processes or products?

More ›