Apparently ASUS Crosshair V (990FX) comes with AMD's version of VT-D hardware visualization. They call it IOMMU. Would be interesting to see someone try it.
All FX series chips have it, none of the rest do. That includes 790FX, 890FX, and 990FX. Its really a downscaled version of their server chipset, not a member of their desktop series. IOMMU is the literal description of what that is, both Intel and AMD (plus Sun's and IBM's) are called that. All it does is allow two way remapping of hardware interrupts and device-side device memory map remapping (such as unified remaps of device->system ram and cpu->device ram). If you're going to do any virtualization of suitably complex devices, you need IOMMU. This includes, but isn't limited to, GPUs, higher end NICs (10gbit and up typically), high end RAID controllers, etc. |
|
|
|
Great thanks, its working now, but is there an more easy way to get it working? i am starting cmd every time and go with "cd c:\..." to the folder and have to type it all.
ohh and the hash rate increase by about 10 Mhash compared to the gui miner.
Write a .bat file with notepad and put it in the same dir with DiabloMiner-Windows.exe, type out the entire command in it. Then just double click on the .bat file and it should work if you did it right. |
|
|
|
Hello, im new to DiabloMiner and want to know how i can tell the miner to run gpu #0 with low aggression (gaming pc) and gpu #1 with high aggression?
right now it looks like this. (im using p2pool)
C:\DiabloMiner>DiabloMiner-Windows.exe -o localhost -r 9332
Run two, use -D 1 -f 120 on one, -D 2 -f 1 on the other. Keep upping the first one until your game stops lagging, you'd probably be better off just shutting the miner off when gaming. -f is best as a multiple or divisor of 60. (1, etc etc, 30, 60, 120, 180, 240, etc etc) |
|
|
|
Do you guys know what is the limit of GPUs that a typical mobo(bios) can initialize? The fastra2 had 13 GPUs and they needed a custom bios to boot.
I have tried to look into this virtualization business and damn there's a little of good howtos available. The esxi could have been an easy solution but 5970s won't play nice with that.
This depends on the motherboard... For instance, my GA-990FXA-UD7 wouldn't even boot with 8 GPUs until I started disabling some devices in the BIOS (USB3, eSata, etc...). After a couple of tries, everything works and this box has been running stable for several months now. Yeah, you need to disable as many devices filling up address space blocks as possible. It might be able to get more than 4 on 32 bit Linux just by doing that, and also by putting less memory in it (but its hard to get less than 1gb DDR3 sticks that aren't crap or volted wrong as the DDR3 manufacturing process was targeting chip sizes for 1gb minimum sticks). |
|
|
|
Sorry, i cant see the power Draw on any tool, but i got an kill-a-watt at the wall, i could turn of the gpu using its feature zero-power core, and then see the difference, hope that this will be enough good enough.
EDIT:
450 Watt - Total System Power Draw with 7870 Mining at 430 mhash/s
273 Watt - Total System Power Draw with 7870 in ZeroCore Power sleep
that would make 177 Watts for the 7870 alone!
Sounds about right, AMD quotes 175 for 7870. |
|
|
|
BAMT is worthless shit if it isn't a 64 bit Debian build already. You can't even get more than 4 cards working in 32 bit Linux, sometimes only 3.
I *think* (not sure though) that the current BAMT is 64-bit, but I had been trying with the previous version which wasn't. Open xterm, uname -a |
|
|
|
Do you guys know what is the limit of GPUs that a typical mobo(bios) can initialize? The fastra2 had 13 GPUs and they needed a custom bios to boot.
I have tried to look into this virtualization business and damn there's a little of good howtos available. The esxi could have been an easy solution but 5970s won't play nice with that.
Depends on the CPU arch's address space (hint: 64 bit processors, internally, do not have 64 bits of address space, its usually something like 40 or 48 bit). It also depends on how the BIOS allocates space for the PCI-E GART window, some blindly select huge address spaces no matter what the device will actually use leaving you enough room for 8-10 devices. My BIOS has an option that I can enable that allows for the use of a 64-bit address table to map devices with, instead of the usual 32-bit table. However, it is an option that I have never seen on any other device, and I do not yet know whether it will work correctly, or even what operating systems support it. When I turned it on and attempted to boot BAMT, the screen filled with corruption and I didn't get too far. Similarly, Windows refused to install, citing some unknown hardware capabilities. I think a 64-bit flavor of linux should work though. BAMT is worthless shit if it isn't a 64 bit Debian build already. You can't even get more than 4 cards working in 32 bit Linux, sometimes only 3. |
|
|
|
Does anyone have any hash tags they use for mining that improve performance with this card?
#firstworldproblems |
|
|
|
Hello, i just got my 7870 Sapphire 1 Ghz Eddition today and overclocked it to 1.1 Ghz (i think it could easy do 1.2 ghz or even 1.3 ghz)
im using 12.3 drivers and 2.6 SDK (for some reason i have no 100% cpu bug, (i use an 5850 and the new 7870).
and right now with " -v -w256 -f0 " in guiminer i get 403 Mhash, and my p2p is showing the increase in speed too!
anyone else can confirm this performance?
Why are you using Guiminer? Use DiabloMiner or cgminer instead. DiabloMiner has a kernel optimized for GCN. Anyhow, by my calculations, a 7870/1ghz should get about 375, at 1.1ghz about 412. |
|
|
|
Do you guys know what is the limit of GPUs that a typical mobo(bios) can initialize? The fastra2 had 13 GPUs and they needed a custom bios to boot.
I have tried to look into this virtualization business and damn there's a little of good howtos available. The esxi could have been an easy solution but 5970s won't play nice with that.
Depends on the CPU arch's address space (hint: 64 bit processors, internally, do not have 64 bits of address space, its usually something like 40 or 48 bit). It also depends on how the BIOS allocates space for the PCI-E GART window, some blindly select huge address spaces no matter what the device will actually use leaving you enough room for 8-10 devices. |
|
|
|
|
Yeah, but when will GLBSE get driven underground because its not Tor proxied and people know who the owner is
|
|
|
|
You might be facing different issues all at once.
First, DiabloMiner does not have a version number as I don't officially release it. Just follow it via git, github, or subscribe to this thread.
Second, remove -f 240. It'll make your desktop lag, but you don't need nearly as much -f to stop that, AMD drivers are much better built than Nvidia's. -f 30 or -f 60 is what most people use on AMD hardware depending on taste.
Third, 290-310 seems to be about right for a 6870 at stock clocks. And try -v 2 over -v 2,1, sometimes its faster, sometimes its not for vliw5 hardware.
Fourth, makes sure you're running Catalyst 12.3 WITHOUT SDK 2.6, 12.3 ships with the OpenCL runtime and its newer than whats in SDK 2.6.
Thanks for the tips. I tried -f 60 with no noticeable lag, and a few extra Mhash/s. Now looking at more like 295-315! -v2,1 was faster for me, but the explanation does help. Now, about the SDK thing. I just cleared my computer of all graphics drivers from Nvidia and then installed Catalyst 12.3... didn't do anything else. Should I have? You should have done nothing else. Unless you are installing Catalyst before 12.1, do _NOT_ install the SDK. |
|
|
|
is it super duper faster yet? Sure, its pretty close to as fast as its going to get. |
|
|
|
Either that, or cut the backplanes and only leave the very top screw mount and the overhang (so it balances on the beam).
This is the part that I want to get fabricated, so I can keep the original brackets for resale. Unless someone wants to sell me a bunch of brackets? They could actually be from any reference card model, as long as it has the 2 screws at the top, since the rest would be getting chopped off. Would this single wide bracket work? EDIT: More info here. Available here for ~$2 each. Well, the top of the card/bracket has 2 screws, and it would be ideal to attach both, since I don't want the card flopping around on one screw. Also I would have to drill some holes in those - is there a double-slot version of the same thing that i could chop the top off of? You know, you wouldnt really need the double slot version. You could trim these, and then use the backplane screw that you normally use when plugging cards into real machines (which exist to do exactly that: keep it from flopping around). |
|
|
|
I've been using a few months old version which would keep having kernel queue errors on my Nvidia GPU. But I upgraded to a 6870 and the kernel queue errors went away... or so it seemed. I got one last night. Very frustrating because I'm trying to mine and I don't want my freaking miner to stop working.
So I upgraded to the most recent build but it dropped by hash rate by 100Mhash/s. WTF? I fiddled with other configurations but the highest I could get was a drop of 40Mhash/s instead. So I reverted to the build I had.
Is there some reason why I'm getting these kernel queue errors, and why upgrading the build drops my hash rate so much? I was told kernel queue errors are driver bugs, but I'm wondering if that's true when I keep changing my drivers to newer/older versions looking for a fix and nothing works.
I'd give the version of DM I'm using but I actually can't find the version number in the docs, seriously WTF?
Anyway my config is -w 128 -v 2,1 -f 240 and it gives about 290-310Mhash/s on this GPU, no overclock, if that helps.
You might be facing different issues all at once. First, DiabloMiner does not have a version number as I don't officially release it. Just follow it via git, github, or subscribe to this thread. Second, remove -f 240. It'll make your desktop lag, but you don't need nearly as much -f to stop that, AMD drivers are much better built than Nvidia's. -f 30 or -f 60 is what most people use on AMD hardware depending on taste. Third, 290-310 seems to be about right for a 6870 at stock clocks. And try -v 2 over -v 2,1, sometimes its faster, sometimes its not for vliw5 hardware. Fourth, makes sure you're running Catalyst 12.3 WITHOUT SDK 2.6, 12.3 ships with the OpenCL runtime and its newer than whats in SDK 2.6. |
|
|
|
Why not just use enterprise PSU modules for redundant PSUs? They typically output only 12v and large amounts of it. You can get upwards of 4000w in a redundant PSU across four modules. |
|
|
|
You are going to want to remove those back plates on the video cards...you'll impede so much airflow with them still attached...
There's tons of room, and they prevent damage to the back of the board due to clumsy mishandling, which I am prone to. Removing the plastic gave me a huge amount of space, and the only thing I really need to do is mod/remove/replace the front brackets with higher flow pieces. Sorry, I meant the slot mounting bracket...right now they're like a wall to your airflow...I removed them on all of my cards and my temps dropped ~5C...back plates are fine. Not entirely they aren't. Hes building a closed air channel that is as large as the cards that go in it. The only way for air to escape, if he properly built it, is through the vents in the card backplates. The worst he'll get here is a pocket of hot air right behind the heatsink, but the heatsink will still perform correctly. The airflow dynamic is no different then how the card would have operated normally. However, I suggest he removes the backplate and use zip ties on the now empty screw holes to secure the card semi-firmly. Either that, or cut the backplanes and only leave the very top screw mount and the overhang (so it balances on the beam). |
|
|
|
|
Huh, now that I see that coming together thats almost exactly like the rig I always wanted to build but never could afford.
|
|
|
|
|
Show Posts
