qyliss changed the topic of #spectrum to: A compartmentalized operating system | https://spectrum-os.org/ | Logs: https://logs.spectrum-os.org/spectrum/
tilpner_ has joined #spectrum
tilpner has quit [Ping timeout: 240 seconds]
tilpner_ is now known as tilpner
<qyliss> I think there are two ways of looking at what I need to do
<qyliss> Either I need a relatively small total fixed amount of growable buffers, or a large amount of allocs/frees, thousands of times per second
<qyliss> i.e., I know up front that there will always be 2^n buffers in total, for some n <= 16
<qyliss> But I do not know in advance what size those buffers will be
<qyliss> They will probably always be kilobytes, as opposed to bytes or megabytes, though.
<qyliss> That's not guaranteed but should be extremely common.
<qyliss> I do have to be prepared to handle a 1GB buffer if it comes up, though.
<qyliss> The absolute maximum buffer size is 2^32 bytes
<qyliss> What I'd really like, though, is to not have to write an allocator, at least not now.
<qyliss> I'd much prefer to use something off-the-shelf, and come back to tune it later
<qyliss> But I haven't really found much in the way of off-the-shelf here
<qyliss> cc edef for allocator discussion
<qyliss> edef has cleverly pointed out elsenet that I don't actually need an allocator
<qyliss> I can allocate n buffers of the maximum possible size up front, and then use madvise/fallocate to tell the kernel when I'm done with a buffer.
<qyliss> And then it can clean up those pages until I next use that buffer
<IdleBot_2e4f9b4b> If you consider a risk of needing 1 GiB sized buffers, sounds like an overcommit-dependent strategy
cole-h has quit [Quit: Goodbye]
amanjeev- has quit [Quit: Logging off! Bbye!]
amanjeev has joined #spectrum
nicoo has quit [Ping timeout: 240 seconds]
nicoo has joined #spectrum
jb55 has quit [Ping timeout: 240 seconds]
nicoo has quit [Ping timeout: 240 seconds]
jb55 has joined #spectrum
nicoo has joined #spectrum
alj[m] has quit [Ping timeout: 260 seconds]
alj[m] has joined #spectrum
mvnetbiz_6 has joined #spectrum
mvnetbiz_ has quit [Ping timeout: 256 seconds]
mvnetbiz_ has joined #spectrum
mvnetbiz_6 has quit [Ping timeout: 240 seconds]
jb55 has quit [Remote host closed the connection]
jb55 has joined #spectrum
mvnetbiz_8 has joined #spectrum
mvnetbiz_ has quit [Ping timeout: 260 seconds]
mvnetbiz_8 is now known as mvnetbiz_
mvnetbiz_ has quit [Client Quit]
mvnetbiz_ has joined #spectrum
E1HC has quit [Ping timeout: 265 seconds]
E1HC has joined #spectrum
qyliss has quit [Quit: bye]
qyliss has joined #spectrum
cole-h has joined #spectrum
jb55 has quit [Remote host closed the connection]
jb55 has joined #spectrum
vilhalmer has quit [Ping timeout: 256 seconds]
vilhalmer has joined #spectrum
cation21 has quit [Quit: Leaving]
cation21 has joined #spectrum
tazjin is now known as edefdef
edefdef is now known as tazjin