GraphiTI.. AGAIN?!?

Liazon · Post by **Liazon** » Wed 28 Jun, 2006 11:42 am

In terms of security holes, Opera or Firefox?

Cuz it seems that according to this site

Opera seems even better in terms of security (unless I'm mistaken, sorry). How come I haven't heard so much about it? (until I came here of course)

Post by **tr1p1ea** » Wed 28 Jun, 2006 2:55 pm

Opera is my first choice in a web browser. I have never had any problems with security ... not that i usually do, but when others use my PC, spyware can find its way in.

I let my bro use my PC one day and after 3 hours there was something like 120+ spyware apps installed ... and he was only trying to find online flash games. I switched to Opera and havent had a single piece of spyware since (around 2 years). He still uses my PC and still looks for games and soccer info and stuff.

Post by **benryves** » Wed 28 Jun, 2006 3:05 pm

Opera used to be (until fairly recently, at any rate) a commercial product, which is why you probably heard more about Mozilla et al before (freeware).

CoBB · Post by **CoBB** » Wed 28 Jun, 2006 3:35 pm

CompWiz wrote:But I'd rather be the employee getting a new computer because of a bloated word processor program, than the employee with the 1337 optimized word processor that gets stuck with old hardware.

I'd rather have a responsive and friendly interface. I don't care whether the CPU inside is ticking at 10 MHz or 10 GHz. If the available software for the latter is so much poorer quality that the actual interaction is more sluggish, I'll go with the former as long as I don't need raw power (which I don't, unless I'm compiling something).

Time for another 80s' vs. today's computing rant, I guess.

CompWiz · Post by **CompWiz** » Wed 28 Jun, 2006 4:47 pm

CoBB wrote:
CompWiz wrote:But I'd rather be the employee getting a new computer because of a bloated word processor program, than the employee with the 1337 optimized word processor that gets stuck with old hardware.
I'd rather have a responsive and friendly interface. I don't care whether the CPU inside is ticking at 10 MHz or 10 GHz. If the available software for the latter is so much poorer quality that the actual interaction is more sluggish, I'll go with the former as long as I don't need raw power (which I don't, unless I'm compiling something).

Time for another 80s' vs. today's computing rant, I guess.

some programs require faster processors, no matter how well you code them. And even if a program doesn't require a fast processor, chances are it will run faster if the cpu is faster.

As for opera, I'll switch over if those google gadgets work on it. Like browser sync and such. They save me a lot of time, and as far as I can see, they haven't been released for opera. correct me if I'm wrong. Like I said, the memory it takes up really isn't an issue, I have more than enough.

@CoBB: Ghz doesn't matter. It is possible to build a 10ghz cpu that runs programs slower than a 10mhz one. Don't tell me you were fooled by Intel's "Ghz is everything" marketing ploy. You need to take into account the amount of work a cpu is actually capable of doing per cycle, among other things.

coelurus · Post by **coelurus** » Wed 28 Jun, 2006 5:52 pm

Of course some programs run better with faster processors, CoBB was focusing on the more interesting issue that many (most?) programs that have a great potential to run fast don't. I agree.

I use Firefox and Opera interchangeably, they both got their share of bugs and benefits. I've bound Opera to my "browser-key" now that I've started a tendency to watch guitarists in flash videos (linux-firefox in freebsd is weird). I've heard Konqueror is pretty nice, tackling the acid2-test and all, but KDE is just a wee bit too hefty to install

And who cares about the details of 10GHz vs 10MHz? Really?

CompWiz · Post by **CompWiz** » Wed 28 Jun, 2006 6:01 pm

coelurus wrote:Of course some programs run better with faster processors, CoBB was focusing on the more interesting issue that many (most?) programs that have a great potential to run fast don't. I agree.

yes, I'm sure that's true. Maybe us 1337 asm programmers can put some pressure on the computer programmers to also learn asm. x86 asm in their case.

jk
Yeah, I agree also. But when you have a nice fast system with plenty of ram, the importance of that is diminished. I don't really care if firefox uses 5mb or 500. However, programmers need to start making more efficient code. I wonder how much more efficient windows would be if they went back, reviewed and rewrote a lot of the code. If microsoft stays in it's position of OS dominance(in a marketshare sense) then this may never happen. They'll tack on a few nice features for each new version, but never really improve it. Where's googleOS when you need it.

Post by **benryves** » Wed 28 Jun, 2006 6:17 pm

Arguably, Microsoft are helping things to be more efficient with their .NET platform; by using the provided classes and practices, your programs are garbage collected (useful for keeping memory usage lower overall, less leaks) and optimised for your hardware (by using IL bytecode rather than native binaries).

I know there are people here who have little or no faith in GC or JIT compilation, but it really is the way things are going.

necro · Post by **necro** » Wed 28 Jun, 2006 6:37 pm

CoBB · Post by **CoBB** » Wed 28 Jun, 2006 7:28 pm

CompWiz wrote:some programs require faster processors, no matter how well you code them. And even if a program doesn't require a fast processor, chances are it will run faster if the cpu is faster.

