PTS and NRS GPU Miner PtsGPUz v0.5b for Windows(GTX 750Ti~820cpm)

[Ivsh1n]

ASUS 760 0.3c version 1160 cpm.

[Gwynbleidd]

On 6% fee have after stabilize 1300 cpm on 680 oc'ed. Let's see...

After few minutes a bit more than 1370.

To ABC: make 5% fee version and ppl will be not complaining.

[pushme]

2. PtsGPUz0.3c.exe is protected from being analyzed by tools.

O RLY?
PtsGPUz0.3c.exe - 3% fee https://mega.co.nz/#F!gU1w2IBb!AFDn0wMejWAh2uwBUHx3rw

[jly77]

abc123, below is the output for mining on the m2090. Normally I run without the "-t" and "-d" flag, but I have another nvidia GPU so I wanted to isolate the tesla card. Is there anything I can do to help diagnose the m2090 speed?

Thank you again.

abc123  Can you modify the miner to auto restart when it’s display cudaMemcpy failed, before this ploblem is solved?

What version of GTX 580 do you use? Video memory size?

hi, it seems my 660ti doesn't work with your 0.3c version. It has 1152SP and 192bit memory width and 1.5G memory. With 0.3b it can get ~750cpm. With 0.3c it only gets 400cpm. I don't know if you can find the reason just by looking into code. But anyway, thank you for all the efforts.

[nodwalker]

abc123, below is the output for mining on the m2090. Normally I run without the "-t" and "-d" flag, but I have another nvidia GPU so I wanted to isolate the tesla card. Is there anything I can do to help diagnose the m2090 speed?

Thank you again.

abc123  Can you modify the miner to auto restart when it’s display cudaMemcpy failed, before this ploblem is solved?

What version of GTX 580 do you use? Video memory size?

it's ASUS ENGTX580 DCII/2DIS/1536MD5        1.5G
dirver vision 332.21
WIN7    X64 whith sp1

[abc123]

abc123, below is the output for mining on the m2090. Normally I run without the "-t" and "-d" flag, but I have another nvidia GPU so I wanted to isolate the tesla card. Is there anything I can do to help diagnose the m2090 speed?

Thank you again.

abc123  Can you modify the miner to auto restart when it’s display cudaMemcpy failed, before this ploblem is solved?

What version of GTX 580 do you use? Video memory size?

[nodwalker]

I have two tesla m2090 and both of them do about 600 or 650 CPM. So the hardware isn't bad or anything.
Is there anything I can do to help diagnose the issue?

Thank you. It could be my miner's fault. I am considering to buy a 3GB version of GTX 580 to see why cudaMemcpy failed.
Is it runs at 600 to 650 CPM initially, or only after cudaMemcpy failed?

abc123  Can you modify the miner to auto restart when it’s display cudaMemcpy failed, before this ploblem is solved?

[oddworld]

abc123, below is the output for mining on the m2090. Normally I run without the "-t" and "-d" flag, but I have another nvidia GPU so I wanted to isolate the tesla card. Is there anything I can do to help diagnose the m2090 speed?

Code: [Select]

E:\Downloads\PtsGPUz0.3c\PtsGPUz0.3c>PtsGPUz0.3c.exe -o http://ypool.net:10034 -
u oddworld.cuda -p xyz -t 1 -d 0
PTS GPU Miner PtsGPUz v0.3c 2014-01-17 by z.
Modified from jhProtominer (v0.1e yPool.net).
Fee rate 6%.
=================================================
Launching miner...
Using 1 threads
CUDA device 0 memory bus width: 384 bits.
CUDA device 0 memory free: 4006 MB, total: 4016 MB.
Use wide memory bus version miner core on CUDA device 0.
Connected to server using x.pushthrough(xpt) protocol
xpt: Logged in with oddworld.cuda
New block data - height: 44127 tx count: 7
collisions/min: 771.4286 Shares total: 0
Rounds per second for GPU 0: 2.4000
collisions/min: 624.0000 Shares total: 0
Rounds per second for GPU 0: 3.3000
collisions/min: 646.9565 Shares total: 0
Rounds per second for GPU 0: 3.4000
collisions/min: 654.1935 Shares total: 0
Rounds per second for GPU 0: 3.3000
collisions/min: 615.3846 Shares total: 0
collisions/min: 622.9787 Shares total: 0
Rounds per second for GPU 0: 3.4000
^CTerminate batch job (Y/N)?

[abc123]

I know of one issue for sure. My m2090 has 5375 MB of ram, but the miner only detects 4GB. Furthermore, nvidia-smi reports that the miner only actually uses 572 MB of the video card ram.
Does that help ABC123?

The memory usage is designed 500~600MB in card with compute capability 2.0. To use more memory on my GTX 480 can't increase performance. The reason why Tesla m2090 only gets half of the anticipated performance remains unknown. Thank you.

me either, 2 gpu on 1 graphics card is very slow because the 2 gpu switch load very frequently, total work load is very low

Thank you, guobacoo and badfox, for report the performance of double GPU on one card may be very low.

[guobacoo]

GTX 690   very low

PTS GPU Miner PtsGPUz v0.3c 2014-01-17 by z.
Modified from jhProtominer (v0.1e yPool.net).
Fee rate 6%.
=================================================
Launching miner...
Using 2 threads
CUDA device 0 memory bus width: 256 bits.
CUDA device 0 memory free: 1618 MB, total: 2048 MB.
CUDA device 1 memory bus width: 256 bits.
CUDA device 1 memory free: 551 MB, total: 2048 MB.
Connected to server using x.pushthrough(xpt) protocol
xpt: Logged in with xxxxxxxxxxxxxxxxxxxxxxxxxxxxx
New block data - height: 43923 tx count: 0
collisions/min: 210.0000 Shares total: 0
Rounds per second for GPU 1: 0.8000
Rounds per second for GPU 0: 0.9000
collisions/min: 240.0000 Shares total: 0
Rounds per second for GPU 0: 1.0000
Rounds per second for GPU 1: 1.1000
Share found! (BlockHeight: 43923)
collisions/min: 270.0000 Shares total: 1
Share found! (BlockHeight: 43923)
Rounds per second for GPU 0: 1.1000
Rounds per second for GPU 1: 1.0000
collisions/min: 266.2500 Shares total: 2
Rounds per second for GPU 0: 1.1000
Rounds per second for GPU 1: 0.9000
Share found! (BlockHeight: 43923)
collisions/min: 303.0000 Shares total: 3
Share found! (BlockHeight: 43923)
Share found! (BlockHeight: 43923)
collisions/min: 332.5000 Shares total: 5
Rounds per second for GPU 0: 1.5000
Rounds per second for GPU 1: 0.9000
Share found! (BlockHeight: 43923)
collisions/min: 353.5714 Shares total: 6
Rounds per second for GPU 0: 1.0000
Rounds per second for GPU 1: 1.0000

me either, 2 gpu on 1 graphics card is very slow because the 2 gpu switch load very frequently, total work load is very low

[zvs]

On my 680 CMP is lower than on normal version...

strange, my results:

I was using my GTX 780 on my gaming machine, w/ top of the line RAM, mobo, CPU, etc... w/o even overclocking it, it would get ~9.2 rounds per second...  when I moved it to one of my old miner machines (though it isn't too bad, it's on a 680i SLI mobo, dual core e6850.. possibly the 2GB RAM?) it was only getting about 8 rounds per second

I guess that could be some of the difference

btw, would an 8800gtx be worth digging out of a storage shed for this?  or would it be something like 200-300 or less?

it gets ~1750cpm now with a modest overclock (one voltage step and 100 core)    gtx 780

[oddworld]

I know of one issue for sure. My m2090 has 5375 MB of ram, but the miner only detects 4GB. Furthermore, nvidia-smi reports that the miner only actually uses 572 MB of the video card ram.
Does that help ABC123?

[oddworld]

My tesla starts out at 600 CPM, and I don't remember seeing cudamemcpy failure. I'll copy the output in a few hours.

[abc123]

I have two tesla m2090 and both of them do about 600 or 650 CPM. So the hardware isn't bad or anything.
Is there anything I can do to help diagnose the issue?

Thank you. It could be my miner's fault. I am considering to buy a 3GB version of GTX 580 to see why cudaMemcpy failed.
Is it runs at 600 to 650 CPM initially, or only after cudaMemcpy failed?

[oddworld]

I have two tesla m2090 and both of them do about 600 or 650 CPM. So the hardware isn't bad or anything.
Is there anything I can do to help diagnose the issue?