AMD OpenCL Momentum miner

[NaN]

I tested Catalyst 14.6RC2 on Windows 7 64bit:
    R9 280X@1100/1500 MHz: 5700+ cpm with -a 0 (and -t 0,0)

[kpematop]

What is the performance of the miner with 14.6 catalyst on win7 64bit for 280x and 7950?

[Ykw]

POW of NRS has ended .

Is it possible to obtain a number of Pool to mine PTS instead on YPOOL , please ?

Thanks in advance.

EDIT

Automatic mining from NRS is now over PTS over YPOOL

[xeroc]

I only know of beeeeer.org as pts mining pool

[jeanjean15]

POW of NRS has ended .

Is it possible to obtain a number of Pool to mine PTS instead on YPOOL , please ?

Thanks in advance.

[Coindgr]

Just found something very appealing for users of 290/X.... and above...

Summer is coming and I am waiting for my AC upgrade... in order to run overclocked GPUs and CPUs.. without burning at home.. so.. super experiments.. come to place.. where you try to lower the temps at all cost!

I have set the 290X (via AMD driver).. to -50% power.. and -47.x% GPU clock... and xarammmmm the results! AMAZIMG!!!

This is... brutal efficiency! Temps when from 72 ºC -> 60 ºC (on die with over 28ºC temps ambient)!! The current AMPs went even further!.. 82 -> 60 ºC!!! Power consumption varies.. as the card is constantly changing clocks... but.. in average.. is a lot less.. I would say.. 40% less power consumption.. So.. guys..! more viability to 290/X GPUs than with the older ones.. the 270X can't do the same...

But I would like to ear from your results... mines are like this:

Code Select Expand

13:37:17 | 4256 c/m | 8.366 sh/m | VL: 5742 (98.98%) | RJ: 56 (0.97%) | ST: 3 (0.05%) | ART: 52.5 ms
(47.62/37.61/23.32)

These results you're getting are similar to mine, but when I want to lower temperature I just use one thread.

[Ykw]

Just found something very appealing for users of 290/X.... and above...

Summer is coming and I am waiting for my AC upgrade... in order to run overclocked GPUs and CPUs.. without burning at home.. so.. super experiments.. come to place.. where you try to lower the temps at all cost!

I have set the 290X (via AMD driver).. to -50% power.. and -47.x% GPU clock... and xarammmmm the results! AMAZIMG!!!

This is... brutal efficiency! Temps when from 72 ºC -> 60 ºC (on die with over 28ºC temps ambient)!! The current AMPs went even further!.. 82 -> 60 ºC!!! Power consumption varies.. as the card is constantly changing clocks... but.. in average.. is a lot less.. I would say.. 40% less power consumption.. So.. guys..! more viability to 290/X GPUs than with the older ones.. the 270X can't do the same...

But I would like to ear from your results... mines are like this:

Code Select Expand

13:37:17 | 4256 c/m | 8.366 sh/m | VL: 5742 (98.98%) | RJ: 56 (0.97%) | ST: 3 (0.05%) | ART: 52.5 ms
(47.62/37.61/23.32)

[Ykw]

For the ones that use the GPU for multiple purposes... (and only for 290/X models) i found out that using "-a 3" it's actually a more performer mode for mining... but also more prioritizing for the same realm. Giving you little usage for other multipurpose activities: games, video accelerated streaming... and everything that uses about less than 50% of your the GPU.

I found that the best option for multipurpose/mining is "-a 0".... as I have been using till now without problems. The observation taken is simple:

"The mining is running over the background.. and when I start watching a film with GPU acceleration, it crackles a bit until it adapts to the new workload.. then it settles just right for it.." - using for example "splash PRO EX".

Using "-a 3" you will experience a millisecond lag.. periodically... let's say.. less than half second period, where your eye experience a freeze over the frame.

Is this true for newer versions of CPU/BOARDS/RAM etc?... I have a X58 with XEON 4Cores and 1600 DDR3.

Disclaimer... This is a result of several software and hardware experiences.. ones more noticeable than others... but.. they all conclude that the most adaptable choice is the one I mention (-a 0).

[Ykw]

What is the ART over the miner?

How is it measured...

[JoeyD]

I can confirm a very big boost in performance from the 14.6 Catalyst on ubuntu linux, last I checked I was getting over 5900 c/m for each 280X with -a 0, it still hasn't stabilized yet, while I was averaging around 5200 c/m with the 14.4 catalyst version.

[Ykw]

you need about 1200 shares for an optimal result

Yes.. for -a 3 specially... for the -a 1 and 0 it was somehow quicker to stabilize or the result approaches faster to the stable one (more than 4h running.. with max/min values)..

[Ykw]

Some tests with Catalyst 14.6 on Windows 7 x86_64 (at least >30 minutes to stabilize or +1200 shares - Done.. aborted. I had to change hardware due to local temperature... test env is different now..):
@Mario241077 Thanks for the nice number =)

[with -a 0]:
-  R9 270X@1120/1400MHz: ~2230 col/m & ~4.7 sh/m
-  R9 290X@1050/1350MHz: ~5500 col/m & ~10.60-11.16 sh/m - 618-647 Share/h @ypool (1h54m)

Code Select Expand

14:44:56 | 5491 c/m | 10.68 sh/m | VL: 1206 (99.26%) | RJ: 9 (0.74%) | ST: 0 (0.00%) | ART: 40.58 ms
(37.49/25.15/18.18)
    EDIT: Interesting result for the 290X (this actually matches my initial runs, where -a 0 was faster for me.. when I had 3k col/m with the 290X)

[with -a 1]:
-  R9 270X@1120/1400MHz: ~2540 col/m & ~4.56-4.73 sh/m - 259-275 Share/h @ypool (4h47m)

Code Select Expand

17:39:03 | 2551 c/m | 4.719 sh/m | VL: 1344 (98.90%) | RJ: 15 (1.10%) | ST: 0 (0.00%) | ART: 4.937e+
008 ms (44.37/21.82/21.4)
-  R9 290X@1050/1350MHz: ~5420 col/m & ~9.63-10.94 sh/m - 567-632 Share/h @ypool (1h57m)

Code Select Expand

18:51:56 | 5418 c/m | 10.94 sh/m | VL: 1280 (99.15%) | RJ: 11 (0.85%) | ST: 0 (0.00%) | ART: 40.89 m
s (40.1/25.16/16.15)

[with -a 2]:
-  R9 270X@1120/1400MHz: ~2215 col/m & ~4.14-4.37 sh/m - 243-252 Share/h @ypool (4h40m)

Code Select Expand

22:20:31 | 2218 c/m | 4.369 sh/m | VL: 1209 (98.77%) | RJ: 15 (1.23%) | ST: 0 (0.00%) | ART: 6.37e+0
08 ms (45.8/28.78/26.64)
-  R9 290X@1050/1350MHz: ~5385 col/m & ~10.07-10.35 sh/m - 577-594 Share/h @ypool (2h03m)

Code Select Expand

16:51:19 | 5362 c/m | 10.34 sh/m | VL: 1259 (98.05%) | RJ: 23 (1.79%) | ST: 2 (0.16%) | ART: 41.82 m
s (37.55/25.29/20.45)

[with -a 3]:
-  R9 270X@1120/1400MHz: ~2580 col/m & ~5.55-5.67 sh/m - 322-327 Share/h @ypool
-  R9 290X@1050/1350MHz: ~5470 col/m & ~10.50-11.24 sh/m - 626-649 Share/h @ypool (2h14m)

Code Select Expand

21:08:51 | 5474 c/m | 11.17 sh/m | VL: 1493 (98.68%) | RJ: 19 (1.26%) | ST: 1 (0.07%) | ART: 41 ms (
38.22/23.96/19.35)
    EDIT: I think Mario241077 is right here.. it takes a lot more time to stabilize.. but at the end.. the result is better, in average I would say (still testing with GPU usage for main adapter).

NOTE: Just noticed that I need no more to switch displays to activate the GPUs without screen connected. =) I am not sure if it was with this new beta. Very handy. This might have been introduced with the Mantle version.

Disclaimer: My results might differ a bit in max/min ration because I am also using CPU to other purposes.. but leaving at least 50% of the logical cores free. I will try to set the max I see and the min.. after stabilization. The results include using the 290X for the display adapter.

[Mario24]

win 7 64x driver 14.6

I use the monitor at the R 290X

I use the command line
setx GPU_FORCE_64BIT_PTR 1
setx GPU_MAX_ALLOC_PRECENT 100
these I have taken from the mines bfgminer description for Scrypt

-t 0,0 -a 0  HD 7970  4041=c/m - 7.86=sh/m - GPU 1000 Mhz - Memory 1375 MHZ - ART=55.2 - Share value/h on ypool.net=456.65 - stable trend after shares total 1483

-t 1,1 -a 0  R 290X  5117=c/m - 10.25=sh/m - GPU 1050 Mhz - Memory 1350 MHZ - ART=2.832e+008 - Share value/h on ypool.net=594.06 - stable trend after shares total 3961 - 22:02:42 | 5117 c/m | 10.25 sh/m | VL: 3961 (98.85%) | RJ: 39 (0.97%) | ST: 7 (0.17%) | ART: 2.832e+008 ms (40.46/27.2/18.85)


-t 0,0 -a 1  HD 7970  4357=c/m - 8.64=sh/m - GPU 1000 Mhz - Memory 1375 MHZ - ART=51.17 - Share value/h on ypool.net=500.62 - stable trend after shares total 3633 - 21:51:21 | 4358 c/m | 8.664 sh/m | VL: 3633 (98.75%) | RJ: 42 (1.14%) | ST: 4 (.11%) | ART: 51.17 ms (56.87/21.57/23.26)

-t 1,1 -a 1  R 290X  5534=c/m - 10.98=sh/m - GPU 1050 Mhz - Memory 1350 MHZ - ART=2.496+008 - Share value/h on ypool.net=635.32 - stable trend after shares total 2597


-t 0,0 -a 2  HD 7970  4008=c/m - 7.78=sh/m - GPU 1000 Mhz - Memory 1375 MHZ - ART=55.92 - Share value/h on ypool.net=449.63 - stable trend after shares total 2286  - 02:51:32 | 4008 c/m | 7.79 sh/m | VL: 2265 (98.65%) | RJ: 29 (1.26%) | ST: 2 (0.09%) | ART: 55.93 ms (53.55/24.97/32.83)

-t 1,1 -a 2  R 290X  5067=c/m - 9.51=sh/m - GPU 1050 Mhz - Memory 1350 MHZ - ART=1.213e+008 - Share value/h on ypool.net=547.37 - stable trend after shares total 1965 - 07:55:00 | 5067 c/m | 9.477 sh/m | VL: 3461 (98.63%) | RJ: 47 (1.34%) | ST: 1 (0.03%) | ART: 1.213e+008 ms (70.35/44.51/37.01)



-t 0,0 -a 3  HD 7970  4400=c/m - 8.43=sh/m - GPU 1000 Mhz - Memory 1375 MHZ - ART=50.5 - Share value/h on ypool.net=491,37 - stable trend after shares total 5730 - 22:49:05 | 4409 c/m | 8.514 sh/m | VL: 5730 (98.76%) | RJ: 68 (1.17%) | ST: 4 (0.07%) | ART: 50.91 ms (56.71/21.36/23.28)

-t 1,1 -a 3  R 290X  5359=c/m - 10.39=sh/m - GPU 1050 Mhz - Memory 1350 MHZ - ART=1.64e+008 - Share value/h on ypool.net=602.42 - stable trend after shares total 3062 - 03:00:47 | 5359 c/m | 10.39 sh/m | VL: 3031 (99.15%) | RJ: 26 (0.85%) | ST: 0 (0.00%) | ART: 1.64e+008 ms (40.99/22.25/19.62)



without command line
setx GPU_FORCE_64BIT_PTR 1
setx GPU_MAX_ALLOC_PRECENT 100


-t 0,0 -a 0  HD 7970  =c/m - =sh/m - GPU 1000 Mhz - Memory 1375 MHZ - ART= - Share value/h on ypool.net= - stable trend after shares total
-t 1,1 -a 0  R 290X  =c/m - =sh/m - GPU 1050 Mhz - Memory 1350 MHZ - ART= - Share value/h on ypool.net= - stable trend after shares total  -


-t 0,0 -a 1  HD 7970  4364=c/m - 8.60=sh/m - GPU 1000 Mhz - Memory 1375 MHZ - ART=51.1 - Share value/h on ypool.net=497.28 - stable trend after shares total 2734 - 08:15:29 | 4364 c/m | 8.588 sh/m | VL: 2702 (98.97%) | RJ: 26 (0.95%) | ST: 2 (0.07%) | ART: 51.1 ms (78.27/29.7/32)

-t 1,1 -a 1  R 290X  5539=c/m - 10.75=sh/m - GPU 1050 Mhz - Memory 1350 MHZ - ART=2.414+008 - Share value/h on ypool.net=621.96 - stable trend after shares total 2706 - 08:43:29 | 5540 c/m | 10.76 sh/m | VL: 3597 (98.82%) | RJ: 43 (1.18%) | ST: 0 (0.00%) | ART: 2.414e+008 ms (53.96/28.07/21.75)


-t 0,0 -a 2  HD 7970  =c/m - =sh/m - GPU 1000 Mhz - Memory 1375 MHZ - ART= - Share value/h on ypool.net=
-t 1,1 -a 2  R 290X  =c/m - =sh/m - GPU 1050 Mhz - Memory 1350 MHZ - ART= - Share value/h on ypool.net=

-t 0,0 -a 3  HD 7970  4406=c/m - 8.70=sh/m - GPU 1000 Mhz - Memory 1375 MHZ - ART=50.92 - Share value/h on ypool.net=504.61 - stable trend after shares total 7612 - 23:09:24 | 4406 c/m | 8.704 sh/m | VL: 7813 (99.00%) | RJ: 74 (0.94%) | ST: 5 (0.06%) | ART: 50.92 ms (56.66/21.34/23.27)

-t 1,1 -a 3  R 290X  =c/m - =sh/m - GPU 1050 Mhz - Memory 1350 MHZ - ART= - Share value/h on ypool.net=


This ART value is important for NAN
the lower is better

you need min 1200 shares for an optimal result, more is better for correct average value on. you reach only after a very long time and many shares. let it run

[NaN]

I tested Catalyst 14.6 on Linux
    R9 280X@1100/1500MHz: >5600 cpm with -a 0
    R9 290@stock: ~5250 cpm with -a 1

This looks promising but the Windows driver for R9 290(X) still seems to be slower than the Linux driver.

[lilunxm12]

The 14.6 is faster than 14.4 for me. 10400 cpm vs 9300 cpm,dual 290x,1000/1250
I didn't test 13.12 though

windows or linux?

windows