 Show Posts
|
|
Pages: [1] 2 »
|
|
hmm... not really... it's a bit complicated. The most gains in the hash are eliminating loops.
For example, the kernel has an #pragma unroll which removes loop and generates the call line by line. If a variable is unknown, #pragma unroll is not done. Also if cgminer is not compiled with -O3 option (most are being compiled -O2, I can assure you this), the kernel will not even be unrolled at all, so it's better for you to do the unrolling manually.
EDIT: oh, and my settings are these:
"intensity" : "19",
"vectors" : "1",
"worksize" : "256",
"kernel" : "scrypt",
"lookup-gap" : "2",
"thread-concurrency" : "24000",
"gpu-engine" : "1140,1140,1020",
"gpu-memclock" : "1250"
The last card was a lemon, doesn't want to get higher than 1020, that's what's you see on the stats above (about 633~ish) still pretty decent for hash/watt ratio.
|
|
|
|
ah-- no interest?  It requires no hardware modification. I was trying to compile cgminer to my Galaxy S4, managed to do it, but the kernel doesn't want to be accepted, so I modified the scrypt kernel code. The good is what you see above, higher hashrate-- you might even need to tune down (like I did, to reach a good hash/power ratio), the bad is the modification is specific for each configuration, so the kernel code I have is specific to mine. Bear in mind that cgminer can do a lot of things and can be configured to suite any GPU settings, but making it (the scrypt kernel) concentrate on a specific configuration will yield higher hashrates. So here's what I did: - edit scrypt130511.cl
- remove, as much as possible, looping structure
- use hardcoded constants as much as possible
- minimize arithmetic operations (addition, subtraction, multiplication, and division) by calculating the exact values from hardcoded settings (thread concurrency, worksize, gap)
- minimize calling custom functions and incorporate them to the function caller itself.
- save!
- remove all scrypt130511......bin files, because this is the compiled kernel that cgminer uses for the current setting
- start cgminer
The readability and maintainability of the code is of course affected, but we don't need this-- we want the kernel to be as efficient as possible! |
|
|
|
A month ago I was hashing like this:  You can see that the highest core-engine/memory there is 1145/1250. Fast forward now, with lower core-engine/memory settings:  Don't mind the Kh/s. I tweaked my BAMT to display it in Kh/s. See the numbers above? That's 700+ on the same MSI 7950 being reached with lower engine clocks, all other settings the same! Got you interested now? Just when you've achieved the maximum stable configuration of your rigs that are running for months now, this is a guide to achieve higher hashing performance out of your GPU. (Stay tuned-- I'm just excited to share this now with all the nights I've spent doing this, so I'll let you hang there for a while now... in the meantime, please post your current scrypt settings below) |
|
|
|
erm.. bump?
...
No answer clearly means you have to sort it out yourself. The GPU code is written to work with ATI library supported cards with specific drivers. If you want to it work with others, you'll be modifying the CL and/or GPU driver. However, it is a complete waste of time, since it will be a complete waste of electricity. well with all the other replies I thought it was missed, having this kind of answer is better since now I know I need to proceed with a different approach in solving it. |
|
|
|
erm.. bump? I managed to compile cgminer in Android and trying to run it in my SGS4 with opencl, and I have some problems running it (spare me the "why bother" response, I think running cgminer in my phone is kinda cool, and with the advancement of GPUs in the mobile space, you'll never know  ) Anyway-- running cgminer -n gives me: [2013-09-13 12:07:15] CL Platform 0 vendor: QUALCOMM
[2013-09-13 12:07:15] CL Platform 0 name : QUALCOMM Snapdragon(TM)
[2013-09-13 12:07:15] CL Platform 0 name : OpenCL 1.1 QUALCOMM build: (CL3579225)
Build Date: 04/22/13 Mon
Local Branch: adreno_sba_20130419
Remote Branch:
Local Patches:
Reconstruct Branch:
[2013-09-13 12:07:15] Platform 0 devices: 1
[2013-09-13 12:07:15] 0 QUALCOMM Adreno(TM)
[2013-09-13 12:07:15] 1 GPU devices max detected
So it should be ok, right? But trying to run cgminer with the usual parameters (--scrypt -o <host:port> -u <username> -p <password>) gives me: [2013-09-13 12:08:12] Started cgminer 3.1.1
[2013-09-13 12:08:12] Probing for an alive pool
Assertion failed: !spillIs.empty() && "No spill intervals?", file vendor/qcom/proprietary/gles/adreno200/llvm/lib/CodeGen/QGPURegAlloc.cpp, line 2628
I compiled cgminer with OpenCL Adreno SDK instead of AMD SDK (for obvious reasons). |
|
|
|
I managed to compile cgminer in Android and trying to run it in my SGS4 with opencl, and I have some problems running it (spare me the "why bother" response, I think running cgminer in my phone is kinda cool, and with the advancement of GPUs in the mobile space, you'll never know ) Anyway-- running cgminer -n gives me: [2013-09-13 12:07:15] CL Platform 0 vendor: QUALCOMM
[2013-09-13 12:07:15] CL Platform 0 name : QUALCOMM Snapdragon(TM)
[2013-09-13 12:07:15] CL Platform 0 name : OpenCL 1.1 QUALCOMM build: (CL3579225)
Build Date: 04/22/13 Mon
Local Branch: adreno_sba_20130419
Remote Branch:
Local Patches:
Reconstruct Branch:
[2013-09-13 12:07:15] Platform 0 devices: 1
[2013-09-13 12:07:15] 0 QUALCOMM Adreno(TM)
[2013-09-13 12:07:15] 1 GPU devices max detected
So it should be ok, right? But trying to run cgminer with the usual parameters (--scrypt -o <host:port> -u <username> -p <password>) gives me: [2013-09-13 12:08:12] Started cgminer 3.1.1
[2013-09-13 12:08:12] Probing for an alive pool
Assertion failed: !spillIs.empty() && "No spill intervals?", file vendor/qcom/proprietary/gles/adreno200/llvm/lib/CodeGen/QGPURegAlloc.cpp, line 2628
I compiled cgminer with OpenCL Adreno SDK instead of AMD SDK (for obvious reasons). |
|
|
|
Hey how many do you need? I still have some 1x-16x and 16x-16x left. We can meet up in Rotterdam. Payment can occur in LTC, BTC or EUR.
