Haven't found a driver that works for me yet. (R9 290x) Tried various 14.x and 13.9. Tried with and without APP SDK in each instance. Every time the final line in cmd is Building OpenCL program and then the .exe crashes.
Tbh I've kinda given up, I've been on this for most of the day now with no success and I'm left thinking that I'm not going to land on the right avenue.
Going to wait and see what the next version brings.
My mistake, I meant DDU driver cleaner not DCCU. Try it: http://www.wagnardmobile.com/DDU/Then install 13.9 and APP SDK at default setting on the gpu, no OC. With only one gpu per rig setup. |
|
|
|
Is there any wiki/repository on plot speed vs cpu?
Not that I know of. Here's some basic performance metrics on Centos 6.5 with dcct's linux plotgen: i5-2400: 3400 nonces/min E5-2560v2: 7000 nonces/min E5-2670v2: 7100 nonces/min 2xE5-2560v2: 13900 nonces/min i7-3770k: 4300 nonces/min FX-8350 @ 4.2GHz: 3500 nonces/min adding to that: Intel® Core™ i7-2600 Quad-Core: ~3900 nonces/min i7-5960X: 7400 nounces/min i7-5930K: 6130 nounces/min Baller status, 3~4 7970 can be bought for the price of that i7-5960X....LOL |
|
|
|
hey guys, I tried the GPU plotter but GPU driver crashes every time it starts plotting  second thing, I've noticed kinda big amount of data being transfered while plotting/mining burst. It can download even about 500MB per day, while mining other coins it was about 10MB/day. It might put me out from burst train to the moon, because I have limited data transfer:( Does it download data only while plotting or while mining? Try with different version of catalyst software. There are still different version of gpu to be tested but my mate successfully plotting with 290x twin frozer x10 times faster than his i7 processor. If you gpu is same model, pm for specs  I tried on 2x 7950 and the driver was crashing. Then I tried on another pc with single 7950 and it doesn't even start to plot, it crashes after 'Building OpenCL program'. what catalyst version does your mate have? I'm pretty sure I have the newest drivers. Supposedly they are on 13.9. I just did a clean uninstall and installing 13.9 on a HD7970. Yes my mate is on 13.9, i told him to download this version as it was mentioned here before and it worked straight on his 290x but he said on same pc it didnt work on 280x, so hope driver is an issue. You can try installing different version and see if one works for you. I used the DCCU gpu driver cleaner to do a clean uninstall of 14.7, then installed 13.9 with APP SDK 2.9.1 to get it working. I used a stagger size of 4096 and 128 thread, that was the max it would let me do on the HD7970. Nonce/min of 12880 at a load of 60%~100% with system ram less then 3GB out of 8GB used. Temps are under 50C~47C also. |
|
|
|
i had bought at 800 and now the price is 400  You have to understand that in the beginning with any coin, sell 25%~50% of what you mine and hold the rest. Then when prices drop 50% or less, buy. So, at least you make something and/or break even with the potential to gain when there's an uptrend in the market. |
|
|
|
dead coin is dead  How long have you been in the crypto game? If you can't handle a temporary downturn, then maybe this isn't the thing for you and also the stock market isn't ideal for you too. In 2 years, I've seen some huge swings at times when the right coin innovation comes out and/or PR on a coin. |
|
|
|
|
Happy to say, I got the GPU plotter working on AMD 13.9 with the newest APP SDK 2.9.1 (32 or 64 bit). Stagger size of 4096 and threads count of 128 on a HD7970. I did a clean uninstall with DCCU gpu driver cleaner. 12288 nonce/min is my current rate at the moment, will update if it's stable. Going to try it with a HD6970 and HD6950 2GB card also.
|
|
|
|
hey guys, I tried the GPU plotter but GPU driver crashes every time it starts plotting second thing, I've noticed kinda big amount of data being transfered while plotting/mining burst. It can download even about 500MB per day, while mining other coins it was about 10MB/day. It might put me out from burst train to the moon, because I have limited data transfer:( Does it download data only while plotting or while mining? Try with different version of catalyst software. There are still different version of gpu to be tested but my mate successfully plotting with 290x twin frozer x10 times faster than his i7 processor. If you gpu is same model, pm for specs I tried on 2x 7950 and the driver was crashing. Then I tried on another pc with single 7950 and it doesn't even start to plot, it crashes after 'Building OpenCL program'. what catalyst version does your mate have? I'm pretty sure I have the newest drivers. Supposedly they are on 13.9. I just did a clean uninstall and installing 13.9 on a HD7970. |
|
|
|
Maybe the 280x and 290x memory system works/handles better with the plotter compared to hd7970/7990, that why it kind of work on those cards.
Maybe it's not just the card. I am running the same card (R9 290) and similar specs to the 1 or 2 reported working gpu plotters. However I can not get mine to start plotting as the .exe crashes. I've tried various definitions for the batch file and various driver versions with and without SDK. There is something else I am missing and until I find it I will sit here and stare at my screen :/ Ava. Did you try 13.x drivers and install APP SDK? |
|
|
|
How much memory do you have in your GPU?
The most I can gen @ is 3000 on a higher end card.
The lower end ones with 1GB constantly fail with the error posted above.
It's a 3GB R9 280x... Hi guys, I am aware of this issue. The fact is that I have to create two full size buffers on the GPU side to reduce thread-local memory consumption. Thus the memory amount needed on the CPU side has to be doubled to get an estimate of what is needed on the GPU side. As an example, for a stagger size of 4000 you will need 1GB RAM on CPU side and more than 2GB (exactly (PLOT_SIZE + 16) x stagger) on GPU side (doesn't include here the local buffers and the kernel code itself). Once I have a stable version (really soon  ), I will work on this particular problem. Please, consider also to test a version for nvidia cards! Afaik there isn't yet a user that was able to start the plot generator on nvidia gpus The next version should work on both NVIDIA an AMD cards. Still need some test but sounds promising. I'm down to be a tester with a HD7970, HD6950 2GB and HD6970. This 16TB is going take a long time to plot anyway, about 30+ days on a old Core i5 750 with all 4 thread plotting. |
|
|
|
How much memory do you have in your GPU?
The most I can gen @ is 3000 on a higher end card.
The lower end ones with 1GB constantly fail with the error posted above.
It's a 3GB R9 280x... Hi guys, I am aware of this issue. The fact is that I have to create two full size buffers on the GPU side to reduce thread-local memory consumption. Thus the memory amount needed on the CPU side has to be doubled to get an estimate of what is needed on the GPU side. As an example, for a stagger size of 4000 you will need 1GB RAM on CPU side and more than 2GB (exactly (PLOT_SIZE + 16) x stagger) on GPU side (doesn't include here the local buffers and the kernel code itself). Once I have a stable version (really soon ), I will work on this particular problem. Maybe the 280x and 290x memory system works/handles better with the plotter compared to hd7970/7990, that why it kind of work on those cards. |
|
|
|
What is going on? Cant get it to run.  vga driver ? 14.7 ? 14.6 ? fix it to 13.x I am on 14.x, ill back down to 12.x and see if that fixes it. Thanks! 13.9 works with 280x I cant even get it to recognize my GPU what am i doing wrong? Same problem with 13.9 EDIT : Gigabyte Radeon HD 7970 OC (GV-R797OC-3GD) is my GPU So far it looks like it works on the R9 series cards: 280x and 290x on 13.x AMD drivers. |
|
|
|
Advices? (GTX 770 2GB) 100 bursts bounty C:\Users\x\Desktop>gpuPlotGenerator.exe C:\Users\x\Desktop\plots 7733xxxxxxxxxxx 5700001 2000000 1000 64
GPU plot generator v1.1.0
Author: Cryo
Bitcoin: 138gMBhCrNkbaiTCmUhP9HLU9xwn5QKZgD
Burst: BURST-YA29-QCEW-QXC3-BKXDL
--------------
Path: C:\Users\x\Desktop\plots
Nonces: 5700001 to 7700001 (488 GB)
Process memory: 250MB
Threads number: 64
--------------
Retrieving OpenCL platform
Retrieving OpenCL GPU device
Creating OpenCL context
Creating OpenCL command queue
Creating CPU buffer
Creating OpenCL GPU generation buffer
Creating OpenCL GPU scoops buffer
Creating OpenCL program
Building OpenCL program
Creating OpenCL kernel
Setting OpenCL kernel arguments
Generating from nonce #5700001
An OpenCL error occured in the generation process, aborting...
>>> [-54] Error in kernel launch
C:\Users\x\Desktop>pause
The -54 error is "CL_INVALID_WORK_GROUP_SIZE". Maybe 64 threads is too high for your GPU. Try a lower number like 32. Tried with 32, 16 and also 1  It doesn't work Seems like only the AMD 290X is the only card that successfully runs the gpu plotter. |
|
|
|
hey guys i have tried to make a bat to run the gpu plotting (i have R9 290x) i downloaded the gpu plotter but when i run the bat i get not a valid win32 application. any ideas?
Put "pause" on the end of the cmd line in the bat file to troubleshoot what is causing it. |
|
|
|
Ok there is tested config for 280x: gpuPlotGenerator.exe plots myaccount 68847637 18500000 4000 64
I'm getting arround 15k per minute. Great job. Thank you.  You put 64 instead of 256? I'll try 64 to see if i got more too Replaced the openCL dll for the error - now Everytime - 'Display driver has stopped responding'. All at stock settings - no OC. FUCK man this is bullshit. I don't understand how a 280X works, but a 7970 and 7990 don't work. They are virtually the same core, minus a couple of minor differences. Totally agree - Its the fucking AMD drivers I bet - Constant problems, one thing fixed, something else unknown gets broken. What version of Catalyst is everyone using who has to WORKING 100%? 14.6 yea but you have a 290x no? better to write that since we are talking about 280X... and they are NOT the same.. If someone with 280x want to try to downgrade to 13.xx that might be a temporarily fix. Hey guy, you are not alone, so calm down, if you're not happy, try to resolve the problem yourself Better yet, everyone should help each other to resolve this issue. I'm semi-confused why this wouldn't work on 7970, since the 280x just has a couple little more features and better ram structure. But they are still the same process and mostly the same main core technologies. Do you try with other threads value, and other stagger size? Yes, I've tried combo's of 1024 to 4096 staggers sizes. 64 to 256 threads values and all end up the same still. Maybe because the Hawaii has 4GB ram could be why it work, I might be wrong. The 7970 has only 3GB and 7970 is 3GB per gpu. |
|
|
|
Ok there is tested config for 280x: gpuPlotGenerator.exe plots myaccount 68847637 18500000 4000 64
I'm getting arround 15k per minute. Great job. Thank you. You put 64 instead of 256? I'll try 64 to see if i got more too Replaced the openCL dll for the error - now Everytime - 'Display driver has stopped responding'. All at stock settings - no OC. FUCK man this is bullshit. I don't understand how a 280X works, but a 7970 and 7990 don't work. They are virtually the same core, minus a couple of minor differences. Totally agree - Its the fucking AMD drivers I bet - Constant problems, one thing fixed, something else unknown gets broken. What version of Catalyst is everyone using who has to WORKING 100%? 14.6 yea but you have a 290x no? better to write that since we are talking about 280X... and they are NOT the same.. If someone with 280x want to try to downgrade to 13.xx that might be a temporarily fix. Hey guy, you are not alone, so calm down, if you're not happy, try to resolve the problem yourself Better yet, everyone should help each other to resolve this issue. I'm semi-confused why this wouldn't work on 7970, since the 280x just has a couple little more features and better ram structure. But they are still the same process and mostly the same main core technologies. |
|
|
|
New client Update 1.1.0Backend is caught up to date with nxt 1.2.8. Nxt new APIs are usable. Hard fork on block 11800 to enable new nxt features(encrypted messaging, DGS). UI is not yet updated for new features. Move your burst_db folder from 1.0.3 into the new folder to avoid a resync. Db upgrade will take a few minutes though. https://mega.co.nz/#!3xZ2DIwR!QQnamRK_JCz-QvFwb_A9gS6_tWCxAiFtVoqI8Unqlhgsha256: f9d2a2d199993088eb9a03f1cb282ecbb36e704a4b7c609d5e878573752dad30 yes yes yes That means Bittrex will be able to enable Burst on their market. |
|
|
|
Ok there is tested config for 280x: gpuPlotGenerator.exe plots myaccount 68847637 18500000 4000 64
