News

Comment by: ZippyDSM (Dec 03, 2007 08:23)

Quote:

---

Originally posted by ZippyDSM:

---

stupid Q but couldn't a 700$ quadcore PC with 2GB ramm do the same with linux?

or is it the point its the cheapest hack tool available?

---

That isn't a stupid question at all.

I have a somewhat limited understanding of parallel processing, but I believe the following is correct (somebody correct me if I'm wrong):

The cell processor is optimized to perform the same operation on multiple pieces of data at the same time.

A quadcore PC is more general purpose. All 4 cores could potentially be running different threads or working on the same problem. They aren't optimized to be working on the same problem all at once like the cell is.

Since cracking a code basically requires doing the same operation ad nauseum, the cell is better suited to the task.

The amount of RAM, in this case really doesn't make much difference.

---

2GB o ramm is cheap 2 GBof DR 2 are going for 50-70$ and some of thats is 4-4-4-15!
(on a side note is 1 2GB stick of 4-4-4 12 worth 120$ or would 4-4-4-15 be as good?
I am moving to a e6600!(got it for 150$ :P) from a 939 3700+ and a P4 945)

ya but with out the software/OS behind the CPU whatever features are available mean nothing, but I see what you are saying with the right software its easier to run circles around consumer class PCs for soem things.

Now only if they can only hack games to get them to run better :P

Comment by: maitland (Dec 03, 2007 08:46)

Actually, that wasn't exactly what I meant.

The cell processor has 8 cores, right? But they are all controlled by one PPC64 (aka G5) processor. It can tell them all 8 to do the same thing at once just on different chunks of the same data, or it can tell half of them to do one thing and half to do the other. It is really fast at doing that because it is all coordinated by the one master processor issuing the instructions, and that is what it's built to do.

With a quad core x86_64/amd64 there is no dedicated master, it just runs multiple threads as scheduled by the operating system. So, they can all be working on the same data at once just like the cell, but since it is controlled by software, it's not as optimized for that purpose. The x86_64's would beat the pants off a Cell at multitasking, though.

haha as for your ram question, I'm not really certain... I should've listened closer in Computer Architecture class. (I still got an A-, and i have no idea about RAM timing... hmmm...)

Comment by: keiram (Dec 03, 2007 09:09)

im new here does anyone now where to download ps2 games

Comment by: ZippyDSM (Dec 03, 2007 09:11)

Originally posted by maitland:

---

Actually, that wasn't exactly what I meant.

The cell processor has 8 cores, right? But they are all controlled by one PPC64 (aka G5) processor. It can tell them all 8 to do the same thing at once just on different chunks of the same data, or it can tell half of them to do one thing and half to do the other. It is really fast at doing that because it is all coordinated by the one master processor issuing the instructions, and that is what it's built to do.

With a quad core x86_64/amd64 there is no dedicated master, it just runs multiple threads as scheduled by the operating system. So, they can all be working on the same data at once just like the cell, but since it is controlled by software, it's not as optimized for that purpose. The x86_64's would beat the pants off a Cell at multitasking, though.

haha as for your ram question, I'm not really certain... I should've listened closer in Computer Architecture class. (I still got an A-, and i have no idea about RAM timing... hmmm...)

---

AH! so thats the big diffrance in the Cell, that makes alot of things clearer! (and whycurrent 2 and 4 X core models need to die and be replaced with soemthing the OS and apps can really use.)

DOH got the timings a bit off, 5-5-5-12 or 4-4-4-15,anyway I can get 4GB of the slower or 2GB of the faster for about 120-130.

From my past experience with ramm more is better than timing when they are that close.

So I ave a question for you, I've been on a 939 3700+ for the past 2 years built a P4 dual core 3.0 mmm 945 is the model I think ran about 30% faster,I built it with 2GB 800Mhz ramm on a asrock 4core sata 2for a friend for Xmas, what kind of performance jump should I see, another 30% over the P4?

sigh I know what the main bottle neck for it,I got 2 of thos asrock boards...its a good starting board at least but the 677mhz mem cap sucks.

Comment by: ZippyDSM (Dec 03, 2007 09:13)

Originally posted by keiram:

---

im new here does anyone now where to download ps2 games

---

remove question and do not ask again not, or meet teh ban stick!

Comment by: keiram (Dec 03, 2007 09:42)

eih fuck you man

Comment by: ZippyDSM (Dec 03, 2007 09:49)

Originally posted by keiram:

---

eih fuck you man

---

Dude understand the rules, you can't ask thos questions here in the "open" without being banded or warned.

This isnot hat kind of site, now if you have gray area questions on how to get backups "you made" workignwe can help.
^^

I suggest you remove both comments before a mod comes and nukes you ,thank you for your time!

Comment by: maitland (Dec 03, 2007 10:02)

Zippy sez:

Quote:

---

AH! so thats the big diffrance in the Cell, that makes alot of things clearer! (and whycurrent 2 and 4 X core models need to die and be replaced with soemthing the OS and apps can really use.)

---

Well, there are a lot of ways to think about the problem of increasing computing power.

For the Cell processor, it makes sense to optimize for working on one or two problems at the same time. That's because in video games the processor is typically working on the same problem from many different angles (calculating collisions between flying bullets and baddies and what-have-you).

However, for a general purpose computer, it's just that, general purpose. So, they need to make it work on a wide range of applications. For instance, you have heard, no doubt, of the various SSE instruction sets. These make use of multiple cores in much the same way as the Cell does (Single in Multiple Data or SIMD). But these cores can also work more or less independently from each other, if need be.

It's not really that either one is better; it's just that the Cell is more specialized and the x86_64's are more generalized.

--

I've been experimenting with a completely radical concept, though. You can multiply and divide more or less instantly in the analog domain, so why not have a processor that can convert to analog and do the calculation in a single clock cycle and then convert back to digital to store the result...!

Of course, this would not be as precise as a digital calculation, but it would be much faster. Digital multiplications and divisions typically take several clock cycles, so you could save those for the really critical tasks like keeping the OS stable, and use analog calculations to do the flying bullets and baddies in a video game or whatever that isn't gonna break the system if it is off by a bit or so.


Zippy continues:

Quote:

---

DOH got the timings a bit off, 5-5-5-12 or 4-4-4-15,anyway I can get 4GB of the slower or 2GB of the faster for about 120-130.

From my past experience with ramm more is better than timing when they are that close.

So I ave a question for you, I've been on a 939 3700+ for the past 2 years built a P4 dual core 3.0 mmm 945 is the model I think ran about 30% faster,I built it with 2GB 800Mhz ramm on a asrock 4core sata 2for a friend for Xmas, what kind of performance jump should I see, another 30% over the P4?

sigh I know what the main bottle neck for it,I got 2 of thos asrock boards...its a good starting board at least but the 677mhz mem cap sucks.

---

It's really impossible to say whether it would be 30% faster without knowing what yours and your friends applications typically are. Even then, it would be better just to test it than to speculate.

As for your speed versus size issue, that is another area where you really need to determine what is more important to you to make a valid decision.

Also, if you can put the 4 gigs in a dual channel configuration, you might get better performance (depending on the application) than if you had 2 gigs with lower latency but they were only configured on one channel.

god I'm rambling my ass off now hahaha

Comment by: maitland (Dec 03, 2007 10:22)

@keiram:

Surely you can find what you seek with only a modicum of effort. I suggest you read some and then type some of the things you read into a search engine and press enter. At such time, you will indubitably be close to your mark.

Comment by: ZippyDSM (Dec 03, 2007 10:24)

Quote:

---

Zippy sez:

Quote:

---

AH! so thats the big diffrance in the Cell, that makes alot of things clearer! (and whycurrent 2 and 4 X core models need to die and be replaced with soemthing the OS and apps can really use.)

---

Well, there are a lot of ways to think about the problem of increasing computing power.

For the Cell processor, it makes sense to optimize for working on one or two problems at the same time. That's because in video games the processor is typically working on the same problem from ma ny different angles (calculating collisions between flying bullets and baddies and what-have-you).

However, for a general purpose computer, it's just that, general purpose. So, they need to make it work on a wide range of applications. For instance, you have heard, no doubt, of the various SSE instruction sets. These make use of multiple cores in much the same way as the Cell does (Single in Multiple Data or SIMD). But these cores can also work more or less independently from each other, if need be.

It's not really that either one is better; it's just that the Cell is more specialized and the x86_64's are more generalized.

--

I've been experimenting with a completely radical concept, though. You can multiply and divide more or less instantly in the analog domain, so why not have a processor that can convert to analog and do the calculation in a single clock cycle and then convert back to digital to store the result...!

Of course, this would not be as precise as a digital calculation, but it would be much faster. Digital multiplications and divisions typically take several clock cycles, so you could save those for the really critical tasks like keeping the OS stable, and use analog calculations to do the flying bullets and baddies in a video game or whatever that isn't gonna break the system if it is off by a bit or so.


Zippy continues:

Quote:

---

DOH got the timings a bit off, 5-5-5-12 or 4-4-4-15,anyway I can get 4GB of the slower or 2GB of the faster for about 120-130.

From my past experience with ramm more is better than timing when they are that close.

So I ave a question for you, I've been on a 939 3700+ for the past 2 years built a P4 dual core 3.0 mmm 945 is the model I think ran about 30% faster,I built it with 2GB 800Mhz ramm on a asrock 4core sata 2for a friend for Xmas, what kind of performance jump should I see, another 30% over the P4?

sigh I know what the main bottle neck for it,I got 2 of thos asrock boards...its a good starting board at least but the 677mhz mem cap sucks.

---

It's really impossible to say whether it would be 30% faster without knowing what yours and your friends applications typically are. Even then, it would be better just to test it than to speculate.

As for your speed versus size issue, that is another area where you really need to determine what is more important to you to make a valid decision.

Also, if you can put the 4 gigs in a dual channel configuration, you might get better performance (depending on the application) than if you had 2 gigs with lower latency but they were only configured on one channel.

god I'm rambling my ass off now hahaha

---

rambling is good :P

I ran the P4 945 for a month before taking ti to my friend I noticed 30% (give or take 10%) on response,loading/sorting data and other odds and ends its not a "drastic" change,hell I am running a XP 3000 with 1GB 400Mhz ramm at 166 across the board (dont ask reboots at 200mhz FBS but its working thats the goal of a backup PC),its not that different than the 3700 at time I benchmarked it it got 2000K less on 3Dmark03.

Bascily I am waiting for that huge jump in speed it never seems to come :P


I guess I will go with the 2GB and get another 2GB stick later,no use to try and save 50$ not with the 400$ 8800GTS I have to put in it in it LOL

Comment by: maitland (Dec 03, 2007 10:37)

@Zippy:

hehe, your wait is almost over. I go to school for Computer Engineering, and all I do is try to think of radical ideas. I try to use my right brain and my left brain in tandem, which is a rare trait among my classmates, I've noticed.

So, what I'm saying is, I'll make your computer faster, Zippy! I swear to you! I SWEAR TO FREAKING GOD I WILL!! Zippy your computer will go so fast that it will finish the calculation before it even starts the calculation! That's how fast it will go zippy!!! THAT'S HOW FREAKING FAST IT WILL GO, ZIPPY!!!

Comment by: ZippyDSM (Dec 03, 2007 10:43)

Originally posted by maitland:

---

@Zippy:

hehe, your wait is almost over. I go to school for Computer Engineering, and all I do is try to think of radical ideas. I try to use my right brain and my left brain in tandem, which is a rare trait among my classmates, I've noticed.

So, what I'm saying is, I'll make your computer faster, Zippy! I swear to you! I SWEAR TO FREAKING GOD I WILL!! Zippy your computer will go so fast that it will finish the calculation before it even starts the calculation! That's how fast it will go zippy!!! THAT'S HOW FREAKING FAST IT WILL GO, ZIPPY!!!

---

but...thats how zippy dose...only all the data gets lost because it moves to fast 0-o

....scary....!

LOL

But really when will ADM and intel get off the current limited multi core kick...

Comment by: varnull (Dec 03, 2007 10:44)

@ Zippy

You don't see a great speed increase because of the way 99% of current software is written..
On xp that amounts to a 30% hit running on 64 bit hardware.. and dual core 64? do you add the 30%'s together?? who knows.. checking an amd dual core 3800+ whatever it is doesn't seem radically faster than a single p4 2.4..

The trick is running truly parallel optimized and timed applications.

On a 25 p2 cluster I can transcode a full dvd (dvd:author) in less than 15 minutes, where the amd monster takes an hour plus.. the speed is in optimized parallel software, but that changes for every system.. it's because of timings.

At the moment the parallel supercomputer is in the domain of the immensely rich, big business, university computer departments, experimenters and redneck junk hoarders like me.. unfortunately due to the general numty level of the average computer user (I blame Bill) this situation isn't likely to change... and who the hell needs a 20 gigaflop machine for staring at idiots on youtube anyway??

Comment by: maitland (Dec 03, 2007 10:57)

A few years ago, one of my professors built a massively parallel SIMD device that was something like 40 processors running at a blistering 20 megahertz (!) and could outperform a P4 running at a Gigahertz or so on certain massively redundant tasks like image convolution and the like.

But Zippy, I really hear your lament. The current trend is to build sideways (parallel, multi-core, whatever) because we have built up (clock speed, pipeline optimizations, smaller transistors etc..) as far as we can go. The problem is nobody is having any new ideas, and nobody really has for decades...

It's sad really, our society has made people forget how to create new things, evidently.

Comment by: ZippyDSM (Dec 03, 2007 11:02)

Originally posted by varnull:

---

@ Zippy

You don't see a great speed increase because of the way 99% of current software is written..
On xp that amounts to a 30% hit running on 64 bit hardware.. and dual core 64? do you add the 30%'s together?? who knows.. checking an amd dual core 3800+ whatever it is doesn't seem radically faster than a single p4 2.4..

The trick is running truly parallel optimized and timed applications.

On a 25 p2 cluster I can transcode a full dvd (dvd:author) in less than 15 minutes, where the amd monster takes an hour plus.. the speed is in optimized parallel software, but that changes for every system.. it's because of timings.

At the moment the parallel supercomputer is in the domain of the immensely rich, big business, university computer departments, experimenters and redneck junk hoarders like me.. unfortunately due to the general numty level of the average computer user (I blame Bill) this situation isn't likely to change... and who the hell needs a 20 gigaflop machine for staring at idiots on youtube anyway??

---

So nothing will make XP and current apps run "faster",since hardware has hardware speed limits, hardware is changing from speed/power to more advanced at running more things at once better,but I donot see the current line of consumer CPUs changing "much" then again the OS's are stuck in the stone age in comaprision.

I know the software is "meek" and hardware is not much better.

I use the "30%" as a telling of the performance I see even a 4200X2 with 400Mhz 3GB CL 3 ramm is not much different than a P4 2 core with 2GB of DDR 667 there is alil diff in them and a 3700 at 2.7Ghz with 2GB cas 2 DDR1 400MZ ramm, all this dcking with sht I have found there lil reason to waste the extra 80 on premium sht 0-o

Bascily I have found there are 2 things you do not want cheap, CPUs and PSUs and maybe CPU heatsink units everything else you can get on the cheap and get buy at reasonable performance for half the cost.

Even vidcards,better to wait a year and get a 200$ card at 100 than to get a 200$ card at 200 0-o
unless you got the cash for powa!

Comment by: ZippyDSM (Dec 03, 2007 11:09)

Originally posted by maitland:

---

A few years ago, one of my professors built a massively parallel SIMD device that was something like 40 processors running at a blistering 20 megahertz (!) and could outperform a P4 running at a Gigahertz or so on certain massively redundant tasks like image convolution and the like.

But Zippy, I really hear your lament. The current trend is to build sideways (parallel, multi-core, whatever) because we have built up (clock speed, pipeline optimizations, smaller transistors etc..) as far as we can go. The problem is nobody is having any new ideas, and nobody really has for decades...

It's sad really, our society has made people forget how to create new things, evidently.

---

its the requiem for a consumer nation, we want things cheaper,smaller and quicker, "better" is a option, and with the goverment stifling industry and creating monopolies and conglomerates suing each other over stupid things nothign will change until it all breaks.

I can see them repacking CPU tech for the next decade untill the industry is ready for some real change, it will be funny if lunix gains ground on parallel and floors windows int the coming years in performance.

Would really like to see linux/unix gain some ground and become better OS's.

Comment by: varnull (Dec 03, 2007 11:51)

hehehe.. for a dare I put damnsmall on a quad core thing..

4 penguins on the boot screen, and so fast you were hard pressed to use the thing. A 50mb os running completely in ram is a wonderful thing.

For multicore machines linux and unix are leaps and bounds ahead of windows.. servers and the like have been running multi cpu architecture for years.

As for the point of this topic.. hacking with a ps3..

Not much point.. there are far better machines without the crazy instruction set and unique architecture. It's a games console with multimedia capabilities.. get real.. there is far more to using that cpu fully in a normal os environment than installing linux... so unless sony release full hardware specs and the full assembler instruction set then...... Wait and see.. I think a good quad core/quad cpu machine with a heap of ram will beat it hands down every time (oh and yellow dog linux is from back in 1999.. it can be upgraded, but sony don't really want people doing anything amazing with it.. just sidetrack all the console hackers a little why not ;-)

Comment by: madman91 (Dec 04, 2007 02:41)

This is cracking.. not hacking.

Learn your terminology before spreading false truths.

Comment by: ZippyDSM (Dec 04, 2007 02:48)

Originally posted by madman91:

---

This is cracking.. not hacking.

Learn your terminology before spreading false truths.

---

cracking a pass into a site/aera of the net is hacking tho and thats the point I believe.

Comment by: agwild99 (Dec 05, 2007 17:47)

Quote:

---

Originally posted by rainofire:

---

and security is always getting better so it think its to hard for simple ppl to break i think that a "real" hacker must hav this $10 000 computer wich is proably faster than a ps3.
cuz he has already made a lot of money. hey why r we talking about a he it could be a she? why can´t we picture a hacker as woman as a fisrt ttought funny eh..

---

Follow the white rabbit Neo!

Comment by: lubricant (Dec 07, 2007 02:02)

wasnt there some sort of news report that cell processors shouldnt be allowed in some countries cause they could be used to pilot missiles or some such jazz? or was that the ps2?

Comment by: ZippyDSM (Dec 07, 2007 02:05)

Originally posted by lubricant:

---

wasnt there some sort of news report that cell processors shouldnt be allowed in some countries cause they could be used to pilot missiles or some such jazz? or was that the ps2?

---

PS2, like they need them when they can just buddy up to our for sale officails....

Comment by: maitland (Dec 07, 2007 02:06)

Originally posted by lubricant:

---

wasnt there some sort of news report that cell processors shouldnt be allowed in some countries cause they could be used to pilot missiles or some such jazz? or was that the ps2?

---

Haha, prolly should have been the US, then. Were the ones doing the most firing these days. I'm more afraid of my own government's executive branch than I am of any terrorist. By a long shot.

Comment by: borhan9 (Dec 21, 2007 00:06)

Quote:

---

To conclude his findings, Breese hopes that the report will encourage software-makers to increase their password security. "That's the reason I'm doing this," he says.

---

I would have to agree with this comment.

Comment by: CKhaleel (Dec 21, 2007 01:02)

must...hack...into U.S. goverment...jk