When it comes to decoding embedded JPGs ("Internal JPEG") then FRV can only match Faststone when at least 3 "RAW decode threads" are used. So with 8 "RAW decode threads" you need to set a minimum of 10 "Decode RAW Cache size" to prevent such delays. The latter is also evidenced by both the total measurable CPU load and the load of the FRV main thread which mostly doesn't max out its core.Ĭuriously FRV limits the number of available threads to "Decode RAW Cache size" -1, so a cache size of 2 (minimum) would limit the number of "RAW decode threads" to 1.įurthermore FRV needs at least threads "RAW decode threads" + 2 for "Decode RAW cache size" to prevent delays from happening every time you change the browsing direction (forwards to backwards and vice versa). FRV with raw files is at least as fast as using Faststone with embedded previews, if not faster.īut: at least 3 "RAW decode threads" are needed for maximum performance and more than 3 threads don't seem to improve performance any further. FRV has no match when it comes to displaying raw files when enough CPU cores/threads are available.
NVidia GTX 780 GHz Edition, using default DX9 mode. XnView was slowest overall and IrfanView is mostly matched or outmatched by Faststone, so I leave these two out from here on.ĬPU is an AMD 1800X running all cores at 3.95 Ghz + 3333-C14 RAM (tight subtimings). I tested this version using NEF, ORF, RW2 and JPG files versus Faststone, Irfanview and XnView (Classic).
Fastrawviewer key license key#
LIcense key (valid till the end of July) If you are still using a release version, you will find the first batch of changes to Preferences and explanations in "Preview I" here: 14.3 fps for "repeat forward browsing" (the raw files from the test set are already decoded and put to cache)įor default video mode (Upsample: Bilinear Downsample: Low res previews) 8.8 fps during sequential forward browsing (limits the unpacking speed)