hashcat Forum
v1.31 and v1.30 are both much slower than v1.21 - Printable Version

+- hashcat Forum (https://hashcat.net/forum)
+-- Forum: Deprecated; Previous versions (https://hashcat.net/forum/forum-29.html)
+--- Forum: Old oclHashcat Support (https://hashcat.net/forum/forum-38.html)
+--- Thread: v1.31 and v1.30 are both much slower than v1.21 (/thread-3744.html)



v1.31 and v1.30 are both much slower than v1.21 - smile - 10-16-2014

My os is windows server 2008 r2, driver is 14.9,gpu are 290 or 290x , I found that v1.31 and v1.30 are both much slower than v1.21. The same happens on driver 14.4.

Command:
oclHashcat64.exe t:\mtask\1015.notfound --force --gpu-loops 1024 --potfile-disable --status --status-timer 300 --remove --remove-timer=3600 -n 200 --session=session5100_1_0 -m 0 -o t:\task\md5-0.out ?1?1?1?1?1?1?1?1 -a 3 -1 ?l?d


v1.29
Quote:Session.Name...: session5100_1_0
Status.........: Exhausted
Input.Mode.....: Mask (?1?1?1?1?1?1?1?1) [8]
Hash.Target....: File (t:\mtask\1015.notfound)
Hash.Type......: MD5
Time.Started...: Thu Oct 16 07:53:47 2014 (3 mins, 55 secs)
Time.Estimated.: 0 secs
Speed.GPU.#1...: 66468.5 kH/s
Speed.GPU.#2...: 5829.7 MH/s
Speed.GPU.#*...: 5896.2 MH/s
Recovered......: 0/237796 (0.00%) Digests, 0/1 (0.00%) Salts
Progress.......: 2821109907456/2821109907456 (100.00%)
Skipped........: 0/2821109907456 (0.00%)
Rejected.......: 0/2821109907456 (0.00%)
HWMon.GPU.#1...: 0% Util, 72c Temp, 92% Fan
HWMon.GPU.#2...: 14% Util, 71c Temp, 92% Fan

Started: Thu Oct 16 07:53:47 2014
Stopped: Thu Oct 16 07:57:47 2014


v1.30
Quote:Session.Name...: session5100_1_0
Status.........: Running
Input.Mode.....: Mask (?1?1?1?1?1?1?1?1) [8]
Hash.Target....: File (t:\mtask\1015.notfound)
Hash.Type......: MD5
Time.Started...: Thu Oct 16 07:59:23 2014 (2 mins, 33 secs)
Time.Estimated.: Thu Oct 16 08:09:51 2014 (7 mins, 48 secs)
Speed.GPU.#1...: 2278.4 MH/s
Speed.GPU.#2...: 2276.8 MH/s
Speed.GPU.#*...: 4555.2 MH/s
Recovered......: 0/237764 (0.00%) Digests, 0/1 (0.00%) Salts
Progress.......: 696123392000/2821109907456 (24.68%)
Skipped........: 0/696123392000 (0.00%)
Rejected.......: 0/696123392000 (0.00%)
HWMon.GPU.#1...: 100% Util, 68c Temp, 89% Fan
HWMon.GPU.#2...: 100% Util, 66c Temp, 88% Fan

v1.31
Quote:Session.Name...: session5100_1_0
Status.........: Running
Input.Mode.....: Mask (?1?1?1?1?1?1?1?1) [8]
Hash.Target....: File (t:\mtask\1015.notfound)
Hash.Type......: MD5
Time.Started...: Thu Oct 16 09:02:09 2014 (3 mins, 54 secs)
Time.Estimated.: Thu Oct 16 09:12:32 2014 (6 mins, 26 secs)
Speed.GPU.#1...: 2275.8 MH/s
Speed.GPU.#2...: 2274.6 MH/s
Speed.GPU.#*...: 4550.4 MH/s
Recovered......: 0/237234 (0.00%) Digests, 0/1 (0.00%) Salts
Progress.......: 1065222144000/2821109907456 (37.76%)
Skipped........: 0/1065222144000 (0.00%)
Rejected.......: 0/1065222144000 (0.00%)
HWMon.GPU.#1...: 100% Util, 68c Temp, N/A Fan
HWMon.GPU.#2...: 14% Util, 67c Temp, N/A Fan

[s]tatus [p]ause [r]esume [b]ypass [q]uit =>



RE: v1.31 and v1.30 are both much slower than v1.21 - atom - 10-16-2014

Especially when it comes to raw MD5 you can be sure that we have a close eye on it's performance. Note that the default workload settings have been changed. If you want to do a good comparison, remove --force and make sure to use the correct driver and use -b -m 0 instead of doing a run. The code used in multihash is the same.

There's was no speed loss between 1.21 and 1.30 as you can see here: https://hashcat.net/misc/p130_img/changes_v130.png


RE: v1.31 and v1.30 are both much slower than v1.21 - smile - 10-16-2014

(10-16-2014, 10:55 AM)atom Wrote: Especially when it comes to raw MD5 you can be sure that we have a close eye on it's performance. Note that the default workload settings have been changed. If you want to do a good comparison, remove --force and make sure to use the correct driver and use -b -m 0 instead of doing a run. The code used in multihash is the same.

There's was no speed loss between 1.21 and 1.30 as you can see here: https://hashcat.net/misc/p130_img/changes_v130.png


Yes, when I run "-b -m 0", the speed is the same.But ,When I run oclexample0.cmd or crack other hash lists, the speed of v1.30 is much slower. I found this when v1.30 released on driver 14.6b. I wait for the new version but the problem exists too. I test this on 4 computers.

Computer1---Windows server 2008 r2,290*2,driver 14.9
Quote:T:\run\oclHashcat-old>oclHashcat64.exe t:\mtask\1015.notfound ?1?1?1?1?1?1?1?
1 -a 3 -1 ?l?d --force
oclHashcat v1.21 starting...

Device #1: Hawaii, 3072MB, 947Mhz, 40MCU
Device #2: Hawaii, 3072MB, 947Mhz, 40MCU

Hashes: 234344 hashes; 234344 unique digests, 1 unique salts
Bitmaps: 21 bits, 2097152 entries, 0x001fffff mask, 8388608 bytes
Applicable Optimizers:
* Zero-Byte
* Precompute-Init
* Precompute-Merkle-Demgard
* Meet-In-The-Middle
* Early-Skip
* Not-Salted
* Not-Iterated
* Single-Salt
* Brute-Force
* Scalar-Mode
* Raw-Hash
Watchdog: Temperature abort trigger set to 97c
Watchdog: Temperature retain trigger set to 95c
Device #1: Kernel ./kernels/4098/m0000_a3.Hawaii_1573.4_1573.4 (VM).kernel (1694
52 bytes)
Device #1: Kernel ./kernels/4098/markov_le_v1.Hawaii_1573.4_1573.4 (VM).kernel (
93720 bytes)
Device #1: Kernel ./kernels/4098/bzero.Hawaii_1573.4_1573.4 (VM).kernel (30484 b
ytes)
Device #2: Kernel ./kernels/4098/m0000_a3.Hawaii_1573.4_1573.4 (VM).kernel (1694
52 bytes)
Device #2: Kernel ./kernels/4098/markov_le_v1.Hawaii_1573.4_1573.4 (VM).kernel (
93720 bytes)
Device #2: Kernel ./kernels/4098/bzero.Hawaii_1573.4_1573.4 (VM).kernel (30484 b
ytes)

[s]tatus [p]ause [r]esume [b]ypass [q]uit =>


Session.Name...: oclHashcat
Status.........: Aborted
Input.Mode.....: Mask (?1?1?1?1?1?1?1?1) [8]
Hash.Target....: File (t:\mtask\1015.notfound)
Hash.Type......: MD5
Time.Started...: Thu Oct 16 19:24:41 2014 (28 secs)
Time.Estimated.: Thu Oct 16 19:28:59 2014 (3 mins, 47 secs)
Speed.GPU.#1...: 5657.2 MH/s
Speed.GPU.#2...: 5592.6 MH/s
Speed.GPU.#*...: 11249.8 MH/s
Recovered......: 0/234344 (0.00%) Digests, 0/1 (0.00%) Salts
Progress.......: 309539635200/2821109907456 (10.97%)
Skipped........: 0/309539635200 (0.00%)
Rejected.......: 0/309539635200 (0.00%)
HWMon.GPU.#1...: 0% Util, 64c Temp, 87% Fan
HWMon.GPU.#2...: 2% Util, 63c Temp, 86% Fan

Started: Thu Oct 16 19:24:41 2014
Stopped: Thu Oct 16 19:25:12 2014

T:\run\oclHashcat-old>cd ..\oclHashcat

T:\run\oclHashcat>oclHashcat64.exe t:\mtask\1015.notfound ?1?1?1?1?1?1?1?1 -
a 3 -1 ?l?d
oclHashcat v1.31 starting...

Device #1: Hawaii, 3072MB, 947Mhz, 40MCU
Device #2: Hawaii, 3072MB, 947Mhz, 40MCU

Hashes: 234344 hashes; 234344 unique digests, 1 unique salts
Bitmaps: 21 bits, 2097152 entries, 0x001fffff mask, 8388608 bytes
Applicable Optimizers:
* Zero-Byte
* Precompute-Init
* Precompute-Merkle-Demgard
* Meet-In-The-Middle
* Early-Skip
* Not-Salted
* Not-Iterated
* Single-Salt
* Brute-Force
* Scalar-Mode
* Raw-Hash
Watchdog: Temperature abort trigger set to 97c
Watchdog: Temperature retain trigger set to 95c
Device #1: Kernel ./kernels/4098/m00000_a3.Hawaii_1573.4_1573.4 (VM).kernel (171
960 bytes)
Device #1: Kernel ./kernels/4098/markov_le_v1.Hawaii_1573.4_1573.4 (VM).kernel (
92400 bytes)
Device #1: Kernel ./kernels/4098/bzero.Hawaii_1573.4_1573.4 (VM).kernel (30492 b
ytes)
Device #2: Kernel ./kernels/4098/m00000_a3.Hawaii_1573.4_1573.4 (VM).kernel (171
960 bytes)
Device #2: Kernel ./kernels/4098/markov_le_v1.Hawaii_1573.4_1573.4 (VM).kernel (
92400 bytes)
Device #2: Kernel ./kernels/4098/bzero.Hawaii_1573.4_1573.4 (VM).kernel (30492 b
ytes)

[s]tatus [p]ause [r]esume [b]ypass [q]uit =>


Session.Name...: oclHashcat
Status.........: Aborted
Input.Mode.....: Mask (?1?1?1?1?1?1?1?1) [8]
Hash.Target....: File (t:\mtask\1015.notfound)
Hash.Type......: MD5
Time.Started...: Thu Oct 16 19:25:28 2014 (29 secs)
Time.Estimated.: Thu Oct 16 19:36:23 2014 (10 mins, 23 secs)
Speed.GPU.#1...: 2192.1 MH/s
Speed.GPU.#2...: 2193.1 MH/s
Speed.GPU.#*...: 4385.2 MH/s
Recovered......: 0/234344 (0.00%) Digests, 0/1 (0.00%) Salts
Progress.......: 125745233920/2821109907456 (4.46%)
Skipped........: 0/125745233920 (0.00%)
Rejected.......: 0/125745233920 (0.00%)
HWMon.GPU.#1...: 0% Util, 65c Temp, N/A Fan
HWMon.GPU.#2...: 14% Util, 63c Temp, N/A Fan

Started: Thu Oct 16 19:25:28 2014
Stopped: Thu Oct 16 19:26:00 2014

T:\run\oclHashcat>


Computer2----Windows server 2008 r2, hd7970*1,driver 14.9
Quote:T:\t1\run\oclHashcat-old>oclHashcat64.exe t:\mtask\1015.notfound ?1?1?1?1?1?1?1?1 -a 3 -1 ?l?
d --force
oclHashcat v1.21 starting...

Device #1: Tahiti, 3072MB, 925Mhz, 32MCU

Hashes: 234344 hashes; 234344 unique digests, 1 unique salts
Bitmaps: 21 bits, 2097152 entries, 0x001fffff mask, 8388608 bytes
Applicable Optimizers:
* Zero-Byte
* Precompute-Init
* Precompute-Merkle-Demgard
* Meet-In-The-Middle
* Early-Skip
* Not-Salted
* Not-Iterated
* Single-Salt
* Brute-Force
* Scalar-Mode
* Raw-Hash
Watchdog: Temperature abort trigger set to 90c
Watchdog: Temperature retain trigger set to 80c
Device #1: Kernel ./kernels/4098/m0000_a3.Tahiti_1573.4_1573.4 (VM).kernel (169500 bytes)
Device #1: Kernel ./kernels/4098/markov_le_v1.Tahiti_1573.4_1573.4 (VM).kernel (93708 bytes)
Device #1: Kernel ./kernels/4098/bzero.Tahiti_1573.4_1573.4 (VM).kernel (30484 bytes)

[s]tatus [p]ause [r]esume [b]ypass [q]uit =>


Session.Name...: oclHashcat
Status.........: Aborted
Input.Mode.....: Mask (?1?1?1?1?1?1?1?1) [8]
Hash.Target....: File (t:\mtask\1015.notfound)
Hash.Type......: MD5
Time.Started...: Thu Oct 16 19:33:46 2014 (33 secs)
Time.Estimated.: Thu Oct 16 19:44:22 2014 (10 mins, 1 sec)
Speed.GPU.#1...: 4536.0 MH/s
Recovered......: 0/234344 (0.00%) Digests, 0/1 (0.00%) Salts
Progress.......: 148511916032/2821109907456 (5.26%)
Skipped........: 0/148511916032 (0.00%)
Rejected.......: 0/148511916032 (0.00%)
HWMon.GPU.#1...: 95% Util, 68c Temp, 55% Fan

Started: Thu Oct 16 19:33:46 2014
Stopped: Thu Oct 16 19:34:21 2014

T:\t1\run\oclHashcat-old>cd ..\oclHashcat

T:\t1\run\oclHashcat>oclHashcat64.exe t:\mtask\1015.notfound ?1?1?1?1?1?1?1?1 -a 3 -1 ?l?d

oclHashcat v1.31 starting...

Device #1: Tahiti, 3072MB, 925Mhz, 32MCU

Hashes: 234339 hashes; 234339 unique digests, 1 unique salts
Bitmaps: 21 bits, 2097152 entries, 0x001fffff mask, 8388608 bytes
Applicable Optimizers:
* Zero-Byte
* Precompute-Init
* Precompute-Merkle-Demgard
* Meet-In-The-Middle
* Early-Skip
* Not-Salted
* Not-Iterated
* Single-Salt
* Brute-Force
* Scalar-Mode
* Raw-Hash
Watchdog: Temperature abort trigger set to 90c
Watchdog: Temperature retain trigger set to 80c
Device #1: Kernel ./kernels/4098/m00000_a3.Tahiti_1573.4_1573.4 (VM).kernel not found in cache! Buil
ding may take a while...
Device #1: Kernel ./kernels/4098/m00000_a3.Tahiti_1573.4_1573.4 (VM).kernel (172008 bytes)
Device #1: Kernel ./kernels/4098/markov_le_v1.Tahiti_1573.4_1573.4 (VM).kernel not found in cache! B
uilding may take a while...
Device #1: Kernel ./kernels/4098/markov_le_v1.Tahiti_1573.4_1573.4 (VM).kernel (92388 bytes)
Device #1: Kernel ./kernels/4098/bzero.Tahiti_1573.4_1573.4 (VM).kernel (30492 bytes)

[s]tatus [p]ause [r]esume [b]ypass [q]uit =>


Session.Name...: oclHashcat
Status.........: Aborted
Input.Mode.....: Mask (?1?1?1?1?1?1?1?1) [8]
Hash.Target....: File (t:\mtask\1015.notfound)
Hash.Type......: MD5
Time.Started...: Thu Oct 16 19:34:37 2014 (30 secs)
Time.Estimated.: Thu Oct 16 19:59:04 2014 (23 mins, 52 secs)
Speed.GPU.#1...: 1978.5 MH/s
Recovered......: 0/234339 (0.00%) Digests, 0/1 (0.00%) Salts
Progress.......: 58653147136/2821109907456 (2.08%)
Skipped........: 0/58653147136 (0.00%)
Rejected.......: 0/58653147136 (0.00%)
HWMon.GPU.#1...: 98% Util, 65c Temp, 55% Fan

WARNING: Failed to restore default fan speed for gpu number: 0:
Started: Thu Oct 16 19:34:37 2014
Stopped: Thu Oct 16 19:35:12 2014

T:\t1\run\oclHashcat>



RE: v1.31 and v1.30 are both much slower than v1.21 - philsmd - 10-16-2014

1. there were a lot of changes and atom already mentioned that the default tuning was changed heavily among those versions. Also to avoid users continously reporting freezes, hangs etc
Therefore, you should make sure that the same tuning is used among the versions you test and/or have a look at the parameter:
-w, --workload-profile=NUM Enable a specific workload profile, see references below
The current workload profiles are:
1 = Reduced performance profile (low latency desktop)
2 = Default performance profile
3 = Tuned performance profile (high latency desktop)

So, you should try w/ -w 3, to use the performance profile (instead of the default performance profile, -w 2).

2. pls do not test/benchmark/compare with --force. oclHashcat even warns you that when you use --force you shouldn't ask for help nor report any problems (including performance differences etc). It is also very likely that if you use --force (and the driver is not supported by that specific oclHashcat version), that you miss some cracks etc (and other strange behavior). So pls do not post any more tests w/ --force . Thx


RE: v1.31 and v1.30 are both much slower than v1.21 - smile - 10-16-2014

(10-16-2014, 03:00 PM)philsmd Wrote: 1. there were a lot of changes and atom already mentioned that the default tuning was changed heavily among those versions. Also to avoid users continously reporting freezes, hangs etc
Therefore, you should make sure that the same tuning is used among the versions you test and/or have a look at the parameter:
-w, --workload-profile=NUM Enable a specific workload profile, see references below
The current workload profiles are:
1 = Reduced performance profile (low latency desktop)
2 = Default performance profile
3 = Tuned performance profile (high latency desktop)

So, you should try w/ -w 3, to use the performance profile (instead of the default performance profile, -w 2).

2. pls do not test/benchmark/compare with --force. oclHashcat even warns you that when you use --force you shouldn't ask for help nor report any problems (including performance differences etc). It is also very likely that if you use --force (and the driver is not supported by that specific oclHashcat version), that you miss some cracks etc (and other strange behavior). So pls do not post any more tests w/ --force . Thx


Please notice that '--force' is only used to v1.21, otherwise the program can't start on driver 14.9 !!
Please notice that even run oclexample0.cmd, it turns out the same result. Even I add -w 3, speed is much slower than v1.21 too!
I don't think there was any problem in my tests.


RE: v1.31 and v1.30 are both much slower than v1.21 - chnhnm - 11-09-2014

can you check if your gpu clock drop when your display goes off. cuz that's the problem i'm seeing on 1.31 with latest 14.9 catalyst. don't know if 1.30 had that behavior but 1.21 didn't have that problem.

if your gpu clock does drop too then either i can submit the bug report or i can do the screencaps and logs and u can turn it in all the same. i used gpu-z to verify. on my platform it drops to 276MHz from 400MHx (A6-3400M radeon 6520g gpu) even though gpu stays fully loaded @ 100%

(10-16-2014, 04:23 PM)smile Wrote:
(10-16-2014, 03:00 PM)philsmd Wrote: 1. there were a lot of changes and atom already mentioned that the default tuning was changed heavily among those versions. Also to avoid users continously reporting freezes, hangs etc
Therefore, you should make sure that the same tuning is used among the versions you test and/or have a look at the parameter:
-w, --workload-profile=NUM Enable a specific workload profile, see references below
The current workload profiles are:
1 = Reduced performance profile (low latency desktop)
2 = Default performance profile
3 = Tuned performance profile (high latency desktop)

So, you should try w/ -w 3, to use the performance profile (instead of the default performance profile, -w 2).

2. pls do not test/benchmark/compare with --force. oclHashcat even warns you that when you use --force you shouldn't ask for help nor report any problems (including performance differences etc). It is also very likely that if you use --force (and the driver is not supported by that specific oclHashcat version), that you miss some cracks etc (and other strange behavior). So pls do not post any more tests w/ --force . Thx


Please notice that '--force' is only used to v1.21, otherwise the program can't start on driver 14.9 !!
Please notice that even run oclexample0.cmd, it turns out the same result. Even I add -w 3, speed is much slower than v1.21 too!
I don't think there was any problem in my tests.



RE: v1.31 and v1.30 are both much slower than v1.21 - smile - 11-10-2014

No,the gpu clock did not drop on v1.31. I checked with msi afterburner and gpuz. Thanks.

(11-09-2014, 02:13 PM)chnhnm Wrote: can you check if your gpu clock drop when your display goes off. cuz that's the problem i'm seeing on 1.31 with latest 14.9 catalyst. don't know if 1.30 had that behavior but 1.21 didn't have that problem.

if your gpu clock does drop too then either i can submit the bug report or i can do the screencaps and logs and u can turn it in all the same. i used gpu-z to verify. on my platform it drops to 276MHz from 400MHx (A6-3400M radeon 6520g gpu) even though gpu stays fully loaded @ 100%

(10-16-2014, 04:23 PM)smile Wrote:
(10-16-2014, 03:00 PM)philsmd Wrote: 1. there were a lot of changes and atom already mentioned that the default tuning was changed heavily among those versions. Also to avoid users continously reporting freezes, hangs etc
Therefore, you should make sure that the same tuning is used among the versions you test and/or have a look at the parameter:
-w, --workload-profile=NUM Enable a specific workload profile, see references below
The current workload profiles are:
1 = Reduced performance profile (low latency desktop)
2 = Default performance profile
3 = Tuned performance profile (high latency desktop)

So, you should try w/ -w 3, to use the performance profile (instead of the default performance profile, -w 2).

2. pls do not test/benchmark/compare with --force. oclHashcat even warns you that when you use --force you shouldn't ask for help nor report any problems (including performance differences etc). It is also very likely that if you use --force (and the driver is not supported by that specific oclHashcat version), that you miss some cracks etc (and other strange behavior). So pls do not post any more tests w/ --force . Thx


Please notice that '--force' is only used to v1.21, otherwise the program can't start on driver 14.9 !!
Please notice that even run oclexample0.cmd, it turns out the same result. Even I add -w 3, speed is much slower than v1.21 too!
I don't think there was any problem in my tests.