Topic: Gigabyte 7950 (GV-R795WF3-3GD) Bitcoin Config Thread (Read 10423 times)

I was having the same sort of problem and I noticed that cgminer was automatically setting 2 threads per gpu which effectively doubles your thread concurrency. try running without anything about thread concurrency in your command line and if it starts tell it to write a configuration ((s) then (w) I think) file then edit the one it produces until you get results you like.

I was running mine with the gpu clock pegged at 1200 and memclock as low as it would go, 1050 I think. ran stable for days at 70c with 70% fan and getting 610Mh/s. I am mining script now and getting 660Kh/s stable.

i also have the same issue work well on rig 1 and rig 2 doest work i dont know what to do

I got it stable @629Mh/s while working on the computer


Info

OS:	Windows 7
Miner:	cgminer 3.1.1
Bios:	F43
Driver:	Catalyst 13.4
GPU clock:	1200 Mhz
Mem clock:	950 Mhz
Voltage:	stock
Fan:	auto (~58%)
Temp:	71°
Avg Hashes:	629 Mhz

config

read https://bitcointalksearch.org/topic/m.2111051
This is post for Gigabyte WF3 7950  help me https://forum.litecoin.net/index.php/topic,2641.0.html
I test mining BTC - 2x7950 Gigabyte WF3,guiminer,1100 core/625 memory 1100~1150MH/s 480W Power "on the wall"
but now i mine  LTC 1140/1420 ~1250kh/s - 590-610W. (open rig,73-78°C)

I just got my Gigabyte 7950s (GV-R795WF3-3GD) rev 2 with 6+8 pins.

Factory Overclock is 1000Mhz which mines 500MH/s at 325watt (+100w for system)

88-92c and SADLY CRASHES after just 10-20 Minutes! It's 25w over what the PCI-E cables + MB supply!

I ran tests at reference clock 800Mhz for an hour which mines 400MH/s at 260watt, 78-80c and seems fine.

I ran tests at rev 1 OC 900Mhz for an hour which mines 450MH/s at 275watt, 81c, and seems fine.

It's not as good as 2x7850s I normally use I get 600MH/s at 260watt, even 2x5830s get 500MH/s for around 325watt.  

I think there was a reason these were cheap with 4 free games on HotUKDeals.

----

I just read about the BIOS switch earlier in this thread, flicked it, now it's running 1000MHz mining 500MH/s at 300watt and 81c.

I tested it for an hour, SUCCESS! No crashes.

I'll try the F43 BIOS later as it sounds more power efficient, but at least the cards are not duff and not need to be returned. 

I don't know if most people paid attention to the voting. It's about Mh/s and not Kh/s. So its obviously about bitcoin mining.
I know this card performs well on LTC mining, but I am trying to figure out the best settings for BTC and also want to know the avg Hash rate on BTC for this card.

So all the votes that are above 650 did you actually mean BTC or LTC mining?

I edited the poll/question and included the word bitcoin

I just use MSI After burner (beta).
I have catalysts overclocking disabled.

The only problem is that MSI Afterburner does not keep the settings after a reboot.

Anyways, I run AB as admin, set the maximum clock speed and then drop the memory down to 600 or 620 depending on the card.
Hit apply, and do it to all of the cards (I mix 7950 and 7870s).  Then I exit
Reload AB and then I drop it again to 300 or 310.
Hit apply, then exit (or leave it running, but I jsut exit it).
My settings are:

-Enable hardware control and monitoring
-Enable low level hardware access interface - set to user mode
-Disable ULPS
-Unofficial overclocking mode without powerplay support

I don't change the voltage on the board, as I haven't needed to as yet (and if I ever need to - I just wont overclock that high)

I don't bother with start with windows etc - as it doesn't remember the settings anyways when unofficial overclocking mode is enabled.
That means that my ram speed always starts at default (although cg can drop it by 150).  I then have to load  up AB and set it to 300/310.
However I can RDP or tight vnc to it from work or from my phone, and its stable enough that it barely resets.  I have caught it blue screening and forcing a reset twice though, and once I had to ring up and get my wife to do a physical reset when its TCPIP stack appeared to have died (couldn't RDP or VNC in, or PSEXEC to it, yet it was still mining)

To get unofficial overclocking mode running, I had to change an ini file.  After that happened, and the EULA popped up, and closed and reopened, I then closed it again and did a reboot.  After that it came good - except that it now no longer stores settings/profiles.

GPU Engine 1150MHz
GPU Memclock 1250MHz
Thread Concurrency 21000
Intensity 18


Getting just under 600Khash/sec from each card.  I could bump it up a little bit more, but would rather play on the safe side.

I had tried that already with no effect (despite cmd confirmations of the changes) on both machines. Thanks for the reply though.

Back on topic of the thread, settings:

F43 BIOS
GPU Engine 1110MHz
GPU Memclock 1400MHz
Powertune +20%
Thread Concurrency 16384
Intensity 19
Worksize 256

Fan speed usually 50 - 70% (Auto)
Temps 65C (Min) - 76C (Max) Depending on time of day and weather
2 cards per rig, open air

Win7x64
Catalyst 13.1
GUI-Miner scrypt

~650kH/s per card

See: https://bitcointalksearch.org/topic/m.1256502


guiminer-scrypt is a bit nicer because it doesn't cause you to permanently alter your environmental variables, though.

Hi Taco,

Funnily enough that's what I am now doing on the 2 that don't work with my .bat file on cgminer. Now happy at ~650kH/s.

However, this doesn't help the important learning process of what I was (or wasn't) doing for it to produce the errors!

you need to set environmental variables

easy mode: use guiminer-scrypt

Win 7x64, Catalyst 13.1, cgminer 3.1.1

I've executed

.bat file is:


This works on rig 1, the error on the other 2 is:


I had just ignored it and have been running for a couple of months at a slightly lower kH/s with 2 threads at TC of 8192, but I can't help but keep coming back to it as I want to figure out why it doesn't work.

F43 bios
engine 1135
intensity 18
lookup-gap 2
-g 1
worksize 256

2 gpu in one rig
win 7 x64
8 gb ram

catalyst 13.1
cgminer 3.1.0

over 610 kh/s

What OS/Catalyst are you running on each rig? What's the specific error you're getting?

So I have 3 rigs running these 2 of these cards each, flashed with F43 BIOS, on one rig (e8400, 4GB DDR2 RAM) I can run at a thread concurrency of 24000 with 1 GPU thread at ~650kH/s per card after using the GPU_MAX_ALLOC_PERCENT 100 command.

On the other two rigs, (Intel G2130, 8GB DDR3 RAM) even after using this command and the exact same settings as the other rig, CGMiner won't start on a TC higher than 8192. The highest I can get these cards at TC 8192 is ~595kH/s using 2 GPU threads.

Can anybody offer advice as to why this only seems to work on one of my rigs and not the other two? It seems although it confirms the setting has been changed this is not being applied, and thus CGMiner reports a memory buffer overload. Reading thread after thread of people pulling 650kH/s> with high TC is really starting to bug me when two of mine don't work for some reason

link?

the problem is that i can't set intensity over 13, or the gpu drives stop working

I guess you guys don't pay much attention to #litecoin, or the litecoin forums. I've been talking about how to get 700kh/s from a 7950 for a few months now, and someone even posted a semi-competent guide on how to get over 650kh/s on the litecoin forums. These cards will easily do more than 610Mh/s with BTC.