I'm getting arround 15k per minute. Great job. Thank you. You put 64 instead of 256? I'll try 64 to see if i got more too Replaced the openCL dll for the error - now Everytime - 'Display driver has stopped responding'. All at stock settings - no OC. FUCK man this is bullshit. I don't understand how a 280X works, but a 7970 and 7990 don't work. They are virtually the same core, minus a couple of minor differences. Totally agree - Its the fucking AMD drivers I bet - Constant problems, one thing fixed, something else unknown gets broken. What version of Catalyst is everyone using who has to WORKING 100%? 14.6 yea but you have a 290x no? better to write that since we are talking about 280X... and they are NOT the same.. If someone with 280x want to try to downgrade to 13.xx that might be a temporarily fix. I have just downgraded after using driversweeper - still 'display driver has stopped responding' on 'generating from nonce' Strange if this plotter is optimized for the R series cards only. Most of the 7xxx's cards are similar with the R series cards. |
|
|
|
Ok there is tested config for 280x: gpuPlotGenerator.exe plots myaccount 68847637 18500000 4000 64
I'm getting arround 15k per minute. Great job. Thank you. You put 64 instead of 256? I'll try 64 to see if i got more too Replaced the openCL dll for the error - now Everytime - 'Display driver has stopped responding'. All at stock settings - no OC. FUCK man this is bullshit. I don't understand how a 280X works, but a 7970 and 7990 don't work. They are virtually the same core, minus a couple of minor differences. |
|
|
|
Hi everyone, After many hours of setup I finally made it. I have a 1Tb generation in progress and 3x100Gb already finished. I would like to test the V2 pool but I haven't any BURST for now. Could someone send me 1 BURST to test it please ? Here is my address : BURST-YA29-QCEW-QXC3-BKXDL. Regarding the plot generation, I found an OpenCL implementation of Shabal ( https://github.com/aznboy84/X15GPU/blob/master/kernel/shabal.cl) that could be used to make a GPU version of the generator. I will try to work on it when I have some free time. Regards Hi everyone, As promised I have been working on a GPU plot generator on the last few days. I made a little program built on top of OpenCL, and it seems to work pretty well in CPU mode. Unfortunately, I can't test the GPU mode as it requires a very powerfull graphic card (with at least 46kB private memory per compute unit, because the algorithm needs at least 4096*64 static bytes to store an entire plot). Here is a preview you can test for now : gpuPlotGenerator-src-1.0.0.7z : https://mega.co.nz/#!bcF2yKKL!3Ud86GaibgvwBehoxkbO4UNdiBgsaixRx7ksHrgNbDI gpuPlotGenerator-bin-win-x86-1.0.0.7z : https://mega.co.nz/#!HJsziTCK!UmAMoEHQ3z34R4RsXoIkYo9rYd4LnFtO_pw-R4KObJs I will build another release in the end of the day with some minor improvements (threads per compute unit selection, output of OpenCL error codes, improvement of the Makefile to generate the distribution directly). I will also try to figure out another mean to dispatch the work between the GPU threads to reduce the amount of private memory needed by the program. For the windows people, you can use the binary version directly. For the linux people, just download the source archive, make sure to modify the OpenCL library and lib path in the makefile (and maybe the executable name), and build the project via "make". To run the program, you need the "kernel" and the "plots" directories beside the executable. The executable usage is : ./gpuPlotGenerator <address> <start nonce> <nonces> <stagger size> The parameters are the same as the original plot generator, without the threads number. If you find bugs or if you want some new features, let me now. If you want to support me, here are my Bitcoin and Burst addresses : Bitcoin: 138gMBhCrNkbaiTCmUhP9HLU9xwn5QKZgD Burst: BURST-YA29-QCEW-QXC3-BKXDL Regards Unfortunately, I can't test the GPU mode as it requires a very powerfull graphic card (with at least 46kB private memory per compute unit, because the algorithm needs at least 4096*64 static bytes to store an entire plot).
It's nice to see someone else working on this, since I seem to have failed in it. Private memory is actually part of global on AMD cards, so storing it in private isn't any better than just using global for everything; it's local that needs to aimed for for the massive speedup. No AMD cards have more than 64KB local per workgroup, which makes storing it all in local impossible however. I haven't tried your implementation yet, but on my own first attempt, I also used global on everything also, and the result was faster than the java plotter, but slower than dcct's c plotter. My 2nd attempt used a 32KB local buffer I rotated through for storing the currently being hashed stuff, however I couldn't figure out how to get it copied also to global fast enough, and the local -> global copy killed the performance. You might be interested in those kernels here: https://bitcointalk.org/index.php?topic=731923.msg8695829#msg8695829Thanks, I will look at your kernels to see if I can find a better solution. Here is the new version. I reduced the amount of memory used from 40KB to about 1KB per unit. The only drawback is that it requires twice the global memory as before. I will search a mean to reduce this overhead later. In CPU mode, it all goes pretty well (when no graphic card is detected). The GPU mode is still kind of buggy on my graphic card (an old GeForce 9300M GS), don't know the exact reason yet. Sometimes it works, sometimes not. I will try to fix this issue tomorrow. Here are the files : gpuPlotGenerator-src-1.1.0.7z : https://mega.co.nz/#!iYFWAL5B!BvtmRQ5qGq4gGwjDglFNtDtNIX4LDaUvATBtClBdTlQ gpuPlotGenerator-bin-win-x86-1.1.0.7z : https://mega.co.nz/#!aBVGBBQD!tBsRtb8VrHR12_anrFTrl41U0fPQu_OqFnxyi5nCyBY For the linux users, the Makefile has a new target named "dist" that builds and copy/paste all the necessary files to the "bin" directory. The executable usage is : ./gpuPlotGenerator <path> <address> <start nonce> <nonces> <stagger size> <threads> <path> : the path to the plots directory <threads> : number of parrallel threads for each work group Found the "randomness" cause. NVIDIA is caching the kernel after the first build and rebuild it from time to time. By cleaning the cache, I can force the kernel build and speed up the debugging process. I will notify you as soon as the crash cause is found and corrected. I just got it to run for a bit, but crashes after a few seconds on the AMD HD7970. Exactly the same here on the HD 7990. App crashes on "building openCL platform". fault name: amdocl.dll Exactly the same error in Windows. I'm on the 14.7 driver. |
|
|
|
Hi everyone, After many hours of setup I finally made it. I have a 1Tb generation in progress and 3x100Gb already finished. I would like to test the V2 pool but I haven't any BURST for now. Could someone send me 1 BURST to test it please ? Here is my address : BURST-YA29-QCEW-QXC3-BKXDL. Regarding the plot generation, I found an OpenCL implementation of Shabal ( https://github.com/aznboy84/X15GPU/blob/master/kernel/shabal.cl) that could be used to make a GPU version of the generator. I will try to work on it when I have some free time. Regards Hi everyone, As promised I have been working on a GPU plot generator on the last few days. I made a little program built on top of OpenCL, and it seems to work pretty well in CPU mode. Unfortunately, I can't test the GPU mode as it requires a very powerfull graphic card (with at least 46kB private memory per compute unit, because the algorithm needs at least 4096*64 static bytes to store an entire plot). Here is a preview you can test for now : gpuPlotGenerator-src-1.0.0.7z : https://mega.co.nz/#!bcF2yKKL!3Ud86GaibgvwBehoxkbO4UNdiBgsaixRx7ksHrgNbDI gpuPlotGenerator-bin-win-x86-1.0.0.7z : https://mega.co.nz/#!HJsziTCK!UmAMoEHQ3z34R4RsXoIkYo9rYd4LnFtO_pw-R4KObJs I will build another release in the end of the day with some minor improvements (threads per compute unit selection, output of OpenCL error codes, improvement of the Makefile to generate the distribution directly). I will also try to figure out another mean to dispatch the work between the GPU threads to reduce the amount of private memory needed by the program. For the windows people, you can use the binary version directly. For the linux people, just download the source archive, make sure to modify the OpenCL library and lib path in the makefile (and maybe the executable name), and build the project via "make". To run the program, you need the "kernel" and the "plots" directories beside the executable. The executable usage is : ./gpuPlotGenerator <address> <start nonce> <nonces> <stagger size> The parameters are the same as the original plot generator, without the threads number. If you find bugs or if you want some new features, let me now. If you want to support me, here are my Bitcoin and Burst addresses : Bitcoin: 138gMBhCrNkbaiTCmUhP9HLU9xwn5QKZgD Burst: BURST-YA29-QCEW-QXC3-BKXDL Regards Unfortunately, I can't test the GPU mode as it requires a very powerfull graphic card (with at least 46kB private memory per compute unit, because the algorithm needs at least 4096*64 static bytes to store an entire plot).
It's nice to see someone else working on this, since I seem to have failed in it. Private memory is actually part of global on AMD cards, so storing it in private isn't any better than just using global for everything; it's local that needs to aimed for for the massive speedup. No AMD cards have more than 64KB local per workgroup, which makes storing it all in local impossible however. I haven't tried your implementation yet, but on my own first attempt, I also used global on everything also, and the result was faster than the java plotter, but slower than dcct's c plotter. My 2nd attempt used a 32KB local buffer I rotated through for storing the currently being hashed stuff, however I couldn't figure out how to get it copied also to global fast enough, and the local -> global copy killed the performance. You might be interested in those kernels here: https://bitcointalk.org/index.php?topic=731923.msg8695829#msg8695829Thanks, I will look at your kernels to see if I can find a better solution. Here is the new version. I reduced the amount of memory used from 40KB to about 1KB per unit. The only drawback is that it requires twice the global memory as before. I will search a mean to reduce this overhead later. In CPU mode, it all goes pretty well (when no graphic card is detected). The GPU mode is still kind of buggy on my graphic card (an old GeForce 9300M GS), don't know the exact reason yet. Sometimes it works, sometimes not. I will try to fix this issue tomorrow. Here are the files : gpuPlotGenerator-src-1.1.0.7z : https://mega.co.nz/#!iYFWAL5B!BvtmRQ5qGq4gGwjDglFNtDtNIX4LDaUvATBtClBdTlQ gpuPlotGenerator-bin-win-x86-1.1.0.7z : https://mega.co.nz/#!aBVGBBQD!tBsRtb8VrHR12_anrFTrl41U0fPQu_OqFnxyi5nCyBY For the linux users, the Makefile has a new target named "dist" that builds and copy/paste all the necessary files to the "bin" directory. The executable usage is : ./gpuPlotGenerator <path> <address> <start nonce> <nonces> <stagger size> <threads> <path> : the path to the plots directory <threads> : number of parrallel threads for each work group Found the "randomness" cause. NVIDIA is caching the kernel after the first build and rebuild it from time to time. By cleaning the cache, I can force the kernel build and speed up the debugging process. I will notify you as soon as the crash cause is found and corrected. I just got it to run for a bit, but crashes after a few seconds on the AMD HD7970. |
|
|
|
|
Show Posts
