[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 • How high can you get - Page 2
Page 2 of 11

Re: How high can you get

Posted: Thu Oct 16, 2008 10:43 am
by BarsMonster
c4p0ne wrote:By the way, question... What kind of performance are we looking forward to in the upcoming Intel CORE i7 Octacore (actually, 4 cores with 2 threads/per core hyperthreading has reappeared) architecture? Also, we know about about the use of SSE2 instruction set to accelerate calculations, but what about SSE3, SSE4 and upcoming SSE4.1 and SSE4.2?
Core i7 is not octacore, this hyperthreading does not improve speed(and may even lower it because of insufficient registers) for applications which utilize CPU parts for 100% (like BarsWF). SSE2 part of Core i7 could hardly be faster than Core2Duo's (3 SSE2 instructions per cycle is already very very fast - that means 12 instructions per cycle). so I would expect performance increase mainly because of frequency increase.

Core i7 might be significantly faster for memory operations - as both latency & bandwidth have increased.

SSE3, 4, 4.1, 4.2 are about floating point calculations with single and double precision, so we cannot use them.

Re: How high can you get

Posted: Thu Oct 16, 2008 7:14 pm
by c4p0ne
Ahh, I see! Thanks for the correction, that clears up some mysteries for me. I was falsely excited about what I thought would be some "significant" increases in calculation performance in this area. :(

A couple more things, 1. Do you have any input on how well current/future AMD CPU's perform vs. Intel w/SSE2? 2. I have seen some benchmarks on the ATI GPU's recently. Many show up to 30% + improvment over their Nvidia counterparts. Do you think this figure will translate well (for hash recovery)? In other words, if we are able to harness ATI GPU's (e.g. 4870x2), are we likely to see 30% improved speed in math calculation?

If so, I will switch BACK to ATI immediately.

Re: How high can you get

Posted: Thu Oct 16, 2008 8:36 pm
by BarsMonster
c4p0ne wrote:A couple more things, 1. Do you have any input on how well current/future AMD CPU's perform vs. Intel w/SSE2? 2. I have seen some benchmarks on the ATI GPU's recently. Many show up to 30% + improvment over their Nvidia counterparts. Do you think this figure will translate well (for hash recovery)? In other words, if we are able to harness ATI GPU's (e.g. 4870x2), are we likely to see 30% improved speed in math calculation?
Right now current AMD processors are very slow at SSE2 (in comparison to Core2Duo, even phenoms are like twice slower). I do not expect AMD to cover that gap and become significantly faster in the near future. (but that's true when we are talking about top CPUs comparison not considering price. Non overclocked 150$ AMD processor would not be that bad in comparison to 150$ Intel one :crazy:)

AMD video cards have an important advantage (AFAIK) - cyclic rotate operation, which might give us 20-30% of extra performance in the same conditions(on top of these 30% you are talking about). So I believe that AMD might be faster, especially if we would measure performance per $ :-)

Re: How high can you get

Posted: Thu Oct 16, 2008 8:39 pm
by the_drag0n
hah a 150$ intel processor what is that ? an atom ? or an single core 3Ghz :P

Re: How high can you get

Posted: Thu Oct 16, 2008 11:02 pm
by x4d3
c4p0ne wrote:I need at least 40 Teraflops... Anyone offering? :wall: :lol:

Hey c4p0ne,

Whats your video card? Stock clocks?

Re: How high can you get

Posted: Thu Oct 16, 2008 11:27 pm
by Sc00bz
SSE3, SSSE3, SSE4a, SSE4.1, and SSE4.2 are not just for floating point operations but most of the instructions are pointless for cracking passwords.

SSE3 (All the other instructions for SSE3 are for floating points):
LDDQU - This is an alternative misaligned integer vector load. It can be helpful for video compression tasks.
MONITOR, MWAIT - These optimize multi-threaded applications, giving processors with Hyper-Threading better performance.

SSSE3 (This added only integer instructions and this one is the best out of them):
PSHUFB - Packed Shuffle Bytes - takes registers of bytes A = [a0 a1 a2 ...] and B = [b0 b1 b2 ...] and replaces A with [a[b0] a[b1] a[b2] ...]; except that it replaces the ith entry with 0 if the top bit of bi is set.

SSE4.1 (There are other integer instructions but these are the best):
PMULLD - Packed signed multiplication, 4 packed sets of 32-bit integers multiplied to give 4 packed 32-bit results.
PTEST - This does the same as the TEST instruction, in that it sets the ZF and CF flags to the result of an AND between its operators ... it sets the Z flag if any of the bits matched, and the C flag if all of them did.

SSE4a and SSE4.2 have integer instructions but they aren't that useful. POPCNT and LZCNT are nice for parity bit check and estimating the log base 2 of a number. EXTRQ and INSERTQ are kinda cool but worthless in most cases since it only works with one 64 bit segment instead of four 32 bit segments.

SSE5: I keep forgetting what these instructions are but it has a few instructions for AES. Right, this has PROTD which will increase speed a lot since this will replace a mov, two shifts, and an or (43 rotates * 3 instructions = 129 instructions less per four MD5s).

AVX: This is the most awesome thing ever, but totally sucks because it could double performance but the initial release in 2010 will not support an 256 bit integer instructions. It still will speed things up because it has 3 operand instructions. This means that you won't need to move a register's contents to another register. Granted there are these floating point instructions ANDPS, ANDPD, ORPS, ORPD, and XORPS, XORPD which I don't know how those work but it looks like they just do and, or, and xor. So you can do DES really fast if it works just like integer and, or, and xor.

* Note that SSE5 will use less instructions than AVX and probably be faster until AVX adds 256 bit integer support because there is no rotate instruction in AVX. Only problem is that AMD is the only one who is doing SSE5 and SSE5 might be dropped by AMD since AVX is much better. This is just like when AMD came up with 3DNow and Intel came up with SSE. Now AMD needs to support 3DNow because old programs still use it. So AMD might drop SSE5 to prevent needing to support SSE5 even though not many people use it.

Re: How high can you get

Posted: Fri Oct 17, 2008 10:20 am
by BarsMonster
That's not fair :sad: Why when Intel makes an instruction set AMD have to support that. If AMD does - Intel just ignores it... (With an exception of x64 set)

Re: How high can you get

Posted: Fri Oct 17, 2008 10:23 am
by c4p0ne
x4d3 wrote:
c4p0ne wrote:I need at least 40 Teraflops... Anyone offering? :wall: :lol:

Hey c4p0ne,

Whats your video card? Stock clocks?
Nvidia GTX260 (Core 216 series), all stock.

Re: How high can you get

Posted: Fri Oct 17, 2008 9:53 pm
by Sc00bz
BarsMonster wrote:That's not fair :sad: Why when Intel makes an instruction set AMD have to support that. If AMD does - Intel just ignores it... (With an exception of x64 set)
AMD does the same to Intel I'm pretty sure SSSE3 (not SSE3) is Intel only, SSE4.1 is Intel only, SSE4a is AMD only, and SSE4.2 is Intel only.
Also POPCNT was stolen from SSE4a for SSE4.2. So Intel doesn't just ignore AMD, just when they think it isn't worth it.

Re: How high can you get

Posted: Fri Oct 17, 2008 10:03 pm
by hardfalcon
After all, they're both companies to maximize their benefits and their shareholder value, so these things are just plain normal everyday business policies. They want to make money, that's their prime goal, designing more (or less) good CPUs is just their way to get there.

Re: How high can you get

Posted: Fri Oct 17, 2008 11:22 pm
by BarsMonster
hardfalcon wrote:After all, they're both companies to maximize their benefits and their shareholder value, so these things are just plain normal everyday business policies. They want to make money, that's their prime goal, designing more (or less) good CPUs is just their way to get there.
And after that they says that capitalism is effective :crazy:

Re: How high can you get

Posted: Sat Oct 18, 2008 1:55 am
by xprog
Image


AMD64 X2 5000+ 2.6G (duel core)
3x 9800GT mfg oc
1x 8800GT mfg oc


I should have built the pc with an intel board/cpu, i get 50mil/sec per core on another pc at similar ghz... I didnt realize AMD would be less then half the speed per core. Ahw well...

Re: How high can you get

Posted: Sat Oct 18, 2008 2:07 pm
by Frankie00rz
did you have got an electicity flatrate ?
how many watts you need to get all the 4 card working 100% = ?
can you please upload a screenshot of your great high speed cracking pc box ?

Re: How high can you get

Posted: Sat Oct 18, 2008 2:45 pm
by the_drag0n
9800GT takes about 110W max
8800GT as well.

makes 440W plus CPU (about 100W) means ~540W. thats still "ok". i only have 250W max i think ;)

Re: How high can you get

Posted: Sun Oct 19, 2008 3:19 am
by xprog
my crack box has a 1100 watt psu, overkill but i didnt want it to be an issue. my cpu is only like 60watts too. I had bought 4x 9800gt (double slot) with an antec300 case for the better cooling but now the bottom pci-e slot is right above the psu and the double slot 9800gt wont fit. So i grabbed my pc's 8800gt single slot and stuck it in there, almost identical performance.


Barswf and LHC only ones i can find that support multi gpu cracking. EGB disappointingly doesnt =(

Re: How high can you get

Posted: Sun Oct 19, 2008 3:25 am
by BarsMonster
2 9800GX2 overclocked should be cheaper in comparison to 4x9800GTX

Re: How high can you get

Posted: Sun Oct 19, 2008 3:42 am
by xprog
9800gt are lil over $100, 9800gx2 are around $400+. you can find refurb ones for over $200 though.

Re: How high can you get

Posted: Sun Oct 19, 2008 12:35 pm
by Frankie00rz
9800gx2 are a lot of cheaper in germany ;-)
but if you need extrem power , this would help you
Image

Re: How high can you get

Posted: Sun Oct 19, 2008 1:05 pm
by the_drag0n
well it coasts about 280€ which is 376$ atm :P

anyway the most expensive thing is all the energy you need... ;)
if you keep that machine running 24/7 you get to pay *calculating*
267 W * 4 = 1068 plus CPU about 100 plus stuff another 50
thats constantly 1218W !

which means 10669,68 kWh / Year
the actual price is 0,2€ /kWh ....

2133,936 € Per Year only for the pc !!!

But still i like your screenshot ;)

Re: How high can you get

Posted: Sun Oct 19, 2008 2:56 pm
by the_drag0n
budden wrote: Hm.. In Russia the cost of kWh is ten times less - about 0.0266€
hmm somehow i get the feeling that many things are less exspensive in the east.
we also pay 35€ for dsl16k.
i know someone in rumania who only pays 9€ for that!

Re: How high can you get

Posted: Sun Oct 19, 2008 7:36 pm
by BarsMonster
What is dsl16k? :-)

In moscow you can get 10 mbit in/out unlimited for 30$ :-)
Unfortunately, only Moscow has that low prices :sad:

Re: How high can you get

Posted: Mon Oct 20, 2008 12:36 pm
by the_drag0n
16k means 16Mbit/s download and 1Mbit/s upload ;)

Re: How high can you get

Posted: Fri Oct 24, 2008 5:42 am
by juggalobeast
gpu: 8800gs clocks-705core 910mem
cpu: e7200 clock 3.17ghz
ram: 4gb ddr2 800
os: vista ultimate 64bit

Re: How high can you get

Posted: Fri Oct 24, 2008 2:49 pm
by B0ff
I've upgraded my system, I now have 2x 9800GT and a Q6600 in a P5N-E MB, all stock. But for some reason the 2nd gpu is running slower than the first:
Barswf_5.PNG
Barswf_5.PNG (15.39 KiB) Viewed 19252 times
I've updated all drivers (Nvidia graphics & chipset, BIOS) and even given the 2nd drive it's own independent PCI-E power supply via a spare PSU but nothing makes any difference. Is this normal?

Re: How high can you get

Posted: Fri Oct 24, 2008 3:12 pm
by the_drag0n
are you sure the second pci-e slot of your board has the same amount of lanes ? because many boards have one with less lanes.