7 ms·
This is about firmware, nothing to do with the performance of GPUs...
by Raphael_Amiard 4y ago
This is about firmware, nothing to do with the performance of GPUs...
- wheybags 4y agoFirmware and drivers have a massive impact on the performance of GPUs. It's not just hardware.
- timbit42 4y agoThe article states they had no performance hit from switching to SPARK.
- sillysaurusx 4y agoIt’s rare to see a thread where everyone is simultaneously correct but talking past each other. None of you are mistaken.
- wheybags 4y agoI noticed this happening and just stopped replying :p
- thatwasunusual 4y agoThis is for _security_. https://www.slideshare.net/AdaCore/securing-the-future-of-safety-and-security-of-embedded-software https://www.slideshare.net/AdaCore/securing-the-future-of-sa...
- VHRanger 4y agoYes, and security has a large performance impact. Just look at the performance costs of bounds-checking array access in C++ code. Or more macro, the performance impacs of AV tools or Windows Defender on your system
- 3a2d29 4y agoWouldn’t the performance costs of bounds checking on arrays be the same if the computer was doing it or if your code was doing it? By that logic C/C++ doing no bounds checking speeds your code up?
- VHRanger 4y agoYes, which is why compiling on different optimization settings will have bounds checking on or off in C++
- Filligree 4y agoWell, yes, it does. Whether or not that’s a good tradeoff is a different question.
- godshatter 4y ago> Wouldn’t the performance costs of bounds checking on arrays be the same if the computer was doing it or if your code was doing it? It depends. The C programmer can choose to do the bounds checking in a for loop by just checking once before the loop begins, or once per iteration even if an array is accessed multiple times in the loop, or the safe language might have more overhead than a simple if statement in the C code. This can, of course, go the opposite direction (the safe language has verified the loop bounds, but the C programmer is checking before every array access). It's a battle between the C programmer and the designer and/or implementer of the safe language. One of the reasons I like C is it gives you more control. This can be a good or a bad thing. This can lead to some really performant code you couldn't do in most languages or it can lead to some gnarly security problems. Maybe both in the same spot of code. I use C to write mostly pet projects at home. I use it at work without having a choice in the matter.
- glacia01 4y ago>Just look at the performance costs of bounds-checking array access in C++ code. If your compiler can prove you dont need bounds-checking it will remove the check and the performance would be the same. Hence, if your program has been proven to have no runtime errors you dont need them.