[ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX]

[AizenSou]

i done  to the instructions, and compiled it, but i got no shares accepted?

did you compile under windows or linux? if linux, don't forget to run ./autogen.sh after you made the changes.

It depends on your card. If your card is old (<680GTX), so change the last line in Makefile.in to compute_20. At least it works for me.

[vlax352]

i done  to the instructions, and compiled it, but i got no shares accepted?

did you compile under windows or linux? if linux, don't forget to run ./autogen.sh after you made the changes.

It depends on your card. If your card is old (<680GTX), so change the last line in Makefile.in to compute_20. At least it works for me.

i have gtx650Ti, compute 2.0 did the trick. TNX!!!!!!

[cvax]

Cachecoin or Fuguecoin seems profitable to me. There was some underdogs too of course if you subscribe yourself to holynerdvana.com. That guy has a lot of experiences regarding crypto.

Btw, I remembered these days when earning 0.5 BTC daily is easy task. But now everyone rushes in so expecting it will become much harder to earn money with crypto. Esp. if the new 100Mhs ASICs in next 6 months. Scrypt coins will be bullied to death with those monsters.

Thanks for the suggestions. Let me add cachecoin to my track list and see how it goes.

[AizenSou]

Cachecoin or Fuguecoin seems profitable to me. There was some underdogs too of course if you subscribe yourself to holynerdvana.com. That guy has a lot of experiences regarding crypto.

Btw, I remembered these days when earning 0.5 BTC daily is easy task. But now everyone rushes in so expecting it will become much harder to earn money with crypto. Esp. if the new 100Mhs ASICs in next 6 months. Scrypt coins will be bullied to death with those monsters.

Thanks for the suggestions. Let me add cachecoin to my track list and see how it goes.

Actually by design such short profit calculation won't work with CACH. This coin needs 5 days at least for a mining block to mature, so if you mine today and after 5 days the exchange rate crashes, it is not profitable anymore. But IMO, CACH and MRC coins are two best coins for mine right now and hold because diff so low. And it worths a look at cachecoin thread to see some hero members talk about this coin . You won't regret it.

[cbuchner1]

Atrides & me found a fix (well, he came up with it, I just found another spot which has to be fixed ^^) for ccminer 0.3 to work with dwarfpool for GroestlCoin:

stratum works now on http://dwarfpool.com/grs Have fun & join the boat ^^

i done  to the instructions, and compiled it, but i got no shares accepted?

thanks for the above fixes!

I've posted a v0.4 release. Stratum seems to work for me. https://github.com/cbuchner1/ccminer/releases/tag/v0.4

Here's how I pool-mine:

ccminer.exe -q -a groestl -o stratum+tcp://erebor.dwarfpool.com:3345  -u FcPdoV3QQbrAHiPs9M9XXAxT2Bv4SzB7PQ.1 -p x

Code:

[2014-03-24 22:54:00] Starting Stratum on stratum+tcp://erebor.dwarfpool.com:3345
[2014-03-24 22:54:00] 4 miner threads started, using 'groestl' algorithm.
[2014-03-24 22:54:20] accepted: 1/1 (100.00%), 13527 khash/s (yay!!!)
[2014-03-24 22:54:31] accepted: 2/2 (100.00%), 13499 khash/s (yay!!!)
[2014-03-24 22:54:32] accepted: 3/3 (100.00%), 13514 khash/s (yay!!!)
[2014-03-24 22:54:44] accepted: 4/4 (100.00%), 13476 khash/s (yay!!!)
[2014-03-24 22:55:06] accepted: 5/5 (100.00%), 13404 khash/s (yay!!!)
[2014-03-24 22:55:25] accepted: 6/6 (100.00%), 13343 khash/s (yay!!!)
[2014-03-24 22:55:43] accepted: 7/7 (100.00%), 13368 khash/s (yay!!!)

