[1 BTC BOUNTY CLAIMED!] Get 6 radeon 7970 installing and mining in Windows

[ssateneth]

Just because it's newer doesn't mean it's better. When the 2.6 SDK came out with the 11.12 driver, Many 5k series radeon users suffered a huge performance hit for mining, because the new SDK was absolutely terrible for mining. It was a headache to roll back to 2.5 SDK because the uninstall does not remove the SDK files, so you are stuck with the newer SDK until you manually found and removed the files.

[wojgp]

Hi
Is there any option to run 5 x 7870XT Tahiti on MSI Z77A-GD65 under win7 64-bit ? I think I've already tried all options, but always have error 43. 7870XT is supported since 12.11 beta 4 (if I remember well).
I tried powered risers, but since the issue is in iddle mode then I think it is not necessary to try with it.
I use dummy plugs for all cards, I DO NOT use pins shorting in PCI-E sockets.
Please help.

[ssateneth]

dont need dummy plugs for anything. that is a rumor for newer hadrware/drivers and not confirmed to help anything as far as windows goes.

you are pretty much screwed btw. you are limited to maximum 4 7870XT in win7 and 5 7870XT in Win8. This is due to driver/OS limitations that cannot be overridden.

There is rumor of a techpowerup 13.1 driver having extra GPUs enabled but no guarantee. I dont have a link for it, search around.

[tdslot]

Hello,

I have problem. Who can help me?

I have mining rig hardware:

• 4x XFX7950 FX-795A-TDJC
• ASUS SABERTOOTH 990FXR2.0 motherboard
• 8GB RAM
• CPU AMD FX8320 8x Cores 3.51 Ghz
• PSU Lepa 1600W

Software:

• Windows 7 SP1 x64
• AMD Radeon HD drivers 13.4 x64
• AMD SDK 2.7 x64
• CGMINER 3.3.1

My cgminer.conf

Code:

"api-allow" : "W:127.0.0.1",
"api-listen" : true,
"api-port" : "4028",
"expiry" : "120",
"failover-only" : true,
"hotplug" : "5",
"kernel-path" : "/usr/local/bin",
"log" : "5",
"no-submit-stale" : true,
"queue" : "1",
"rotate" : "10",
"scan-time" : "30",
"scrypt" : true,
"shares" : "0",
"auto-fan" : true,
"gpu-threads" : "1",
"gpu-dyninterval" : "7",
"gpu-engine" : "0-1025",
"gpu-fan" : "0-100",
"gpu-platform" : "0",
"gpu-memclock" : "1250",
"gpu-memdiff" : "0",
"gpu-powertune" : "0",
"gpu-vddc" : "1.087",
"intensity" : "20",
"temp-target" : "75",
"temp-overheat" : "88",
"temp-cutoff" : "95",
"temp-hysteresis" : "3",
"vectors" : "1",
"worksize" : "256",
"lookup-gap" : "2",
"shaders" : "2048",
"thread-concurrency" : "21712",
"no-pool-disable" : true

With this configuration 3x GPU cards running stable. But when I connect 4 card and run cgminer works very bad and turnoff motherboard. But PSU still running.

Who can help me with this problem? Maybe not impossible run rig with 4x GPU's? Maybe my configuration is wrong? Or I forget something to install?

Than you in advance,

TDS

[DiamondHunter]

There is rumor of a techpowerup 13.1 driver having extra GPUs enabled
It's not a rumor, I got 6 rigs running with that driver with 6 MSI 7950 in each. Clean Windows 8 Pro install followed by the above diver install. Nothing else.
Other people reported it working as well.


Question about GPUs going to power saving mode. On some of those 6 rigs I have, on and off one of the cards will go down to 60 KHs and restarting cgminer will not fix that. I noticed sometimes just moving the monitor cable from one gpu to another will instantly "wake up the card" running at 60kHs and fix it back to 630ih range.

I'ts been bugging me a while now and I can't find reliable fix. Does dummy DVI/VGA connectors work, or is there other stable solution that fixes that issue?

[BabylonDown]

Well it's been months since the thread was last updated. Has there been a permanent fix to running 6 GPU's reliably?

[jasinlee]

Nope, not that I have seen.

[lbr]

There is rumor of a techpowerup 13.1 driver having extra GPUs enabled
It's not a rumor, I got 6 rigs running with that driver with 6 MSI 7950 in each. Clean Windows 8 Pro install followed by the above diver install. Nothing else.
Other people reported it working as well.


Question about GPUs going to power saving mode. On some of those 6 rigs I have, on and off one of the cards will go down to 60 KHs and restarting cgminer will not fix that. I noticed sometimes just moving the monitor cable from one gpu to another will instantly "wake up the card" running at 60kHs and fix it back to 630ih range.

I'ts been bugging me a while now and I can't find reliable fix. Does dummy DVI/VGA connectors work, or is there other stable solution that fixes that issue?

Try disabling "AMD External Events" service nad "AMD Fuel" if you installed full Catalyst. And all power saving features, like 'Display Off', Screen Saver, PCIE link.

Also I have use
setx DISPLAY ":0"
or dummies on windows 2008 or all cards are always off except one with monitor connected.

[lbr]

Well it's been months since the thread was last updated. Has there been a permanent fix to running 6 GPU's reliably?

If someone will give me remote access to the 6GPU rig, I will find permanent fix ; )
Assuming that it can be fixed by modifing .inf files, I've posted differences in working and non-working infs ealier in this thread.

[MIN3R]

There is rumor of a techpowerup 13.1 driver having extra GPUs enabled
It's not a rumor, I got 6 rigs running with that driver with 6 MSI 7950 in each. Clean Windows 8 Pro install followed by the above diver install. Nothing else.
Other people reported it working as well.


Question about GPUs going to power saving mode. On some of those 6 rigs I have, on and off one of the cards will go down to 60 KHs and restarting cgminer will not fix that. I noticed sometimes just moving the monitor cable from one gpu to another will instantly "wake up the card" running at 60kHs and fix it back to 630ih range.

I'ts been bugging me a while now and I can't find reliable fix. Does dummy DVI/VGA connectors work, or is there other stable solution that fixes that issue?

To fix this problem I manually go into the registry (regedit), and change all "EnableULPS" entries to 0 instead of 1... Fixes the issue for me

* Note that if you change drivers, or add another card you'll have to go through and change them all again. I should just write a file to do this for me, but I'm too lazy, lol

[lbr]

Well it's been months since the thread was last updated. Has there been a permanent fix to running 6 GPU's reliably?

If someone will give me remote access to the 6GPU rig, I will find permanent fix ; )
Assuming that it can be fixed by modifing .inf files, I've posted differences in working and non-working infs ealier in this thread.

Modded 13.12 driver works for me for 6xGPU
https://litecointalk.org/index.php?topic=11761.0

[Timmieman]

Kinda old topic,Though still have a question.

I`m currently running a rig with 7 GPU cores (2x 6990 and 3x 5870) though I`d like to stick in another 5870.
I`m using the new H81 pro btc mobo with ivy bridge processor turned off in BIOS.

When I stick in the last 5870 in the 16x PCI-E port, the computer doesn't boot or even POST.
Is this a driver / OS problem? In other words, would for instance other drivers or OS be able to work with 8 GPU cores?

I`m putting up a 0.15BTC bounty for the one who gets me to mine with 8 cores in my setup.

Cheers

[ssateneth]

Well its not a tahiti system. I've used tried and true 11.12 or 12.1 driver for 8 GPU's (mix of 5870 and 5970). Those drivers should also support the 6k series. Lemme know how it works out. My donation address is in my sig.

[Timmieman]

I tried out both, the oldest one doesnt support 6990
Though my problem lies in the chipset I think.
When I put in the 8th GPU, the computer doesn't even pass POST.
It just hangs on a black screen, I have to clear CMOS to start up with a GPU less.
Could someone comment on this?

I'm gonna wipe my SSD clean and try again with fresh windows and drivers.

[lbr]

When I stick in the last 5870 in the 16x PCI-E port, the computer doesn't boot or even POST.

Maybe put it on x1 riser?
Also there is a setting in BIOS controlling PCIe rev. for this slot, maybe try them..
For 6xGPU on this motherboard it does not affect anything, but for 8x.. who knows.
There are also 'Multi Monitor' and Shared Memory(something like that, set to Auto by default), try changing them and disabling everything like Audio/USB/LAN or enabling(once I saw mobo disabling one PCIe slot if iGPU was disabled).
No POST - deffo not the OS issue, prly BIOS is not able to map all GPUs, maybe increase RAM.

[ssateneth]

Bad risers or bad power connections can attribute to weird system glitches/hangs if the connections are spotty. Also, apparently there are some modded drivers out there of recent driver revisions that allow more GPUs per W7 box.

https://litecointalk.org/index.php?topic=11761.0 Check those drivers out, see what it does for you.

BTW, it's impossible to use ivy bridge anything in H81. It's a haswell only motherboard with a different socket. If you forced an ivy bridge CPU in there, you probably fried your system.

Fake edit: after re-reading your post, are having a POST issue. Test each card and each riser individually. If they all check out, use a known good card and good riser and try them in each PCI-E slot.

[Timmieman]

You`re right, my original plan was a Z77a-gd65 with ivy bridge (would be the better choice i guess), but in order to save some cash I choose the H81 pro btc with G1820 (hashwell).

I`ll try to put in one GPU straight in the 16x PCI-e slot, since the mod as reported doesn't help for any POST problem, only code 43.

All risers and cards have been tested and shown in full working order.

Will post results!

[Timmieman]

Update!!

I've plugged in the 8th card while the pc as on (Was very carefull not to short anything  
And now it states the following error:

Code:

This device cannot find enough free resources that it can use. (Code 12)

If you want to use this device, you will need to disable one of the other devices on this system.

Perhaps one can contemplate on this?

The diagnosis of this error code is:

Two devices have been assigned the same input/output (I/O) ports, the same interrupt, or the same Direct Memory Access channel. The assignment was made by either the basic input/output system (BIOS), the operating system, or a combination of the two.


Cheers,

[koelen3]

Update!!

I've plugged in the 8th card while the pc as on (Was very carefull not to short anything  
And now it states the following error:

Code:

This device cannot find enough free resources that it can use. (Code 12)

If you want to use this device, you will need to disable one of the other devices on this system.

Perhaps one can contemplate on this?

The diagnosis of this error code is:

Two devices have been assigned the same input/output (I/O) ports, the same interrupt, or the same Direct Memory Access channel. The assignment was made by either the basic input/output system (BIOS), the operating system, or a combination of the two.


Cheers,

You need more RAM, especially when using windows.
8 to 12 GB should do just fine

[Timmieman]

@Koelen,

Thanks for the option, tried it out with both 8GB and 16GB, though it still issues the same error.