Topic: [ANN]**** NEW THEKEV VERTMINER ****Ultimate Vertminer guide for Windows**** - page 3. (Read 115646 times)
February 26, 2014, 03:56:22 PM
Will this work with normal Scrypt coins if I change it in the config from Scrypt-vert to just Scrypt? I love the raw intensity option. I have 2 280xs in an enclosed case that run well at I 13, but together are just too hot for the enclosure. I've had success with this miner setting them at TC and raw intensity 7168 which is essentially I 12.807. They stay at temps that I like and I'm fine with 87.5% throughput. However, if like to point one of them to a Scrypt coin and am not sure if there are other miners that allow for such fine tuning of intensity.
February 26, 2014, 03:09:45 AM
Quote
rawintensity @ (2048)*3=6144
Drops my hashrate from 380 to 270.
Quote
or (2048)*5=10240
HW errors from go
Quote
"thread-concurrency" : "9728",
Isn't a multiple of shaders and gave me some issues in the logs.
Still screwing with it. Thus far, this is the best with 1-2 HW error on one card an hour at 280 kH/s. Hope to get it to zero.
"rawintensity": "8192",
"thread-concurrency": "8193",
"gpu-threads": "2",
"gpu-engine": "1070",
"gpu-memclock": "1800",
February 25, 2014, 08:24:33 AM
Nice bit of software, made my 270's fly with 2 threads, still working on my 280s and 290s.
I've noticed all but my 1st card starts as "OFF" and I have to manually enable each one, is this an expected result?
I've noticed all but my 1st card starts as "OFF" and I have to manually enable each one, is this an expected result?
February 24, 2014, 05:30:43 AM
@ Bajawah
You could try the following:
{
"pools" : [
]
,
"rawintensity" : "8192",
"vectors" : "1",
"worksize" : "256",
"kernel" : "scrypt",
"lookup-gap" : "2",
"thread-concurrency" : "9728",
"shaders" : "2048",
"gpu-engine" : "1000",
"gpu-fan" : "60-85",
"gpu-memclock" : "1500",
"gpu-memdiff" : "0",
"gpu-powertune" : "0",
"gpu-vddc" : "0.000",
"temp-cutoff" : "85",
"temp-overheat" : "80",
"temp-target" : "70",
-other stuff-
On the 280x Gigabyte this gives me 340-350 kHash, no Hardware errors.
EDIT
I see you have set your memclock to 1800. try setting it to 1500 as i think the 280's are not really stable above that point.
You could try the following:
{
"pools" : [
]
,
"rawintensity" : "8192",
"vectors" : "1",
"worksize" : "256",
"kernel" : "scrypt",
"lookup-gap" : "2",
"thread-concurrency" : "9728",
"shaders" : "2048",
"gpu-engine" : "1000",
"gpu-fan" : "60-85",
"gpu-memclock" : "1500",
"gpu-memdiff" : "0",
"gpu-powertune" : "0",
"gpu-vddc" : "0.000",
"temp-cutoff" : "85",
"temp-overheat" : "80",
"temp-target" : "70",
-other stuff-
On the 280x Gigabyte this gives me 340-350 kHash, no Hardware errors.
EDIT
I see you have set your memclock to 1800. try setting it to 1500 as i think the 280's are not really stable above that point.
February 24, 2014, 04:55:13 AM
February 24, 2014, 12:09:08 AM
What are some good settings for a Asus 280x?
I am having quite a bit of hardware errors. Never had any on 0.5.3.
(4) Asus DC2 280x
8GB of Ram
The settings below were the same as used on 0.5.3, but I changed intensity: 13 to it's equivalent rawintesnsity: 8192.
Did I do that right?
{
"pools": [
],
"gpu-reorder": true,
"auto-fan": true,
"gpu-fan": "50-100",
"temp-cutoff": "95",
"temp-overheat": "85",
"temp-target": "75",
"rawintensity": "8192",
"vectors": "1",
"worksize": "256",
"lookup-gap": "2",
"thread-concurrency": "8192",
"shaders": "0",
"api-listen": true,
"api-port": "4028",
"api-allow": "W:127.0.0.1",
"gpu-dyninterval": "7",
"gpu-platform": "0",
"gpu-threads": "2",
"gpu-engine": "1070",
"gpu-memclock": "1800",
"gpu-powertune": "20",
"log": "5",
"no-pool-disable": true,
"no-submit-stale": true,
"queue": "0",
"scan-time": "1",
"expiry": "30",
"scrypt": true,
"shares": "0",
"kernel-path": "/usr/local/bin"
}
I am having quite a bit of hardware errors. Never had any on 0.5.3.
(4) Asus DC2 280x
8GB of Ram
The settings below were the same as used on 0.5.3, but I changed intensity: 13 to it's equivalent rawintesnsity: 8192.
Did I do that right?
{
"pools": [
],
"gpu-reorder": true,
"auto-fan": true,
"gpu-fan": "50-100",
"temp-cutoff": "95",
"temp-overheat": "85",
"temp-target": "75",
"rawintensity": "8192",
"vectors": "1",
"worksize": "256",
"lookup-gap": "2",
"thread-concurrency": "8192",
"shaders": "0",
"api-listen": true,
"api-port": "4028",
"api-allow": "W:127.0.0.1",
"gpu-dyninterval": "7",
"gpu-platform": "0",
"gpu-threads": "2",
"gpu-engine": "1070",
"gpu-memclock": "1800",
"gpu-powertune": "20",
"log": "5",
"no-pool-disable": true,
"no-submit-stale": true,
"queue": "0",
"scan-time": "1",
"expiry": "30",
"scrypt": true,
"shares": "0",
"kernel-path": "/usr/local/bin"
}
February 21, 2014, 06:47:48 AM
Has any 7950 owner observed any performance boost while using TheKev's build over classic ones? If so, using what kind of values on their .conf file?
As far as I can see the Kevminer is more stable then the official one. As to your question what value's I am using (I've got an Asus 7950):
rawintensity: 1048576
threadconc: 19712
gpu-engine: 1000
gpu-memcl: 1500
gives me a nice 290+ khash, without any hardware errors whatsoever.
Hope I could be of help.
Thanks a lot for your input! For the past ~12 hours I am running stable 4x 7950 (Sapphire) @ 290Kh/s all undervolted @ 1.094v, using the following:
engine: 950
memory: 1325
rawintensity: 7168
tc: 8193
Temps are (without any extra fans):
gpu0: 73,9
gpu1: 58,9
gpu2: 73,4
gpu3: 70,8
I will keep those setting for a few more hours and then adjust a bit, since it looks like there is a lot of room for testing using rawintensity.
February 21, 2014, 06:26:39 AM
Has any 7950 owner observed any performance boost while using TheKev's build over classic ones? If so, using what kind of values on their .conf file?
As far as I can see the Kevminer is more stable then the official one. As to your question what value's I am using (I've got an Asus 7950):
rawintensity: 1048576
threadconc: 19712
gpu-engine: 1000
gpu-memcl: 1500
gives me a nice 290+ khash, without any hardware errors whatsoever.
Hope I could be of help.
February 21, 2014, 12:07:32 AM
It appears that with Scrypt-N You need to have at least same or more amount of system RAM as is on Your VGA, so If You have 2G of ram and some R9 VGA with 3G, Youll get this error with higher thread concurrency. Eg (approximates)
R9 280X w 3G video RAM + 2GB system RAM on scrypt with TC 8192: 700khs
R9 280X w 3G video RAM + 4GB system RAM on scrypt-n with TC 8192: 350khs
R9 280X w 3G video RAM + 2GB system RAM on scrypt-n with TC 8192: Error -5: Enqueueing kernel onto command queue. (clEnqueueNDRangeKernel)
R9 280X w 3G video RAM + 2GB system RAM on scrypt-n with TC 4096: 200khs
R9 280X w 3G video RAM + 2GB system RAM on scrypt with TC 8192: 700khs
R9 280X w 3G video RAM + 4GB system RAM on scrypt-n with TC 8192: 350khs
R9 280X w 3G video RAM + 2GB system RAM on scrypt-n with TC 8192: Error -5: Enqueueing kernel onto command queue. (clEnqueueNDRangeKernel)
R9 280X w 3G video RAM + 2GB system RAM on scrypt-n with TC 4096: 200khs
I have 2gb of GPU ram and 6gb of system ram... so... can't imagine that's the problem. The same card is doing 250 on windows, but 191 on linux...
February 20, 2014, 02:59:16 PM
anyway of using Vertikev with cgwatcher?
Yes put these two arguments in the batch file of vertikev.
--api-listen --api-allow W:127.0.0.1
Sorry don't have time to explain more. But I am using it right now. And Can tell you that it works just like CGminer.
February 20, 2014, 02:41:02 PM
Has any 7950 owner observed any performance boost while using TheKev's build over classic ones? If so, using what kind of values on their .conf file?
February 20, 2014, 02:13:10 PM
anyway of using Vertikev with cgwatcher?
February 20, 2014, 10:25:46 AM
Also got it compiled and all, but as soon as I put (gpu-threads : 2 )in my config, I get the following
Error -5: Enqueueing kernel onto command queue. (clEnqueueNDRangeKernel)
GPU 0 failure, disabling!
And using gpu-threads : 1 makes it run about 30% slower than it should...
Error -5: Enqueueing kernel onto command queue. (clEnqueueNDRangeKernel)
GPU 0 failure, disabling!
And using gpu-threads : 1 makes it run about 30% slower than it should...
It appears that with Scrypt-N You need to have at least same or more amount of system RAM as is on Your VGA, so If You have 2G of ram and some R9 VGA with 3G, Youll get this error with higher thread concurrency. Eg (approximates)
R9 280X w 3G video RAM + 2GB system RAM on scrypt with TC 8192: 700khs
R9 280X w 3G video RAM + 4GB system RAM on scrypt-n with TC 8192: 350khs
R9 280X w 3G video RAM + 2GB system RAM on scrypt-n with TC 8192: Error -5: Enqueueing kernel onto command queue. (clEnqueueNDRangeKernel)
R9 280X w 3G video RAM + 2GB system RAM on scrypt-n with TC 4096: 200khs
February 20, 2014, 12:01:32 AM
Also got it compiled and all, but as soon as I put (gpu-threads : 2 )in my config, I get the following
Error -5: Enqueueing kernel onto command queue. (clEnqueueNDRangeKernel)
GPU 0 failure, disabling!
And using gpu-threads : 1 makes it run about 30% slower than it should...
Error -5: Enqueueing kernel onto command queue. (clEnqueueNDRangeKernel)
GPU 0 failure, disabling!
And using gpu-threads : 1 makes it run about 30% slower than it should...
February 19, 2014, 11:58:13 PM
Quote
Ok, I got it compiled and its slow to start up but when it does, it see's my four gpu's but doesnt start them up...then come the errors. Any clue? Also, since you are a fellow linux guy, is there a better, more efficient way to reach you? If you dont mind, I would like to get this up and running very soon. Thx.
PM Sent
February 19, 2014, 07:57:46 PM
I reviewed this thread and seen no mention of setups for Linux. Is there support for Linux users as of present for this new miner program?
The title does say "...for Windows". But I'm currently using the "TheKev" version in Debian Linux. You can try this guide:
http://vertcoin.org/gpu-miner-build-howto.html
Just substitute the URL for the miner:
git clone https://github.com/thekev/vertminer-gpu
instead of:
git clone https://github.com/Bufius/vertminer-gpu
Hey man, I ran thru those steps and at the very end...when you run ./buildit.sh.....i get an error that says ./configure does not exist. Did you run into this problem? How did you fix it? FYI - I also have cgminer running on this linux distro as well (Xubuntu 13.10). My vision was to have both cgminer and vertminer. Cgminer is currently up and running so I skipped most of the sdk and amd driver parts. Please help!
Do:
./autogen.sh
Before:
./buildit.sh
Ok, I got it compiled and its slow to start up but when it does, it see's my four gpu's but doesnt start them up...then come the errors. Any clue? Also, since you are a fellow linux guy, is there a better, more efficient way to reach you? If you dont mind, I would like to get this up and running very soon. Thx.
February 19, 2014, 03:42:31 PM
are hash rates expected to be 1/2 -1/3 of regular scrypt mining? just tried this on one of my rigs, and i'm getting 1.05Mh from 3 290's, vs 2.67-2.72 Mh from the same cards.
Code:
......
I have exactly 1/2 of scrypt hashrate on Vert too, is this correct? Also be sure to delete API acces from TheKev's Fork config file, otherwise You can expect some bad surprise..
did i really leave that in there? my bad xD i guess i forgot one of the config files. I will remove it immedietly.
Yes you should expect 50% vs normal scrypt. If you can achieve more then 50% your doing really well. The general average is generally about 40-50% using scrypt-n vs scrypt.
update: yeah im a idiot i forgot to change that config in thekevs vertminer... The API was still setup for my monitoring software
. Apologies, if you download the miner now the config will be like the one in the OP. February 19, 2014, 01:56:58 PM
What do you have the API doing?
February 19, 2014, 10:20:16 AM
I reviewed this thread and seen no mention of setups for Linux. Is there support for Linux users as of present for this new miner program?
The title does say "...for Windows". But I'm currently using the "TheKev" version in Debian Linux. You can try this guide:
http://vertcoin.org/gpu-miner-build-howto.html
Just substitute the URL for the miner:
git clone https://github.com/thekev/vertminer-gpu
instead of:
git clone https://github.com/Bufius/vertminer-gpu
Hey man, I ran thru those steps and at the very end...when you run ./buildit.sh.....i get an error that says ./configure does not exist. Did you run into this problem? How did you fix it? FYI - I also have cgminer running on this linux distro as well (Xubuntu 13.10). My vision was to have both cgminer and vertminer. Cgminer is currently up and running so I skipped most of the sdk and amd driver parts. Please help!
Do:
./autogen.sh
Before:
./buildit.sh
February 18, 2014, 08:05:30 PM
I reviewed this thread and seen no mention of setups for Linux. Is there support for Linux users as of present for this new miner program?
The title does say "...for Windows". But I'm currently using the "TheKev" version in Debian Linux. You can try this guide:
http://vertcoin.org/gpu-miner-build-howto.html
Just substitute the URL for the miner:
git clone https://github.com/thekev/vertminer-gpu
instead of:
git clone https://github.com/Bufius/vertminer-gpu
Hey man, I ran thru those steps and at the very end...when you run ./buildit.sh.....i get an error that says ./configure does not exist. Did you run into this problem? How did you fix it? FYI - I also have cgminer running on this linux distro as well (Xubuntu 13.10). My vision was to have both cgminer and vertminer. Cgminer is currently up and running so I skipped most of the sdk and amd driver parts. Please help!
Jump to: