Pages:
Author

Topic: [ANN] QQCoin- Scrypt Jane, ASIC Resistant, N-Factor, Multipool Resistant - page 17. (Read 187229 times)

newbie
Activity: 3
Merit: 0
Thank you! But it seems not helpful for my case. I tried all low/mid/high combinations of TC between 8192 and 32768 with lookup-gap between 2 and 8 while rawintensity was < 1000  Cry

As I use BAMT which is 32 Bit only 2GB of 8GB are available. When I find some spare time I'll setup a 64Bit system and see if that works better for scrypt-jane

Just throw in a 64 bit system.. 2 GB are never enough for scrypt-jane at higher N. You can try though TC of 4096 or 2048 to see which one is working. Raw intensity 1024/2048/4096 at lg 3 or higher. Then you just increase the TC until it will not start anymore. Then you go higher with R in increments of w (if w is 128 just add 128 each time).

You don't need much system RAM on Linux, 256-512MB should be enough if you are running some lite distribution such as Lubuntu. Windows drivers have a bug, it queries amount of system RAM instead of VRAM.

Below are my settings for 280X, about 135khs per card.

Code:
"intensity" : "12,12,12,12",
"vectors" : "1,1,1,1",
"worksize" : "256,256,256,256",
"kernel" : "scrypt-jane,scrypt-jane,scrypt-jane,scrypt-jane",
"lookup-gap" : "0,0,0,0",
"thread-concurrency" : "16384,16384,16384,16384",
"shaders" : "0,0,0,0",
"gpu-engine" : "0-1050,0-1050,0-1050,0-1050",
"gpu-fan" : "0-85,0-85,0-85,0-85",
"gpu-memclock" : "0,0,0,0",
"gpu-memdiff" : "0,0,0,0",
"gpu-powertune" : "0,0,0,0",
"gpu-vddc" : "0.000,0.000,0.000,0.000",
"temp-cutoff" : "95,95,95,95",
"temp-overheat" : "85,85,85,85",
"temp-target" : "75,75,75,75",
"gpu-reorder" : true,
"api-mcast-port" : "4028",
"api-port" : "4028",
"auto-fan" : true,
"expiry" : "120",
"gpu-dyninterval" : "7",
"gpu-platform" : "0",
"gpu-threads" : "2",
"gpu-reorder" : true,
"log" : "5",
"no-pool-disable" : true,
"queue" : "1",
"scan-time" : "60",
"scrypt" : true,
"scrypt-jane" : true,
"temp-hysteresis" : "3",
"shares" : "0",
"kernel-path" : "/usr/local/bin"


Are you getting no hardware errors with that ?
No HW errors on 4 Gigabyte and 4 Powercolor cards.
legendary
Activity: 2100
Merit: 1167
MY RED TRUST LEFT BY SCUMBAGS - READ MY SIG
i don't think people realise how rare 30M minting is.  The price for this coin is below value for a no premine coin which is this fast. I have liked it from the start. If a billion new coins a day were not coming out qqc will do way better.
full member
Activity: 210
Merit: 100
Thank you! But it seems not helpful for my case. I tried all low/mid/high combinations of TC between 8192 and 32768 with lookup-gap between 2 and 8 while rawintensity was < 1000  Cry

As I use BAMT which is 32 Bit only 2GB of 8GB are available. When I find some spare time I'll setup a 64Bit system and see if that works better for scrypt-jane

Just throw in a 64 bit system.. 2 GB are never enough for scrypt-jane at higher N. You can try though TC of 4096 or 2048 to see which one is working. Raw intensity 1024/2048/4096 at lg 3 or higher. Then you just increase the TC until it will not start anymore. Then you go higher with R in increments of w (if w is 128 just add 128 each time).

You don't need much system RAM on Linux, 256-512MB should be enough if you are running some lite distribution such as Lubuntu. Windows drivers have a bug, it queries amount of system RAM instead of VRAM.

Below are my settings for 280X, about 135khs per card.

Code:
"intensity" : "12,12,12,12",
"vectors" : "1,1,1,1",
"worksize" : "256,256,256,256",
"kernel" : "scrypt-jane,scrypt-jane,scrypt-jane,scrypt-jane",
"lookup-gap" : "0,0,0,0",
"thread-concurrency" : "16384,16384,16384,16384",
"shaders" : "0,0,0,0",
"gpu-engine" : "0-1050,0-1050,0-1050,0-1050",
"gpu-fan" : "0-85,0-85,0-85,0-85",
"gpu-memclock" : "0,0,0,0",
"gpu-memdiff" : "0,0,0,0",
"gpu-powertune" : "0,0,0,0",
"gpu-vddc" : "0.000,0.000,0.000,0.000",
"temp-cutoff" : "95,95,95,95",
"temp-overheat" : "85,85,85,85",
"temp-target" : "75,75,75,75",
"gpu-reorder" : true,
"api-mcast-port" : "4028",
"api-port" : "4028",
"auto-fan" : true,
"expiry" : "120",
"gpu-dyninterval" : "7",
"gpu-platform" : "0",
"gpu-threads" : "2",
"gpu-reorder" : true,
"log" : "5",
"no-pool-disable" : true,
"queue" : "1",
"scan-time" : "60",
"scrypt" : true,
"scrypt-jane" : true,
"temp-hysteresis" : "3",
"shares" : "0",
"kernel-path" : "/usr/local/bin"


Are you getting no hardware errors with that ?
newbie
Activity: 3
Merit: 0
Thank you! But it seems not helpful for my case. I tried all low/mid/high combinations of TC between 8192 and 32768 with lookup-gap between 2 and 8 while rawintensity was < 1000  Cry

As I use BAMT which is 32 Bit only 2GB of 8GB are available. When I find some spare time I'll setup a 64Bit system and see if that works better for scrypt-jane

Just throw in a 64 bit system.. 2 GB are never enough for scrypt-jane at higher N. You can try though TC of 4096 or 2048 to see which one is working. Raw intensity 1024/2048/4096 at lg 3 or higher. Then you just increase the TC until it will not start anymore. Then you go higher with R in increments of w (if w is 128 just add 128 each time).

You don't need much system RAM on Linux, 256-512MB should be enough if you are running some lite distribution such as Lubuntu. Windows drivers have a bug, it queries amount of system RAM instead of VRAM.

Below are my settings for 280X, about 135khs per card.

Code:
"intensity" : "12,12,12,12",
"vectors" : "1,1,1,1",
"worksize" : "256,256,256,256",
"kernel" : "scrypt-jane,scrypt-jane,scrypt-jane,scrypt-jane",
"lookup-gap" : "0,0,0,0",
"thread-concurrency" : "16384,16384,16384,16384",
"shaders" : "0,0,0,0",
"gpu-engine" : "0-1050,0-1050,0-1050,0-1050",
"gpu-fan" : "0-85,0-85,0-85,0-85",
"gpu-memclock" : "0,0,0,0",
"gpu-memdiff" : "0,0,0,0",
"gpu-powertune" : "0,0,0,0",
"gpu-vddc" : "0.000,0.000,0.000,0.000",
"temp-cutoff" : "95,95,95,95",
"temp-overheat" : "85,85,85,85",
"temp-target" : "75,75,75,75",
"gpu-reorder" : true,
"api-mcast-port" : "4028",
"api-port" : "4028",
"auto-fan" : true,
"expiry" : "120",
"gpu-dyninterval" : "7",
"gpu-platform" : "0",
"gpu-threads" : "2",
"gpu-reorder" : true,
"log" : "5",
"no-pool-disable" : true,
"queue" : "1",
"scan-time" : "60",
"scrypt" : true,
"scrypt-jane" : true,
"temp-hysteresis" : "3",
"shares" : "0",
"kernel-path" : "/usr/local/bin"
full member
Activity: 210
Merit: 100
Can you just stop dumping .. whoever mines big amount ??
This coin needs to go high. If you never stop dumping you will never see it high !
sr. member
Activity: 425
Merit: 250
I just tried buying more on coinedup and it says trading has been temporarily halted on this market.  Huh

same, we need another exchange, get voting on mintpal everyone.... you can vote 3x per hour and don't even need to be registered there. Tell your friends to also.

qqc on https://www.allcoin.com/trade/qqc_btc too
full member
Activity: 143
Merit: 100
quickpay
news: allcrypt  exChange https://www.allcrypt.com accept QQC/BTC QQC/LTC QQC/DOGE
A new web site ,about this exchange We are still in the query details.
full member
Activity: 143
Merit: 100
quickpay
community also can join the development ,as coding as you can  Grin
legendary
Activity: 2100
Merit: 1167
MY RED TRUST LEFT BY SCUMBAGS - READ MY SIG
I just tried buying more on coinedup and it says trading has been temporarily halted on this market.  Huh

same, we need another exchange, get voting on mintpal everyone.... you can vote 3x per hour and don't even need to be registered there. Tell your friends to also.
legendary
Activity: 2632
Merit: 2790
Shitcoin Minimalist
I just tried buying more on coinedup and it says trading has been temporarily halted on this market.  Huh
full member
Activity: 210
Merit: 100
Community is in China.
Heard pumping is coming soon, everybody should buy this coin as it is still so much undervalued, take a look at what happened to VTC.
full member
Activity: 133
Merit: 100
Does this coin have any active developers, community or founding?
full member
Activity: 143
Merit: 100
quickpay
i7=150khs
955=100Khs
use the CPU mine
These are only rough mining specs.

when you look here ↗Welcome to the QQC community Grin

I think you are wrong. More like 10-15 kh/sec for that CPU as of now, not 150 kh/sec

10-15 ?  my cpu is AMD 955 so I went to give it a try
full member
Activity: 210
Merit: 100
i7=150khs
955=100Khs
use the CPU mine
These are only rough mining specs.

when you look here ↗Welcome to the QQC community Grin

I think you are wrong. More like 10-15 kh/sec for that CPU as of now, not 150 kh/sec
full member
Activity: 143
Merit: 100
quickpay
i7=150khs
955=100Khs
use the CPU mine
These are only rough mining specs.

when you look here ↗Welcome to the QQC community Grin
full member
Activity: 210
Merit: 100
Thank you! But it seems not helpful for my case. I tried all low/mid/high combinations of TC between 8192 and 32768 with lookup-gap between 2 and 8 while rawintensity was < 1000  Cry

As I use BAMT which is 32 Bit only 2GB of 8GB are available. When I find some spare time I'll setup a 64Bit system and see if that works better for scrypt-jane

Just throw in a 64 bit system.. 2 GB are never enough for scrypt-jane at higher N. You can try though TC of 4096 or 2048 to see which one is working. Raw intensity 1024/2048/4096 at lg 3 or higher. Then you just increase the TC until it will not start anymore. Then you go higher with R in increments of w (if w is 128 just add 128 each time).
hero member
Activity: 984
Merit: 1000
Thank you! But it seems not helpful for my case. I tried all low/mid/high combinations of TC between 8192 and 32768 with lookup-gap between 2 and 8 while rawintensity was < 1000  Cry

As I use BAMT which is 32 Bit only 2GB of 8GB are available. When I find some spare time I'll setup a 64Bit system and see if that works better for scrypt-jane
full member
Activity: 210
Merit: 100
These are two settings I use for the R9 290 and working at n-factor of 11:

1st Computer with R9 290
-------------------------------------------------
setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS 1
yacminer -o x.x.x.x:20112 -u lumberinvestments -p qdwq2133s --scrypt --nfmin 4 --nfmax 30 --starttime 1387769316 --lookup-gap 2 --thread-concurrency 28600 -w 128 -R 6528 -g 1 --expiry 10 --gpu-engine 1070 --gpu-memclock 1500 --gpu-powertune 20,20 --auto-fan --scan-time 1
---------------------------------------------------

Second computer with R9 290 GPU 0 and R9 270x with 4 gigs of memory as GPU 1
-------------------------------------------------
setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS 1
yacminer -o x.x.x.x:20112 -u lumberinvestments -p lsdq22oos --scrypt --nfmin 4 --nfmax 30 --starttime 1387769316 --lookup-gap 3,2 --thread-concurrency 42000,28600 -w 128 -R 9728,10240 -g 1 --auto-fan --scan-time 1 --expiry 10 --gpu-engine 1070 --gpu-powertune 20
-------------------------------------------------

In this second computer I get almost 130 kh/sec with no HW for the R9 290 and almost 90 kh/sec for the 4 Gb R9 270x
You can see that the hashing speed differences between the cards start to get lower as the n-factor goes up.
In no way in windows you will be able to access all 4 Gb of the card, just download HWinfo and use it, look at the dedicated and dynamic memory of the videocard.
The higher the lookup-gap the higher thread concurrency you can go. Beware that there is a point when you increase the raw intensity and your hashrate will drop dramatically. That is the sweet spot. First you set up lookup-gap and thread concurrency until you can't go up anymore with thread concurrency. Have a low raw intensity. Then you increase it slowly and you will see your hashrate increasing, but stop when you will see the HW or sudden drop in hashrate.

Everything depends on your other configuration, like system RAM. If you have too little, you will not be able to go so high TC. I use 4 GB of system Ram and I assume I might be able to get higher if I put 8 GB total.

If this has been proven useful then you can donate to me some QQC and I will donate further share to Thirtybird Smiley

Here is my QQC address: QhACLXR1SbQu354aVJWf5EYYS8dPfFgVKN
We need to raise QQC to the sky by increasing the hashrate !
Have a nice day !
hero member
Activity: 984
Merit: 1000
Hm, I'm now fiddeling with the YACminer settings for hours and still didn't get at least one accepted share - only HW errors.

The README says the following settings are important:

- rawintensity
- thread-concurrency
- lookup-gap

The formular for the OpenCL Buffer allocated per thread is:
MB per thread = thread-concurrency * 128 / (1024 * lookup-gap)

My R9 290 has 4096MB, so the optimal values should be:
TC: 65536
LG: 2

But then the card is DEAD right from the beginning. Since this might just be a performance optimization, I go for now with TC=16384 and LG=4.


Further, the rawintensity seems to be important to avoid HW errors.
Code:
Valid values for raw-intensity need to be in multiples of --worksize.  Cards in the
R7 and R9 family may allow for adjustment in smaller increments - the lowest found
has been adjustment in multiples of 8.  The following is a guideline on where to
start with your card and will most likely not be the exact numbers you end up with.

Memory # of shaders LG Raw Intensity
1GB Low Shaders 4 Shaders*4
2GB Low Shaders 2 Shaders*2
2GB Low Shaders 4 Shaders*8
1GB High Shaders 8 Shaders*1
2GB High Shaders 4 Shaders*2

Adjust Raw Intensity up and/or down until you no longer receive HW errors in the GUI.

The R9 290 has 2560 shaders. Lookup-Gap is 4, so I start with a rawintensity of 5120 and went down to 128 in steps of my worksize (128). But nothing. It only influences the speed of the HW errors coming in.  Huh

Scrypt-Jane is kind of rocket science. Please help! What are your settings for GPU mining this?
hero member
Activity: 984
Merit: 1000
Thanks for all your responses regarding linux mining! Smiley

I finally got Thirtybirds build compiled (with his help) and now just need to get rid of the HW errors. Seems like scrypt-jane is much harder to configure than scrypt.

https://github.com/Thirtybird/YACMiner/issues/4
Pages:
Jump to: