Overclocking has become too easy

[3line]

Well, I don't mean to gripe seriously, but it's just something of an observation.

I've been overclocking for a long, long time now (by computing industry standards).

The first thing I ever overclocked was a Supermicro 440gx server board for slot1 pentium3s; there were no official settings so I had to dig through the bios to find the manufacturer settings. And once I overclocked my AGP freq went to 89 and my pci to 44, which sent me searching for cards that would work with overclocked buses.

Then came the Athlon generation. I remember pencil modding bridges and u-wiring multipliers. I got an NF7-s. The default bios kinda sucked, but fortunately there was an entire cottage industry of hacked bioses that let me squeeze the last drop of performance from this platform with a mobile barton.

I never really got on the p4 bandwagon, but I moved to a pentium M (with a P4P800-E), which I must admit incredibly easy to overclock if it weren't for the fact that the board had no AGP/PCI lock between 166 and 199mhz fsb. Plus, the cores were so fragile that I must have crushed two of them by clamping down too tight, grinding off the corners. Fortunately the little buggers were dirt cheap by the time I got replacements. Nothing that a little masking tape shim wouldn't fix.

And then there was the nforce4 opteron dual server board (Asus K8N-DL) that I used for the past 2-3 years. Due to nvidia's horrible chipset, I couldn't set a bios frequency higher than 215mhz without blankscreening. I had to use clockgen in windows, and even then I had to move the base frequency in 5mhz increments because moving them too fast would cause windows to hard lock. Registered pc3200 was too expensive, so I overvolted and overclocked pc2700. The sticks got uncomfortably hot, so I put extra cooling on them.

And now we have the core i7 920 on a Gigabyte UD5. Overclocks to near 4ghz out of the box and stays that way. Slap on the Venomous X and tighten the bolt all the way, and you don't have a thing to worry about. Don't have to worry about crushing the core or balancing the tension or any of that. Thermalright's new retention system kicks the crap out of my old cooler's. The board even comes with a built-in power switch and backpanel CMOS reset. No more fiddling for jumpers and popping out batteries. No more tapping the screwdriver into the power pins. No more wire tricking--all voltage options you could ever possibly want are in the bios, in .00625v increments.

I understand that this is all a refinement of technology, but there's still a part of me that enjoyed the hair-pulling challenge of finding out all the little quirks of a system, learning its behavior inside and out. Or fiddling around with non-standard stuff like mobile chips on the desktop. Everything is just so streamlined now.

Know what I'm sayin?

 

[dgb303]

It sure has become easier. But don't let that slow you down, raise the bar! It might be easy to hit 4ghz out of the box with lots of procs now but why not go for 5ghz - that certainly requires more of that hair-pulling fun!

Start doing more radical things!

 

[EarthDog]

If its still easy, you arent overclocking high enough.

 

[silent bob]

It sure has become easier. But don't let that slow you down, raise the bar! It might be easy to hit 4ghz out of the box with lots of procs now but why not go for 5ghz - that certainly requires more of that hair-pulling fun!
Start doing more radical things!

If its still easy, you arent overclocking high enough.

I will admit overclocking has gotten easier , but as time rolls on I now care more about stability , than the ultimate OC . Yes I overclock , but not to the extreme . My q6600 is at 2.8 not much of an overclock but still overclocking . So go as far as you want , sky is the limit , but some choose stability .I am NOT saying that most OC'd systems are not stable , or that your oc is not stable( you and your applies to anyone , not specifically you) .

1 mhz is overclocking so

 

[3line]

I will admit overclocking has gotten easier , but as time rolls on I now care more about stability , than the ultimate OC . Yes I overclock , but not to the extreme . My q6600 is at 2.8 not much of an overclock but still overclocking . So go as far as you want , sky is the limit , but some choose stability .I am NOT saying that most OC'd systems are not stable , or that your oc is not stable( you and your applies to anyone , not specifically you) .

1 mhz is overclocking so

Bah, I still say that the Supermicro P6DGU was the finest motherboard I ever bought. That thing is going on 11 years and it still works without a hiccup. Ran it for years almost 24/7 with a 44mhz PCI. Rock solid to this day.

In general these days I care more about stability for my main rig. I know I can push higher with these volts/temps, but I'm going to risk it for the bragging rights of getting that extra 200mhz to top 4ghz.

What this trend has me doing is it makes me want to go retro, to take a look at old platforms now that parts for them are dirt cheap. I just picked up a couple of CCBBWE dual core opties for my old rig (the fleabay market is glutted with dual core opterons right now)--I'm going to turn it into effectively a quad core and shoot for 2.8-3ghz. No sense in letting perfectly good platforms go to waste.

 

[EarthDog]

Ok I'll bite, how are you going to turn a dual core opty into a quad core? Server mobo with dual sockets?

 

[3line]

Ok I'll bite, how are you going to turn a dual core opty into a quad core? Server mobo with dual sockets?

Yep, my old board was an Asus K8N-DL

 

[dfonda]

I disagree. I think it was easier back in the day. My first OC was going from a 166 to a 200 Hrtz Pentium and all I had to do was change some jumpers.

 

[4GHZ_or_bust]

You have it easy. My first real overclocking didn't involve jumpers as there were none! I had to solder in bridge to disabled on board oscillator and drop in a faster clock to make my old Mac 650 go from 25MHz to 40MHz. Try finding the oscillator back in 1996 or '97 when internet hasn't been "invented" yet by Al Gore.

 

[3line]

I disagree. I think it was easier back in the day. My first OC was going from a 166 to a 200 Hrtz Pentium and all I had to do was change some jumpers.

Pssh. We live in an age where we expect 40-50% overclock without touching a jumper or a switch.

 

[SteveLord]

I think its easier only because the industry has done certain things to enable enthusiasts to do so much out of the box. (higher clocking memory, motherboard tools/features, and CPUs essentially hitting a brickwall with regards to clock speed.) CPUs have been selling at 2ghz-3ghz for how long now?

 

[dropadrop]

Ever since soldering and blowing bridges was no longer needed it became boring.

 

[dfonda]

You have it easy. My first real overclocking didn't involve jumpers as there were none! I had to solder in bridge to disabled on board oscillator and drop in a faster clock to make my old Mac 650 go from 25MHz to 40MHz. Try finding the oscillator back in 1996 or '97 when internet hasn't been "invented" yet by Al Gore.

You mean 1986 0r 87....I was on the Internet in 96. AOL started in 97

 

[Shiggity]

One just has to look at Intel's turbo technology to know that overclocking will be obsolete in a short time.

Cores will soon overclock themselves perfectly for the task at hand, maximizing performance : watt and shutting themselves down completely when not being used.

Dynamic frequency scaling is just getting started.

 

[Old Thrashbarg]

Yeah, I'm with dfonda here. It used to be you just had to change a jumper for either the multiplier or bus speed (or both), and maybe change a jumper to bump the voltage. And even on voltage it was just a couple options, like changing it directly from 3.3 to 3.6V or whatever. Even in the much older days when you had to switch out clock crystals, it was, at most, two pins to solder, and the really old ones were often socketed. Just pull the old one, plug in or solder the new one, and you were back on your way.

Nowadays you have to screw around with all kinds of different timings and dividers, miniscule voltage changes, power management settings, all sorts of crap. You're usually looking at a few hours of ironing out the settings and stability testing. While all that stuff is generally pretty well documented, and it offers a much finer-grain control over things making it relatively easy to maximize performance, it's still a far cry from the simple 'switch two jumpers, or switch one clock crystal, and voila, faster computer' that we did 15 years ago.

 

[3line]

Nowadays you have to screw around with all kinds of different timings and dividers, miniscule voltage changes, power management settings, all sorts of crap. You're usually looking at a few hours of ironing out the settings and stability testing. While all that stuff is generally pretty well documented, and it offers a much finer-grain control over things making it relatively easy to maximize performance, it's still a far cry from the simple 'switch two jumpers, or switch one clock crystal, and voila, faster computer' that we did 15 years ago.

Documentation makes all the difference. Take for instance an old PPro mobo, the PR440FX, which I fleabayed a couple of years ago for almost nothing.

That had no documentation on jumper settings, and the only information that exists got on the internet because some guy played with it and found out all the multis and fsbs through trial and error. I believe that information is no longer up online--fortunately I saved a copy.

I remember how I did my own u-wire v-mods by studying processor pin-out charts for server motherboards that had no voltage options.

All the paths today are laid out right out of the box; there isn't the sense of risk and discovery that you had in the old days. I'm not complaining about this higher stable overclock, but there's considerably less skill and research involved.

 

[4GHZ_or_bust]

Don't forget memory speed and timing. Gone are the day where you can overclock CPU and not need to do anything to RAM. Now you got to adjust the multipliers to keep RAM from going over its rated speed (like mine setting to 1066 rather than stock 1600) and there's about 50 different numbers in RAM timing to tweak for better speed or compatibility.

 

[turbohans]

I agree with 4gz there ^ memory can also play a huge part in overclocking somtimes.
Not to mention somtimes you can just be one adjustment off and not even post at an otherwise stable speed.

On these newer boards I amalso noticing a lot more settings to learn and use than just overclocking some Core 2 stuff...

 

[TollhouseFrank]

overclocking a computer has become like 'ricing' a car. Anyone can do it and get decent results, but only the truly dedicated will be able to stay ahead of the curve and get the most bang for the buck.

 

[twEEkerAreUs]

Depends what kind of ocing you're doing, if you want a casual OC it's super easy. If you want an extremely high OC you end up doing a lot of fine tuning and sometimes actual mods.