Signed-off-by: Tanguy Pruvot <[email protected]>

changes only tested/applied to SM 5+ with some cleanup

+ small klausT cleanup..

dynamically alloc the hardcoded gpu clocks list array size.

Unlike the 9xx, the 750ti api call requires to get them all (361.62)

Something was "fixed" recently for the 750 Ti in linux drivers...

+ some new functions to test on pascal (if supported)

Allow to get/set missing infos like the power limit on x86

squashed for a better min/max and device mapping

Signed-off-by: Tanguy Pruvot <[email protected]>

not for GTX pascal neither...

optimized for the 9xx and more recent, same results on the 750 Ti
+ restore second nonce support not present in nicehash published version

Better on linux at least...

seems better on windows and a bit easier to read...

these informations are shown with ccminer -D -n

boost clocks and the thermal limit are shared with afterburner
beware with your settings, not as safe as application clocks!

Note: both nvapi and nvml are now used on windows x64
Signed-off-by: Tanguy Pruvot <[email protected]>

also add perf stats api, unsure what it is exactly..

cuda 6.5 one seems to crash on pascal or report invalid mem sizes

seen on yiimp benchs, PNY duplicate could be for usb keys only

also query the led level and fix OV delta output

note: the Gigabyte 1080 G1 seems not compatible with the current nvapi

you can now use --led=0xFF00FF to change the color/level (windows only)

for non-rvb standard nvidia devices, you can use a level value (0 to 100)

ps: i really need to find how to turn them off on linux, not found yet...

and rename RVB to RGB, french typo...

disabled by default, require --led=100 on nvidia compatible devices

Gigabyte RGB led hack is not perfect for the moment, can fail and require a reboot.

sample usage if GPU #0 is RVB and GPU #1 a generic led:

ccminer --led=0x00ff00,100,mining to put the led on while the gpus are scanning

+ cleanup...

rewrote almost properly ;)

1 MH/s reached on the 1070 ...

without that, no way to read sizes > 4GB on x86 binaries

Signed-off-by: Tanguy Pruvot <[email protected]>

sha512/ripemd swab

this was preview 2

drop ripemd.cu and remove unused kernels

also reduce the few errors on coins using shared mem

zotac cards doesnt allow a real RGB value, so i converted manually
the card seems to handle it, but seems to lock sometimes for some time...

colors can be tested with "ccminer -d 2 --led=0x1f1f00 -Dn"

but... unsure the card handle it fast enough to be fully usable

seems to be too slow to handle all wanted modes

big cleanup...

105 LBC tipped ;)

Signed-off-by: Tanguy Pruvot <[email protected]>

-o http://xxxx:port -u <addr> -p <worker>

Signed-off-by: Tanguy Pruvot <[email protected]>

also known as "Thor's Riddle"... yes sure ;)

Credits to ocminer who found and "implemented" it.

Note: tested "ok" on x64 and CUDA 6.5 x86, not on 7.5 and 8.0 x86

PS: Don't have the time for a more proper CUDA implementation of Streebog

Based on alexis78 work and sponsored by LBRY.IO team (thanks)

Release 1.8.2, use cuda 8 for x86

change was for the 1.8.2 tag (optimal for lbry)

and prepare next version...

build tip: autoreconf && make -j

+ veltor free
+ some missed/extra log things...

This will allow later more gpu candidates.

Note: This is an unfinished work, we keep the previous behavior for now
To finish this, all algos solutions should be migrated and submitted nonces attributes stored.
Its required to handle the different share diff per nonce and fix the possible solved count error (if 1/2 nonces is solved).

when 2 nonces were found, the next scan was not at the right value

Doesn't really affect mining performance...

The double loop is not useful, and prefer the __thread attribute
to enhance the code readability (remove the 2D host arrays).

squashed: return to host 2D array to allow the free

size_t can be a bit... imprevisible on x86

part 1: longpoll stuff (nanopool)

Colorful (and Inno3D) only set their vid, with an empty product id

From Alexis work, sib hash rate 200% also..

based on alexis work, with the new work->nonces

may be required for very fast algos, like blake2s

scratchpad delete fix and redownload, reduce rejects

(work in progress)

todo: fix jh cuda and wrong decimal diff (0xffff problem ?)

change default launch config to -l 32x16 to handle the 750 Ti better
not definitive, doing tests..

Signed-off-by: Tanguy Pruvot <[email protected]>

rpc2: handle properly secondary nonce(s) + api ping time fix

be sure to fully recompile, structures are changed

based on alexis monitoring thread idea, but which only use one thread

note: other api changes will come soon, related to that