pm'ed  |
|
|
|
Still looking To be specific, somewhere I can drive in the range of Rotterdam/Zoetermeer/Leiden |
|
|
|
I broke one when I was moving my setup to a new box  , so I need a new one quickly. Let me know if someone has an extra that I can pick up immediately. I live in Den Haag. |
|
|
|
(I'll just do it in English for you) Well, you have a pretty good price there, I currently have 0.21 Euro per Kwh day and 0.18 night (which I think is about the cheapest you canget, and free, offer is not active anymore, sorry). This site is Dutch but you can put in your annual usages and stuff and it will show you the cheapest opstion: http://www.gaslicht.com/energie-vergelijkenI advise you to just look at that site again every year, becouse deals like 21 cents are most of the time only for one year, and choose the option which is then the cheapest. I hope that helped you, if you have any question, feel free to ask me. (But keep in mind I am not any sort of electrician  ) I can understand as long as I know the context, the problem is still speaking From there, it seems now the difference for my current provider is only .00xx, so not really worth the hassle in switching. Dank u wel! |
|
|
|
Excuse me, but I will ask in English (but you can reply in Dutch  ) Which electricity provider has the best tariff for mining? I'm currently with Nuon which is about 0.2281 p/kwh. |
|
|
|
|
shame, should have been perfect for me-- but I plan to run at least 3x 7950s and the motherboard doesn't support more than 2..!
|
|
|
|
No no lol your missing the point crosstools package totally not exits from the packages goto the link provided as in my post and browse through the packages it is not there the second error is because it can not find the needed crosstools hence it error out http://ipkg.nslu2-linux.org/feeds/optware/cs08q1armel/ try to find the crosstools No no no-- because you can't run "ipkg update" without ipkg-- and from the message you sent: That means you got "ipkg" already. There's nothing there in "/opt/lib/ipkg/lists/cross", but there is something in "/opt/lib/ipkg/lists" which is the cross file (that contains packages that you can download). EDIT: ah-- I see what you mean on the error. Ignore this:Cannot find package crosstool-native.
Check the spelling or perhaps run 'ipkg update'
ERROR: The following packages conflict with wget-ssl:
wget and do this: cd /volume1/@tmp
ipkg install -verbose_wget libidn
it will show you something like this (example for sysfsutils): DiskStation> ipkg install -verbose_wget sysfsutils
Installing sysfsutils (2.1.0-1) to root...
Downloading http://ipkg.nslu2-linux.org/feeds/optware/cs08q1armel/cross/unstable/sysfsutils_2.1.0-1_arm.ipk
take note of the url and do wget <url you got from the libidn> do the same thing for wget-ssl: ipkg install -verbose_wget wget-ssl
wget <url you got from the wget-ssl>
then do ipkg install filename1
ipkg install filename2
where filename1 and filename2 is the name of the .ipk file of libidn and wget-ssl. I'll put this in the guide above. |
|
|
|
First: which DSM version do you have? I see you got until ipkg update, that means you can get the precompiled binaries for the toolchain. When you try these commands nothing happens? ipkg install gcc
ipkg install make automake autoconf libtool binutils
ipkg install crosstool-native optware-devel
type in ipkg list_installed
and paste here the output. |
|
|
|
|
Since BitFurys are priced in euros, will you accept payment in euros (SEPA/paypal)?
|
|
|
|
Now this is the kind of shit I like to see. Are you using it with USB devices? How much power does it use? Good work!
The NAS I have (410j) consumes about 42w, but since it's always on, it's "free" electricity so to speak. The native compiling of cgminer/bfgminer should have been straightforward, but there are just some nags that you would encounter and I merely pointed the solution to it |
|
|
|
Hi Marto,
What's the 'theoretical' timeline on designing the custom board + production when you get the specs from Punin?
What's the expected cost of the custom board (with 16 chips)?
Finished design end of august. Specs as far the order is palced( bye E-mail from Punin) Production - ready tested design with placed orders for PCB and components end of September. Very rough estimate would be in 60 EUR range. Regards: Martin So theoretically when chips are received in October the custom boards are ready for assembly (1 week?) and delivery (another week?) |
|
|
|
|
Hi Marto,
What's the 'theoretical' timeline on designing the custom board + production when you get the specs from Punin?
What's the expected cost of the custom board (with 16 chips)?
|
|
|
|
|
