Forums before death by AOL, social media and spammers... "We can't have nice things"
|    comp.lang.asm.x86    |    Ahh, the lost art of x86 assembly    |    4,675 messages    |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
|    Message 3,822 of 4,675    |
|    Rick C. Hodgin to wolfgang kern    |
|    Re: benchmarks..    |
|    11 Mar 19 12:56:23    |
      From: rick.c.hodgin@gmail.com              On Monday, March 11, 2019 at 5:41:04 AM UTC-4, wolfgang kern wrote:       > [Regarding RDTSCP]       >       > Yes, but only if more than one core is active.       > I mainly have just the boot-core active and make all others sleep.              RDTSCP is serializing. It can affect real-world performance       comparisons in testing because of those side effects.              You'd be better off for benchmarking to set thread / process       affinity to a single core and use RDTSC. Such an affinity       will keep data in the cache and present more realistic real-       world performance values.              You may also be better off for raw performance testing in       choosing a core not inundated with ISR handling and the like.              --       Rick C. Hodgin              --- SoupGate-Win32 v1.05        * Origin: you cannot sedate... all the things you hate (1:229/2)    |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
(c) 1994, bbs@darkrealms.ca