Pages:
Author

Topic: ### A ChainWorks Industries (CWI) Project - Granite | Simple Solid Stable | GRN - page 17. (Read 74297 times)

legendary
Activity: 2912
Merit: 1091
--- ChainWorks Industries ---
hi all ...

thefarm has been successfully upgraded to fedora 22 x 64 and cuda 7.5 ...

one machine has issues - but is currently manually running ... a full reinstall may be necesary on that system ...

the amdbranch of thefarm is almost completely upgraded to fedora 22 x64 and catalyst 15.7 ( fglrx 15.20 ) using kde ...

ill start posting to the granite forum ( currently still testing at http://granitecoin.com/forum/ ) very shortly so that we can keep focussed on 'all things granite' ...

tanx for your patience ...

#crysx
legendary
Activity: 2912
Merit: 1091
--- ChainWorks Industries ---
Hi Crysx,

Do you have this result in x11 mining ?
Because with ccminer 1.5.7 i have 3.2 / gpu (750ti 2go gainward) on win 7 64bits

Huh  Shocked

i can set a test now for that on the machine that was mining purely x11 ...

f22x64c7.0.28 was hashing at approx 13100KH ...

this machine is now on c7.5 - so will test that over the next hour ( which seems to be the most reliable timeframe for the miner to settle ) ... it will be using ccminer-spmod v70 ( as it was prior ) ...

back with a the results ...

i couldnt tell you what the results were with x11 on c6.5 - as i wasnt running x11 with c6.5 - but i can say that the hashrate was around the 2800KH per card on the gigabyte 750ti oc lp ...

back in an about an hour with the result ...

#crysx

silly me ...

i already had another test system doing this same thing ...

it is f22x64c7.5 with 2 x gigabyte 750ti oc ( one lp card - one 'standard' oc which is the powered card ) ... smodv70 at xintensity -X 29 - mining granite on grn-pool.granitecoin.com ...

-------

[2015-10-11 11:15:02] Stratum difficulty set to 0.03125
[2015-10-11 11:15:24] GPU #1: GeForce GTX 750 Ti, 2815 (T= 74C F= 60% C=1189/2700)
[2015-10-11 11:15:25] accepted: 8218/8429 (97.50%), 5582 kH/s yes!
[2015-10-11 11:15:30] GPU #0: GeForce GTX 750 Ti, 2767 (T= 55C F= 38% C=1176/2700)
[2015-10-11 11:15:41] GPU #0: GeForce GTX 750 Ti, 2767 (T= 55C F= 38% C=1176/2700)
[2015-10-11 11:15:42] accepted: 8219/8430 (97.50%), 5582 kH/s yes!
[2015-10-11 11:15:53] GPU #1: GeForce GTX 750 Ti, 2815 (T= 74C F= 60% C=1189/2700)
[2015-10-11 11:15:54] accepted: 8220/8431 (97.50%), 5582 kH/s yes!
[2015-10-11 11:16:11] GPU #0: GeForce GTX 750 Ti, 2767 (T= 55C F= 38% C=1176/2700)
[2015-10-11 11:16:11] accepted: 8221/8432 (97.50%), 5582 kH/s yes!
[2015-10-11 11:16:23] GPU #1: GeForce GTX 750 Ti, 2815 (T= 74C F= 60% C=1189/2700)
[2015-10-11 11:16:41] GPU #0: GeForce GTX 750 Ti, 2767 (T= 55C F= 38% C=1176/2700)
[2015-10-11 11:16:47] GPU #1: GeForce GTX 750 Ti, 2816 (T= 74C F= 60% C=1189/2700)
[2015-10-11 11:16:48] accepted: 8222/8433 (97.50%), 5582 kH/s yes!
[2015-10-11 11:16:57] mining-ace.granitecoin.com:6001/ x11 block 225042

-------

as you can see - c7.5 is running stable with slightly better than f20x64c6.5 performance ... but not by very much ...

i honestly believe that if the devs optimize their code for c7.5 - we will see much better improvement in hashrates across the baord ...

#crysx
legendary
Activity: 2912
Merit: 1091
--- ChainWorks Industries ---
Hi Crysx,

Do you have this result in x11 mining ?
Because with ccminer 1.5.7 i have 3.2 / gpu (750ti 2go gainward) on win 7 64bits

Huh  Shocked

i can set a test now for that on the machine that was mining purely x11 ...

f22x64c7.0.28 was hashing at approx 13100KH ...

this machine is now on c7.5 - so will test that over the next hour ( which seems to be the most reliable timeframe for the miner to settle ) ... it will be using ccminer-spmod v70 ( as it was prior ) ...

back with a the results ...

i couldnt tell you what the results were with x11 on c6.5 - as i wasnt running x11 with c6.5 - but i can say that the hashrate was around the 2800KH per card on the gigabyte 750ti oc lp ...

back in an about an hour with the result ...

#crysx
member
Activity: 225
Merit: 17
Hi Crysx,

Do you have this result in x11 mining ?
Because with ccminer 1.5.7 i have 3.2 / gpu (750ti 2go gainward) on win 7 64bits

Huh  Shocked
legendary
Activity: 2912
Merit: 1091
--- ChainWorks Industries ---
has cuda 7.5 finally caught up to cuda 6.5 speeds? i thought 6.5 was most efficient in mining.

the first of the systems are completing now ...

compiling ccminer-spmod ( which we currently use for the nvidia branch of thefarm ) has compiled flawlessly ...

the performance seems to be slightly better than the c6.5 compiles - in that - they have 'slightly' better numbers at the miner side ...

the rest of the machines are yet to finalize the full upgrade to f22x64 - which will then be upgraded to c7.5 ... so the testing will be ongoing for the rest of the day when it is completed ...

so far though - the results look quite promising AND the compilations are not broken ... ie - there seem to be no 'cpu validation errors' compiling with c7.5 as opposed to compiling with c7.0.28 - which was broken in more ways that just 'cpu validation errors' ... that version gave poor hashrates and the majority of algos just refused to work properly - being erroneous in many different ways ...

an example - using thefarm gigabyte 750ti oc lp cards ... test machine ( 6 x gigabyte 750ti oc lp - stock with NO oc - as that is how we test ALL the units ) using ccminer-spmod v70 compiled in f20x64c6.5 gave an average hashrate of 34700KH ... f22x64c7.5 is giving an average hashrate of ( and this is current as of writing ) 34770KH ... as seen below ...

[2015-10-10 12:20:37] mining-ace.granitecoin.com:6002/ quark block 2287534
[2015-10-10 12:20:37] GPU #3: GeForce GTX 750 Ti, 5799 (T= 79C F= 65% C=1176/2700)
[2015-10-10 12:20:37] GPU #1: GeForce GTX 750 Ti, 5824 (T= 79C F= 64% C=1176/2700)
[2015-10-10 12:20:37] GPU #0: GeForce GTX 750 Ti, 5631 (T= 80C F= 65% C=1097/2700)
[2015-10-10 12:20:37] GPU #2: GeForce GTX 750 Ti, 5886 (T= 79C F= 65% C=1137/2700)
[2015-10-10 12:20:37] GPU #4: GeForce GTX 750 Ti, 5861 (T= 80C F= 65% C=1150/2700)
[2015-10-10 12:20:37] GPU #5: GeForce GTX 750 Ti, 5787 (T= 79C F= 64% C=1150/2700)
[2015-10-10 12:20:43] GPU #4: GeForce GTX 750 Ti, 5845 (T= 80C F= 65% C=1137/2700)
[2015-10-10 12:20:44] accepted: 507/510 (99.41%), 34770 kH/s yes!
[2015-10-10 12:20:49] GPU #1: GeForce GTX 750 Ti, 5834 (T= 79C F= 64% C=1163/2700)
[2015-10-10 12:20:49] accepted: 508/511 (99.41%), 34771 kH/s yes!

i must admit - that the test machine surprised me - as i did not expect that it would perform so well ... as c7.0 was a broken release ...

when the devs release code specific to c7.5 - i believe it will be a far better performance level that what we are seeing at the moment ... especially when you consider that nvidia themselves have resolved a great many issues within the compilation sphere of cuda 7.5 ...

so in answer to your question - yes ... i believe that c7.5 has not only caught up to c6.5 hashing speeds - but if this test is anything to go by - has surpassed them ... and that is still using the current code optimizations that are intended for c6.5 that the devs release - as far as i know ...

which is why after 12 hours of testing - thefarm is being fully upgraded to take advantage of not only the optimizations - but also the current compilation advances ...

and in fedora 22 x64 - it seems to work flawlessly ...

btw - the amd branch of thefarm is no longer being used publicly for rental ... it is now exclusively internal - and for the projects that are assigned to it by the dev involved ...

the amd branch is also currently being updated to f22x64 with catalyst v15.7 and sgminer 5.1.1-19-g8bf1-dirty ( modified sgminer-dev git - for use with the specialized / optimized binaries ) ...

the amd branch is a little more complex to upgrade incomparison to nvidia - as amd are notorious for a slow ( if ever ) release of linux drivers ... as such - the install needs to be 'hacked' into existence using kde instead of gnome as well as an earlier version of x11 ( the xorg version - not the algo ) and the earlier version of the catalyst driver - not the latest ...

that being said - the full upgrade should be finished by tonight - and hopefully fully functional by tomorrow ( sunday ) morning ... when i will hopefully get some sleep Wink ...

#crysx
sr. member
Activity: 427
Merit: 250
has cuda 7.5 finally caught up to cuda 6.5 speeds? i thought 6.5 was most efficient in mining.
legendary
Activity: 2912
Merit: 1091
--- ChainWorks Industries ---
hi all ...

thefarm is currently being upgraded to fedora 22 x64 and cuda 7.5 ...

downtime expected - 7hours ...

#crysx
legendary
Activity: 2912
Merit: 1091
--- ChainWorks Industries ---


Hey guys, Granite has been included in our new 20% off offer on all hosting orders.
Get your dedicated website hosting now at CryptoCloudHosting.org/order
very much appreciated bates ...

this is why we rely on your services for most things ... apart from the reliability and great backup service you give ...

#crysx
legendary
Activity: 2424
Merit: 1148


Hey guys, Granite has been included in our new 20% off offer on all hosting orders.
Get your dedicated website hosting now at CryptoCloudHosting.org/order
legendary
Activity: 2912
Merit: 1091
--- ChainWorks Industries ---
hi all ...

first ...

donation access have been slightly changed so that when you do mine for the devs and look at the nicehash link - you will not be confused by whether you are mining the right link ...

when the stratum proxies are active - under the nicehash graph is the worker for the dev selected WITH their name now ... so if you scroll down below the graph on the nicehash page - you will see the name of the dev and the current hashrate that is being pushed through that link ...

for example ... if you are mining scrypt for doantion to wolf0 - and check the nicehash site - you will see his name under the graph ...

https://www.nicehash.com/?p=miners&addr=1Jp2R7dF9gYr872FkXiap2MaYaHNEfwWhZ&a=0&l=0 ...


second ...

some of the servers have been changed over to the newer systems and has been completed a week ago ...

the whitelist for the servers on nicehash needed to be changed - and a request sent to nicehash support ... unfortunately ( by a mistake on my part ) the whitelist was not complete ... which means there are a couple of servers ( including ace / crysx - so im not exclusive to this ) that are now not functional on the nicehash system ...

these are ...

- donate-ace.granitecoin.com / donate-crysx.granitecoin.com ( same server ) ...
- donate-pallas.granitecoin.com ...

sorry pallas ... you are stuck with me on this one ...

nicehash support have been emailed with all the correct details - and i await their reply ...


third ...

two new servers have been commissioned and are active - but are awaiting confirmation from nicehash that they also will be added to the nicehash whitelist ...

these are ...

- donate-tnelson ...
- donate-stratehm ...

the op and the devs details willl be updated shortly ...


fourth ...

due to the massive attacks we have recently ( killing two of our servers ) - we have been looking at various ways to combat such issues as ddos attacks ...

as of 12 hours ago - we are now running ALL our servers and systems including the website through cloudflare ...

as we grow - and the systems become more complex and more accessible - we will expand the security further ...


fifth ...

slack channel #granite ...

our new slack channel is now open - and i will be frequenting the channel often ...

pm me the email addresses of those who want to be a part of the discussions ...


sixth ...

thefarm is undergoing changes ... big ones ...

- new cooling for thefarm ...
- new design for thefarm ...
- splitting of thefarm ... amd / nvidia ...
- new and improved premises for thefarm ...
- new systems in place for thefarm ...

all these things and a few more are being implemented currently ...

the amd section of thefarm has been separated - will now be private - and will be taken off the rental systems ... it is now being used exclusively inetrnally and for the development of 'bigger and better things' ...

the nvidia section of thefarm will continue the way it has been going - with the exception of a few upgrades ... i am working on a design for a frame to contain the machine that will house and protect the new gpus ... this design will encompass the requirement for the gpus that use more power than that of the gigabyte 750ti oc lp cards we are currently using in thefarm ... yes - i am talking about housing the gigabyte 980ti g1 gpus - 6 at a time - with room for upgradability ... which will allow for other larger cards to be slot right in also ...

this is a longer term project - but will be the one that replaces all the machines in thefarm eventually ... it is being designed from the ground up - to be modular AND easy to connect - install - and move ...

plug the network cable and power cables - setup where to mine - and start mining ... simple - solid - stable ...


more updates soon ...

ill be copying this same message on the dbm thread - as ive melded the updates together ...

have fun mining ...

#crysx
legendary
Activity: 2912
Merit: 1091
--- ChainWorks Industries ---
hi all ...

we are currently testing the new forum that has been setup for the granite site / coin / development ...

the donation by mining ( dbm ) thread will also be included there and will post here from time to time with the updates ...

we want to make it an enjoyable time in the forum ... so being a test forum currently - it will go through changes as we get to make them ...

the test site is here - http://granitecoin.com/forum ... but when in full working order and out of test - it will be here - grn-forum.granitecoin.com ...

please have a look and register and have a look around ... it is currently bare until we start getting some flow in there - but we will continue with posts and updates regarding granite on there ...

your constructive feedback is always welcome - and encouraged ...

help us make the granite forum a part of the granite ecosystem we are building - with all that is granite AND crypto ...

tanx ...

#crysx
legendary
Activity: 2912
Merit: 1091
--- ChainWorks Industries ---
I don't have as many cards as you, but had a bunch of amd cards (sold most of them) but none of them ever failed. Nor did the nvidia ones.
What's failing for me are ssd disks and, mostly, network equipment... a lot of routers, wifi cards, etc.
Just my experience :-)

BTW great work with all the "ecosystem", keep it going!

tanx pallas ...

im actually migrating the servers now instead of next week ... the new servers are quick - which will make for a good response to the miners ...

the next stage is the stratum proxy upgrade ... that is the one thing that will take a whole weekend to do ... so starting on that tomorrow ...

the granite ecosystem has started - and will keep it going and improving as best i can ...

most of the network equipment i have on occasion fail - but not often ... im a network and infrastructure consultant - so i source the best i can get for the infrastructure ... though recently ive opted for netgear products becuase they are cheap ( er ) than the cisco and highend server components ... spending most of the money on thefarm and the vps systems and servers ...

the amd cards have been a huge disappointment to me - as i REALLY like amd ... gigabyte especially - but those cards have just fallen over way too many times ... some cards are dead - but thats not the problem ... most cards die when they have someone like me hammering them ...

its the fan systems they have for cooling the cards that gigabyte designed - its crap ...

this leads to the cards overheating and making the enitre fan system unstable - and ultimately non functional ... some fans work - some dont ...

and come the warranty claims ... an eternity before we get them back ... missing 6-8 weeks of mining due to it ...

anyway ... i have sourced the gigabyte 980ti g1 cards for the new system ... they are about $1100.00AUD each ... quite expensive ...

for a single machine - 6 are required with two decent psu's ... corsair make a 1200w one that is platinum rated - AX1200i ( https://www.youtube.com/watch?v=7SdDL-A28Sg ) ... depending on how much power each card draws will determine whether two psu like this are necessary ... it will mean that this machien will be almost $10000.00AUD when its finished - but hash fast like a cheetah with the raw power of a lion ...

im unsure what to do with the amds when they get back - so when i send them off on monday - i will have some time to think it over ... they will more than likely be sold - but - may be kept operational as a different part of thefarm for one of the operations i have in mind with the integration of the multipools and donation links Wink ...

i really appreciate the positive feedback - especially from one of the devs like you who have continued to work with other devs - and the communtiy as a whole ...

#crysx
legendary
Activity: 2716
Merit: 1094
Black Belt Developer
I don't have as many cards as you, but had a bunch of amd cards (sold most of them) but none of them ever failed. Nor did the nvidia ones.
What's failing for me are ssd disks and, mostly, network equipment... a lot of routers, wifi cards, etc.
Just my experience :-)

BTW great work with all the "ecosystem", keep it going!
legendary
Activity: 2912
Merit: 1091
--- ChainWorks Industries ---
hi all ...

thefarm will be getting an overhaul very shortly ...

the amd cards are ALL ( thats 26 of them ) going in for maintenance - and ONE nvidia card is going also ...

this shows the difference in reliability in cards from the same manufacturer - gigabyte ...

the amd crads hash VERY hard - especially when being used with the optimized bins and miners that are around ( like wolf0 bins and djm34 lyra2rev2 miner ) ...

unfortunately - this raises some very big questions about what to do with the cards when they all come back from warranty repair ... and these questions are raised for two reasons ...

1 - reliability and longevity ...
2 - miner / optimization availability ...

firstly - reliability has been quite questionable with the amd cards over the period of time we have been mining ... they seem to last about the 3 month period before a warranty repair is required ... nvidia cards have been hashing now for more than 7 months and only one card has failed so far ...

secondly - miner / optimization availability ... amd cards have had some good turns with hashrates and optimization for the miners and kernels have been forthcoming ... but more often that not - optimized miners are MOSTLY private - cost a great deal to purchase most times - and are MOSTLY unavailable to the general public ... withnvidia - the tables are similar except for the fact that the majority of miners and optimizations - including the progressive enhancements to the code by many of the community members seem to outweigh amd buy a huge amount ... even the private miners are not that 'pricey' if one was to purchase the private miners ...

if we were to collate the pros and cons of whether we should keep the amd side of thefarm or the nvidia side of thefarm - the nvidia seem to be winning on both counts ... and that is not even including other factors - like the power consumption rates and systems build tech ...

we want thefram to become a long term facility for both granite and the donation system( as well as other coins ) - all integrated into the same web facility ( which we are working on currently ) ...

so there are quite big questions to be pondered here ... and decisions that need to be made ...

after the pull of the amd side of thefarm for repair - there is also a 6 week wait for gigabyte to get anything done ... that is the experience we have had so far here in australia ...

ho hum ... that is the realities of corporate dealings ...

thefarm has been pulled off the rental market - so that this maintenance ( and subsequent rebuild ) can take place ...

tanx ...

#crysx
legendary
Activity: 2912
Merit: 1091
--- ChainWorks Industries ---
i agree with you both ...

we have both amd and nvidia cards in thefarm - and most times split the mining to the algo most proficient for the cards ...

but x11 and quark have always been two of my fav algos ...

there are many other algos out there - but i highly doubt it will be beneficial to put the work into the code to be able to mine something that most wont even look at ...

we are looking at a great many options and i personally think that the most popular algos - as well some of the less popular ones - really need to be considered for addition ...

but there are more ways than just coding algos to be able to accomplish this - so work has already begun to cater for the change ... if any change is needed at all Wink ...

tanx for the input ... it really is appreciated ...

#crysx
hero member
Activity: 1064
Merit: 500
MOBU
Yeah,

Well...I woulda voted a nVidia compat algo, but I have a coulpla scrypt miners that leave me to mine something else /w the Vid, so...there ya go. Up to the group!! Ya know what I voted for...

Later-

      Cool
member
Activity: 225
Merit: 17
Hi everyone !

I voted Quark !

I also like Qubit or Neoscrypt

Because I mine with nvidia cards... 5 X 950GTX + 11 X 750TI

THANKS
legendary
Activity: 2912
Merit: 1091
--- ChainWorks Industries ---
hi all ...

ive setup a poll posing the question -

if you were able to mine granite using any number of algorithms - not just x11 - which ones would they be? ...

ill keep the poll going for a couple of weeks - while we are in the process of setting up the new base for the new pool ...

your input would be VERY much appreciated ...

tanx ...

#crysx
legendary
Activity: 2912
Merit: 1091
--- ChainWorks Industries ---
still holding mine long term here  Grin

glad to hear ...

all the nodes are currently being updated to the latest build - v1.0.1 granitecoind ...

they will be offline for about an hour - before they come back on line with v1.0.1 ...

only a few small changes were done with the wallet - but that is an ongoing development - which is also v1.0.1 now ...

keeping at it ... Wink ...

tanx for your support ...

#crysx
hero member
Activity: 671
Merit: 1000
still holding mine long term here  Grin
Pages:
Jump to: