rev120 lowers SSE2-performance when Stream-core is installed

[GoogleCodeExporter]

What steps will reproduce the problem?
1. run 'pyrit benchmark' without and with Stream-core is installed
2.
3.

What is the expected output? What do you see instead?
test from rev118

Pyrit 0.2.3-dev (svn r118) (C) 2008, 2009 Lukas Lueg
http://pyrit.googlecode.com
This code is distributed under the GNU General Public License v3

The ESSID-blobspace seems to be empty; you should create an ESSID...

Running benchmark for at least 60 seconds...

CPU-Core (SSE2): 646.06 PMKs/s (99.71% occupancy)
CPU-Core (SSE2): 647.56 PMKs/s (96.07% occupancy)
CPU-Core (SSE2): 648.46 PMKs/s (99.77% occupancy)
CPU-Core (SSE2): 643.58 PMKs/s (99.78% occupancy)

Benchmark done. 2585.66 PMKs/s total.

test from rev119

Pyrit 0.2.3-dev (svn r119) (C) 2008, 2009 Lukas Lueg
http://pyrit.googlecode.com
This code is distributed under the GNU General Public License v3

The ESSID-blobspace seems to be empty; you should create an ESSID...

Running benchmark for at least 60 seconds...

ATI-Stream device 1: 9665.21 PMKs/s (92.57% occupancy)
CPU-Core (SSE2): 634.58 PMKs/s (99.84% occupancy)
CPU-Core (SSE2): 634.37 PMKs/s (98.94% occupancy)
CPU-Core (SSE2): 633.92 PMKs/s (99.70% occupancy)

Benchmark done. 11568.07 PMKs/s total.

test from rev120

Pyrit 0.2.3-dev (svn r120) (C) 2008, 2009 Lukas Lueg
http://pyrit.googlecode.com
This code is distributed under the GNU General Public License v3

The ESSID-blobspace seems to be empty; you should create an ESSID...

Running benchmark for at least 60 seconds...

ATI-Stream device 1: 9411.93 PMKs/s (93.82% occupancy)
CPU-Core (SSE2): 372.91 PMKs/s (98.79% occupancy)
CPU-Core (SSE2): 376.89 PMKs/s (99.70% occupancy)
CPU-Core (SSE2): 379.54 PMKs/s (97.34% occupancy)

Benchmark done. 10541.26 PMKs/s total.

rev120 without Stream-core

Pyrit 0.2.3-dev (svn r120) (C) 2008, 2009 Lukas Lueg
http://pyrit.googlecode.com
This code is distributed under the GNU General Public License v3

The ESSID-blobspace seems to be empty; you should create an ESSID...

Running benchmark for at least 60 seconds...

CPU-Core (SSE2): 625.91 PMKs/s (99.69% occupancy)
CPU-Core (SSE2): 591.09 PMKs/s (99.73% occupancy)
CPU-Core (SSE2): 613.17 PMKs/s (97.50% occupancy)
CPU-Core (SSE2): 622.54 PMKs/s (98.47% occupancy)

Benchmark done. 2452.71 PMKs/s total.

What version of the product are you using? On what operating system?

pyrit rev120 on Ubuntu 9.04-64bit Phenom II 920, ATI 4850

Please provide any additional information below.

Original issue reported on code.google.com by datapirates@googlemail.com on 13 Jul 2009 at 11:08

[GoogleCodeExporter]

I've a ATI 4850 now and will look into this next week..

Original comment by lukas.l...@gmail.com on 18 Jul 2009 at 3:31

[GoogleCodeExporter]

I can't confirm the problem on my box. Is it still present with r130 ?

Original comment by lukas.l...@gmail.com on 22 Jul 2009 at 8:46

[GoogleCodeExporter]

Gave this a test on my laptop with r130 and got as below

Pyrit 0.2.3-dev (svn r130) (C) 2008, 2009 Lukas Lueg http://pyrit.googlecode.com
This code is distributed under the GNU General Public License v3

Running benchmark for at least 60 seconds...

#1: 'CPU-Core (SSE2)': 716.4 PMKs/s (Occ. 99.9%; RTT 2.9)
#2: 'CPU-Core (SSE2)': 702.7 PMKs/s (Occ. 96.9%; RTT 2.8)

Benchmark done. 1419 PMKs/s total.

Pyrit 0.2.3-dev (svn r130) (C) 2008, 2009 Lukas Lueg http://pyrit.googlecode.com
This code is distributed under the GNU General Public License v3

Running benchmark for at least 60 seconds...

#1: 'ATI-Stream device 1': 779.2 PMKs/s (Occ. 99.3%; RTT 2.9)
#2: 'CPU-Core (SSE2)': 602.7 PMKs/s (Occ. 100.0%; RTT 2.8)

Benchmark done. 1381 PMKs/s total.

Pretty bad ati card, its a radeon 3650 HD mobility and think its an intel T9400
running on ubuntu 9.04 64bit. On r120 the results were worse than r130, with it
running about 500PMKs/s

Original comment by adam.k...@gmail.com on 23 Jul 2009 at 6:28

[GoogleCodeExporter]

The ATI-Core sucks away some of the CPU-time - that's why you get only 600 
PMKs/s
instead of 700 PMKs/s from the CPU while the ATI-Core is active. Usually the 
GPU is
multiple times faster than the CPU so the burden put upon the CPU is no 
problem. 

I consider this closed for now, please report back if you have further problems 
:-)

Original comment by lukas.l...@gmail.com on 23 Jul 2009 at 6:48

• Changed state: Invalid