Thanks Bombadil for the quick upload!
BTW, what settings are you using for your 750ti? (core/mem clocks and do you over volt?)
+60core, +200mem, +31mv. I don't know if going lower could work too, but this is the highest I could get for both my EVGA SC one and Gigabyte one (same main desktop rig). So give it a go if you're into OC
|
|
|
Does someone have a link to a compiled version of the "killer groestl" release? I don't know the first thing about compiling from github... Yep, I have a compiled version for compute 3.0, 3.5 and 5.0: https://bitcointalk.org/index.php?topic=167229.msg7299983#msg7299983(Compute 2.0 isn't possible anymore, compute 3.5 & 5.0 are merged into one exe ^^" ccminer35-50.exe, this gets confusing)
|
|
|
So, does this seem correct for one 750ti?: -X11:2.6MH/s -Jackpot:5.5MH/s -NIST5: 8.4MH/s -Quark: 4.7MH/s -Groestl: 8MH/s (!!!!!! was 3.1MH/s before !!!!!) I used these numbers to generate following list for one 750ti: http://pastebin.com/xcFqtX9EStill old hashrates for X13. Want to see Christian's magic first Again, this list is only a snapshot for this moment, with current prices and difficulties, so lower your own estimates. 5.25MH seems to be a better estimate on jackpot unless my 3 brands are just all bad at overclocking. My EVGA FTW cards come overclocked with tdp raised so i'm able to get 5.5 out of those with +10 core and +400 memory, but my MSI Gaming get close and most of my Gigabyte 750Ti's look like they're averaging 5.25. Could do more but everytime I use normally stable overclocks with ccminer it spits out large hash numbers and stops working again. Happens on any of my rigs so I know it's not hardware or anything else. I've seen most people here with that issue as well, but hey if there's a fix I'd love to hear about so I can just leave them at the perfect overclock. Hah, yeah, should've taken the overclock down. Anyways, ratio's are shifting, as you can see GRS is becoming 'interesting' again I get such a freak-out only when it's too hot where my rig is ^^" Else +60core, +200mem, +31mV works fine for me (one EVGA SC and one Gigabyte)
|
|
|
So, does this seem correct for one 750ti?: -X11:2.6MH/s -Jackpot:5.5MH/s -NIST5: 8.4MH/s -Quark: 4.7MH/s -Groestl: 8MH/s (!!!!!! was 3.1MH/s before !!!!!) I used these numbers to generate following list for one 750ti: http://pastebin.com/xcFqtX9EStill old hashrates for X13. Want to see Christian's magic first Again, this list is only a snapshot for this moment, with current prices and difficulties, so lower your own estimates.
|
|
|
so I guess now jpc algo is using a lot more groestl Previously Groestl was also eliminated in the 1st and 2nd rounds, leaving only 1/4th of the hashes to be evaluated. Now with the faster Groestl I made it run Groestl in the 2nd round too. So half the hashes are fully evaluated now. It's faster in the end - this is what counts. Christian I'm <3'ing it, thank you so much, Christian! You're god
|
|
|
Looking pretty nice
|
|
|
Killer hangover is cured, so I added Killer Groestl in the quad version. This cudaminer revision is source code only and called V1.1
the following algos should now mine faster -quark -anime -x11 -jackpot -groestl -myr-gr
Tomorrow I will try to get the X13 algo from the forked version back into my code and call it V1.2, including a binary release.
Would you be so kind to test this V1.1 release out and report the speed-ups you're getting? I've only tried Windows compilation - will the Linux version still build fine?
What other patches would you guys like to have included in V1.2 or V1.3?
Christian
Ooh, nice! Busy compiling it Seems like you've already got rid of the MAXWELL_OR_FERMI constant, superb So ccminer35 and ccminer50 are the same now? Guess it will be ccminer35-50.exe from now on, no? Well, the diamond-groestl (dmd-gr) one is simple, you could add it for V1.1/1.2: https://github.com/KBomba/ccminer-KBomba/commit/d4541dbd77aa8a9585287234cf11f72aa1ace854
|
|
|
How can I mine same time on CPU and GPUs?
I'd suggest you run the CPU mining with -t 3, to use just 3 cores, leaving one free for GPU miner duties (relevant if that CPU coin miner is already optimized for AES-NI). In any case, you can play with the number of threads for CPU mining all the way up to 7, just not the default (which uses all 8 threads). I usually go into the task manager details and set the affinity of the GPU miner to be on my 1st core/thread (or the 1st two), and set the affinity of the CPU miner to the other ones, thus isolating one from the other. (if you're not using windows, please ignore ) ~ Myagui This makes me think that ccminer just has difficulties with cross-threading on the cpu. If work A is on core 0 and work B is on core 1, but you're also CPU-mining on those cores, ccminer will have to pause a bit, and he doesn't like that Also, Intel knows how to deal with threads, so this might be why a Celeron is a winner for ccminer. So yeah, try setting affinity to lock the processes on cores. Please report back if that did or didn't do it If it did it, it's something I will implement in my ccminer gui. Should you set the affinity per ccminer, like ccminer30 on core0, ccminer50 on core1, etc... or does it matter at all? I see no real difference when I put them in one thread on my system (i5-3470) But before you ask, I won't do a gui for minerd
|
|
|
x11 seems to be best right now as x13 profit has dropped like 3x since few days back..
X11 and X13 are uber crap. Not even profitable to mine with GPU's.... Yeah, right, so you suggest we'd all start mining scrypt again? Or why not SHA? Seriously, X11/X13/Jackpotcoin/Talkcoin/any other exotic algo, are the most profitable ones for GPU miners ATM. Not only does it use less power and generate less heat, they also don't have the competition of überasics, maybe an FPGA here and there, but that's about it. Actually, the newest trend is just to keep on inventing algo's. Let's see where that will bring us Mining still has a bright future, even for you, if you know what you're doing.
|
|
|
Hi, I'm trying to mine Darkcoin on GPUs and Quazarcoin on CPU - please help I have CPU i7-4470k and two GPU cards nVidia 750 Ti When I run the ccminer50 I get around 3.1MH for GPU both cards, for Darkcoin at pool, then when I run minerd for Quazarcoin, which is CPU only, it runs ok, but the mining of ccminer50 (Darkcoin, GPUs) drop down to very low numbers Is it because ccminer still uses CPU? On performance monitor the CPU usage is not bigger than 15% - ? Also, I tried to run minerd (Quazar, CPU only) with only 2 logical threads, it does not help at all. How can I mine same time on CPU and GPUs? Yep, for some unknown reason (probably pretty low-level) ccminer doesn't like other minerds or other cpu intensive processes, even with low cpu usages. Some people with very low-end AMD cpu's also reported problems, but I have yet to find a recent Celeron that can't handle ccminer
|
|
|
but one thing is true, amd gpu are dead
+1
|
|
|
But I got wasted today in a Munich Biergarten, so the killer groestl will have to wait until tomorrow. What awaits me is a killer hangover.
hey christian...whats about your headache? hope you feeling ok today. hope you upload the K-gröstl not this evening - or I have to compile it in the same condition as you were in yesterday. would be a mess Just wait max half an hour after he released it ^^" Compiles everywhere * Bombadil keeps on hitting refresh at the commit page
|
|
|
Yeah, stop mining it so it becomes profitable again The first X13 coin, Snakexcoin, appears at #7 on a profitability list for my GTX750TI's, next up is #10, Burnercoin ^^" (First and second are Jackpotcoin & Talkcoin)
|
|
|
Depending on who built that ccminer/cudaminer release, you might need instead the Visual Studio 2012 redistributable (Christian's releases I think would always just need the 2010 version 'cos that's what he builds with).
Cheers, ~ Myagui
My compiles were made in VS2010 too I already guessed problems could arise if I used 2012
|
|
|
Has anyone been able to successfully incorporate an API into ccminer? So that we can run it with something like cgwatcher, for instance.
I'm making a cgwatcher-like GUI for ccminer, creatively called CudaWatch. It will just redirect the console output on windows to my program, so that I can read all written strings and interpret them. This works rather well, but I'm busy with linking everything together and building a nice GUI (which takes up way too much time ) An API would be nice, but isn't necessary to me
|
|
|
i have put this link in the download and named it "v1.02-X13-Mod" hopefully your compiles work I'm testing with a 760 (3.0) and 2 750ti's, so those work (for me at least ) So Christian, if you're planning on incorporating X13 (and if possible, dmd-gr) in the next official release, together with killer groestl, it would truly be an awesome release
|
|
|
|