Author

Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 631. (Read 3426922 times)

hero member
Activity: 812
Merit: 1000
Nope, that's groestl myriad.

Damn that sucks, well if anyone can help me with Heavycoin config for a 6970 i would be grateful. Wish it was as easy as it is on my nvidia rig.

Just posted a conf file there, copy it and modify it as you see fit! :p

Pretty much the same card, or pretty similar at least.
member
Activity: 91
Merit: 10
Hey Christian & Christian, did you just add myriad groestl capability in your super secret test lab? Couldn't help but notice that the difficulty which has been around 200 for days, but as low as 125 a few hours ago, just jumped to 287.

EDIT: For some reason, running minerd for groestl on 6 CPU threads bumps my 780ti from 21,950 khash/s to 22,520 khash/s on ccminer35.
newbie
Activity: 9
Merit: 0
Nope, that's groestl myriad.

Damn that sucks, well if anyone can help me with Heavycoin config for a 6970 i would be grateful. Wish it was as easy as it is on my nvidia rig.
hero member
Activity: 812
Merit: 1000
Nope, that's groestl myriad.
newbie
Activity: 9
Merit: 0
8mh/s for the 780ti (non shared memory, the one with shared memory gives 9.5mh)
Did you recompile with compute 3.5, it may-be a little faster.
Actually, this coin is really a cpu coin, my 3770 gets around 1.1mh/s (6 threads)

groestl, as implemented now, does a massive amount of random access lookups in S-box tables derived from AES.

CPUs have a bit of an advantage due to their big data caches. So the absolute speed-up of a GPU isn't
that great. Maybe factor 10.

By the way, according to some information I got an AMD R9 290X only does around 6.5 MHash/s. Can anyone confirm this? If so, we've got another nVidia coin here.

Christian

Groestl runs at 7.5MH/s on 6970 (900/500) and 5.3MH/s on 5850 (850/300). Low mem for running cool


what are your config/bat settings for your 6970's? i cant figure out how to get them past 1.5mh/s  Huh

I got my 6950 to 10mh/s on groestl...so yeah.

{
"pools" : [
   {
      "url" : "stratum+tcp://stablehash.com:4502",
      "user" : "ivanlabrie.1",
      "pass" : "x"
   }
],
"gpu-platform" : "1",
"intensity" : "16",
"kernel" : "myriadcoin-groestl",
"gpu-engine" : "0-850",
"gpu-fan" : "70-100",
"gpu-memclock" : "150",
"gpu-powertune" : "20",
"temp-cutoff" : "95",
"temp-overheat" : "85",
"temp-target" : "75",
"api-listen" : false,
"api-mcast-port" : "4028",
"api-port" : "4028",
"auto-fan" : true,
"auto-gpu" : true,
"expiry" : "120",
"failover-only" : true,
"gpu-threads" : "4",
"worksize" : "64",
"log" : "5",
"queue" : "1",
"scan-time" : "60",
"temp-hysteresis" : "3",
"failover-switch-delay" : "60",
"no-pool-disable" : true,
"kernel-path" : "/usr/local/bin"
}


I will try that,This will work for heavycoin right?
hero member
Activity: 812
Merit: 1000
8mh/s for the 780ti (non shared memory, the one with shared memory gives 9.5mh)
Did you recompile with compute 3.5, it may-be a little faster.
Actually, this coin is really a cpu coin, my 3770 gets around 1.1mh/s (6 threads)

groestl, as implemented now, does a massive amount of random access lookups in S-box tables derived from AES.

CPUs have a bit of an advantage due to their big data caches. So the absolute speed-up of a GPU isn't
that great. Maybe factor 10.

By the way, according to some information I got an AMD R9 290X only does around 6.5 MHash/s. Can anyone confirm this? If so, we've got another nVidia coin here.

Christian

Groestl runs at 7.5MH/s on 6970 (900/500) and 5.3MH/s on 5850 (850/300). Low mem for running cool


what are your config/bat settings for your 6970's? i cant figure out how to get them past 1.5mh/s  Huh

I got my 6950 to 10mh/s on groestl...so yeah.

{
"pools" : [
   {
      "url" : "stratum+tcp://stablehash.com:4502",
      "user" : "ivanlabrie.1",
      "pass" : "x"
   }
],
"gpu-platform" : "1",
"intensity" : "16",
"kernel" : "myriadcoin-groestl",
"gpu-engine" : "0-850",
"gpu-fan" : "70-100",
"gpu-memclock" : "150",
"gpu-powertune" : "20",
"temp-cutoff" : "95",
"temp-overheat" : "85",
"temp-target" : "75",
"api-listen" : false,
"api-mcast-port" : "4028",
"api-port" : "4028",
"auto-fan" : true,
"auto-gpu" : true,
"expiry" : "120",
"failover-only" : true,
"gpu-threads" : "4",
"worksize" : "64",
"log" : "5",
"queue" : "1",
"scan-time" : "60",
"temp-hysteresis" : "3",
"failover-switch-delay" : "60",
"no-pool-disable" : true,
"kernel-path" : "/usr/local/bin"
}
newbie
Activity: 9
Merit: 0
8mh/s for the 780ti (non shared memory, the one with shared memory gives 9.5mh)
Did you recompile with compute 3.5, it may-be a little faster.
Actually, this coin is really a cpu coin, my 3770 gets around 1.1mh/s (6 threads)

groestl, as implemented now, does a massive amount of random access lookups in S-box tables derived from AES.

CPUs have a bit of an advantage due to their big data caches. So the absolute speed-up of a GPU isn't
that great. Maybe factor 10.

By the way, according to some information I got an AMD R9 290X only does around 6.5 MHash/s. Can anyone confirm this? If so, we've got another nVidia coin here.

Christian

Groestl runs at 7.5MH/s on 6970 (900/500) and 5.3MH/s on 5850 (850/300). Low mem for running cool


what are your config/bat settings for your 6970's? i cant figure out how to get them past 1.5mh/s  Huh
full member
Activity: 146
Merit: 100
More ccminer weirdness, haven't tried with cudaminer.

On my 8 core 8350 machine with 3 750 ti's, if I run Boinc on 7 threads (haven't tried less, that is just what I run it at), the hashrate immediately goes up, from 11.x megahashes to the mid 12 megahashes. Strange Stuff, this is witchcraft, more than science it appears.

Don't understand that, as my boinc is set to CPU only. Anyway, maybe some can try 1 thread using boinc, and see if that helps hash rate.
full member
Activity: 263
Merit: 100


mining HVC
upper cmd is ccminer30 0.5 msi 660ti
down cmd is ccminer35 0.5 2 x msi 750ti tf
newbie
Activity: 19
Merit: 0
I want to play titanfall but wont pay the stupid price
£25 using a voucher here if that's any use to you: http://www.postabargain.co.uk/categories/pc-mac/titanfall-pc/041558
full member
Activity: 189
Merit: 100
does 10Mh/s sound about right for a single 750Ti? or am i way off.

I'm having 12.7MHs with a single Gigabyte 750 Ti "OCable" direct at x16 PCi-E.

Thanks, i can't seem to get over 10, and both are directly on the board.. I'll try some different OC setting..
sr. member
Activity: 328
Merit: 250
I might be a nag with this but...

Any progress with X11?

I'm currently using CUDAminer (big hat-tip) and found a recent interest in the X11 algos.  I'm seeing glorious hashes using AMD without anything Nvidia comparative.

I have 4 - 750ti showing up at my door any minute.  Finding myself scratching my head and considering sending them back.  Undecided

 
yep... send them back they are completely useless without x11  Roll Eyes

Touche.  Grin  Looking forward to checking out CUDAminer / Maxwell regardless.
legendary
Activity: 1400
Merit: 1050
I might be a nag with this but...

Any progress with X11?

I'm currently using CUDAminer (big hat-tip) and found a recent interest in the X11 algos.  I'm seeing glorious hashes using AMD without anything Nvidia comparative.

I have 4 - 750ti showing up at my door any minute.  Finding myself scratching my head and considering sending them back.  Undecided

 
yep... send them back they are completely useless without x11  Roll Eyes
sr. member
Activity: 328
Merit: 250
I might be a nag with this but...

Any progress with X11?

I'm currently using CUDAminer (big hat-tip) and found a recent interest in the X11 algos.  I'm seeing glorious hashes using AMD without anything Nvidia comparative.

I have 4 - 750ti showing up at my door any minute.  Finding myself scratching my head and considering sending them back.  Undecided

 
member
Activity: 112
Merit: 10
imho: i do not see any benefit in +500 mem for GPU algo like heavycoin (SHA3), you can put there -500 mem to safe some power and keep the hashrate Wink
very good observation
Reducing my Mem Clock OC from +650 to -500 I ketp the same rate.
hero member
Activity: 756
Merit: 502
will compute 5.0 do any boost for 750Ti cards? Smiley

didn't measure any boost so far when I tried.

Christian

sr. member
Activity: 280
Merit: 250
For heavycoin my 2 750ti's

http://www.newegg.com/Product/Product.aspx?Item=N82E16814487024

are holding around 13.5mh each with CCminer5 and compute 3.5.  Haven't had to drop the overclocks at all (+100 core, 500 mem)
55c @ 60% fan

imho: i do not see any benefit in +500 mem for GPU algo like heavycoin, you can put there -500 mem to safe some power and keep the hashrate Wink


Good tip.  If i stay on heavy for any length ill go ahead and do that.
sr. member
Activity: 300
Merit: 250
For heavycoin my 2 750ti's

http://www.newegg.com/Product/Product.aspx?Item=N82E16814487024

are holding around 13.5mh each with CCminer5 and compute 3.5.  Haven't had to drop the overclocks at all (+100 core, 500 mem)
55c @ 60% fan

imho: i do not see any benefit in +500 mem for GPU algo like heavycoin (SHA3), you can put there -500 mem to safe some power and keep the hashrate Wink
member
Activity: 112
Merit: 10
does 10Mh/s sound about right for a single 750Ti? or am i way off.

I'm having 12.7MHs with a single Gigabyte 750 Ti "OCable" direct at x16 PCi-E.
Can you post your settings and a Screenshot while mining of GPU-Z please?

Already did:
https://bitcointalksearch.org/topic/m.5922764

With x86 ccminer35 i achieved peaks of 12.8MHs.

Now I'm testing x64 ccminer35 version because people says it crashes less.
My average rate stays around 12.5~12.6MHs.

I'm using win 7 x64.
sr. member
Activity: 280
Merit: 250
For heavycoin my 2 750ti's

http://www.newegg.com/Product/Product.aspx?Item=N82E16814487024

are holding around 13.5mh each with CCminer5 and compute 3.5.  Haven't had to drop the overclocks at all (+100 core, 500 mem)
55c @ 60% fan
Jump to: