Why just clock when you can overclock

[Creaphis]

So, I'm still browsing Doomworld with the same piece of shit that refuses to work with the new video card that I bought and paid for some months ago. But, now I'm posting about a completely different issue. I need a few more processor cycles than this computer's one entire CPU core is giving me (and I can't justify the expense of upgrading right now) so I'm asking you, the resident tech experts of Doomworld: is overclocking this thing a possibility? You'll find the system specs and components in the link up there. Nothing's changed, except now it has more RAM and a better PSU. It runs reasonably cool right now (30 degrees Celsius, maybe 40 under extended full load). I don't know what else you need to know.

PS: This is the first time in three and a half thousand posts that I've remembered to set the message icon to something other than "DUMB."

[John Smith]

Overclocking is often not a BIOS feature on most OEM vendors motherboards. You can feel free to check but I wouldn't get your hopes up.

[Creaphis]

So it has to be done through the BIOS eh? I can't just open this case up and, say, move a jumper? Or tighten some screws?

I guess overclocking this thing isn't an option. Thank you for letting my hopes down as humanely as possible.

[Aliotroph?]

I don't like overclocking except on a comp I can waste for experimentation. It's almost impossible to tell if you're going to just suffer pain and frustration while maybe even hurting the thing.

[caco_killer]

I had a computer very similar to yours for four years. I sincerely doubt the single 80mm case fan and the cpu heatsink (no fan!) could handle overclocking.

[Doom Marine]

Why are y'all treating overclocking like it's open heart surgery?

[Csonicgo]

Doom Marine said:

Why are y'all treating overclocking like it's open heart surgery?

because it used to be. for years we were told that overclocking could fry processors, burst capacitors, cause precision errors, and once the damage was done, it was permanent.

Everything that happens in the PC involves the processor in one way or another, and everything relies on the CPU being able to compute properly: 1+1+1 must always be 3, etc. When this can no longer be guaranteed, basically all bets are off.

This is the main reason that I do not overclock. I'm not worried about damage as I am worried that my data will not be safe, and that any time I have a problem I will have to guess about whether or not it was the result of my overclocking.

[Doom Marine]

Csonicgo said:

because it used to be. for years we were told that overclocking could fry processors, burst capacitors, cause precision errors, and once the damage was done, it was permanent.

It sure can, if you shoot the voltage like a child offing a shotgun he just found in his parent's closet, then yeah. In other words, you'd have to be a complete retard for this to happen, just like how you have to be a complete retard to drive into a lamp post.

If you're competent and studied the technical specifications of a given CPU, such as its maximum voltage and thermal tolerance, then the overclock should operate within the safe technical specification of that chip.

Csonicgo said:

Everything that happens in the PC involves the processor in one way or another, and everything relies on the CPU being able to compute properly: 1+1+1 must always be 3, etc. When this can no longer be guaranteed, basically all bets are off.

You don't just overclock and wait for errors to occur, you find them pro-actively. There are programs such as the Intel Burn Test, Orthos, and SuperPi to test the overclock rigorously, far beyond everyday usage, for calculation consistency and stability.

Csonicgo said:

This is the main reason that I do not overclock. I'm not worried about damage as I am worried that my data will not be safe, and that any time I have a problem I will have to guess about whether or not it was the result of my overclocking.

Maybe your data is safe, maybe not. From a bad power supply to high humidity, there's all kinds of ways to fuck it up.

Data corruption can happen at any time, once again, there are programs out there that can help rule out the possibility of data corruption via overclocking.

[DoomUK]

Overclocking? I've never really bought into the culture of spending oodles of time (and money) on adjusting settings and replacing existing cooling methods, all for a virtually unnoticeable increase in system performance. Entire online communities are dedicated to this sort of thing, and other than learning more about computers that is a by-product of fiddling around with this stuff, I find it all a bit dorky.

The benefits don't justify any risk there is of damaging your hardware, if you ask me. But... to each their own, etc.

[xepop]

DoomUK said:

Overclocking? I've never really bought into the culture of spending oodles of time (and money) on adjusting settings and replacing existing cooling methods, all for a virtually unnoticeable increase in system performance. Entire online communities are dedicated to this sort of thing, and other than learning more about computers that is a by-product of fiddling around with this stuff, I find it all a bit dorky.

The benefits don't justify any risk there is of damaging your hardware, if you ask me. But... to each their own, etc.

Well, you clearly don't know what you are talking about. I myself got just for cpu 700Mhz increase easily and it runs cooler (less than stock voltages), too. FPS in games like TF2 went significantly up. It didn't take oodles of time nor money, actually it only took couple days (I never before OCd anything) and zero money. Now I don't have to touch it anymore if I don't want to.

But... to each their own, etc.

[Quasar]

DoomUK said:

The benefits don't justify any risk there is of damaging your hardware, if you ask me. But... to each their own, etc.

It made more sense, to me at least, back when we were still behind the steep upward slope of the exponential increase in microprocessor speeds - you know, back when it took a year to get the next 25 MHz increase out of Intel's official product line.

Things were so slow back then, that the difference between a stock processor and an overclocked one was enough to matter significantly. Now, if you have anything Core 2 Quad or better, you're not waiting around to do much of anything. With the TwineCompile plugin, my work machine, a Core i7 with 3 GHz cores, can compile 90 million lines of code in 50 seconds - yikes >_> Overclocking would not speed that up by much as I'm sure it's IO bound at this point, and that's a point people obsessed with CPU speeds miss - and have always missed - the rest of the system architecture matters more after a certain point. For example no matter how much GHz you throw at the DOOM engine, it's always going to be slow when dealing with too many 2S lines and vissprites because that algorithm is cache-bound - you need better cache performance to see any benefit.

Cache RAM remains a major bottleneck on modern computers. I personally think the companies should invest more R&D into improving that than into additional GHz at this point ;)

[exp(x)]

Quasar said:

I personally think the companies should invest more R&D into improving that than into additional GHz at this point ;)

Tell that to the marketing department.

[Maes]

Quasar said:

Cache RAM remains a major bottleneck on modern computers.

Well, this begs the question: why don't they make main RAM of the same quality as cache RAM or CPU registers by now? Is it is still prohibitively expensive to make all RAM static pure flip-flops rather than that "dynamic RAM" bullshit with its refresh cycles?

That would only leave the bus/propagation delay to put an upper limit to processing speed, at which point we'd have to move to asynchronous CPUs (with no external clock) and then to fully optical computers.

...ok, back to reality. Stamping out a bunch of capacitors by photolithographic process will always be cheaper than making 6-transistor latching flip-flops for each bit of storage, no matter what.

Still...what happened to upgradeable cache modules or "Cache on a stick" ? I'm sure many ov3rcl0xx0rz d00dz would be more than happy to burn $1000 on a 64MB fully static ram cache module, if they were for sale :-p

Or better yet, 16 MB module of register-grade storage selling for $5000 :-p

Creaphis said:

So it has to be done through the BIOS eh? I can't just open this case up and, say, move a jumper? Or tighten some screws?

Welcome to the wonderful 21st century, where jumpers on mobos are mostly a thing of the past :-p

That being said...you could try swapping the CPU for another. The specs say it's an AMD Socket 939, so that means you can go all the way up to an Athlon X2 4800+, if you can still find one at a reasonable price (they are highly sought after, though). That should surely give some breathing room.

[Phml]

It didn't take oodles of time nor money, actually it only took couple days (I never before OCd anything) and zero money. Now I don't have to touch it anymore if I don't want to.

But... to each their own, etc.

To each their own indeed, because here's the kicker : how much do you earn ? Even at minimum wage levels, spending two days to get an overclock going costs more than simply buying a faster processor that will yield better results ; so unless you actually enjoy the process, it is definitely costly.

Same thing with people building their own PC "to save money". When I see some of these guys spending days because a certain piece of hardware isn't working so they have to first test stuff to find out exactly what is the culprit, then return it to the store / buy another one, as opposed to simply paying $50 extra and having your PC built, tested and shipped without losing more than the few minutes it takes to open up the box.

There's points in one's life when that kind of stuff might make sense, like a teen who isn't entreprising enough to get a job and have plenty of time to spare ; and, again, if you enjoy the whole building your PC / overclocking thing then it's different, as it becomes a hobby. However, from a pure efficiency perspective, I can't fathom why you'd trade time for money when talking about such relatively low amounts that you actually lose both time and money in the end.

[John Smith]

I'm siding with Doom Marine here, you people know nothing about overclocking for the most part. It isn't incredibly time consuming, And the poster directly above me knows nothing about anything at all from what I can tell (hurr hurr 2 days how much wage loss b/cuz every1 works 7 days a week!, hurr hurr I know retards who cant select hardware and it takes them days to fix a problem that should never have happened, but that makes PC building pointless ofc!!!).

Anyways kids it doesn't take that long to do overclocking on average. It's relatively easy when building a PC to go from the parts being in the box to the parts being assembled and ready to go in an hour. Add to that the time to install windows, which I'll generously give an hour to. The time to overclock your system to get the most out of it usually only takes about 2-3 hours, though you usually want to sleep on Prime95 doing a very heavy "burn-in" testing once you think you've gotten it right. So that's, what, five hours and the computer running overnight to get the computer from start to OC'd and ready. You people post on a fucking doom forum, there is no way your time is so valuable you cant find 5 hours somewhere on a weekend or something.

Also, you people have some serious misconceptions about what overclocking is about. Overclocking is not about spending tons of money on hardware just to make it more ridiculously powerful, it's about spending less money to get better results. Secondly, like Doom Marine said, overclocking only permanently damages your hardware if you're a retard, or just trying to push it a bit too far. It's a pretty safe process if you do it even remotely right.

And for Quasar, Maes regarding the importance of cache in current processors see: http://www.tomshardware.com/reviews/athlon-l3-cache,2416.html

[deathbringer]

I remember when i first heard of overclocking. It involved my brother going "I found this thing on the internet that makes ur computar faster harhar". This being the same brother who used to pummel the "degauss" button on the monitor, and when i told him that it could fuck it up, which i heard from somebody who used to service computers for the navy, his response was "no it doesn't".

So yeah, i'm glad he never tried it.

[Belial]

It took me a day to figure out the top CPU and GPU speeds I could set and the best RAM timings.

Then I reverted everything and underclocked my CPU when I realized how pointless it was.

Phml said:

Same thing with people building their own PC "to save money". When I see some of these guys spending days because a certain piece of hardware isn't working so they have to first test stuff to find out exactly what is the culprit, then return it to the store / buy another one, as opposed to simply paying $50 extra and having your PC built, tested and shipped without losing more than the few minutes it takes to open up the box.

When I first built my own PC it took me 2 hours. After another 2 hours I was playing Far Cry.

[Maes]

John Smith said:

http://www.tomshardware.com/reviews/athlon-l3-cache,2416.html

That was enlightening. So at least for typical applications, even "heavy" ones like video, audio and games, it seems that after a certain point having more L3 cache doesn't help (and might even hurt). Hmm so a 64-MB L3 module doesn't sound like such a good investment....

...then again, the target group of such a product would be people who like boasting about numbers or getting a fraction of a hair of improvement in [insert e-peen metric here], and of course somebody could always construct some degenerate application or benchmark that DESPERATELY needs an X amount of L3 cache...or they'd move to pluggable L2 cache (the COAST modules you could plug into some Pentium mobos were actually L2, I have such a mobo with a whoopping total of of 512K of L2 cache and...nothing else :-p )

Just to think about how sick having 4 MB or more of cache is....a machine running vanilla DOOM under DOS would work almost exclusively from the cache :-p

[printz]

Usually when my computer is too slow, I get the message it's time to buy a whole new machine :)

[Maes]

printz said:

Usually when my computer is too slow, I get the message it's time to buy a whole new machine :)

With windows, there's always the "disk fragmentation", "software bloat" and "installation aging" factor to take into consideration. Gotta find a way to cut down those 100 processes during idle, huh?

With the amounts of RAM one typically has today (with XP, anything above 2 GB) virtual memory page swapping should no longer be an issue unless you use stuff that eats up 4 GB...or you're using Vista/7. Even then, with 3 or 4 GB you should be set.

Therefore, most "my PC is slow" complains often refer to some specific kind of slow:

• It's slow when trying to open a large, fragmented file.
  
• The hard disk trashes a lot when trying to do pretty much anything.
  
• It's slow when you're trying to stuff it with something larger than its physical RAM e.g. running Kane & Lynch 2 with 1 GB surely won't be blazing fast!
  
• It's slow when starting up because it has to load 10 different messenger programs, 20 "calendar applications" and 40 sssorted "Hotkey" and "Update" manager, and on top of that, the Antivirus has to scan all of that.
  

So no, before I decide that my machine is to be replaced, I check each of these steps. Their solutions are, respectively:

• De-fragment.
  
• De-fragment, and delete useless shit before doing so or get a bigger hard disk, ghost your old ones' contents on it, and you won't lose any of your precious crappy files.
  
• Get more RAM, or remove useless shit services.
  
• Remove useless shit services. IMO, if you have more than 40-50 processes when idling then you're doing something wrong. Remove that Quicktime Bonjour service, remove iTunes, remove those Dell Assistants and online helpers/hotkeys etc.
  

There you have it. Also you can try updating some drivers like the chipset or graphics card or the BIOS itself. There's quite a lot of stuff you can do and fat you can trim before "burying" a machine as a gone.

[DoomUK]

Maes said:

De-fragment

...with a robust tool, like PerfectDisk. Windows' defrag tool is pretty shit, in my experience (in it's XP incarnation at least, haven't had enough experience with Vista/7 to really comment on it, but I would bet that certain system files still refuse to cooperate and remain fragmented clutter on your disk after performing the operation with the integrated software).

[Maes]

JkDefrag 3.36 is also pretty neat, and if you defrag with forced by-folder alphabetic sort, you can get some pretty fast starting times for windows itself and its applications.

If you end up having fragmented hibernation or worse. page files, then AFAIK the only tool that can defragment those is pagedefrag by sysinternals, and only during boot. After you took care of that you can use "conventional" defragging.

Sometimes however, if shit is TOO fragmented, you're better off copying everything (or at least the badly fragmented shit) to another drive, deleting them, and copying them back: this way they will be present in sorted order.

[printz]

Maes said:

There's quite a lot of stuff you can do and fat you can trim before "burying" a machine as a gone.

Usually it takes enough years until they run too slow, and even then they're still operable. They just can't stand modern antivirus software very well, or use modern versions of applications, but I blame that on CPU speed. And as long as I use them to play old stuff like Doom, I'm fine (though ZDoom and especially GZDoom are pushing).

EDIT anyway: How am I supposed to expect applications demanding 2.5GHz (multicore recommended) on my old Celeron single-core 2.6GHz, for example (save from overclocking or replacing the CPU)?

[Maes]

printz said:

Usually it takes enough years until they run too slow, and even then they're still operable.

With an older OS like XP, maybe.

I've seen far too many -even dual core- laptops struggling with Vista because they were factory-laden with about 200 useless applications and widgets, "helpers", toolbars, and a lot of times, also crap like bonzi buddy and those malware smileys central. These can fuck up even the beefiest machine pretty quickly, as they don't give a fuck about being well-behaved or having a minimal system imprint or whatever. You could as well be running 20 SuperPI's in the background!

[printz]

Maes said:

I've seen far too many -even dual core- laptops struggling with Vista because they were factory-laden with about 200 useless applications and widgets, "helpers", toolbars, and a lot of times, also crap like bonzi buddy and those malware smileys central. These can fuck up even the beefiest machine pretty quickly, as they don't give a fuck about being well-behaved or having a minimal system imprint or whatever. You could as well be running 20 SuperPI's in the background!

Heh, I probably know better not to allow those apps running, they look too conspicuous :P

[Doom Marine]

I disabled my Page File a long time ago, all it does is create virtual memory on the disk, which slows processes down considerably. If anyone has 4 or more GB of RAM, consider getting rid of that useless piece of junk.