Author

Topic: [SOLVED] Radeon 280X Toxic (mining) and nVidia GTX570 (gaming) - same rig (Read 1086 times)

member
Activity: 84
Merit: 10
I tried it two different ways. Firstly I installed the Radeon in the 16x PCI-slot. That worked fine, but I wanted to play games while it mined, so I reinstalled the nVidia card in the next slot it would fit in, which happened to be a 4x slot. This didn't work very well, because any app I used seemed to want to see the Radeon card and use it, which caused horrible input lag and frame stuttering.

So I swapped them around. The nVidia is back in the 16x slot now (slot 1 if you like) and the Radeon is in the 4x slot (slot 3 on my mobo - couldn't fit it into slot 2 because both cards are huge).

Both sets of drivers installed - no issues with this, no conflicts, nothing.
Monitor is plugged into the nVidia card
The Radeon card does NOT have a dummy plug (not needed)

Works great Smiley

Next step is to buy a cheap mobo+cpu+ram setup and get the Radeon out and on it's own so that I can leave it mine and ignore it.

RIt.
full member
Activity: 166
Merit: 100
Can you tell me how you got both cards running on the same box? I didn't think you could mix them like that. I only have a single monitor and maybe that is the problem. I assume you install both sets of drivers and thats it?
member
Activity: 84
Merit: 10
SOLVED!!! Smiley

I got this solution from the Litecoin boards, and I believe credit needs to go to TheMage.

Screen REsolution -> Detect
Choose second display and force VGA output
Set second display as "Output from R9-280x"
Extend the desktop

It fixed both the monitoring issue AND the hashrate.

Maybe this will help someone in future Smiley

Rit.
member
Activity: 84
Merit: 10
That' a great suggestion - I'll try that today!

Thank you mate Smiley
sr. member
Activity: 840
Merit: 255
SportsIcon - Connect With Your Sports Heroes
change the PCI-e frequency with Asus TurboV, slowly from 95 to 105 Mhz, 1 at a time. See if this influences hashrate. If TurboV doesn't allow it, you need to set values on the BIOS.

This change may help with the HW errors:  --gpu-fan 70-90 --gpu-engine 1030 --gpu-memclock 1500
member
Activity: 84
Merit: 10
Hiya,

Thanks for your reply!

I had initially put the Radeon in the 16x slot, but what was happening is I couldn't persuade Windows and Games not to "see" it and use it. So the moment I attempted to use the machine for anything but mining, it tanked the hash rate and I had severe input and framerate lag - basically the one card was trying to do both jobs. Even though the monitor was plugged into the nVidia card, which at the time was in the 4x slot.

Here's the link to the board: http://www.asus.com/Motherboards/P8P67M_PRO/

At the moment, the GTX570 is plugged into the blue slot - the true 16x slot. The Toxic is plugged into the black slot, which is only a 4x.

It does work - gaming sees only the nVidia card, and the Toxic mines away. However, as I mentioned, I can't get the hash rate above about 590Khash (typically it's running at about 560 Khash) with the Toxic in this slot, and when I use the other card the Hardware errors start to appear.

I've tried disabling non-essentials in the BIOS, but it hasn't made any difference that I can see.

One thing I have noticed is this:

I am using the argument -d 0 in the bat file to start mining. I'd have thought that changing the cards around would have necessitated me changing this to -d 1, but the GPUs seem to appear in the same order regardless of the slots they're in. I suspect this is part of the issue tbh.

The last issue I have is the ADL thing. I've done a bit of light reading on it, and it appears I can pretty much ignore it. However, the side-effect is that I can no longer monitor the card stats in cgminer.

Any more info I can provide, just let me know!

Rit.
sr. member
Activity: 840
Merit: 255
SportsIcon - Connect With Your Sports Heroes
See what's the behaviour if you connect the GTX570 on the 4x slot and the Toxic on the 16x, regarding gaming and image stability

Go to the bios and disable all peripherals you don't need. Temporarily disable sound and even ethernet (if you have an wifi dongle). Maybe it allow you to alleviate issues with the southbridge/platform controller of the motherboard. Performance may tank on that 4x slot, because it doesn't offer 4x PCI-e directly, there's a busy chip in between the Intel CPU and the slot, just good enough for SATA, USB and other devices.

Could you post a link to your exact Motherboard model?
member
Activity: 84
Merit: 10
Just to add to the above, because it's probably related....

I also get a message on cgminer that the ADL doesn't match the number of OCL devices (or something like that). I had thought that the -d argument and the --gpu-platform would sort that out, but it hasn't.

Also, to be more specific about the setup:

nVidia GTX570 in 16x PCIe slot.
Sapphire R9-280X Toxic in 3rd PCIe slot, which is a 4x slot.

There is a second 16x slot on the board, but it's too close to the nVidia card to fit the R9 in there. There's also a x1 slot, but I don't have a riser on hand to try it.

Any ideas?

Rit.
member
Activity: 84
Merit: 10
Hi all,

Before I start, yes, I googled. Extensively. All damn day Smiley It helped me get the thing actually -working- but I have some questions here.

I am running:

- nVidia GTX570 (16x PCIe Slot)
- Sapphire Radeon R9-280X Toxic (4x PCIe Slot - slot 3 if it matters)
- 740W Corsair Gold
- i5-2500K
- 4 GB RAM (generic)
- Windows 7 64 bit
- single monitor

I have posted a previous thread about serious issues with artefacting on the 280X. It looked in the end like my only option was to RMA it, but the thing is hashing like mad, so I decided instead to keep it and hash the life out of it, then RMA it when and if it dies. So I plugged my GTX570 back in to deal with Windows/gaming.

The whole setup is working....kinda. I have basically 2 issues, which I could use help with:

1) In the PCIe 4x slot, the R9-280X has gone from approx 730Khash to about 590Khash. This doesn't seem to be dependant on anything else which happens with the machine. It maintains this regardless of what I do on the machine. The config file I am using is as follows:

Code:
cgminer.exe --scrypt -o stratum+tcp://eu.clevermining.com:3333 -u 1LaWk8EYHXXHHbuMywbmfcutMFT25apisG -p x -d 0 --gpu-platform 1 -I 13 -g 2 -w 256 --thread-concurrency 8192 --auto-fan --temp-cutoff 90 --temp-overheat 85 --temp-target 72 --gpu-fan 30-75 --gpu-engine 1050 --gpu-memclock 1500 --shaders 2048 --gpu-powertune 20 --gpu-vddc 1.100

Why does it tank so badly? It can't be PCIe bandwidth, 4x is MORE than enough.

2) Hardware errors - when I run the miner by itself, and/or just browse the web, HW errors are 0. When I start gaming on the second card (the nVidia), the hardware errors just get crazy. Example: 0 in 12 hours to 146 in 2 hours. I'm thinking this might be heat, but I'd love any advice on it.

Can anyone offer any pointers here? Thanks in advance!

Rit.
Jump to: