Topic: [ANN] Yet another GPU miner release! [YAC] - page 22. (Read 53310 times)
mega never worked for me either until I installed the firefox extension
It seems that under this weird circumstance, the GPU miner should be released.
What weird circumstance?
Here's what I compiled for x86 - It works on my 5770 and 7850, but does not work on my 5870 (gpu disabled - but his 64 bit binary doesn't do that) - every caveat hanzac said applies.
https://mega.co.nz/#!K8o2VR5a!GbdWik4AYFMRUWoFv96fcy1xvNs79IQzS1uwxPIT8vY
https://mega.co.nz/#!K8o2VR5a!GbdWik4AYFMRUWoFv96fcy1xvNs79IQzS1uwxPIT8vY
Can you upload somewhere else as well, MEGA is totally screwed for me and I don't feel like installing Google Chrome just for 1 download. Thanks!
http://www.speedyshare.com maybe?
and it was always possible to use it before you just had to click past a popup
Im using the windows version on a gtx 680 i have, and its flipping out but im getting a heap of rejects and a few accepted
and the hash rate at least registers on the p2pool
and the hash rate at least registers on the p2pool
Really, good news. I don't have a nvidia card, how about the hash rate.
i found a nice spot at about 35kh/s lol but its something at least. I got higher but it wasn't running right.
Ive got a mild clock of 1333mhz 1.2v i can max it and game at about 1555mhz 1.4v but as i said it wasn't running right for this
Code:
cgminer --scrypt --worksize 256 --lookup-gap 2 --thread-concurrency 16400 -I 10
Anybody figure out how to make the 7970 go faster? still getting 150Kh/s with I 11 8192, anything over intensity 11 causes crash.
are you mining solo? got a block(s)? i let it run for about 5 hours with 2 x 144kh/s but only 2 stales nothing more. i think we are using the wrong settings.
Mining here http://yac.coinmine.pl/index
Yeah I agree, if 7950s are doing 240s we should do at least double what we are
Anybody figure out how to make the 7970 go faster? still getting 150Kh/s with I 11 8192, anything over intensity 11 causes crash.
are you mining solo? got a block(s)? i let it run for about 5 hours with 2 x 144kh/s but only 2 stales nothing more. i think we are using the wrong settings.
Here's what I compiled for x86 - It works on my 5770 and 7850, but does not work on my 5870 (gpu disabled - but his 64 bit binary doesn't do that) - every caveat hanzac said applies.
NEW 3.2.1 x86 compilation link:
https://mega.co.nz/#!e1YVDBYZ!YSzn40ba5ciOVUeJjIcQnFxbvr0JhdYwkzqempL_l8Y
original source compile link (depreciated)
https://mega.co.nz/#!K8o2VR5a!GbdWik4AYFMRUWoFv96fcy1xvNs79IQzS1uwxPIT8vY
NEW 3.2.1 x86 compilation link:
https://mega.co.nz/#!e1YVDBYZ!YSzn40ba5ciOVUeJjIcQnFxbvr0JhdYwkzqempL_l8Y
original source compile link (depreciated)
https://mega.co.nz/#!K8o2VR5a!GbdWik4AYFMRUWoFv96fcy1xvNs79IQzS1uwxPIT8vY
Anybody figure out how to make the 7970 go faster? still getting 150Kh/s with I 11 8192, anything over intensity 11 causes crash.
Caicos? what's this GPU brand?
Can you also post the command line? the thread-concurrency? Maybe you set the parameter not match with your hardware.
Can you also post the command line? the thread-concurrency? Maybe you set the parameter not match with your hardware.
It's AMD Radeon 7470M. Works in Windows, though... Tried many options in Linux, even omitting all of the HW-specific ones but to no avail. In Windoze I'm using
Code:
cgminer.exe --device 0 --scrypt --worksize 256 --lookup-gap 2 -I 5 --thread-concurrency 8192
waiting for your new version for x86.
Regarding rejects, just FYI I'm getting about 33% rejects across both 7950's I'm running after running after a few hours. Would this be down to the pool?
I'm getting 6% of rejects with a HD6770, using yac.ltcoin.net. Regarding rejects, just FYI I'm getting about 33% rejects across both 7950's I'm running after running after a few hours. Would this be down to the pool?
Me too, the rejects are quite high (1/5 rejects), back to when I used it mining on p2pool, the rejects are not that high. So I assume the pool is blocking fast, make a lots of stales. But considering if you use p2pool, the fee caused by small payments. It's all really trade-offs. I think pushpools should be better than p2pools.
Regarding rejects, just FYI I'm getting about 33% rejects across both 7950's I'm running after running after a few hours. Would this be down to the pool?
What hashrate has 6930 ?
What optimal memory clocks?
Is it profitable to mininig yac or ltc?
What optimal memory clocks?
Is it profitable to mininig yac or ltc?
hanzac, what's the relationship between intensity and thread concurrency, and should there be a way to run this at > I=12? I tried 13 and it seemed to drop the hash rate and 14 died. My TC is 32768
Currently the intensity is used to control the global worksize (2^I, if I = 12, global worksize is 4096) of the opencl kernel execution. This means the concurrent executions pushed to the GPU kernel. The total memory buffer for chacha mixing will be 2^I * 128 kB (nfactor = 9, check the tech info here: http://yacexplorer.tk/graphs.htm), if I = 12, it will need 512MB of GPU's global memory. If you use I = 13, it will consume 1GB GPU memory, but considering some GPU's local memory need, it will need more than 1GB.
Even if you have enough memory, you still have limitation on the GPU stream cores. For 7970, it has 2048 GPU stream cores, so set the intensity to make concurrent executions to 2x GPU cores will be approaching optimal, that's why I = 12 works better than I = 11. I think concurrent executions works well in range from 2x GPU cores - 4x GPU cores.
The reason it died, it because memory used up, pointer to some memory address not exist. So if you insist on achieve higher intensity, the cl kernel needs to support look-up gap or some other memory shrink/share technique, but current cl kernel is quite plain & simple and doesn't support these.
The thread-concurrency(=32768) here is for allocating the memory pad buffer, 32768 * 64K, is allocated 2GB memory. To allocate more, you can try some other lookup-gap setting to allocate the maximized memory. For example, your card total memory is 3GB you set the lookup-gap = 2, and set thread-concurrency to 49152? I think I have tried once, but not succeed for my card.
Can anyone suggest a good pool? I was using yac.coinmine.pl but I think it's a scam as my "estimated per payout" was really low (0.1) after a couple of hours at 500 kh/s
I think it's legit. YAC needs 520 confirmations to payout. So maybe you need to wait for tomorrow to start getting payouts.
Just check Unpaid Shares - Your Valid: -----, if it's normal.
Or you go to yac.ltcoin.net, it seems having the highest hash rate so far.
Again I have to definitely disagree of calling our pool scam - as mentioned earlier - to confirm first block you started mining we need ~8-10 hours - and this is approx the time you should have full payouts. Please note that even if you stop mining you will be receiving payouts until you are out of last Nshares. This has been proved by me several times here...
feeleep
My sincere apologies, I've only ever mined with p2pool so was not expecting the long wait. All is working now and I'm in your top 10 for contributors / hashrate.
Can anyone suggest a good pool? I was using yac.coinmine.pl but I think it's a scam as my "estimated per payout" was really low (0.1) after a couple of hours at 500 kh/s
I think it's legit. YAC needs 520 confirmations to payout. So maybe you need to wait for tomorrow to start getting payouts.
Just check Unpaid Shares - Your Valid: -----, if it's normal.
Or you go to yac.ltcoin.net, it seems having the highest hash rate so far.
Again I have to definitely disagree of calling our pool scam - as mentioned earlier - to confirm first block you started mining we need ~8-10 hours - and this is approx the time you should have full payouts. Please note that even if you stop mining you will be receiving payouts until you are out of last Nshares. This has been proved by me several times here...
feeleep
hanzac, what's the relationship between intensity and thread concurrency, and should there be a way to run this at > I=12? I tried 13 and it seemed to drop the hash rate and 14 died. My TC is 32768
I'm getting segfaults 
The reason for segfault seems to be a dangling pointer issue, or some sort of buffer overflow.
Code:
[2013-06-12 11:37:45] Started cgminer 3.2.0
[2013-06-12 11:37:45] Started cgminer 3.2.0
[2013-06-12 11:37:45] CL Platform 0 vendor: Advanced Micro Devices, Inc.
[2013-06-12 11:37:45] CL Platform 0 name: AMD Accelerated Parallel Processing
[2013-06-12 11:37:45] CL Platform 0 version: OpenCL 1.2 AMD-APP (1214.3)
[2013-06-12 11:37:45] Platform 0 devices: 1
[2013-06-12 11:37:45] 0 Caicos
[2013-06-12 11:37:45] Probing for an alive pool
[2013-06-12 11:37:45] Popping work to stage thread
[2013-06-12 11:37:45] Testing pool [...]
[2013-06-12 11:37:45] Probing for GBT support
[2013-06-12 11:37:45] JSON-RPC call failed: {
"data": null,
"code": -32601,
"message": "Method not found"
}
[2013-06-12 11:37:45] Failed to connect in json_rpc_call
[2013-06-12 11:37:45] No GBT coinbase + append support found, using getwork protocol
[2013-06-12 11:37:45] X-Roll-Ntime expiry set to 100
[2013-06-12 11:37:45] Successfully retrieved and deciphered work from pool 0 [...]
[2013-06-12 11:37:45] Pushing pooltest work to base pool
[2013-06-12 11:37:45] Network diff set to 49.1K
[2013-06-12 11:37:45] New block: d9c3b6dc1e97ac28... diff 49.1K
[2013-06-12 11:37:45] Pushing work to getwork queue
[2013-06-12 11:37:45] Popping work to stage thread
[2013-06-12 11:37:45] Pool 0 [...] alive
[2013-06-12 11:37:45] Long-polling activated for [...]Segmentation fault (core dumped)
The reason for segfault seems to be a dangling pointer issue, or some sort of buffer overflow.
Code:
--- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_MAPERR, si_addr=0x2904908} ---Caicos? what's this GPU brand?
Can you also post the command line? the thread-concurrency? Maybe you set the parameter not match with your hardware.
I'm getting segfaults
The reason for segfault seems to be a dangling pointer issue, or some sort of buffer overflow.
Code:
[2013-06-12 11:37:45] Started cgminer 3.2.0
[2013-06-12 11:37:45] Started cgminer 3.2.0
[2013-06-12 11:37:45] CL Platform 0 vendor: Advanced Micro Devices, Inc.
[2013-06-12 11:37:45] CL Platform 0 name: AMD Accelerated Parallel Processing
[2013-06-12 11:37:45] CL Platform 0 version: OpenCL 1.2 AMD-APP (1214.3)
[2013-06-12 11:37:45] Platform 0 devices: 1
[2013-06-12 11:37:45] 0 Caicos
[2013-06-12 11:37:45] Probing for an alive pool
[2013-06-12 11:37:45] Popping work to stage thread
[2013-06-12 11:37:45] Testing pool [...]
[2013-06-12 11:37:45] Probing for GBT support
[2013-06-12 11:37:45] JSON-RPC call failed: {
"data": null,
"code": -32601,
"message": "Method not found"
}
[2013-06-12 11:37:45] Failed to connect in json_rpc_call
[2013-06-12 11:37:45] No GBT coinbase + append support found, using getwork protocol
[2013-06-12 11:37:45] X-Roll-Ntime expiry set to 100
[2013-06-12 11:37:45] Successfully retrieved and deciphered work from pool 0 [...]
[2013-06-12 11:37:45] Pushing pooltest work to base pool
[2013-06-12 11:37:45] Network diff set to 49.1K
[2013-06-12 11:37:45] New block: d9c3b6dc1e97ac28... diff 49.1K
[2013-06-12 11:37:45] Pushing work to getwork queue
[2013-06-12 11:37:45] Popping work to stage thread
[2013-06-12 11:37:45] Pool 0 [...] alive
[2013-06-12 11:37:45] Long-polling activated for [...]Segmentation fault (core dumped)
The reason for segfault seems to be a dangling pointer issue, or some sort of buffer overflow.
Code:
--- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_MAPERR, si_addr=0x2904908} ---Jump to: