Oh sorry, I see what you are saying. I understand that it would be tricky to report the accumulated compressed mipped tiled IO.
I really just meant it would be good to know the total potential size of all of the textures files on disk VS. the total potential raw pixels in memory which is what I'm assuming the "Total size of all images referenced" stat is referring to? Mainly, we were just confused as we were looking at the "Total size of all images referenced" stat thinking that it was talking about the total size on disk and that wasn't marrying up to the measured disk file sizes. On 6 September 2016 at 20:35, Larry Gritz <[email protected]> wrote: > I'm not sure I can give you that. > > We can easily report the amount of (uncompressed) texture data read, as we > do. > > When we open each file, we can note the total disk size of it > (compressed), and then report that in the per-file stats and the total of > all files we ever touched. > > But I don't think I can give any kind of accurate count of true I/O > because I'm letting the underlying format libraries (such as libtiff or > libIlmImf) handle all the decompression. Neither the TIFF nor OpenEXR > libraries have a way for an app to find out the compressed size of a > particular tile that is read, let alone also account for the fact that to > read the tile, some amount (how much, exactly?) of other I/O is necessary > to read headers, seek around, etc. > > But if you have a good idea for how I can somehow account for this, I am > all ears. > > > On Sep 6, 2016, at 12:15 PM, Dan Kripac <[email protected]> wrote: > > Thanks Larry, > > Yes it would be really good to differentiate between what we are pulling > over the network compared to what the native raw pixels consume in memory. > > Sent from my iThingy > > On 6 Sep 2016, at 18:09, Larry Gritz <[email protected]> wrote: > > It seems reasonable to also collect data on the compressed file size on > disk, I'll add this stat. > > -- lg > > > On Sep 6, 2016, at 9:43 AM, Larry Gritz <[email protected]> wrote: > > The OIIO stats are the *uncompressed* equivalent -- i.e., the amount of > texture data, not the file size. > > > On Sep 6, 2016, at 5:12 AM, Søren Ragsdale <[email protected]> wrote: > > I'm looking at the OIIO logs for an OSL render: > > OSL ShadingSystem statistics (0x5ab6360) ver 1.7.3 > [...] > OpenImageIO ImageCache statistics (shared) ver 1.7.3dev > [...] > Images : 1909 unique > ImageInputs : 17441 created, 100 current, 470 peak > Total size of all images referenced : 161.7 GB > > The problem is that I've isolated the images I'm accessing into a single > directory, and I'm getting very different numbers: > > > du -c -s -h TEXBL_character_arm* > 65M TEXBL_character_arm-cable_01_texture_default_v002 > 151M TEXBL_character_arm-cable_01_texture_default_v005 > 1.2G TEXBL_character_arm-muscles_texture_default_v020 > 46G TEXBL_character_arm_texture_default_v111 > 48G total > > OIIO seems to be overstating 48 GB of textures as 161 GB. > > I've double checked that there aren't any textures being accessed outside > this directory. If anything, some of the textures in that directory should > never be accessed so the 'du' total should be even larger. > _______________________________________________ > Oiio-dev mailing list > [email protected] > http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org > > > -- > Larry Gritz > [email protected] > > > _______________________________________________ > Oiio-dev mailing list > [email protected] > http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org > > > -- > Larry Gritz > [email protected] > > > _______________________________________________ > Oiio-dev mailing list > [email protected] > http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org > > _______________________________________________ > Oiio-dev mailing list > [email protected] > http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org > > > -- > Larry Gritz > [email protected] > > > > _______________________________________________ > Oiio-dev mailing list > [email protected] > http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org > >
_______________________________________________ Oiio-dev mailing list [email protected] http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org
