Author

Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 167. (Read 3426944 times)

full member
Activity: 182
Merit: 100
Yes, powered risers dont get power from the motherboard so without power from a molex the cards wont work, or atleast shouldnt work well

Ok..so i have to connect molex and then try with cards all to 1300mhz.. if ccminer crash again, i have to try to use only 1 to 16 risers, even into 16pin ports.. if ccminer crash again, it is better to spend time with my wife Wink all right?? Smiley
sr. member
Activity: 350
Merit: 250
Yes, powered risers dont get power from the motherboard so without power from a molex the cards wont work, or atleast shouldnt work well
full member
Activity: 182
Merit: 100
Just my 2 satoshis, but those unpowered ribbon risers are an accident waiting to happen. I strongly advise you to get the powered 1x to 16x ones (usb type cable). You can fit the 1x risers in x16 slots, no problem.
~ Myagui

Agreed

I cannot use only 1 to 16 because my MoBo has only 4x 1pin ports and 3x 16pin ports..This is the motivation cause i bought 16 to 16 risers..

you can use x1 risers in x16 slots, i have and so have many others

Many thanks.. and what about powered riser not connected? Is it necessary to connect molex?
sr. member
Activity: 350
Merit: 250
Just my 2 satoshis, but those unpowered ribbon risers are an accident waiting to happen. I strongly advise you to get the powered 1x to 16x ones (usb type cable). You can fit the 1x risers in x16 slots, no problem.
~ Myagui

Agreed

I cannot use only 1 to 16 because my MoBo has only 4x 1pin ports and 3x 16pin ports..This is the motivation cause i bought 16 to 16 risers..

you can use x1 risers in x16 slots, i have and so have many others
legendary
Activity: 1154
Merit: 1001
Just my 2 satoshis, but those unpowered ribbon risers are an accident waiting to happen. I strongly advise you to get the powered 1x to 16x ones (usb type cable). You can fit the 1x risers in x16 slots, no problem.
~ Myagui
full member
Activity: 182
Merit: 100
First of all, many thanks for your interest in my problem.. my rig functioned so well when:

- 3 Cards are directly pushed into MoBo;
- 2 cards with 1 to 16 risers;

When 6th cards arrived, my rig was began:

- 1 card directly to MoBo;
- 2 cards with 16 to 16 risers;
- 3 cards with 1 to 16 risers.

At this star, nothing goes well..I tried change risers and invert cards.. nothing..

Can you explain what you meant by invert cards?
A lot of people have had trouble getting 6 cards to work, 5 usually does but 6 is the killer number.

Ok here's some things to ask right now, may have been covered but worth asking again

Are the risers usb or ribbon?
Powered or non-powered?

one last one, are the 2 cards that keep crashing the ones on the 16 to 16 risers?
These close to always give issued with so many cards, i recommend you switch them for 1 to 16 risers

Sorry for my bad English, with "invert" i meant switch cards that crash..

Risers 16 to 16 are powered but not connected, i tried to connect it with another PSU cause my has no Molex.. but nothing changed.. anyway the situation is:

- 2 risers 16 to 16 powerd but not connected;
- 3 risers 1 to 16 non-powered

I cannot use only 1 to 16 because my MoBo has only 4x 1pin ports and 3x 16pin ports..This is the motivation cause i bought 16 to 16 risers..

All risers are ribbon.

Many many thanks again bro'
sr. member
Activity: 350
Merit: 250
First of all, many thanks for your interest in my problem.. my rig functioned so well when:

- 3 Cards are directly pushed into MoBo;
- 2 cards with 1 to 16 risers;

When 6th cards arrived, my rig was began:

- 1 card directly to MoBo;
- 2 cards with 16 to 16 risers;
- 3 cards with 1 to 16 risers.

At this star, nothing goes well..I tried change risers and invert cards.. nothing..

Can you explain what you meant by invert cards?
A lot of people have had trouble getting 6 cards to work, 5 usually does but 6 is the killer number.

Ok here's some things to ask right now, may have been covered but worth asking again

Are the risers usb or ribbon?
Powered or non-powered?

one last one, are the 2 cards that keep crashing the ones on the 16 to 16 risers?
These close to always give issued with so many cards, i recommend you switch them for 1 to 16 risers
full member
Activity: 182
Merit: 100
ok thats strange, have you tried removing the 2 cards entirely?
You said you had 6 in right?

First of all, many thanks for your interest in my problem.. my rig functioned so well when:

- 3 Cards are directly pushed into MoBo;
- 2 cards with 1 to 16 risers;

When 6th cards arrived, my rig was began:

- 1 card directly to MoBo;
- 2 cards with 16 to 16 risers;
- 3 cards with 1 to 16 risers.

At this star, nothing goes well..I tried change risers and invert cards.. nothing..
legendary
Activity: 1537
Merit: 1005
Are there any plans to add API so ccminer can communicate with software like cgwatcher?

My 750ti are rock solid, but still would like to be super sure Cheesy
Also is there any solution to CPU mine, while using ccminer? Right now the gpu hashrate plummets to like 30%.
hero member
Activity: 553
Merit: 500
Solo Miner Legend
300khash/s is perfectly normal for Scrypt on a 750Ti and isnt really possible to increase

Ahh ok thanks, I thought there was a way to get it to 340-350 KH/s now, guess I read that wrong.

My first time mining on a GTX, have mined for years on 4000 and 7000 series Radeon, though.
This 750 Ti is my nephew's new gaming maching. Wink Just wanted to give it a hard spin.
sr. member
Activity: 350
Merit: 250
ok thats strange, have you tried removing the 2 cards entirely?
You said you had 6 in right?
full member
Activity: 182
Merit: 100
the clock limit of the graphics cards change depending on what they are doing. Typically a very high memory usage algorithm uses little processing power and so can have a higher clock.

Algorithms that use little memory and generate large amounts of hash's will often need a lower clock because the card is working much harder and can not handle the clock for long.

Your CPU would be the same, overclock to 4.9GHz and it runs whilst browsing and general use, load a game and it dies. Drop it down a little and its stable at the lower clock and the game plays fine

All cards at 1300mhz, +0 memory clock, +0 Voltage..
Result: crash.
sr. member
Activity: 350
Merit: 250
300khash/s is perfectly normal for Scrypt on a 750Ti and isnt really possible to increase
hero member
Activity: 553
Merit: 500
Solo Miner Legend
Just a quick question, been reading through some of the pages of this thread, not all of them ofc, too many to go through.

I'm running MSI GTX 750 Ti gaming edition on a Haswell i3, Win 7 64 Ultimate with Det 340.52 drivers.

Core about 1311 (EVGA Precision) and cuda settings: -l T5x24 -c 1 -m 1 -i 0 -s 10

Getting about 302 - 304 KH/s, anything I can do, or settings to add make it go faster or tune it a bit?

Thanks Smiley
sr. member
Activity: 350
Merit: 250
the clock limit of the graphics cards change depending on what they are doing. Typically a very high memory usage algorithm uses little processing power and so can have a higher clock.

Algorithms that use little memory and generate large amounts of hash's will often need a lower clock because the card is working much harder and can not handle the clock for long.

Your CPU would be the same, overclock to 4.9GHz and it runs whilst browsing and general use, load a game and it dies. Drop it down a little and its stable at the lower clock and the game plays fine
full member
Activity: 182
Merit: 100
actual Stock clocks for the cards is around 1100mhz, the "Stock" your referring to is the factory overclock which the card deems as stock. Most 750Ti's wont do anything at more then 1350mhz unless certain algorithms are being mined. But most algorithms prefer 1300mhz or less
I know the difference between stock clock and factory overclock.. but i thought that the only limit of ccminer is the clock limit of graphics cards.. so.. I was wrong Smiley
sr. member
Activity: 350
Merit: 250
actual Stock clocks for the cards is around 1100mhz, the "Stock" your referring to is the factory overclock which the card deems as stock. Most 750Ti's wont do anything at more then 1350mhz unless certain algorithms are being mined. But most algorithms prefer 1300mhz or less
full member
Activity: 182
Merit: 100
1400mhz is simply too high, its very rare the cards will run properly stable above 1350mhz

honestly clock them down to 1300mhz, with +0 voltage and i will have bets that they all run perfectly. Your simply trying to get too much out of the cards and WILL kill them if you dont stop it
It is 50mhz over stock clock.. not a crazy thing i think, this cards have 6pin direct to PSU..moreover it crash even with stock clock, so why the problem would be too high clock?

Anyway u believe in u and my cards goes immediatly to 1300mhz Smiley
sr. member
Activity: 350
Merit: 250
1400mhz is simply too high, its very rare the cards will run properly stable above 1350mhz

honestly clock them down to 1300mhz, with +0 voltage and i will have bets that they all run perfectly. Your simply trying to get too much out of the cards and WILL kill them if you dont stop it
full member
Activity: 182
Merit: 100
In precision x make sure that sync on box in unchecked. Then select GPU 1 and lower the GPU clock by 10mhz, hit apply, the same for GPU2. If still crashes lower for another 10mhz.
What is the GPU frequency and voltage of your crashing cards?

"Poor" card is when factory overclocked chip occasionally crashes under constant heavy load. 

Yep, and some factory overclocked cards may actually need to be underclocked to run properly.
On a side note, i found a small toggle switch on my MSI 750ti's the other day which switches bios's i believe between factory overclock and stock, haven't messed or looked into it but the switchs may be on other cards aswell

This is correct, but only twin frozr gaming series have that switch. Regular MSI single fan 750ti's v1 and v2 does not have it:

http://www.tweaktown.com/reviews/6148/msi-geforce-gtx-750-ti-2gb-twin-frozr-gaming-oc-video-card-review/index2.html



Cards crashes at stock clock of evga.. i tried a little manual  OC without sync option to make all card at the same clock..

GPU1: +81mhz (stock 1319mhz)
GPU2: +55mhz (stock 1345mhz)
GPU3,4,5,6: +68mhz (stock 1332mhz)

GPU1 and GPU2 are those ones that crash, as you can see they run at different clock compared to stock (1332mhz).

With my manual OC now all cards run at 1400mhz but ccminer and nvminer still crashes.. Voltage is +18mV.. please.. help me Sad
Jump to: