Slow hashing with R9 290
#1
Command: clHashcat64.exe -m 2500 -a 3 capture.hccap ?d?d?d?d?d?d?d?d?d?d

Code:
Session.Name...: oclHashcat
Status.........: Running
Input.Mode.....: Mask (84?d?d?d?d?d?d?d?d) [10]
Hash.Target....: <deleted>
Hash.Type......: WPA/WPA2
Time.Started...: Mon Mar 14 19:52:53 2016 (12 secs)
Time.Estimated.: Mon Mar 14 20:14:15 2016 (21 mins, 9 secs)
Speed.GPU.#1...:    78965 H/s
Recovered......: 0/1 (0.00%) Digests, 0/1 (0.00%) Salts
Progress.......: 983040/100000000 (0.98%)
Rejected.......: 0/983040 (0.00%)
Restore.Point..: 983040/100000000 (0.98%)
HWMon.GPU.#1...: 100% Util, 81c Temp, 97% Fan

I know it's a short run and the GPU temperature/fan might look a little off, but that's because the GPU was already warm from a previous hashing session. After letting it settle, the fan speed comes down and the temperatures drop to an acceptable 80 C. AMD control panel shows me that the clock speed is the stock 947 MHz, so no throttling going on there. Is it me, or does this say I'm doing barely 79 kH/s?
#2
1. you mentioned the mask ?d?d?d?d?d?d?d?d?d?d but the status screen of oclHashcat clearly and undoubtedly shows us that you are lying about it. What I mean is, that it should be very clear to all of us that ?d?d?d?d?d?d?d?d?d?d is different from 84?d?d?d?d?d?d?d?d :
Code:
Input.Mode.....: Mask (84?d?d?d?d?d?d?d?d) [10]
The problem here is that, as mentioned in dozens if not in hundreds of other forum threads mentioning exactly the same "problem", the left-hand side is very small. Since the beginning of your mask, i.e. "84...", is static, oclHashcat cannot reach full accereration (in general. This of course is particularly noticeable with so-called fast hashes). oclHashcat will always update/change the left side most frequently, and because it is hard-coded/fixed, it won't vary a lot. As said, the problem is that oclHashcat can't reach full accereration if the mask (or left-hand side of the mask) is constant (or part of it).
The before mentioned forum threads that deal with exactly that problem also mention several workarounds and tricks (which I won't repeat here, use the search function). In some cases, even using a pipe makes sense and could be faster (this is especially true if the hashing algorithm is a so-called slow hash, as is -m 2500 = WPA/WPA2)
Also note: there is sometimes a tradeoff between "the smallest mask as possible" (with the minimum of keyspace) and "the mask with best acceleration". This depends from case to case. For instance, having a "smaller mask" (less keyspace or hard-coded left hand side) could result in less speed (particular noticable with fast hashes) but at the very end it could still be faster compared to a run with a much larger mask (where the keyspace is larger).

2. R9 290 is not the same as R9 290x

3. there is a special benchmark mode which you can use to test the speed of your GPU(s), e.g. -m 2500 --benchmark

4. you forgot to specify a workload profile, i.e. -w 3 for "full speed". It doesn't make much sense to speak about "Slow hashing" without even specifying the -w 3 command line option
#3
(03-15-2016, 12:12 PM)philsmd Wrote: 1. you mentioned the mask ?d?d?d?d?d?d?d?d?d?d but the status screen of oclHashcat clearly and undoubtedly shows us that you are lying about it. What I mean is, that it should be very clear to all of us that ?d?d?d?d?d?d?d?d?d?d is different from 84?d?d?d?d?d?d?d?d :
Code:
Input.Mode.....: Mask (84?d?d?d?d?d?d?d?d) [10]
The problem here is that, as mentioned in dozens if not in hundreds of other forum threads mentioning exactly the same "problem", the left-hand side is very small. Since the beginning of your mask, i.e. "84...", is static, oclHashcat cannot reach full accereration (in general. This of course is particularly noticeable with so-called fast hashes). oclHashcat will always update/change the left side most frequently, and because it is hard-coded/fixed, it won't vary a lot. As said, the problem is that oclHashcat can't reach full accereration if the mask (or left-hand side of the mask) is constant (or part of it).
The before mentioned forum threads that deal with exactly that problem also mention several workarounds and tricks (which I won't repeat here, use the search function). In some cases, even using a pipe makes sense and could be faster (this is especially true if the hashing algorithm is a so-called slow hash, as is -m 2500 = WPA/WPA2)
Also note: there is sometimes a tradeoff between "the smallest mask as possible" (with the minimum of keyspace) and "the mask with best acceleration". This depends from case to case. For instance, having a "smaller mask" (less keyspace or hard-coded left hand side) could result in less speed (particular noticable with fast hashes) but at the very end it could still be faster compared to a run with a much larger mask (where the keyspace is larger).

2. R9 290 is not the same as R9 290x

3. there is a special benchmark mode which you can use to test the speed of your GPU(s), e.g. -m 2500 --benchmark

4. you forgot to specify a workload profile, i.e. -w 3 for "full speed". It doesn't make much sense to speak about "Slow hashing" without even specifying the -w 3 command line option

Thanks for your detailed explanation; it looks like -w 3 was the switch I had missed. It performs exactly the same (at least to my untrained eye) no matter which mask I use. I'm not looking to eke out every single bit of performance from my GPU - I was just wondering why there's such a huge difference between the numbers I was seeing vs. what I was expecting. Thanks again!
#4
no problem, you are welcome.

... as said, the speed difference with small masks or static left parts of the mask is more or less only noticeable with fast hashes, but the important thing to remember is that this could make a huge difference and even only slightly different masks are different and could result in (huge) "speed drops".

The other important thing you can learn here is that using the forum search function (or your favorite search engine) you will find exactly those forum threads that mention e.g. "don't forget to use -w 3 before you claim it is not as fast as expected"... many questions were already answered (several times), there even is a FAQ, https://hashcat.net/wiki/frequently_asked_questions, which is very worth reading).
cheers
#5
I did do a search on the forum, but I guess I wasn't using the correct keywords. I had tunnel vision because I assumed it was the R9 290 that had issues, which limited my search results. I also missed the FAQ; thanks for the link!