[ltcnim]

Atrides & me found a fix (well, he came up with it, I just found another spot which has to be fixed ^^) for ccminer 0.3 to work with dwarfpool for GroestlCoin:

stratum works now on http://dwarfpool.com/grs Have fun & join the boat ^^

i done  to the instructions, and compiled it, but i got no shares accepted?

thanks for the above fixes!

I've posted a v0.4 release. Stratum seems to work for me. https://github.com/cbuchner1/ccminer/releases/tag/v0.4

Here's how I pool-mine:

ccminer.exe -q -a groestl -o stratum+tcp://erebor.dwarfpool.com:3345  -u FcPdoV3QQbrAHiPs9M9XXAxT2Bv4SzB7PQ.1 -p x

Code:

[2014-03-24 22:54:00] Starting Stratum on stratum+tcp://erebor.dwarfpool.com:3345
[2014-03-24 22:54:00] 4 miner threads started, using 'groestl' algorithm.
[2014-03-24 22:54:20] accepted: 1/1 (100.00%), 13527 khash/s (yay!!!)
[2014-03-24 22:54:31] accepted: 2/2 (100.00%), 13499 khash/s (yay!!!)
[2014-03-24 22:54:32] accepted: 3/3 (100.00%), 13514 khash/s (yay!!!)
[2014-03-24 22:54:44] accepted: 4/4 (100.00%), 13476 khash/s (yay!!!)
[2014-03-24 22:55:06] accepted: 5/5 (100.00%), 13404 khash/s (yay!!!)
[2014-03-24 22:55:25] accepted: 6/6 (100.00%), 13343 khash/s (yay!!!)
[2014-03-24 22:55:43] accepted: 7/7 (100.00%), 13368 khash/s (yay!!!)

great, thanks for the fast update. I already got payouts, so everything seems to work :>

[bromstar]

Hi Christian,

is there any update regarding the progress on the 1x riser improvements?

Best regards

[bigjme]

Hi Christian,

is there any update regarding the progress on the 1x riser improvements?

Best regards

Im interested in this aswell. Especially with new cards and risers coming this week

[unrulycow]

I'm trying to setup CudaMiner on Ubuntu 13.10, but I'm getting the following error compiling it.  Any ideas?

Code:

make[1]: Entering directory `/home/berniehglee/CudaMiner'
Making all in compat
make[2]: Entering directory `/home/berniehglee/CudaMiner/compat'
Making all in jansson
make[3]: Entering directory `/home/berniehglee/CudaMiner/compat/jansson'
make[3]: Nothing to be done for `all'.
make[3]: Leaving directory `/home/berniehglee/CudaMiner/compat/jansson'
make[3]: Entering directory `/home/berniehglee/CudaMiner/compat'
make[3]: Nothing to be done for `all-am'.
make[3]: Leaving directory `/home/berniehglee/CudaMiner/compat'
make[2]: Leaving directory `/home/berniehglee/CudaMiner/compat'
make[2]: Entering directory `/home/berniehglee/CudaMiner'
g++  -g -O2 -pthread -L/usr/local/cuda/lib64  -o cudaminer cudaminer-cpu-miner.o cudaminer-util.o wrapnvml.o cudaminer-sha2.o cudaminer-scrypt.o cudaminer-maxcoin.o cudaminer-blakecoin.o cudaminer-sha3.o cudaminer-scrypt-jane.o salsa_kernel.o sha256.o keccak.o blake.o cudaminer-blake.o fermi_kernel.o kepler_kernel.o test_kernel.o nv_kernel.o nv_kernel2.o titan_kernel.o -L/usr/lib/x86_64-linux-gnu -lcurl compat/jansson/libjansson.a -lpthread  -lcudart -fopenmp -lcrypto -lssl  -lcrypto -lssl
/usr/bin/ld: wrapnvml.o: undefined reference to symbol 'dlclose@@GLIBC_2.2.5'
/lib/x86_64-linux-gnu/libdl.so.2: error adding symbols: DSO missing from command line

	

[flyingdog]

Atrides & me found a fix (well, he came up with it, I just found another spot which has to be fixed ^^) for ccminer 0.3 to work with dwarfpool for GroestlCoin:

stratum works now on http://dwarfpool.com/grs Have fun & join the boat ^^

i done  to the instructions, and compiled it, but i got no shares accepted?

thanks for the above fixes!

I've posted a v0.4 release. Stratum seems to work for me. https://github.com/cbuchner1/ccminer/releases/tag/v0.4

Here's how I pool-mine:

ccminer.exe -q -a groestl -o stratum+tcp://erebor.dwarfpool.com:3345  -u FcPdoV3QQbrAHiPs9M9XXAxT2Bv4SzB7PQ.1 -p x

Code:

[2014-03-24 22:54:00] Starting Stratum on stratum+tcp://erebor.dwarfpool.com:3345
[2014-03-24 22:54:00] 4 miner threads started, using 'groestl' algorithm.
[2014-03-24 22:54:20] accepted: 1/1 (100.00%), 13527 khash/s (yay!!!)
[2014-03-24 22:54:31] accepted: 2/2 (100.00%), 13499 khash/s (yay!!!)
[2014-03-24 22:54:32] accepted: 3/3 (100.00%), 13514 khash/s (yay!!!)
[2014-03-24 22:54:44] accepted: 4/4 (100.00%), 13476 khash/s (yay!!!)
[2014-03-24 22:55:06] accepted: 5/5 (100.00%), 13404 khash/s (yay!!!)
[2014-03-24 22:55:25] accepted: 6/6 (100.00%), 13343 khash/s (yay!!!)
[2014-03-24 22:55:43] accepted: 7/7 (100.00%), 13368 khash/s (yay!!!)

wonderful job but cuda compute_10 is not valid for cuda_fugue256.cu
line 604
Error: Unaligned memory access not supported

I went back to only support compute_20 and compile is fine now...

[tachyon_john]

I'm trying to setup CudaMiner on Ubuntu 13.10, but I'm getting the following error compiling it.  Any ideas?

Code:

make[1]: Entering directory `/home/berniehglee/CudaMiner'
Making all in compat
make[2]: Entering directory `/home/berniehglee/CudaMiner/compat'
Making all in jansson
make[3]: Entering directory `/home/berniehglee/CudaMiner/compat/jansson'
make[3]: Nothing to be done for `all'.
make[3]: Leaving directory `/home/berniehglee/CudaMiner/compat/jansson'
make[3]: Entering directory `/home/berniehglee/CudaMiner/compat'
make[3]: Nothing to be done for `all-am'.
make[3]: Leaving directory `/home/berniehglee/CudaMiner/compat'
make[2]: Leaving directory `/home/berniehglee/CudaMiner/compat'
make[2]: Entering directory `/home/berniehglee/CudaMiner'
g++  -g -O2 -pthread -L/usr/local/cuda/lib64  -o cudaminer cudaminer-cpu-miner.o cudaminer-util.o wrapnvml.o cudaminer-sha2.o cudaminer-scrypt.o cudaminer-maxcoin.o cudaminer-blakecoin.o cudaminer-sha3.o cudaminer-scrypt-jane.o salsa_kernel.o sha256.o keccak.o blake.o cudaminer-blake.o fermi_kernel.o kepler_kernel.o test_kernel.o nv_kernel.o nv_kernel2.o titan_kernel.o -L/usr/lib/x86_64-linux-gnu -lcurl compat/jansson/libjansson.a -lpthread  -lcudart -fopenmp -lcrypto -lssl  -lcrypto -lssl
/usr/bin/ld: wrapnvml.o: undefined reference to symbol 'dlclose@@GLIBC_2.2.5'
/lib/x86_64-linux-gnu/libdl.so.2: error adding symbols: DSO missing from command line

	

Hi,
  If you edit the Makefile.am and remove the "-ldl" and rerun the autogen.sh, and make, does it link, or not?  Does your system have a libdl.so or libdl.a in /lib, /lib64, or /usr/lib or /usr/lib64?

Cheers,
  John

[Dotcommie]

Here are my results with risers:

ASRock H81 BTC mobo with 4gb of ram on Win7 x64= only 5 EVGA FTW 750ti cards run stable
ASRock H81 BTC mobo with 8gb of ram on Win7 x64= all 6 EVGA FTW 750ti cards run stable

I get consistent 295-300khs on all cards with +20 core and +410 mem.  I think if I go too much more than that it crashes.  Since they're FTW cards, that comes out to 1209(1288 boost) clock and 1555 mem.  I can probably go higher if tweaking them separately since I probably have at least 1 card that's a weak overclocker, but that is stable so I left it.  I'm thinking I will get more without risers, but I have no way to test more than 1 on this board since it's only got 1 16x slot.  **Just an FYI, but GPU-Z is saying boost is 1288 but Precision X is actually reporting 1330-1350 for clocks**

On my other rig with Gigabyte 750ti's, I should have researched more instead of jumping on the cheapest board simply because it came with a freebie mpcie ssd, but that one seems to experience the degradation people are talking about when adding more cards.

That mobo is an ECS Z77H2-A4 with a Celeron G1620.  Adding 4 cards is fine, but it seems to choose PCI-E 1.1 1x mode for every card on the 1x slots and only PCI-E 2.0 for the one plugged in to the first x16 slot. It boots with the 5th plugged in but only detects 3 cards then. I've tried the x16 slot with x1 and x16 risers and it still only did PCI-E 2.0, which is the only card getting full speed.  ***I just checked the processor specs while I was typing this and realized the problem.... "PCI Express Revision 2.0 / Supports up to 1x16, 2x8, 1x8 & 2x4"

I wonder how many other people cheaped out on processors and are getting similar declines in hashrate or incompatibilities with multiple cards. I surely can't be the only one.  The ASRock board is running a socket 1150 i3 4130, so I'm guessing that's why I don't see a decline in khs. (Only 1155 cpus i5 and up supported PCIE v3)

[cvax]

Here are my results with risers:

ASRock H81 BTC mobo with 4gb of ram on Win7 x64= only 5 EVGA FTW 750ti cards run stable
ASRock H81 BTC mobo with 8gb of ram on Win7 x64= all 6 EVGA FTW 750ti cards run stable

I get consistent 295-300khs on all cards with +20 core and +410 mem.  I think if I go too much more than that it crashes.  Since they're FTW cards, that comes out to 1209(1288 boost) clock and 1555 mem.  I can probably go higher if tweaking them separately since I probably have at least 1 card that's a weak overclocker, but that is stable so I left it.  I'm thinking I will get more without risers, but I have no way to test more than 1 on this board since it's only got 1 16x slot.  **Just an FYI, but GPU-Z is saying boost is 1288 but Precision X is actually reporting 1330-1350 for clocks**

On my other rig with Gigabyte 750ti's, I should have researched more instead of jumping on the cheapest board simply because it came with a freebie mpcie ssd, but that one seems to experience the degradation people are talking about when adding more cards.

That mobo is an ECS Z77H2-A4 with a Celeron G1620.  Adding 4 cards is fine, but it seems to choose PCI-E 1.1 1x mode for every card on the 1x slots and only PCI-E 2.0 for the one plugged in to the first x16 slot. It boots with the 5th plugged in but only detects 3 cards then. I've tried the x16 slot with x1 and x16 risers and it still only did PCI-E 2.0, which is the only card getting full speed.  ***I just checked the processor specs while I was typing this and realized the problem.... "PCI Express Revision 2.0 / Supports up to 1x16, 2x8, 1x8 & 2x4"

I wonder how many other people cheaped out on processors and are getting similar declines in hashrate or incompatibilities with multiple cards. I surely can't be the only one.  The ASRock board is running a socket 1150 i3 4130, so I'm guessing that's why I don't see a decline in khs. (Only 1155 cpus i5 and up supported PCIE v3)

I have an ASRock 970 Extreme 4 and some shit 170u CPUs. I have 4x Zotac 750Tis in my rigs and they don't slow down except for thermal slow downs. But then again my cards average around 280-295 to begin with.

[DrAlco]

Am I running faster than Christian's 4 cards with only 2 here?    

Code:

[2014-03-25 16:30:31] Starting Stratum on stratum+tcp://eredluin.dwarfpool.com:3
345
[2014-03-25 16:30:31] 2 miner threads started, using 'groestl' algorithm.
[2014-03-25 16:30:34] accepted: 1/1 (100.00%), 10621 khash/s (yay!!!)
[2014-03-25 16:31:36] accepted: 2/2 (100.00%), 16893 khash/s (yay!!!)
[2014-03-25 16:31:37] accepted: 3/3 (100.00%), 16828 khash/s (yay!!!)
[2014-03-25 16:31:45] accepted: 4/4 (100.00%), 16924 khash/s (yay!!!)
[2014-03-25 16:33:23] accepted: 5/5 (100.00%), 16428 khash/s (yay!!!)
[2014-03-25 16:33:59] accepted: 6/6 (100.00%), 16524 khash/s (yay!!!)
[2014-03-25 16:34:38] accepted: 7/7 (100.00%), 16487 khash/s (yay!!!)
[2014-03-25 16:34:50] accepted: 8/8 (100.00%), 16530 khash/s (yay!!!)
[2014-03-25 16:35:16] accepted: 9/9 (100.00%), 16565 khash/s (yay!!!)
[2014-03-25 16:35:22] accepted: 10/10 (100.00%), 16569 khash/s (yay!!!)

/edit: I'm running 0.4 ccminer win x64 compute 3.5 compile.
                                                                              

[Dotcommie]

I have an ASRock 970 Extreme 4 and some shit 170u CPUs. I have 4x Zotac 750Tis in my rigs and they don't slow down except for thermal slow downs. But then again my cards average around 280-295 to begin with.

I don't think I saw many people with AMD having a problem because the motherboard is still controlling the pcie spec.  With intel it's mostly controlled by cpu now, so even if my board theoretically should be at least pcie2.0, the cpu is going to run it at what it is programmed to.  At least that's my theory.  Maybe AMD guys were having the same problem, but I'm pretty sure their only problem was getting multiple cards to run properly on certain boards.

[reznor]

Here are my results with risers:

ASRock H81 BTC mobo with 4gb of ram on Win7 x64= only 5 EVGA FTW 750ti cards run stable
ASRock H81 BTC mobo with 8gb of ram on Win7 x64= all 6 EVGA FTW 750ti cards run stable

I get consistent 295-300khs on all cards with +20 core and +410 mem.  I think if I go too much more than that it crashes.  Since they're FTW cards, that comes out to 1209(1288 boost) clock and 1555 mem.  I can probably go higher if tweaking them separately since I probably have at least 1 card that's a weak overclocker, but that is stable so I left it.  I'm thinking I will get more without risers, but I have no way to test more than 1 on this board since it's only got 1 16x slot.  **Just an FYI, but GPU-Z is saying boost is 1288 but Precision X is actually reporting 1330-1350 for clocks**

On my other rig with Gigabyte 750ti's, I should have researched more instead of jumping on the cheapest board simply because it came with a freebie mpcie ssd, but that one seems to experience the degradation people are talking about when adding more cards.

That mobo is an ECS Z77H2-A4 with a Celeron G1620.  Adding 4 cards is fine, but it seems to choose PCI-E 1.1 1x mode for every card on the 1x slots and only PCI-E 2.0 for the one plugged in to the first x16 slot. It boots with the 5th plugged in but only detects 3 cards then. I've tried the x16 slot with x1 and x16 risers and it still only did PCI-E 2.0, which is the only card getting full speed.  ***I just checked the processor specs while I was typing this and realized the problem.... "PCI Express Revision 2.0 / Supports up to 1x16, 2x8, 1x8 & 2x4"

I wonder how many other people cheaped out on processors and are getting similar declines in hashrate or incompatibilities with multiple cards. I surely can't be the only one.  The ASRock board is running a socket 1150 i3 4130, so I'm guessing that's why I don't see a decline in khs. (Only 1155 cpus i5 and up supported PCIE v3)

I have an ASRock 970 Extreme 4 and some shit 170u CPUs. I have 4x Zotac 750Tis in my rigs and they don't slow down except for thermal slow downs. But then again my cards average around 280-295 to begin with.

So if that's the case then the issues of slow hashrates using pcie/usb riser depends on the hardware. So make sure your processor supports PCI-E ver. 3.0 but of course more hardware testing should support this claim. One thing if youre motherboard supports the PCI-E 3.0 then a graphics card supporting that bus speeds should be use when using risers.

Tech Specs: quick comparison)

PCI-E 2.0   5.0GT/s (raw bitrate) - 500MB/s (bandwidth per lane direction)
PCI-E 3.0   8.0GT/s (raw bitrate) - 1Gb/s   (bandwidth per lane direction)

 

[Dotcommie]

Here are my results with risers:

ASRock H81 BTC mobo with 4gb of ram on Win7 x64= only 5 EVGA FTW 750ti cards run stable
ASRock H81 BTC mobo with 8gb of ram on Win7 x64= all 6 EVGA FTW 750ti cards run stable

I get consistent 295-300khs on all cards with +20 core and +410 mem.  I think if I go too much more than that it crashes.  Since they're FTW cards, that comes out to 1209(1288 boost) clock and 1555 mem.  I can probably go higher if tweaking them separately since I probably have at least 1 card that's a weak overclocker, but that is stable so I left it.  I'm thinking I will get more without risers, but I have no way to test more than 1 on this board since it's only got 1 16x slot.  **Just an FYI, but GPU-Z is saying boost is 1288 but Precision X is actually reporting 1330-1350 for clocks**

On my other rig with Gigabyte 750ti's, I should have researched more instead of jumping on the cheapest board simply because it came with a freebie mpcie ssd, but that one seems to experience the degradation people are talking about when adding more cards.

That mobo is an ECS Z77H2-A4 with a Celeron G1620.  Adding 4 cards is fine, but it seems to choose PCI-E 1.1 1x mode for every card on the 1x slots and only PCI-E 2.0 for the one plugged in to the first x16 slot. It boots with the 5th plugged in but only detects 3 cards then. I've tried the x16 slot with x1 and x16 risers and it still only did PCI-E 2.0, which is the only card getting full speed.  ***I just checked the processor specs while I was typing this and realized the problem.... "PCI Express Revision 2.0 / Supports up to 1x16, 2x8, 1x8 & 2x4"

I wonder how many other people cheaped out on processors and are getting similar declines in hashrate or incompatibilities with multiple cards. I surely can't be the only one.  The ASRock board is running a socket 1150 i3 4130, so I'm guessing that's why I don't see a decline in khs. (Only 1155 cpus i5 and up supported PCIE v3)

I have an ASRock 970 Extreme 4 and some shit 170u CPUs. I have 4x Zotac 750Tis in my rigs and they don't slow down except for thermal slow downs. But then again my cards average around 280-295 to begin with.

So if that's the case then the issues of slow hashrates using pcie/usb riser depends on the hardware. So make sure your processor supports PCI-E ver. 3.0 but of course more hardware testing should support this claim. One thing if youre motherboard supports the PCI-E 3.0 then a graphics card supporting that bus speeds should be use when using risers.

Tech Specs: quick comparison)

PCI-E 2.0   5.0GT/s (raw bitrate) - 500MB/s (bandwidth per lane direction)
PCI-E 3.0   8.0GT/s (raw bitrate) - 1Gb/s   (bandwidth per lane direction)

 

If you're running AMD I don't think the CPU bit applies, but it would be great if we could confirm the drastic slowdowns are only on Intel rigs running CPUs that don't support PCI-E 3.0.

I still don't know how people are getting 330khs out of their cards though.  Anything close to 310 on my gigabyte and evga cards crashes cudaminer or the driver in a few minutes. I might be able to get it stable at 310 with lots of trial and error but haven't had the time or patience lately.

[cbuchner1]

Am I running faster than Christian's 4 cards with only 2 here?    

/edit: I'm running 0.4 ccminer win x64 compute 3.5 compile.
                                                                              

well to be fair, these 4 cards were 1 GT 640, 2 x GTX 750Ti, 1x GTX 780...

[DrAlco]

Am I running faster than Christian's 4 cards with only 2 here?    

/edit: I'm running 0.4 ccminer win x64 compute 3.5 compile.
                                                                              

well to be fair, these 4 cards were 1 GT 640, 2 x GTX 750Ti, 1x GTX 780...

Ah that makes much more sense! I thought it was based off your 780ti's 

[christian1980]

Here are my results with risers:

ASRock H81 BTC mobo with 4gb of ram on Win7 x64= only 5 EVGA FTW 750ti cards run stable
ASRock H81 BTC mobo with 8gb of ram on Win7 x64= all 6 EVGA FTW 750ti cards run stable

I get consistent 295-300khs on all cards with +20 core and +410 mem.  I think if I go too much more than that it crashes.  Since they're FTW cards, that comes out to 1209(1288 boost) clock and 1555 mem.  I can probably go higher if tweaking them separately since I probably have at least 1 card that's a weak overclocker, but that is stable so I left it.  I'm thinking I will get more without risers, but I have no way to test more than 1 on this board since it's only got 1 16x slot.  **Just an FYI, but GPU-Z is saying boost is 1288 but Precision X is actually reporting 1330-1350 for clocks**

On my other rig with Gigabyte 750ti's, I should have researched more instead of jumping on the cheapest board simply because it came with a freebie mpcie ssd, but that one seems to experience the degradation people are talking about when adding more cards.

That mobo is an ECS Z77H2-A4 with a Celeron G1620.  Adding 4 cards is fine, but it seems to choose PCI-E 1.1 1x mode for every card on the 1x slots and only PCI-E 2.0 for the one plugged in to the first x16 slot. It boots with the 5th plugged in but only detects 3 cards then. I've tried the x16 slot with x1 and x16 risers and it still only did PCI-E 2.0, which is the only card getting full speed.  ***I just checked the processor specs while I was typing this and realized the problem.... "PCI Express Revision 2.0 / Supports up to 1x16, 2x8, 1x8 & 2x4"

I wonder how many other people cheaped out on processors and are getting similar declines in hashrate or incompatibilities with multiple cards. I surely can't be the only one.  The ASRock board is running a socket 1150 i3 4130, so I'm guessing that's why I don't see a decline in khs. (Only 1155 cpus i5 and up supported PCIE v3)

I have an ASRock 970 Extreme 4 and some shit 170u CPUs. I have 4x Zotac 750Tis in my rigs and they don't slow down except for thermal slow downs. But then again my cards average around 280-295 to begin with.

So if that's the case then the issues of slow hashrates using pcie/usb riser depends on the hardware. So make sure your processor supports PCI-E ver. 3.0 but of course more hardware testing should support this claim. One thing if youre motherboard supports the PCI-E 3.0 then a graphics card supporting that bus speeds should be use when using risers.

Tech Specs: quick comparison)

PCI-E 2.0   5.0GT/s (raw bitrate) - 500MB/s (bandwidth per lane direction)
PCI-E 3.0   8.0GT/s (raw bitrate) - 1Gb/s   (bandwidth per lane direction)

 

If you're running AMD I don't think the CPU bit applies, but it would be great if we could confirm the drastic slowdowns are only on Intel rigs running CPUs that don't support PCI-E 3.0.

I still don't know how people are getting 330khs out of their cards though.  Anything close to 310 on my gigabyte and evga cards crashes cudaminer or the driver in a few minutes. I might be able to get it stable at 310 with lots of trial and error but haven't had the time or patience lately.

What are your settings for 310khs? I have got 2xZotac 750tiOC and 4xGigabyte 750tiOC in Asrock H81 Pro BTC and i was not able to break 300khs with risers.