Pages:
Author

Topic: DiabloMiner GPU Miner - page 53. (Read 866596 times)

legendary
Activity: 1162
Merit: 1000
DiabloMiner author
June 18, 2011, 12:07:19 PM
My math gives about 469 combined for two 6850s if you can find the right settings, but its more likely to be in the 450s because you're on SDK 2.4. A 6870 is about as fast as a 5850 at their respective stock clocks, a 6850 is about 1/3rd slower than a 6870 at their respective stock clocks, and a 5850 is about 1/4th slower than a 5870.

So 450s for two overclocked 6850s is in the right ballpark if a overclocked 5870 is doing 450.

Thanks for the info man. I needed some closure lol. I think my main problem is the power supply in my computer under volting causing fluctuation performance (600w coolermaster). I'm going to try unplugging my 2 dvd burners and my second hdd to see what happens. So do you recommend reverting to an earlier revision of sdk?

BTW, yes, 0 accepts and 0 rejects are valid for solo mining. At the current difficulty it will take you about 14 weeks on average to find a block, which then accepts will turn to 1.

Disconnect your crossfire cable if you have crossfire disabled. If you're solo mining, the only thing you need to give to DiabloMiner is -u and -p, -o and -r are already set correctly by default for local solo mining.

From your description in the other thread, it almost sounds like its failing to connect to Bitcoin (which eventually will timeout) during startup.

Also, are you using a dummy plug on your second GPU? With crossfire off, Windows turns cards off, even if they're in use, when there is no monitor plugged in.
newbie
Activity: 20
Merit: 0
June 18, 2011, 10:48:32 AM
My math gives about 469 combined for two 6850s if you can find the right settings, but its more likely to be in the 450s because you're on SDK 2.4. A 6870 is about as fast as a 5850 at their respective stock clocks, a 6850 is about 1/3rd slower than a 6870 at their respective stock clocks, and a 5850 is about 1/4th slower than a 5870.

So 450s for two overclocked 6850s is in the right ballpark if a overclocked 5870 is doing 450.

Thanks for the info man. I needed some closure lol. I think my main problem is the power supply in my computer under volting causing fluctuation performance (600w coolermaster). I'm going to try unplugging my 2 dvd burners and my second hdd to see what happens. So do you recommend reverting to an earlier revision of sdk?
legendary
Activity: 1162
Merit: 1000
DiabloMiner author
June 18, 2011, 10:20:57 AM
I'm pushing ~450mh/s on 2 6850's oc'd to 875 on the cores. Does that sound right? Based on what I've read on this forum, people with single 5870's push the same thing. So, in theory, I should be pushing at least 650 and at most 800. I'm running one of your recommended settings -v 19 -w 192 after hours of fishing for the sweet spot. I made a post in the newbie section, http://forum.bitcoin.org/index.php?topic=18632.0, because I was unable to post here. There's more info on the way I have things set up, in that thread. Another question I have is, is it safe to assume that your program reporting accept: 0 reject: 0 hw error: 0 is normal on solo mining? I'm guessing that it wont start processing a block until it finds one, so those numbers will remain at zero, right? I know, I'm a nub, but, you gotta crawl before you walk, lol.

My math gives about 469 combined for two 6850s if you can find the right settings, but its more likely to be in the 450s because you're on SDK 2.4. A 6870 is about as fast as a 5850 at their respective stock clocks, a 6850 is about 1/3rd slower than a 6870 at their respective stock clocks, and a 5850 is about 1/4th slower than a 5870.

So 450s for two overclocked 6850s is in the right ballpark if a overclocked 5870 is doing 450.
newbie
Activity: 20
Merit: 0
June 18, 2011, 10:03:37 AM
I'm pushing ~450mh/s on 2 6850's oc'd to 875 on the cores. Does that sound right? Based on what I've read on this forum, people with single 5870's push the same thing. So, in theory, I should be pushing at least 650 and at most 800. I'm running one of your recommended settings -v 19 -w 192 after hours of fishing for the sweet spot. I made a post in the newbie section, http://forum.bitcoin.org/index.php?topic=18632.0, because I was unable to post here. There's more info on the way I have things set up, in that thread. Another question I have is, is it safe to assume that your program reporting accept: 0 reject: 0 hw error: 0 is normal on solo mining? I'm guessing that it wont start processing a block until it finds one, so those numbers will remain at zero, right? I know, I'm a nub, but, you gotta crawl before you walk, lol.
legendary
Activity: 1162
Merit: 1000
DiabloMiner author
June 17, 2011, 01:09:34 PM
Another suggestion is to put the -f settings (with a short explanation) under the optimal configuration section of your post on the first page. I think a lot of people are doing tests without it leading them to think that your miner is slower. On my crappy 6750M without -f1 I get about 72-73Mhash, with -f1 I get about 77-78Mhash. My desktop interactivity is fine because of the ability of my onboard graphics I think. However, people with dedicated mining rigs probably wouldn't care about desktop interactivity anyway. Thanks for all your hard work!

I only recommend -f 1 for dedicated miners. For desktop users, -f 1 usually causes excessive lag.
Yeah I would probably agree, it is just the only mention of -f in the first post is to decrease performance. I didn't even know there was a -f command that could increase performance until a while after I started mining. Without -f, I probably would have went with phoenix miner because it displayed a higher hash rate.

Theres actually a lengthy argument about that. There is a large group of people, me included, that believe Phoenix displays hash rates about 5% too high. So, ymmv.
member
Activity: 78
Merit: 10
June 17, 2011, 01:02:48 PM
Another suggestion is to put the -f settings (with a short explanation) under the optimal configuration section of your post on the first page. I think a lot of people are doing tests without it leading them to think that your miner is slower. On my crappy 6750M without -f1 I get about 72-73Mhash, with -f1 I get about 77-78Mhash. My desktop interactivity is fine because of the ability of my onboard graphics I think. However, people with dedicated mining rigs probably wouldn't care about desktop interactivity anyway. Thanks for all your hard work!

I only recommend -f 1 for dedicated miners. For desktop users, -f 1 usually causes excessive lag.
Yeah I would probably agree, it is just the only mention of -f in the first post is to decrease performance. I didn't even know there was a -f command that could increase performance until a while after I started mining. Without -f, I probably would have went with phoenix miner because it displayed a higher hash rate.
legendary
Activity: 1162
Merit: 1000
DiabloMiner author
June 17, 2011, 12:06:55 PM
Another suggestion is to put the -f settings (with a short explanation) under the optimal configuration section of your post on the first page. I think a lot of people are doing tests without it leading them to think that your miner is slower. On my crappy 6750M without -f1 I get about 72-73Mhash, with -f1 I get about 77-78Mhash. My desktop interactivity is fine because of the ability of my onboard graphics I think. However, people with dedicated mining rigs probably wouldn't care about desktop interactivity anyway. Thanks for all your hard work!

I only recommend -f 1 for dedicated miners. For desktop users, -f 1 usually causes excessive lag.
member
Activity: 78
Merit: 10
June 17, 2011, 12:04:32 PM
Another suggestion is to put the -f settings (with a short explanation) under the optimal configuration section of your post on the first page. I think a lot of people are doing tests without it leading them to think that your miner is slower. On my crappy 6750M without -f1 I get about 72-73Mhash, with -f1 I get about 77-78Mhash. My desktop interactivity is fine because of the ability of my onboard graphics I think. However, people with dedicated mining rigs probably wouldn't care about desktop interactivity anyway. Thanks for all your hard work!
legendary
Activity: 1162
Merit: 1000
DiabloMiner author
June 17, 2011, 05:07:45 AM
Update: Fixes to deal with pushpool bugs: Added async sendwork, obsessive LP failure retrying, and time incrementing on nonce saturation getwork failure
legendary
Activity: 1162
Merit: 1000
DiabloMiner author
June 16, 2011, 11:36:56 PM
Hey Diablo,

I like your miner the best because it is one of the fastest, rather easy to setup and play around with settings, and it is compatible with multiple operating systems. However, virtually all pools have lately, had a very difficult time staying up and assigning work because of things such as DDoS, rampant increase in users, etc. I was wondering if you had any plans to implement a backup pool option (I have heard you can do this on GUIminer, but I have not ever used GUIminer, so not sure) where you can enter info for a backup pool in case your primary one goes down. I saw an interesting post where someone said they mine for two pools at the same time on each GPU. If one does go down, the other miner just ramps up from 50% to 100% (and I think this way the pools won't dog you for pool hopping because you have been doing some mining with either pool all along). Also, less importantly, something like a queue size (as in Phoenix, but hopefully your queue would be flushed properly).

Thanks,
Druas

I'm considering adding a feature to mine with multiple pools.

Also, running multiple miners does not work with my miner to do fallover... my miner does not stop mining during severe error conditions like that.
member
Activity: 78
Merit: 10
June 16, 2011, 05:10:38 PM
Hey Diablo,

I like your miner the best because it is one of the fastest, rather easy to setup and play around with settings, and it is compatible with multiple operating systems. However, virtually all pools have lately, had a very difficult time staying up and assigning work because of things such as DDoS, rampant increase in users, etc. I was wondering if you had any plans to implement a backup pool option (I have heard you can do this on GUIminer, but I have not ever used GUIminer, so not sure) where you can enter info for a backup pool in case your primary one goes down. I saw an interesting post where someone said they mine for two pools at the same time on each GPU. If one does go down, the other miner just ramps up from 50% to 100% (and I think this way the pools won't dog you for pool hopping because you have been doing some mining with either pool all along). Also, less importantly, something like a queue size (as in Phoenix, but hopefully your queue would be flushed properly).

Thanks,
Druas
legendary
Activity: 1162
Merit: 1000
DiabloMiner author
June 16, 2011, 11:45:52 AM
Hello!
Sorry for a newbie question... how can I build DiabloMiner from the source? (on Ubuntu)

Thanks,
Tom

Its in the op post: install maven, download from git, run mvn package

Oh! Sorry!  I read past that line many times!  Thanks!

The source seems to work slightly differently than the last distributed.  For instance, -v 19 -w 128 worked on the distributed, and on the source it produces HW errors.  Hashrate seems to be slower too.  Do you want to know?

-Tom

Like I told the guy above this post, the HW error is a pita. I'm trying to increase error reporting to catch this bug, so you're probably seeing something thats always been going on but was not sufficiently reported.
newbie
Activity: 10
Merit: 0
June 16, 2011, 08:12:36 AM
Hello!
Sorry for a newbie question... how can I build DiabloMiner from the source? (on Ubuntu)

Thanks,
Tom

Its in the op post: install maven, download from git, run mvn package

Oh! Sorry!  I read past that line many times!  Thanks!

The source seems to work slightly differently than the last distributed.  For instance, -v 19 -w 128 worked on the distributed, and on the source it produces HW errors.  Hashrate seems to be slower too.  Do you want to know?

-Tom
legendary
Activity: 1162
Merit: 1000
DiabloMiner author
June 16, 2011, 01:50:41 AM
DiabloD3, I just downloaded the latest DiabloMiner and I am trying to run DiabloMiner-Windows.exe with various -v options to see what's best for my GTX 480. I had been running phoenixminer with the VECTORS option, so I assume that it works using uint2s, but DiabloMiner-Windows errors out, then crashes my video driver with -v 2, -v 18 or -v 19. -v 1 works and I have yet to try any other values.

Here's the error I get:
Code:
[6/15/11 9:02:30 PM] Started
[6/15/11 9:02:30 PM] Connecting to: http://localhost:9378/
[6/15/11 9:02:31 PM] Using NVIDIA CUDA OpenCL 1.0 CUDA 4.0.1
[6/15/11 9:02:38 PM] Added GeForce GTX 480 (#1) (15 CU, local work size of 512)
[6/15/11 9:02:39 PM] ERROR: CL_INVALID_COMMAND_QUEUE error executing CL_COMMAND_READ_BUFFER on GeForce GTX 480 (Device 0).

[6/15/11 9:02:39 PM] ERROR: Failed to queue read buffer, error -36
[6/15/11 9:02:39 PM] ERROR: CL_OUT_OF_RESOURCES error waiting for idle on GeForce GTX 480 (Device 0).

Waiting...:02:39 PM] ERROR: Failed to queue read buffer, error -36
[6/15/11 9:02:39 PM] ERROR: CL_INVALID_COMMAND_QUEUE error executing CL_COMMAND_READ_BUFFER on GeForce GTX 480 (Device 0).

Waiting...:02:39 PM] ERROR: Failed to queue read buffer, error -36
Waiting...:02:39 PM] ERROR: Failed to queue read buffer, error -36
Waiting...:02:39 PM] ERROR: Failed to queue read buffer, error -36
[6/15/11 9:02:39 PM] ERROR: CL_OUT_OF_RESOURCES error waiting for idle on GeForce GTX 480 (Device 0).
Waiting...
The errors continue to repeat until the video driver crashes and Windows restarts the video driver.

Is -v 2 or higher expected to work with nVidia cards?

-v does not work on nvidia due to driver bugs.
newbie
Activity: 10
Merit: 0
June 15, 2011, 11:29:48 PM
DiabloD3, I just downloaded the latest DiabloMiner and I am trying to run DiabloMiner-Windows.exe with various -v options to see what's best for my GTX 480. I had been running phoenixminer with the VECTORS option, so I assume that it works using uint2s, but DiabloMiner-Windows errors out, then crashes my video driver with -v 2, -v 18 or -v 19. -v 1 works and I have yet to try any other values.

Here's the error I get:
Code:
[6/15/11 9:02:30 PM] Started
[6/15/11 9:02:30 PM] Connecting to: http://localhost:9378/
[6/15/11 9:02:31 PM] Using NVIDIA CUDA OpenCL 1.0 CUDA 4.0.1
[6/15/11 9:02:38 PM] Added GeForce GTX 480 (#1) (15 CU, local work size of 512)
[6/15/11 9:02:39 PM] ERROR: CL_INVALID_COMMAND_QUEUE error executing CL_COMMAND_READ_BUFFER on GeForce GTX 480 (Device 0).

[6/15/11 9:02:39 PM] ERROR: Failed to queue read buffer, error -36
[6/15/11 9:02:39 PM] ERROR: CL_OUT_OF_RESOURCES error waiting for idle on GeForce GTX 480 (Device 0).

Waiting...:02:39 PM] ERROR: Failed to queue read buffer, error -36
[6/15/11 9:02:39 PM] ERROR: CL_INVALID_COMMAND_QUEUE error executing CL_COMMAND_READ_BUFFER on GeForce GTX 480 (Device 0).

Waiting...:02:39 PM] ERROR: Failed to queue read buffer, error -36
Waiting...:02:39 PM] ERROR: Failed to queue read buffer, error -36
Waiting...:02:39 PM] ERROR: Failed to queue read buffer, error -36
[6/15/11 9:02:39 PM] ERROR: CL_OUT_OF_RESOURCES error waiting for idle on GeForce GTX 480 (Device 0).
Waiting...
The errors continue to repeat until the video driver crashes and Windows restarts the video driver.

Is -v 2 or higher expected to work with nVidia cards?
legendary
Activity: 1162
Merit: 1000
DiabloMiner author
June 15, 2011, 07:53:53 PM
Hello!
Sorry for a newbie question... how can I build DiabloMiner from the source? (on Ubuntu)

Thanks,
Tom

Its in the op post: install maven, download from git, run mvn package
newbie
Activity: 10
Merit: 0
June 15, 2011, 07:04:58 PM
Hello!
Sorry for a newbie question... how can I build DiabloMiner from the source? (on Ubuntu)

Thanks,
Tom
legendary
Activity: 1162
Merit: 1000
DiabloMiner author
June 15, 2011, 12:29:04 AM
Update: I seem to have fixed all the Nvidia problems for now. Also, connection errors on send work are no longer counted as rejects.

If anyone is having a high number of hardware errors on Radeons, speak up.
legendary
Activity: 1162
Merit: 1000
DiabloMiner author
June 12, 2011, 10:34:33 PM
So why should I not enable crossfire when mining I have heard to enable it and now not to. Now that I am not running crossfire and connecting to each card to start the miner my second card is down by nearly 1000 blocks. I am new to this stuff and all the research I try doing dosen't come up with much, I know people have talked about the dummy plugs and that sounds to be what I need to do. And just guessing but if I run a dummy plug to my other card I can start both miners from one card without having crossfire enabled correct? BTW I edited my post above yours this morning because I could not reply for some reason.

AMD themselves say if you want to use OpenCL and use all your cards effectively, you need to turn Crossfire off. The usual side effect of leaving it up is that all the GPUs after the first one are either stuck at 0% usage or produce absolute garbage.

In Windows, you need dummy plugs to make Windows turn the card on if its not in a Crossfire pair.
newbie
Activity: 10
Merit: 0
June 12, 2011, 04:05:12 PM
So why should I not enable crossfire when mining I have heard to enable it and now not to. Now that I am not running crossfire and connecting to each card to start the miner my second card is down by nearly 1000 blocks. I am new to this stuff and all the research I try doing dosen't come up with much, I know people have talked about the dummy plugs and that sounds to be what I need to do. And just guessing but if I run a dummy plug to my other card I can start both miners from one card without having crossfire enabled correct? BTW I edited my post above yours this morning because I could not reply for some reason.
Pages:
Jump to: