If Your Holding UltraCoin for the longterm!
oh yea
Topic: [ANN] [ASIC-RESISTANT] UltraCoin (UTC) - Ultrafast 6 second transactions!! - page 182. (Read 946659 times)
May 02, 2014, 04:53:05 PM
oh yea
May 02, 2014, 04:39:46 PM
If Your Holding UltraCoin for the longterm!
May 02, 2014, 04:01:54 PM
Quote
It's not a false rumor and has been proven here many times. Thirtybird has explained it better than I can a few times in this thread, but the more cards you have, the more RAM you need to run it with a low lookup gap. My lack of system RAM only allows me to get about 80% out of each 290x I have because I can't run them with the higher lookup gap without one card refusing to start.
Also, you assume all new miners are using low end gear. What if a new miner comes in like me with 4x 290x cards and only 8 GB system RAM and has the same issues I do? I was already at a 25% profitability disadvantage against scrypt coins because of this, and that was before the orphan problem. It does work out good for CPU miners and 750 Ti miners, but it doesn't hold true for everyone.
That will be true with ultracoin miner. But yacminer, with its fine intensity tweaking should give you the same results on whatever lookup gap you mine.
This right here is why I've advocated people using the "buffer-size" setting instead of thread-concurrency - it confuses what people think the problem may be. They both will accomplish the same things - allocating the OpenCL buffer for the card - but thread-concurrency is factored on lookup-gap, which gets fiddly when trying to test config changes. buffer-size just says "I'm allocating this much memory" and the rest of the settings work with that to do their hashing. So, it's not a lookup-gap problem that's preventing you from maxxing out the cards.
System RAM can be important, depending on your config. If you've only got 4GB of system memory, you're going to be hurting yourself if you try to work with _any_ cards that have 3GB or 4GB of VRAM. Is it possible, absolutely. Is it easy, hell no. I've examined Beave's setup quite some time ago - he's banged it out pretty good - high effeciency with low cost and decent hashpower. He's gone with a large number of 2GB cards per system, and absolutely can run 6 x 2GB cards on just 4 GB of system RAM. I know it can work, and I personally run a 4 x 2GB system on Windows on just 4GB RAM. My other mining setup uses 4GB cards, and needed 8GB of system ram to go beyond 2 cards.
Now, with that said, you've got R9 290x's with (guessing) 4 GB of GPU memory. It is absolutely quite possible that 8GB is not enough to allocate 3.6GB per card on just 8 GB of system memory. I can allocate 3GB per card, but I don't have as many shaders on my R7 240's to keep busy as you do so even at N=14 I don't need to go higher. The efficiency of your card will suffer by trying to cram more work to those shaders as you crank up the lookup-gap. One nice thing is, so long as you can allocate as much memory as I can on your card (--buffer-size 3040), you can still get ~ 3KH/sec by running the same lookup-gap (2) and a reduced raw intensity (-R 1280). From there, any additional memory you can allocate will give you a higher -R to use. The key is to find out how high of a buffer-size you can allocate on each card before you have problems. On my system with 4 x 2GB cards and 4GB, I am allocating a buffer-size of 1753 for each card - any higher than that, and I will get cards fail to start up. This is also pushing the limits of what can be allocated per card anyway (from single card testing), so I'm happy to be there
Configuring miners for high end cards on high NFactor coins is hard. I don't think anyone can say it's not intimidating for new miners, or experienced miners alike. I've tried to put as much of my knowledge and findings out there as possible, but not everyone has the patience to read it all or the time to test a myriad of settings. I've got something in the testing fork of YACMiner that will make finding working settings much easer for new miners, but I've got some issues with timing and the multiple threads that I haven't had time to look into solving.
We actually went over this when I posted the problem over a month ago. I may have changed it to just messing with the buffer-size. I'm not currently mining it, nor have I touched my settings in over a month, so I dont' know what I settled on, I only remember that it was still only about 80% of what others could get, and even what I could get if I only left 3 cards active. The point I was making still stands, changing the N-factor can make it a pain for people who aren't all geeked out over this and reading thread posts all day about problems people are having. Thanks for the information assist again Thirtybird.
Well, with work comes reward. If people are not willing to put some man hours into it, then they are better off buying coins.
May 02, 2014, 04:01:06 PM
4GB cards, ThirtyBird, 6 x 4GB cards!!! The point is... NFactor does not effect system RAM. ThirtyBird, you have windows, and when I used windows, I remember needing more than 4GB of system RAM.
I can tell you, using xubunutu OS, buffersize is not a factor for YAC for my 4GB cards at NFactor 14. Obviously, it will be an issue for subsequent NFactors. But a buffersize of 3500 vs 3750 yields no different results at this time.
I don't know how the buffer loads, but could it have something to do with the controller only being 128 bit on r7? You can't push near the data down the pipe that I can with an r9 with a 512 bit controller. Or does it purely have to do with the different shader counts? Are the two inter-related? I'd need more tech support again, that's beyond my talents.
May 02, 2014, 03:56:38 PM
Hi all,
has somebody experience with KNCminer ?
Did some of you buy or ordered a asic with them ?
has somebody experience with KNCminer ?
Did some of you buy or ordered a asic with them ?
May 02, 2014, 03:51:22 PM
Quote
It's not a false rumor and has been proven here many times. Thirtybird has explained it better than I can a few times in this thread, but the more cards you have, the more RAM you need to run it with a low lookup gap. My lack of system RAM only allows me to get about 80% out of each 290x I have because I can't run them with the higher lookup gap without one card refusing to start.
Also, you assume all new miners are using low end gear. What if a new miner comes in like me with 4x 290x cards and only 8 GB system RAM and has the same issues I do? I was already at a 25% profitability disadvantage against scrypt coins because of this, and that was before the orphan problem. It does work out good for CPU miners and 750 Ti miners, but it doesn't hold true for everyone.
That will be true with ultracoin miner. But yacminer, with its fine intensity tweaking should give you the same results on whatever lookup gap you mine.
This right here is why I've advocated people using the "buffer-size" setting instead of thread-concurrency - it confuses what people think the problem may be. They both will accomplish the same things - allocating the OpenCL buffer for the card - but thread-concurrency is factored on lookup-gap, which gets fiddly when trying to test config changes. buffer-size just says "I'm allocating this much memory" and the rest of the settings work with that to do their hashing. So, it's not a lookup-gap problem that's preventing you from maxxing out the cards.
System RAM can be important, depending on your config. If you've only got 4GB of system memory, you're going to be hurting yourself if you try to work with _any_ cards that have 3GB or 4GB of VRAM. Is it possible, absolutely. Is it easy, hell no. I've examined Beave's setup quite some time ago - he's banged it out pretty good - high effeciency with low cost and decent hashpower. He's gone with a large number of 2GB cards per system, and absolutely can run 6 x 2GB cards on just 4 GB of system RAM. I know it can work, and I personally run a 4 x 2GB system on Windows on just 4GB RAM. My other mining setup uses 4GB cards, and needed 8GB of system ram to go beyond 2 cards.
Now, with that said, you've got R9 290x's with (guessing) 4 GB of GPU memory. It is absolutely quite possible that 8GB is not enough to allocate 3.6GB per card on just 8 GB of system memory. I can allocate 3GB per card, but I don't have as many shaders on my R7 240's to keep busy as you do so even at N=14 I don't need to go higher. The efficiency of your card will suffer by trying to cram more work to those shaders as you crank up the lookup-gap. One nice thing is, so long as you can allocate as much memory as I can on your card (--buffer-size 3040), you can still get ~ 3KH/sec by running the same lookup-gap (2) and a reduced raw intensity (-R 1280). From there, any additional memory you can allocate will give you a higher -R to use. The key is to find out how high of a buffer-size you can allocate on each card before you have problems. On my system with 4 x 2GB cards and 4GB, I am allocating a buffer-size of 1753 for each card - any higher than that, and I will get cards fail to start up. This is also pushing the limits of what can be allocated per card anyway (from single card testing), so I'm happy to be there
Configuring miners for high end cards on high NFactor coins is hard. I don't think anyone can say it's not intimidating for new miners, or experienced miners alike. I've tried to put as much of my knowledge and findings out there as possible, but not everyone has the patience to read it all or the time to test a myriad of settings. I've got something in the testing fork of YACMiner that will make finding working settings much easer for new miners, but I've got some issues with timing and the multiple threads that I haven't had time to look into solving.
We actually went over this when I posted the problem over a month ago. I may have changed it to just messing with the buffer-size. I'm not currently mining it, nor have I touched my settings in over a month, so I dont' know what I settled on, I only remember that it was still only about 80% of what others could get, and even what I could get if I only left 3 cards active. The point I was making still stands, changing the N-factor can make it a pain for people who aren't all geeked out over this and reading thread posts all day about problems people are having. Thanks for the information assist again Thirtybird.
May 02, 2014, 03:44:02 PM
Quote
It's not a false rumor and has been proven here many times. Thirtybird has explained it better than I can a few times in this thread, but the more cards you have, the more RAM you need to run it with a low lookup gap. My lack of system RAM only allows me to get about 80% out of each 290x I have because I can't run them with the higher lookup gap without one card refusing to start.
Also, you assume all new miners are using low end gear. What if a new miner comes in like me with 4x 290x cards and only 8 GB system RAM and has the same issues I do? I was already at a 25% profitability disadvantage against scrypt coins because of this, and that was before the orphan problem. It does work out good for CPU miners and 750 Ti miners, but it doesn't hold true for everyone.
That will be true with ultracoin miner. But yacminer, with its fine intensity tweaking should give you the same results on whatever lookup gap you mine.
This right here is why I've advocated people using the "buffer-size" setting instead of thread-concurrency - it confuses what people think the problem may be. They both will accomplish the same things - allocating the OpenCL buffer for the card - but thread-concurrency is factored on lookup-gap, which gets fiddly when trying to test config changes. buffer-size just says "I'm allocating this much memory" and the rest of the settings work with that to do their hashing. So, it's not a lookup-gap problem that's preventing you from maxxing out the cards.
System RAM can be important, depending on your config. If you've only got 4GB of system memory, you're going to be hurting yourself if you try to work with _any_ cards that have 3GB or 4GB of VRAM. Is it possible, absolutely. Is it easy, hell no. I've examined Beave's setup quite some time ago - he's banged it out pretty good - high effeciency with low cost and decent hashpower. He's gone with a large number of 2GB cards per system, and absolutely can run 6 x 2GB cards on just 4 GB of system RAM. I know it can work, and I personally run a 4 x 2GB system on Windows on just 4GB RAM. My other mining setup uses 4GB cards, and needed 8GB of system ram to go beyond 2 cards.
Now, with that said, you've got R9 290x's with (guessing) 4 GB of GPU memory. It is absolutely quite possible that 8GB is not enough to allocate 3.6GB per card on just 8 GB of system memory. I can allocate 3GB per card, but I don't have as many shaders on my R7 240's to keep busy as you do so even at N=14 I don't need to go higher. The efficiency of your card will suffer by trying to cram more work to those shaders as you crank up the lookup-gap. One nice thing is, so long as you can allocate as much memory as I can on your card (--buffer-size 3040), you can still get ~ 3KH/sec by running the same lookup-gap (2) and a reduced raw intensity (-R 1280). From there, any additional memory you can allocate will give you a higher -R to use. The key is to find out how high of a buffer-size you can allocate on each card before you have problems. On my system with 4 x 2GB cards and 4GB, I am allocating a buffer-size of 1753 for each card - any higher than that, and I will get cards fail to start up. This is also pushing the limits of what can be allocated per card anyway (from single card testing), so I'm happy to be there
Configuring miners for high end cards on high NFactor coins is hard. I don't think anyone can say it's not intimidating for new miners, or experienced miners alike. I've tried to put as much of my knowledge and findings out there as possible, but not everyone has the patience to read it all or the time to test a myriad of settings. I've got something in the testing fork of YACMiner that will make finding working settings much easer for new miners, but I've got some issues with timing and the multiple threads that I haven't had time to look into solving.
4GB cards, ThirtyBird, 6 x 4GB cards!!! The point is... NFactor does not effect system RAM. ThirtyBird, you have windows, and when I used windows, I remember needing more than 4GB of system RAM.
I can tell you, using xubunutu OS, buffersize is not a factor for YAC for my 4GB cards at NFactor 14. Obviously, it will be an issue for subsequent NFactors. But a buffersize of 3500 vs 3750 yields no different results at this time.
May 02, 2014, 03:38:09 PM
Sometime in the future. Im going to need some UltraCoin brochures made up. So I can put a street team together to pass them out along with my mixtape.
You can send a pm to Bumface, Jakoblight or me and we can provide you with any materials you need. And you should work with Jakoblight, he's our video producer.
ok thanks dcGirl. :-)
May 02, 2014, 03:35:38 PM
I'll be doing promotions on my twitter all month. Feel free to donate. UTC address in my sig. Shoutout to everyone who has donated so far. I had about 1400 when I first joined you guys now im up to 4000 UTC. I love this community and you guys will have my support forever. Hmm If I ever get an award with my musical efforts. I might have to walk on live tv with a UltraCoin t-shirt on while accepting my award.. haha
May 02, 2014, 03:34:56 PM
Sometime in the future. Im going to need some UltraCoin brochures made up. So I can put a street team together to pass them out along with my mixtape.
You can send a pm to Bumface, Jakoblight or me and we can provide you with any materials you need. And you should work with Jakoblight, he's our media/video producer.
May 02, 2014, 03:26:19 PM
Sometime in the future. Im going to need some UltraCoin brochures made up. So I can put a street team together to pass them out along with my mixtape.
May 02, 2014, 03:24:19 PM
Quote
It's not a false rumor and has been proven here many times. Thirtybird has explained it better than I can a few times in this thread, but the more cards you have, the more RAM you need to run it with a low lookup gap. My lack of system RAM only allows me to get about 80% out of each 290x I have because I can't run them with the higher lookup gap without one card refusing to start.
Also, you assume all new miners are using low end gear. What if a new miner comes in like me with 4x 290x cards and only 8 GB system RAM and has the same issues I do? I was already at a 25% profitability disadvantage against scrypt coins because of this, and that was before the orphan problem. It does work out good for CPU miners and 750 Ti miners, but it doesn't hold true for everyone.
That will be true with ultracoin miner. But yacminer, with its fine intensity tweaking should give you the same results on whatever lookup gap you mine.
This right here is why I've advocated people using the "buffer-size" setting instead of thread-concurrency - it confuses what people think the problem may be. They both will accomplish the same things - allocating the OpenCL buffer for the card - but thread-concurrency is factored on lookup-gap, which gets fiddly when trying to test config changes. buffer-size just says "I'm allocating this much memory" and the rest of the settings work with that to do their hashing. So, it's not a lookup-gap problem that's preventing you from maxxing out the cards.
System RAM can be important, depending on your config. If you've only got 4GB of system memory, you're going to be hurting yourself if you try to work with _any_ cards that have 3GB or 4GB of VRAM. Is it possible, absolutely. Is it easy, hell no. I've examined Beave's setup quite some time ago - he's banged it out pretty good - high effeciency with low cost and decent hashpower. He's gone with a large number of 2GB cards per system, and absolutely can run 6 x 2GB cards on just 4 GB of system RAM. I know it can work, and I personally run a 4 x 2GB system on Windows on just 4GB RAM. My other mining setup uses 4GB cards, and needed 8GB of system ram to go beyond 2 cards.
Now, with that said, you've got R9 290x's with (guessing) 4 GB of GPU memory. It is absolutely quite possible that 8GB is not enough to allocate 3.6GB per card on just 8 GB of system memory. I can allocate 3GB per card, but I don't have as many shaders on my R7 240's to keep busy as you do so even at N=14 I don't need to go higher. The efficiency of your card will suffer by trying to cram more work to those shaders as you crank up the lookup-gap. One nice thing is, so long as you can allocate as much memory as I can on your card (--buffer-size 3040), you can still get ~ 3KH/sec by running the same lookup-gap (2) and a reduced raw intensity (-R 1280). From there, any additional memory you can allocate will give you a higher -R to use. The key is to find out how high of a buffer-size you can allocate on each card before you have problems. On my system with 4 x 2GB cards and 4GB, I am allocating a buffer-size of 1753 for each card - any higher than that, and I will get cards fail to start up. This is also pushing the limits of what can be allocated per card anyway (from single card testing), so I'm happy to be there
Configuring miners for high end cards on high NFactor coins is hard. I don't think anyone can say it's not intimidating for new miners, or experienced miners alike. I've tried to put as much of my knowledge and findings out there as possible, but not everyone has the patience to read it all or the time to test a myriad of settings. I've got something in the testing fork of YACMiner that will make finding working settings much easer for new miners, but I've got some issues with timing and the multiple threads that I haven't had time to look into solving.
May 02, 2014, 03:23:39 PM
bikini's, thongs, casual clothes. But overall I let the women decide. haha If they want to do a nudey it's up to them. lmao
bikini's, thongs or nude all sounds good to me
Hmmm it is (very) tempting guys.... but think clear....
I am no moral knight but I don't believe this is the right way we are sailing with UTC
yea we just messing about Dcgirl would kick our ass
Shut up, trolls. It's called 'Money Maker' for a reason.
May 02, 2014, 03:22:50 PM
That will build up hype, or get random folks to stare at the pics...don't think they'll buy UTC or use it. :p
Use yacminer, I'd suggest using 8gb of ram if using windows and trying these settings:
--scrypt-chacha --worksize 256 -g 1 --lookup-gap 4 --buffer-size 2560 -R 8192 -o stratum+tcp://stratum.ultracoinpool.info:3333 -u xxxx -p xxx --nfmin 4 --nfmax 30 --starttime 1388361600
That's a good starting point, from there you can try increasing the -R value gradually till you get HW errors, then reduce it till they go away.
no one to help me any one can help me i have this setup --scrypt-jane --worksize 128 -g 2 --lookup-gap 3 --thread-concurrency 41440 -I 11 -o stratum+tcp://stratum.ultracoinpool.info:3333 -u xxxx -p xxx --sj-nfmin 4 --sj-nfmax 30 --sj-time 1388361600
i have r9 280 and when i launch ultracoinminer.exe i can see an error but can't read he shut down to fast i have pick my estup here http://ultracoin.net/configgen_raw.html
any one can help me i have this setup --scrypt-jane --worksize 128 -g 2 --lookup-gap 3 --thread-concurrency 41440 -I 11 -o stratum+tcp://stratum.ultracoinpool.info:3333 -u xxxx -p xxx --sj-nfmin 4 --sj-nfmax 30 --sj-time 1388361600i have r9 280 and when i launch ultracoinminer.exe i can see an error but can't read he shut down to fast i have pick my estup here http://ultracoin.net/configgen_raw.html
Use yacminer, I'd suggest using 8gb of ram if using windows and trying these settings:
--scrypt-chacha --worksize 256 -g 1 --lookup-gap 4 --buffer-size 2560 -R 8192 -o stratum+tcp://stratum.ultracoinpool.info:3333 -u xxxx -p xxx --nfmin 4 --nfmax 30 --starttime 1388361600
That's a good starting point, from there you can try increasing the -R value gradually till you get HW errors, then reduce it till they go away.
May 02, 2014, 03:21:56 PM
bikini's, thongs, casual clothes. But overall I let the women decide. haha If they want to do a nudey it's up to them. lmao
bikini's, thongs or nude all sounds good to me
Hmmm it is (very) tempting guys.... but think clear....
I am no moral knight but I don't believe this is the right way we are sailing with UTC
yea we just messing about Dcgirl would kick our ass
May 02, 2014, 03:18:44 PM
bikini's, thongs, casual clothes. But overall I let the women decide. haha If they want to do a nudey it's up to them. lmao
bikini's, thongs or nude all sounds good to me
Hmmm it is (very) tempting guys.... but think clear....
I am no moral knight but I don't believe this is the right way we are sailing with UTC
May 02, 2014, 03:14:15 PM
May 02, 2014, 03:12:08 PM
bikini's, thongs, casual clothes. But overall I let the women decide. haha If they want to do a nudey it's up to them. lmao
bikini's, thongs or nude all sounds good to me
May 02, 2014, 03:06:53 PM
bikini's, thongs, casual clothes. But overall I let the women decide. haha If they want to do a nudey it's up to them. lmao
You are the reason why I would invest in UTC.
May 02, 2014, 02:48:41 PM
bikini's, thongs, casual clothes. But overall I let the women decide. haha If they want to do a nudey it's up to them. lmao
Jump to: