Newbie--why o'clock?

[mact]

I've used computers for 25+ years, but am not now nor ever have been a gamer.

I use "graphic arts" software, from Adobe/Corel/and when necessary, Microsoft. (I am in the "prepress" business)

I'm in the early throes of deciding specs for my next Win-box build...the last one was about 3 yrs ago, a P4/3.0. Works well, no complaints.

My next box will likely be the last one I build before I friggin' give it up and retire.

In the interim there has been a factor applied to "Moore's Theorem" (was "Law" but I dispute..). And shit is so fast now I am not confident which way to turn.

So the question is: is overclocking going to provide a good risk:benefit ratio in this situation (a freeze or lockup that kills customer files is not the same as having a stupid (IMHO) game abort.

Opinions, please.

I realize this may have been explored _ad infinitum_ in the past, but I just beamed up and am getting used to the ship.<G> (OK, ST:NG is on the tube just now) (I saw the originals, when they were b&w and the acting was terrible!)

TIA.

 

[ChipDeath]

I'm generalising here, and I've not actually been keeping up-to-date for 7 months or more, so please no-one jump on me for anything here..

Well, the AMDs and Intels of this world generally only produce one actual core for each line of processors. These are all speed-binned according to the quality of each core (or possibly each batch of cores).

Neither company would do well if they only sold high-end products though, so they realise they NEED to fill the low-end market, with slower, lower-rated chips.

These days, almost all of the cores they produce are high enough quality to run at 90% the speed of the fastest models, so they have to restrict cores that are otherwise identical to the hideously expensive top-of-the-line chips, purely to fill a lower market segment.

So basically, most low-end chips nowadays are perfectly safe and stable running like mid or high-end parts from their family.

I fully agree about losing customer's files though. I simply wouldn't risk it for anything that important. While you could stress-test to your heart's content and be happy its stable, at the end of the day you're exceeding the official specs of your CPU, so you don't have any form of guarantee or leg to stand on if it does crash (even if not down to the overclock itself).

My PC at home has been running its 25% overclock for yonks, and I can't remember the last time I had a crash. I still wouldn't use it at those settings for doing anything vitally important though.

Overclocking is two things: 1) Faster CPUs for less money and 2)Something fun to do if you're an enthusiast - it's a hobby, just like people who modify cars I guess.

 

[gomerpile]

OC=heat=power=moremoney,
gamer OC = more kills per second = firstplace every time
that sums up the Laws of ocing

 

[fsalazar90]

Because nothing is more exciting than gaining 2 seconds in SuperPi!!

Be careful though, if you overclock too much everything will be so fast that in games you'll walk and shoot at turbo speeds.
Oh and it's a problem when you try to view slideshows; each photo appears for milliseconds.

 

[Rashind]

Well, let's break it down a little bit.

Pro: Faster operation: 10 second script can become a 7-8 second script, without spending more money... you can do even better if you spend a bit on cooling. It's probably not a huge boost in productivity, but if you're doing the design work yourself it's a huge subjective difference.

Con: More power drawn, more bills to pay.
More heat, need cooling. (Though, the stock C2D fan will take a modest overclock in stride at the low end.)
Higher statistical rate of failure. This seems to be your biggest concern. You seem concerned about losing customer data, but this can be avoided by performing regular backups. If you backup nightly, and do RAID 1 or 1+0 (the latter is not a ridiculous idea for graphic design...), then you're not going to lose more than a day's worth of work. (RAID does not excuse you from making backups. If the CPU eats it while the HDD is working, whatever it's working on is going to turn to crap.) A day's worth of work isn't a disaster, it's just a day's worth of work. It costs you a little time and money, but hopefully not a customer or a lawsuit. If you're in an arrangement where a day's worth of work CAN cost you a customer or lawsuit, then you need to reconsider your contract practices, because whether you OC or not, you run a risk of failure.

 

[p05esto]

Man, I have a P4/3.0 myself right now and it's fast enough for prepress work/photoshop/etc... I don't work fast enough to be waiting for my computer much (2GB RAM helps).

If you buy buy a new Dual Core processor, a couple GB of RAM you'll be so amazinggly happy. These new processors are 2/3 times faster than our p4/3.0 is.

I would NOT overclock. You probably have better things to do for one thing and if you're not playing games then there's really no reason at all, not even a close reason in my opinion. Go for stability. The little overclock would never be noticed in real life (work) - not at all! Go ahead and bash that opinion but it's true.

 

[Crashman]

Some applications run sooooo slow that any increase in performance directly translates into increased productivity.

Some systems crash without overclocking.

You increase the likelyhood your system will crash when you overclock. Your risk to benefit ratio will depend on numerous factors, since every piece of hardware has a different clock ceiling.

I run my Athlon 64 20% overclocked and haven't had a crash yet. My friend has the same configuration, his crashes occasionally. I'd bet neither of us would have a crash at 15%.

Finding an overclock that doesn't cause a crash is a bit of an art form. Experience helps, I wouldn't risk anyone's data until I had confidence in my level of experience.

 

[CompTIA_Rep]

You mention that you have no complaints with the computer you are using now... so why change? If the computer does what you want, as fast as you need, then really thats the best you can hope for... the rest is just a fun factor.


An experienced overclocker with the right amount patience can OC a computer with stable results (I think most would agree). However, since this is a business computer, you need to consider the added risk of frying a component (granted not likely with minimal OCing). How much $$$ will be lost, do you have proper backup solutions (off site, or off computer anyway). How much will down time affect your work? And do you have a backup computer.

Point being, I would consider looking at this from a business perspective first.

 

[AlaskaFox]

Another advantage I like is if you get a cheaper processor, with even a mild OC you can achieve similar performance to a processor double the price. for example, I can get an E6300 to match the performance of an E6600 without spending an extra dime, even using the stock cooler. even if you go with a slightly beefier cooler to extend the life and cut down on heat, you can push that further.
the downfalls as mentioned are more heat, slightly higher power bill (I think an oc'd c2d can still stay below a P4 toaster in terms of power draw), and possible stability issues. alot of OCers run utilities like Prime95 and other stress test utilities to get the max load out of the processor and check for stability. if it passes, usually stability is not a concern. if it fails, turn it down a hair and check again.

 

[Syntonic]

For the majority of us, I think yes it does give us a great risk:benefit ratio since for example, the core 2 duos are great overclockers and more or less proven to be stable at those higher rated speeds through countless people. It's not as if c2ds appeared on the market yesterday or anything and people, regardless if they are running a business or not, value their data.

That being said, I think you should have retired from box building awhile ago and shouldn't even consider overclocking seriously. If you are interested in stability and not fighting with your machine to inch every last ounce of performance, it helps to have someone to scream at when everything goes wrong. Just dig up that 1-800 number and go at it especially if you are in a business setting. You may get better parts for your money if you build it yourself but you also have the headache of putting it together and RMAing/testing the parts if you are lucky to get it to POST. Not sure if that's really acceptable in your case or if you have the patience for that anymore. It just sounds like you just want a faster machine that works out of the box. So... dude, are you getting a Dell?

 

[mact]

Thanks for all your suggestions.

I think I'll likely not "fool with Mother Nature/Intel/AMD" and let things run per spec.

After a little peeling at the iMac -- it is pretty darn quick -- I can see where I can build a solid box for $1500 that should last me for 3-5 years and only be behind the curve fo about half that (which is a winner, I think<G>)

Now I'm studying the Beginners Guide to MoBo selection.