Possible bug - Utilization reporting when specifying devices

[MackWage]

Greetings all! My rig has 3 graphics cards in it and due to cooling issues, I changed 2 of them to have hybrid cooling. So I start rerunning some cracking tests specifying to use only those two cards and not the one with stock cooler. 

I noticed when I specify devices, the utilization shows as 100% on one but 0% on the other. Below is a censored version of the command I used followed by Hashcat output. Is this a bug or is it expected behavior and I am maybe missing something? 

Also - i noted my cracking performance is about 1/3 that of when running the benchmark. Is this an example of when I need to generate work? I thought it would not be the case as I am doing pure bruteforce and the need to generate work comes with performance loss when waiting on the wordlist to copy across the pci-x bus. If I am incorrect, happy to be pointed in the right direction.  

Code:

hashcat64.bin -m 100 -w 3 -d 1,2 --remove hashes5.txt -o cracked.txt -a 3 ?a?a?a?a?a?a

Code:

Session.Name...: hashcat
Status.........: Running
Input.Mode.....: Mask (?a?a?a?a?a?a) [6]
Hash.Target....: File (hashes5.txt)
Hash.Type......: SHA1
Time.Started...: Mon Jul 25 13:44:58 2016 (12 secs)
Time.Estimated.: Mon Jul 25 13:49:41 2016 (3 mins, 28 secs)
Speed.Dev.#1...:  1663.0 MH/s (349.34ms)
Speed.Dev.#2...:  1663.4 MH/s (353.87ms)
Speed.Dev.#*...:  3326.4 MH/s
Recovered......: 0/31700081 (0.00%) Digests, 0/1 (0.00%) Salts
Recovered/Time.: CUR:N/A,N/A,N/A AVG:0.00,0.00,0.00 (Min,Hour,Day)
Progress.......: 42141696000/735091890625 (5.73%)
Rejected.......: 0/42141696000 (0.00%)
Restore.Point..: 4081920/81450625 (5.01%)
HWMon.Dev.#1...: Temp: 37c Fan: 27% Util:100% Core:1995Mhz Mem:3802Mhz Lanes:16
HWMon.Dev.#2...: Temp: 39c Fan: 30% Util:  0% Core:1500Mhz Mem:4006Mhz Lanes:16

[MackWage]

Perhaps it goes without saying but just for clarity, when I run the same command without specifying devices, all 3 devices will show 100% utilization.

[atom]

This looks like a typical xorg.conf issue. Note that OpenCL devices and NVML devices (NVML is used to query utilization etc). do not share any unique identifier, so there's no way for hashcat to match those devices. IOW, you propably have a different ordering. Try to exchange the pci-e bus in the xorg.conf for the different devices.

The performance of the benchmark should be exactly the same performance you have with a -a 3 attack with -w 3 and a single hash. Also read this: https://hashcat.net/wiki/doku.php?id=fre...full_speed