Pipe is slow

[shodan]

In my setup -w 3

Speed.GPU.#1...: 153.8 kH/s
Speed.GPU.#2...: 0 H/s
Speed.GPU.#3...: 0 H/s
Speed.GPU.#*...: 153.8 kH/s

UPD:
i think i found! "CPU hashcat" is slow.
if i send all password to file; file pipe to oclhashcat via stdin, speed oclHashcat is normal.

..\hashcat-0.49\hashcat-cliAVX2.exe -r bf.rule "D:\wordlist\concentrate" --stdout > manual

type manual | cudaHashcat64.exe -a 0 -m 2500 -w 2 --session=s6 --gpu-temp-abort=90 --gpu-temp-retain=90 -o "D:\keys\hash_found.txt" --outfile-format=3 D:\keys\step6\out80.hccap

Session.Name...: s6
Status.........: Running
Input.Mode.....: Pipe
Hash.Target....: ********* (********* <-> *********)
Hash.Type......: WPA/WPA2
Time.Started...: Sat Mar 14 22:18:08 2015 (20 secs)
Speed.GPU.#1...: 144.1 kH/s
Speed.GPU.#2...: 121.6 kH/s
Speed.GPU.#3...: 126.2 kH/s
Speed.GPU.#*...: 391.9 kH/s
Recovered......: 0/1 (0.00%) Digests, 0/1 (0.00%) Salts
Progress.......: 8671920
Skipped........: 0
Rejected.......: 874680
HWMon.GPU.#1...: 99% Util, 66c Temp, N/A Fan
HWMon.GPU.#2...: 95% Util, 69c Temp, N/A Fan
HWMon.GPU.#3...: 95% Util, 68c Temp, N/A Fan

Rule file

$0$0$0$0
.....
$9$9$9$9

Measured "CPU hashcat" with chronometer: ~260 000 words / sec goes out via stdout. (cli64, cliXOP,cliAVX speed is similar)

How i can accelerate "CPU hashcat" for fastest --stdout generated?

[undeath]

why not let oclhashcat apply the rules and just cat the wordlist?

oclhashcat -r rules.file < wordlist

[shodan]

cudaHashcat64.exe -r bf.rule -a 0 -m 2500 D:\in.hccap <"D:\wordlist\concentrate.txt"

Speed.GPU.#1...: 151.3 kH/s
Speed.GPU.#2...: 0 H/s
Speed.GPU.#3...: 0 H/s
Speed.GPU.#*...: 151.3 kH/s

[atom]

Sometimes, with a lot of GPU power, it takes a bit of manual support. Please optimize your attack. Here's an example:

root@et:~/hashcat-0.50# head -10000 /root/dict/untouched/rockyou.txt > rockyou10k.txt
root@et:~/hashcat-0.50# /root/maskprocessor/mp64.bin '$?d $?d $?d $?d' > hybrid.rule
root@et:~/hashcat-0.50# wc -l rockyou10k.txt
10000 rockyou10k.txt
root@et:~/hashcat-0.50# wc -l hybrid.rule
10000 hybrid.rule
root@et:~/hashcat-0.50# time ./hashcat-cli64.bin --stdout rockyou10k.txt -r hybrid.rule | wc -l
100000000

real 1m39.119s
user 0m18.515s
sys 2m45.179s

That's compareable to what you did. It's just 1 M/s on my 4770k. A better approach would be this:

root@et:~/hashcat-utils-1.2# head -10000 /root/dict/untouched/rockyou.txt > rockyou10k.txt
root@et:~/hashcat-utils-1.2# /root/maskprocessor/mp64.bin ?d?d?d?d > right.txt
root@et:~/hashcat-utils-1.2# wc -l rockyou10k.txt
10000 rockyou10k.txt
root@et:~/hashcat-utils-1.2# wc -l right.txt
10000 right.txt
root@et:~/hashcat-utils-1.2# time ./combinator.bin rockyou10k.txt right.txt | wc -l
100000000

real 0m1.906s
user 0m1.785s
sys 0m0.382s

That's 52 M/s which should be sufficient to feed your GPUs.

[shodan]

Atom, you are best! Thanks!

..\hashcat-utils-1.1\combinator.exe "D:\pass.txt" right.txt | cudaHashcat64.exe -a 0 -m 2500 -w 2 D:\in.hccap

Speed.GPU.#1...: 154.3 kH/s
Speed.GPU.#2...: 126.2 kH/s
Speed.GPU.#3...: 134.1 kH/s
Speed.GPU.#*...: 414.6 kH/s