kimixa

34 ms·

451.
by kimixa 3y ago
RAII mutex locks are a pretty good tool. But like all "modern" c++ features still require programmer diligence to use correctly. Though there are likely other tools outside a compiler error to help enforce this, like static analyz
452.
by kimixa 3y ago
Zip files don't need to be compressed, if there's a need maybe you can promote adding the option upstream to whatever apps you use (if it's not already there somewhere). Or a .zip equivalent for zgrep. Though pattern matchi
453.
by kimixa 3y ago
Much of the cache "management" can be done with specialist load/store instructions that skip the cache rather than being OS managed like a mapping.
454.
by kimixa 3y ago
Registered memory has a buffer for communication between the dram and the memory controller. So the DDR bus is attached to an intermediate buffer chip, rather than directly to the dram chips on the DIMM. This can give better electrical char
455.
by kimixa 3y ago
And also heavily used in non-oss drivers. DMA blocks can only do so much when the (often older but still well used) APIs don't map well to the synchronization required - either allowing the user to immediately free and/or reuse th
456.
by kimixa 3y ago
Also useful for memory you know you won't be touching multiple times and just thrash the cache. For example GPU drivers and texture copying - even on integrated GPUs that have cache coherency with the CPU (so don't "need&quot
457.
by kimixa 3y ago
2.4 is still the best choice for range and penetration (through walls etc.) if you don't need the speed of 5ghz. I'm honestly surprised if they're shipping 5ghz-only consumer access points.
458.
by kimixa 3y ago
The 2018 macbook pros weren't even using the best silicon of the time - they were in the middle of Intel's "14nm skylake again" period, and an AMD GPU from 2016. I suspect one of the reasons why Apple silicon looks so
459.
by kimixa 3y ago
Modern GPUs allow mapping memory through PCIe, so you can treat memory as the same. Resize-able BAR effectively allows the same the other way, where all the GPU vram can be mapped into the CPU. The issue is that suddenly accessing an addres
460.
by kimixa 3y ago
I might expand this to "Being a Tech supplier to Apple is a mistake" - they seem quite happy with exporting their labor-intensive costs, and whitewashing themselves of any "questionable" activities in the process. TSMC i
461.
by kimixa 3y ago
While not "sue-ing" but instead the other way - being trying to avoid paying for patented tech. They tried to stop paying licenses for their PowerVR-derived GPUs, which they were explicitly developed from so no "convergent ev
462.
by kimixa 3y ago
I very much agree - and I think some people misunderstood what I was trying to say too - that was that I don't think switching to SYCL would fix any of the problems that rocm currently has. I personally feel many of the problems with r
463.
by kimixa 3y ago
I mean... Maybe? But Intel proved their success of owning fabs throughout the 90s and 2000s, only 20-30 years after that statement were Intel even considered to be possibly dis-advantaged by owning it's fabs. But that was also a very d
464.
by kimixa 3y ago
There's a natural compromise between "Efficiency" and "Flexibility" - training AIs is currently a fast enough moving field that by the time the really efficient targeted hardware could be released, the state of the
465.
by kimixa 3y ago
I feel this is falling into the rewrite trap - rather than continuing to develop an existing but imperfect system, throw it all away for something that is unproven and currently in a notably worse state. What about rocm makes it fundamental
466.
by kimixa 3y ago
I used to work on mobile graphics and the android HWC stack. The scanout-time hardware was often less useful that you might think - only in dynamic scenes where the GPU is otherwise idle (like playing video possibly with a static UI overlay
467.
by kimixa 3y ago
The linked article shows the "standard" pinout for a 3.5" floppy drive, and it's clear the interface is below the level of anything like DMA. I assume even if you were using something that looked like a standard floppy
468.
by kimixa 3y ago
Was that referring to the Primary Glide Slope Indicator being out of order at the time though? I find the article a little unclear if that's a general recommendation even if all the aids are functional for a visual approach.
469.
by kimixa 3y ago
I also find vcpkg extremely fragile. Not in that pinning versions or the APIs or the actual contents of the packages we rely on, so much that every time we update a (rather small) set of dependencies there's nearly always some weirdnes
470.
by kimixa 3y ago
It's the same reason that students get free MSDN access and unlimited windows keys. It's a pretty solid strategy, but naturally favors the current incumbent who has the $$$ to throw around.
471.
by kimixa 3y ago
I generally felt that AMD(/ATI) were a hardware company first, while Nvidia were a software company first. Though now Nvidia are outselling them in dGPUs by nearly 10x, so it's pretty easy for them to do "both".
472.
by kimixa 3y ago
VLIW is also still heavily used in DSPs and similar. And arguably the dual issue instructions in both AMD's and Nvidia's latest GPUs is a step back towards that. It feels like a lot of these things are cyclical - the core ideas ar
473.
by kimixa 3y ago
There was also Inner Space (1992) [0] - possibly a similar idea, where levels are directories on your computer, and the goal to collect the "icons" (normally executables IIRC) inside. I remember having it on some demo disk as a ki
474.
by kimixa 3y ago
Doesn't zen4 have 2x 256-bit FADD and 2x 256-bit FMA, and with avx512 ops it double-pumps the ALU (a good overview here [0]). If you count FADD as a single flop and FMA as 2, that's 48 "1 flop cores" per core. I think it
475.
by kimixa 3y ago
You can make the "superior" user a member of the default group for the user, and set the umask of the sandboxed user to allow group write by default. Though that doesn't help malicious/"odd" use cases that can
476.
by kimixa 3y ago
A someone with two middle names "only" totaling 19 characters I still run into issues with many forms, both online and offline. I'm never quite sure what to do on offline forms that have boxes for characters that run out, I n
477.
by kimixa 3y ago
I think you completely missed the point here. Your statement categorically denied one proposed cause, and presented another as the sole truth for an incredibly complex social result. I don't know if you're not a native english spe
478.
by kimixa 3y ago
Stating one person's theory about complex social behaviors as "fact" seems a little far, even if they did happen to write it down in a book.
479.
by kimixa 3y ago
Sure, but were (relatively) limited compared to later releases. And the on-disk format was completely changed for windows 2000, hence me saying what you call "NTFS" today is different to what it was originally. If you differentiat
480.
by kimixa 3y ago
Also some of the features listed (Compression, ACLs, volume sizes) were incrementally added over years and not available at first release, which somewhat invalidates the comparisons with other filesystems "Of the era". The NTFS of

More ›