[ANN] PTS.1GH.COM - GPU, CPU mining ★ New version 1.4! Please upgrade ASAP! ★

[fishrat]

Wow,so cool!

[reorder]

By long run, do you mean a week, a few weeks, a month?

Several days at least, to let the pool luck average out.

[mav2000]

By long run, do you mean a week, a few weeks, a month?

[reorder]

I have been mining on this pool for a few hours now and payouts seem to be very low as compared to where I was mining earlier. I am at around 1000 cpm but I just dont see any returns. Am I reading too much into the numbers?

It only pays out in the long run: the pool is finding blocks rarely due to 95% hashrate being on *cough* another pool. But when it does, you get more.

[mav2000]

I have been mining on this pool for a few hours now and payouts seem to be very low as compared to where I was mining earlier. I am at around 1000 cpm but I just dont see any returns. Am I reading too much into the numbers?

[5chdn]

Here, here, where's the beer?

[Gwynbleidd]

How to use more than 1,2GB VRAM?

[reorder]

For example GTX 780 he writes a maximum of 24 warp, but in fact their for her 48
App for check it: http://a0.sderni.ru/d/558165/deviceQuery.exe

Probably he has done it to avoid register spilling: his kernel could not run 48 threads in a warp on 780 without overusing the register file. Of course it could be in mistake too.

The 1gh miner uses workgroup size that the driver tells it it can use.

[old_pioner]

For example GTX 780 he writes a maximum of 24 warp, but in fact their for her 48
App for check it: http://a0.sderni.ru/d/558165/deviceQuery.exe

[reorder]

Did u used cuda optimisation like here? https://bitcointalk.org/index.php?topic=167229.0
But developer used only 50% cuda kit.

No, the miner uses OpenCL. Not sure what 50% cuda kit means, though.

[old_pioner]

Did u used cuda optimisation like here? https://bitcointalk.org/index.php?topic=167229.0
But developer used only 50% cuda kit.

[reorder]

768M actually, but OS needs some for its own needs too. Like, to show windows on the screen etc.

And if the second card, 500 MB is not enough too?

No, 768M is the minimum miner tries to allocate. It could be reduced further at a price of performance, but not below 512M - and then it makes no sense as there is only one model with 768M I believe, and it is 8800-ish Nvidia which is not great for mining anyway.

[old_pioner]

768M actually, but OS needs some for its own needs too. Like, to show windows on the screen etc.

And if the second card, 500 MB is not enough too?

[reorder]

why should they make people use a different pool than their own... strange idea ^^

neeeeed a 512mb vram version tho :S (why not accept parameters like other miners??)

That's because the algorithm needs full 512M just for hashes, and then some to search for solutions. Host-device memory transfers are quite expensive in OpenCL, if the job cannot be done in just VRAM, it is going to be real slow.

Seems min 800mb

768M actually, but OS needs some for its own needs too. Like, to show windows on the screen etc.

[old_pioner]

why should they make people use a different pool than their own... strange idea ^^

neeeeed a 512mb vram version tho :S (why not accept parameters like other miners??)

That's because the algorithm needs full 512M just for hashes, and then some to search for solutions. Host-device memory transfers are quite expensive in OpenCL, if the job cannot be done in just VRAM, it is going to be real slow.

Seems min 800mb,     

but the memory controller loaded little, can there be provision for an upgrade?