I can understand pascal fix but other than that, what is the point for bbr-ccminer?
Dual mining.. Bbr+decred, bbr+lbry, bbr+pascal, bbr+sia
|
|
|
DASH is $78. I used to own a masternode when dash was down at $5... 1000 dash. Would had been worth $78,000 today... ![Sad](https://bitcointalk.org/Smileys/default/sad.gif)
|
|
|
yeah they are pumping everything? more people into crypto maybe? well fuck fiat world i hope everyone come abroad in the crypto world, if that happen you would see at least 10x current profit if not 100...
Thats why you should keep some of the coins you mine. If you always dump, you will miss profit..
|
|
|
Prices are going skyhigh now..
ethereum $27.5(+21.5%) Zcash $46.5 (+20%) monero $17.55(+6%)
etc...
|
|
|
a joke there are coin that are giving me $6 per day currently
But this coin cannot take 100 rigs ![Wink](https://bitcointalk.org/Smileys/default/wink.gif) Wich coin btw?
|
|
|
Decred $5.74 (+62.89%)
($2.42 per day on the gtx 1070)
|
|
|
That's why I thought it was weird when Pallas was talking about having to mess with random read/writes. It's basically the same talk we got when Pascal first came out, then you guys figured out what to tweak and on almost every algo we ended up with the normal speeds, which is a 1070 is about 80% faster then a 970 and a 1060 is almost the same speed.
The 970 and 1070 both have a 256bit bus. so the difference will be smaller. It would be similar to mining Ethereum with a small dagger size. The gtx 970 does around 23MHASH, and the 1070 does around 30MHASH. (the gtx 1070 is only around 30% faster than the gtx 970 in the ethereum algo) So if the gtx 970 does 1400 in boolberry, the 1070 should do around 1820. My modded kernel is doing 1860..
|
|
|
I wasn't talking about the scratchpad. If you ran the miner on a 970 it got 1400.
Gpu #3 is running @ 1025, is it a bug or is this gpu clocked lower than the others? (Asus gpu #0 seems to be doing fine) Boolberry need a good coreclock/memclock ratio. from 1000 to 1400 is a 40% increase. Adjusting mem/core clock, and finding the optimal launchconfig is the key here. Modding the code is not enough...
|
|
|
Original miner got 1400 on a 970, 2500 on a 1070 is about right with almost no changes to the code and updating it to Pascal.
But ccminer 2.0 only does 1500 and not 2500 on the gtx 1070. The scratchpad is 230MB now. In the early days (2014), the scratchpad was smaller, and you got more cachehits in the random access and more speed. The gtx 970 is doing around 1000. ccminer v2.0 RC2 opensource - Boolberry Palit GTX 1070 Core Clock +0 - 1500 kh/s Core Clock +60 - 1800 kh/s (stable work) ![](https://ip.bitcointalk.org/?u=https%3A%2F%2Fpp.userapi.com%2Fc626129%2Fv626129062%2F4bfe3%2FNO60T7zeQPc.jpg&t=663&c=JsLV13NcdVXOpw)
|
|
|
3000mhash ( $2.17) per day
|
|
|
Decred $5.01 (+45.37%)...
|
|
|
What does the opensource miner do?
ccminer 2.0 opensource does 850khash on the gtx 1060 3gb. sp-mod #1 is 60% faster..
|
|
|
[2017-03-10 13:46:57] GPU #2: GeForce GTX 1070: 34603008 hashes, 1596.13 kh/s [2017-03-10 13:46:57] accepted: 19/20 (95.00%), 4255 khash/s (yay!!!) [2017-03-10 13:47:08] GPU #2: GeForce GTX 1070: 16777216 hashes, 1595.91 kh/s [2017-03-10 13:47:08] accepted: 20/21 (95.24%), 4255 khash/s (yay!!!) [2017-03-10 13:47:12] GPU #1: GeForce GTX 1070: 57671680 hashes, 1595.77 kh/s [2017-03-10 13:47:12] accepted: 21/22 (95.45%), 4782 khash/s (yay!!!) [2017-03-10 13:47:16] GPU #1: GeForce GTX 1070: 6291456 hashes, 1596.09 kh/s [2017-03-10 13:47:16] accepted: 22/23 (95.65%), 4783 khash/s (yay!!!) [2017-03-10 13:47:31] GPU #2: GeForce GTX 1070: 36700160 hashes, 1595.99 kh/s [2017-03-10 13:47:31] accepted: 23/24 (95.83%), 4783 khash/s (yay!!!) [2017-03-10 13:47:37] GPU #0: GeForce GTX 1070: 95420417 hashes, 1592.41 kh/s
TDP 100W. slight code modification.
Is this with overclocking?
|
|
|
I don't think I will sell this kernel. Bether to harvest with my farm... nice speed on the 1060 3gb. the 1070 need some more work..
What does the opensource miner do?
|
|
|
Boolberry sp-mod #1 early results.. 1070 1850Khash/s (60% tdp) 1060 3gb 1360Khash/s (60% tdp) ![](https://ip.bitcointalk.org/?u=http%3A%2F%2Fi65.tinypic.com%2Fso7f51.png&t=663&c=vCr3LANSLcTFUQ)
|
|
|
Found a bug.
If you mine boolberry and the scratchpad needs to be downloaded on a slow internet connection, ccminer will exit without any error messages after a few minutes..
|
|
|
Did you really try the last one ? And to answer to Super Parasite... The hard part of wildkeccak is not in the kernel... me likey name Super Parasite you really hit the nail with this, will use from now, everyone should ![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif) and pliiz don't give up with developing, i will cry ![Sad](https://bitcointalk.org/Smileys/default/sad.gif) I will opensource my parasite donation mining sourcecode, with a binary kernel launcher.. ccminer 2.0 sp-mod opensource with binary assembly kernels. He-he
|
|
|
So many words without purpose..
chrysophylax has moved to the other thread. I did a vector instruction rewrite of the boolberry kernel. Nice speedup. I am modding ccminer to load cubin kernels directly. Then I can opensource ccminer, and sell bin files. The binfiles will work on linux too. No need for exe files..
|
|
|
it ALWAYS matters - and is ALWAYS important how the optimizations are done - and HOW you present them ... as this proves the way you work - and proves your character as well ...
Why should I show your development team of dickheads how to do boolberry faster. They can disassemble my work and understand what I did.. boolberry sp-mod #1 0.05btc available today.. hahaha ... see - dickheads ... back atcha mate ... you have the optimizations of OTHER devs work already in your hands ... and all you do is tweak ... eitherway - its fine ... a dickhead is a dickhead is a dickhead - that calls everyone else a dickhead but himself ... ive been through it before with you mate - and burned no more from it ... what a pity ... #crysx Here is the Aussie version: Why should I teach your development team of cunts how to do boolberry faster. They can disassemble my work and understand what I did.. boolberry sp-mod #1 0.05btc available today.. only a fuckwit like you - can say such things ... only a fuckwit like you - can copy code and call it his own ... only a fuckwit like you - can be the fuckwit of all fuckwits ... hows that for language YOU understand ... fuckwit cunt ![Wink](https://bitcointalk.org/Smileys/default/wink.gif) ... #crysx Here is the british version: Why should I teach your development team of wankers how to do boolberry faster. They can disassemble my work and understand what I did.. boolberry sp-mod #1 0.05btc available today..
|
|
|
it ALWAYS matters - and is ALWAYS important how the optimizations are done - and HOW you present them ... as this proves the way you work - and proves your character as well ...
Why should I show your development team of dickheads how to do boolberry faster. They can disassemble my work and understand what I did.. boolberry sp-mod #1 0.05btc available today.. hahaha ... see - dickheads ... back atcha mate ... you have the optimizations of OTHER devs work already in your hands ... and all you do is tweak ... eitherway - its fine ... a dickhead is a dickhead is a dickhead - that calls everyone else a dickhead but himself ... ive been through it before with you mate - and burned no more from it ... what a pity ... #crysx Here is the Aussie version: Why should I teach your development team of cunts how to do boolberry faster. They can disassemble my work and understand what I did.. boolberry sp-mod #1 0.05btc available today..
|
|
|
|