[phpBB Debug] PHP Warning: in file [ROOT]/phpbb/session.php on line 580: sizeof(): Parameter must be an array or an object that implements Countable
[phpBB Debug] PHP Warning: in file [ROOT]/phpbb/session.php on line 636: sizeof(): Parameter must be an array or an object that implements Countable
3.14.by forum • AMD Brook version slow with hd3870?
Page 1 of 1

AMD Brook version slow with hd3870?

Posted: Wed Nov 11, 2009 11:10 pm
by Crucifix
Attached is a screenshot of my system running the 32-bit Brook version of barsWF. My hardware is the following

--Pentium D 920 (stock 2.8GHz, overclocked to 3.4GHz)
--2GB DDR-2 800 ram
--Radeon HD 3870 512MB GDDR3 (stock speeds, no overclock)

Also I run catalyst 9.8.

As you can see I'm getting roughly 7.5 million hash/second out of the GPU...given that a 3870 is roughly equal to a 9600GT which, as we all know, gets 350+m/second, shouldn't I be seeing FAR better numbers here? Why does this seem so slow?

Also noteworthy is the fact that in the regular SSE2 version of Bars i get 35 million/second because it recognizes two cores of CPU...however this brook version seems to only recognize one core?
BARSbench.JPG

Re: AMD Brook version slow with hd3870?

Posted: Wed Nov 11, 2009 11:32 pm
by BarsMonster
It is supposed to work with 8.12 drivers only :-)
1 core is reserved for managing Brook thread.

This all is kinda sucks.

Re: AMD Brook version slow with hd3870?

Posted: Wed Nov 11, 2009 11:39 pm
by Crucifix
BarsMonster wrote:It is supposed to work with 8.12 drivers only :-)
1 core is reserved for managing Brook thread.

This all is kinda sucks.
So you suggest I go down to 8.12?

Re: AMD Brook version slow with hd3870?

Posted: Thu Nov 12, 2009 12:09 am
by BarsMonster
Crucifix wrote:
BarsMonster wrote:It is supposed to work with 8.12 drivers only :-)
1 core is reserved for managing Brook thread.

This all is kinda sucks.
So you suggest I go down to 8.12?
Yes. It should not even start on post-8.12 drivers. Also, I admin that BarsWF kinda sucks from ATI now. You might want to try ighashgpu

Re: AMD Brook version slow with hd3870?

Posted: Thu Nov 12, 2009 12:41 am
by Crucifix
BarsMonster wrote:
Crucifix wrote:
BarsMonster wrote:It is supposed to work with 8.12 drivers only :-)
1 core is reserved for managing Brook thread.

This all is kinda sucks.
So you suggest I go down to 8.12?
Yes. It should not even start on post-8.12 drivers. Also, I admin that BarsWF kinda sucks from ATI now. You might want to try ighashgpu
Okay, installed 8.12.

I looked into ighash, unfortunately it only supports HD 4000 series and up.

However, i'm getting 310 million hash/sec now since driver switch, which is good enough for me.

Also, I had another question...is it true this program supports max of 15 character hash?

Re: AMD Brook version slow with hd3870?

Posted: Thu Nov 12, 2009 1:00 am
by DarkPrince
Crucifix wrote:
Okay, installed 8.12.

I looked into ighash, unfortunately it only supports HD 4000 series and up.

However, i'm getting 310 million hash/sec now since driver switch, which is good enough for me.

Also, I had another question...is it true this program supports max of 15 character hash?
You wouldn't need more when it's not salted

Re: AMD Brook version slow with hd3870?

Posted: Thu Nov 12, 2009 1:05 am
by BarsMonster
Yes, it is 15 max. Can't see how you can reach 16+ except trying "01" charset :-D

Re: AMD Brook version slow with hd3870?

Posted: Thu Nov 12, 2009 1:45 am
by Crucifix
BarsMonster wrote:Yes, it is 15 max. Can't see how you can reach 16+ except trying "01" charset :-D
Well, i'm just thinking about the people who have rigs that do like 5000m/second.

A friend of mine downloaded this program because he had the idea that it could crack CD key hashes, but in reality, most CD keys are 20+ digits long which, even if this program did support 15+ char sets, it would probably take years to crack a 25 character string.

Re: AMD Brook version slow with hd3870?

Posted: Thu Nov 12, 2009 2:01 am
by BarsMonster
Crucifix wrote:
BarsMonster wrote:Yes, it is 15 max. Can't see how you can reach 16+ except trying "01" charset :-D
Well, i'm just thinking about the people who have rigs that do like 5000m/second.

A friend of mine downloaded this program because he had the idea that it could crack CD key hashes, but in reality, most CD keys are 20+ digits long which, even if this program did support 15+ char sets, it would probably take years to crack a 25 character string.
Yeah, to crack 25x[az] you will need 1'501'609'779'350'377'910 years at 5000m/second :crazy:

Re: AMD Brook version slow with hd3870?

Posted: Thu Nov 12, 2009 4:11 am
by Crucifix
BarsMonster wrote:
Crucifix wrote:
BarsMonster wrote:Yes, it is 15 max. Can't see how you can reach 16+ except trying "01" charset :-D
Well, i'm just thinking about the people who have rigs that do like 5000m/second.

A friend of mine downloaded this program because he had the idea that it could crack CD key hashes, but in reality, most CD keys are 20+ digits long which, even if this program did support 15+ char sets, it would probably take years to crack a 25 character string.
Yeah, to crack 25x[az] you will need 1'501'609'779'350'377'910 years at 5000m/second :crazy:
I'm waiting for someone to try 4 5870s and get 10000m/second.

Even then though...way too low to ever crack a CD key hash.

Re: AMD Brook version slow with hd3870?

Posted: Thu Nov 12, 2009 9:54 am
by LordMike
Crucifix wrote:
BarsMonster wrote:
Crucifix wrote:
Well, i'm just thinking about the people who have rigs that do like 5000m/second.

A friend of mine downloaded this program because he had the idea that it could crack CD key hashes, but in reality, most CD keys are 20+ digits long which, even if this program did support 15+ char sets, it would probably take years to crack a 25 character string.
Yeah, to crack 25x[az] you will need 1'501'609'779'350'377'910 years at 5000m/second :crazy:
I'm waiting for someone to try 4 5870s and get 10000m/second.

Even then though...way too low to ever crack a CD key hash.
Uh.. How the heck will you crack a CD Key?!?.. Is it MD5 Encoded?.. o.0

Re: AMD Brook version slow with hd3870?

Posted: Thu Nov 12, 2009 12:19 pm
by Crucifix
LordMike wrote:
Uh.. How the heck will you crack a CD Key?!?.. Is it MD5 Encoded?.. o.0
Yes --

let's say you host an online game. Most online games track people based on their registered CD keys.

So, you open an online game of Halo or Call of Duty or something like that. As people walk in, you ban them from the game. The game bans based on an MD5 hash of their CD key, so theoretically all you would have to is go into your server's small database of banned users, decrypt the MD5s stored there and, bam, you would have a bunch of CD keys.

This is all neglecting the fact that most CD keys are 20+ characters in length, and use a combination of capital letters and numbers, which makes their respective MD5s basically impossible to crack.

Re: AMD Brook version slow with hd3870?

Posted: Thu Nov 12, 2009 1:33 pm
by BarsMonster
Crucifix wrote:
LordMike wrote:
Uh.. How the heck will you crack a CD Key?!?.. Is it MD5 Encoded?.. o.0
Yes --

let's say you host an online game. Most online games track people based on their registered CD keys.

So, you open an online game of Halo or Call of Duty or something like that. As people walk in, you ban them from the game. The game bans based on an MD5 hash of their CD key, so theoretically all you would have to is go into your server's small database of banned users, decrypt the MD5s stored there and, bam, you would have a bunch of CD keys.

This is all neglecting the fact that most CD keys are 20+ characters in length, and use a combination of capital letters and numbers, which makes their respective MD5s basically impossible to crack.
How about sniffing hashes out of network, and injecting them into the game response?

Re: AMD Brook version slow with hd3870?

Posted: Thu Nov 12, 2009 3:12 pm
by DarkPrince
What if 2 cd keys have the same md5 hash? Then both would get banned if only one of them get banned. Doesn't sound right

Re: AMD Brook version slow with hd3870?

Posted: Thu Nov 12, 2009 5:55 pm
by LordMike
DarkPrince wrote:What if 2 cd keys have the same md5 hash? Then both would get banned if only one of them get banned. Doesn't sound right
Chance of collision is minimal. Seriously. :P
Crucifix wrote:
LordMike wrote:
Uh.. How the heck will you crack a CD Key?!?.. Is it MD5 Encoded?.. o.0
Yes --

let's say you host an online game. Most online games track people based on their registered CD keys.

So, you open an online game of Halo or Call of Duty or something like that. As people walk in, you ban them from the game. The game bans based on an MD5 hash of their CD key, so theoretically all you would have to is go into your server's small database of banned users, decrypt the MD5s stored there and, bam, you would have a bunch of CD keys.

This is all neglecting the fact that most CD keys are 20+ characters in length, and use a combination of capital letters and numbers, which makes their respective MD5s basically impossible to crack.
lol. Did not know that. :)

Ok. So the hash of the CD key is then 'salted'. Where 5 characters consist of "-" (Often).

xxxxx-xxxxx-xxxxx-xxxxx-xxxxx

Giving 20 characters (Exactly 20 :P).
(26+10)^20 = 13.367.494.538.843.734.067.838.845.976.576

Damn.. :(

6B/s = 2.227.915.756.473.955.677.973,140996096 seconds
~ 70.646.745.195.140,65 years.. :(