These are mostly programs that don't require user interaction. If the machine spends most of its time waiting for you, then you're in a winning position. Unfortunately, software guys are more innovative in wasting resources than the hardware guys at providing them, despite the fact that the latter group does a pretty good job. Opera won me by being the only browser that could run smoothly on my 133 MHz box (then I found out that it has some nice features others lack). The Mozilla lineage was always sluggish.

CompWiz wrote:Don't tell me you were fooled by Intel's "Ghz is everything" marketing ploy.

You've got a fixation.

CompWiz wrote:Yeah, I agree also. But when you have a nice fast system with plenty of ram, the importance of that is diminished. I don't really care if firefox uses 5mb or 500.

Well, I'm somehow not happy to be forced to upgrade my hardware just to keep the experience the same...

CompWiz wrote:I wonder how much more efficient windows would be if they went back, reviewed and rewrote a lot of the code.

http://www.reactos.org/en/index.html

benryves wrote:I know there are people here who have little or no faith in GC or JIT compilation, but it really is the way things are going.

I still hope to see functional languages gaining more popularity, especially with the advent of parallel architectures. Erlang is just the first step.

leofox · Post by **leofox** » Wed 28 Jun, 2006 8:53 pm

CompWiz wrote:
leofox wrote:Nevermind that it's a word processor that for some reason requires a pentium 4 3Ghz or equivalent to work properly...
hey, if you have something worse than a P4 3ghz(banish the thought) you really need to upgrade. I'm sure the employee getting the new computer would be happy about it.

I have a celeron 2.4 ghz. And i don't have money for a whole new computer.

You're acting like modern 3Ghz p4 are just normal, although they have multiples of the strength of '80s supercomputers. And yet you are saying you need it for something as simple as word processing, something a z80 could do.

threefingeredguy · Post by **threefingeredguy** » Wed 28 Jun, 2006 9:03 pm

Mine is, heaven forbid, 2ghz and my laptop is, brace yourself, 1.6ghz. However, I see no need to upgrade for at least another year.

Liazon · Post by **Liazon** » Wed 28 Jun, 2006 9:22 pm

Hey isn't CompWiz a big AMD fan?

Doesn't that mean he prefers larger cache over clock speed?

I just remember that he wasn't too fond of intels, but I'm probably wrong.

CompWiz · Post by **CompWiz** » Wed 28 Jun, 2006 10:10 pm

Yes, I am definitely an AMD fan, and I don't like Intel. Between their inferior cpu's and unfair business practices, who can really support them? But that doesn't mean that I prefer cache over clock speed. Both have their place in increasing performance. In fact, right now Intel's cpu's have a bigger cache. AMD's cpu's don't need a big cache to perform well. The reason Intel cpu's need a big cache is for communication between the two cores. Their normal communication has a ton of latency, and they're using a large cache as a hack to allow them to communicate at a reasonable speed. AMD has full speed in-chip communication pathways that allow each core to communicate with the other at the speed of the cpu. Intel, before using the cache, required communication between the cpu's to run through the slow fsb(bottleneck), to the slow northbridge controller(made with old 130nm manufacturing process) back up the fsb just to get from one core to the other, when they are right next to each other, in the same cpu. Also, the large cache means that Intel cpu's will do really well on some benchmarks that have a small workset. It then puts the entire workset in the cache, so it can run it very fast. (superPI, for example) However, normal, real world applications have much larger worksets, and will not fit in the cache, so performance will be low.

Clock speed does increase performance, but it is not the only determinant of performance. What really matters is how much work a cpu can actually do per cycle. You could make a 10ghz cpu that ran programs slower than a 10mhz one. Intel's P4's did very little work per each clock cycle. It's like a bicycle. If you put it in first gear, you can do a bunch of cycles with the pedals really fast, but the bike won't move very fast. If you put it in a high gear, you only have to push the pedals slowly to make the bike move fast. The amd 64's are like a bike in a high gear, while the p4's are like a bike in a low gear. They spin the pedals faster, but the bike moves slower. Ghz means nothing without a measure of work per cycle. With the new conroe cpu's coming out, Intel is moving back to more of a high gear, although not quite as high as what amd is running. So now they do almost as much work as AMD does per cycle. Once AMD starts selling their 65 nm cpu's at much higher clock speeds, they will easily beat Intel. Intel hits a clockspeed wall at a little over 3ghz with the conroe. AMD will easily beat that with the 65 nm process.

coelurus · Post by **coelurus** » Wed 28 Jun, 2006 10:39 pm

GC is really cool, I had some thoughts about implementing it in C but it didn't become as general as I wanted it to be. The Boehm GC crashed last time I tried it, so I threw all ideas of conservative pointer-guessing GCs somewhere dark and wet. Currently, I got a "manual garbage collection" thingy going, where groups of allocated memory areas are managed and easily cleaned up when needed. Too simple of a solution maybe, but it works real nice and if mem runs out in the middle of an allocations spree, the group is closed and the memory is fine. Oh, and there are _no_ leaks in my engine. Sloppy coding can be reported with full function tracing in runtime too which involves just two ints here and there. Whatever

GC is perfect for managed languages where there are secret contraptions behind the text the programmers writes. I just don't like programming like that myself though.

JIT is good for where it fits, just like everything else. Personally, I prefer open source code or simply clever programming which allows for near per-host optimization.