One big discriminator among the new A64 (and P4) boards seems to be the relative overclocking performance. As a new member of the PC-builder/"enthusiast" ranks, I'm curious about the real benefits and costs of OCing. Here's a reviewthat shows only minute (<5%) differences in real performance, and I'm confused about the combined effects of OCing the CPU, motherboard and memory. Is that useful, healthy for the PC, and does it add up to a difference that I'm likely to notice?
For those that OC, is this a something that continues to be a hobby, or do you find your system's limits pretty quickly and then just back off a little and stay there until the next upgrade?
I'm also confused reading reviews about astounding OC capabilities of certain motherboards, then looking at forums where individual experiences with that board are quite different. Any ideas about that?
I realize that SCR folks are less likely to be hardcore OCers since OC=>heat=>noise. But you folks have the most balanced, mature approach to computer building and use of any of the forums that I'm following these days, and I suspect that some of you have worthwhile answers to these questions.
Please share your overclocking opinions
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
There was a discussion on this point some time ago, I'm too lazy to look it up.
For me, I usually buy the processor at the sweet spot and overclock by up to 10%. This usually gives me a processor one up on the list at a lower price. Usually 10% is very doable on just about any hardware without serious heat/noise/instability penalties
The link you gave is running along a fine edge. Both stock & overclocked setups are running at the same MHz so it's more of a balance with memory timings vs higher FSB. I wouldn't consider that an overclock. 300x10, now, THAT would be an achievement
Since I'm not a benchmark freak, I really don't care about the minute differences between faster FSBs or tighter memory timings. As long as it works....
A little faster clock has visible benefits for me as I run Folding@Home 24/7 which is very much dependent on clock speed.
As for my Athlon64, it's turning out to be a nice overclocker. Right now I'm at 15%, looking to pushing it to 20%. That's a lot of overclock and I can do it without getting too hot or speeding up my fans so no noise penalty either (For those who've been following my o/c attempts, I found out that I forgot to adjust the Hypertransport 
)
For me, I usually buy the processor at the sweet spot and overclock by up to 10%. This usually gives me a processor one up on the list at a lower price. Usually 10% is very doable on just about any hardware without serious heat/noise/instability penalties
The link you gave is running along a fine edge. Both stock & overclocked setups are running at the same MHz so it's more of a balance with memory timings vs higher FSB. I wouldn't consider that an overclock. 300x10, now, THAT would be an achievement
Since I'm not a benchmark freak, I really don't care about the minute differences between faster FSBs or tighter memory timings. As long as it works....A little faster clock has visible benefits for me as I run Folding@Home 24/7 which is very much dependent on clock speed.
As for my Athlon64, it's turning out to be a nice overclocker. Right now I'm at 15%, looking to pushing it to 20%. That's a lot of overclock and I can do it without getting too hot or speeding up my fans so no noise penalty either
(For those who've been following my o/c attempts, I found out that I forgot to adjust the Hypertransport )-
Green Shoes
- *Lifetime Patron*
- Posts: 477
- Joined: Thu Feb 17, 2005 6:41 am
- Location: Nashville, TN
Unless you're doing heavy numbers-crunching or gaming, it's probably worthless, other than a hobby to tinker around with. If you're trying to run Half-Life 2 and it just looks a little choppy, with a good CPU & HT overclock you can squeeze out another 10-15 frames, which can be a pretty huge difference (maybe 10 frames from the CPU and 5 from the RAM). But other than doing it "just 'cause", that's the only real reason I can think of.
But then, I'm just one of those "just 'cause" people
But then, I'm just one of those "just 'cause" people
It really depends
Much of what I think has already been stated above.
The problem is that overclocking and quiet computers aren't the most compatible concepts in the world.
I've had CPU's that I was able to overlock by a full 1 ghz, but the noise required to cool it isn't something that's worth it in my opinion.
I've gone the whole spectrum from air-cooled to liquid cooled to phase-change cooled and back to air. If I could find a way to render my Vapochill unit silent (hard to do and still get sufficient airflow to make the system work properly), I would use it again, as it is now, it sits in a corner.
BUT, I can say that the difference between my old system using the Vapochill (an Athlon 2500+ Mobile chip) that I had oc'ed to run at 2.6 ghz and my current system utilizing an Athlon64 3000 at stock speed (1.
is negligible.
You may notice the difference in benchmarks, but in real-world use, you probably would not notice it (especially when you're talking about minimal overclocking).
Now, if you compare the performance of my overclocked Athlon 2500 and it's stock speed, the difference is drastic, but you can currently build an Athlon 64 system that will outperform it for much less money, and much less hassle and noise from the cooling required to achieve such high overclocks.
I don't know if any of that made sense, just kind of stream of semi-consciousness rambling.
The problem is that overclocking and quiet computers aren't the most compatible concepts in the world.
I've had CPU's that I was able to overlock by a full 1 ghz, but the noise required to cool it isn't something that's worth it in my opinion.
I've gone the whole spectrum from air-cooled to liquid cooled to phase-change cooled and back to air. If I could find a way to render my Vapochill unit silent (hard to do and still get sufficient airflow to make the system work properly), I would use it again, as it is now, it sits in a corner.
BUT, I can say that the difference between my old system using the Vapochill (an Athlon 2500+ Mobile chip) that I had oc'ed to run at 2.6 ghz and my current system utilizing an Athlon64 3000 at stock speed (1.
is negligible.You may notice the difference in benchmarks, but in real-world use, you probably would not notice it (especially when you're talking about minimal overclocking).
Now, if you compare the performance of my overclocked Athlon 2500 and it's stock speed, the difference is drastic, but you can currently build an Athlon 64 system that will outperform it for much less money, and much less hassle and noise from the cooling required to achieve such high overclocks.
I don't know if any of that made sense, just kind of stream of semi-consciousness rambling
.-
SometimesWarrior
- Patron of SPCR
- Posts: 700
- Joined: Thu Mar 13, 2003 2:38 pm
- Location: California, US
- Contact:
When a hardware site does a review of an OC'ing motherboard, they often use the fastest memory available on the market, combined with the fastest CPU. If an individual tries to replicate the overclock in the review, they also need good components, or they will not be as successful.
I've had good experience with overclocking. I ran the Celeron 300A at 450MHz (50% overclock), a Duron 600 at 750MHz (25% overclock), and a Celeron 1.2 at 1.45GHz (20% overclock). Certain programs, like MP3 encoders, video encoders, or distributed-computing programs like Folding@Home, are almost entirely CPU-dependent, so they speed up appreciably. Games also depend on the video card, so they might speed up at half the rate of the CPU. A few programs (WinRAR?) are also heavily dependent on memory speed (bandwidth, latency, or both).
Let's say a 25% CPU overclock buys you 17% better frame-rates in Half-Life 2. If you have a fast video card, this is about what you should expect, according to another Xbit-labs article on CPU performance in HL2. That brings you from 70FPS to 82FPS, or 30FPS to 35FPS. That 30->35 boost can be noticeable, especially if you have vertical sync enabled on your screen. V-sync keeps the screen from tearing, but it also rounds down the FPS to the next-greatest factor of your screen refresh rate (ex: 100->50->33.3->25->20...), so going from 30 to 35 in the benchmark is like going from 25 to 33 when you are actually playing, and that is definitely noticeable.
You don't always have to sacrifice quietness for overclocking. I overclocked a 1.8GHz Athlon 64 90nm to 2.3GHz with the stock heatsink. This was doable even with PC2700 RAM, by telling the system to run the RAM as if it was PC2100, and then bringing the bus speed up from 200MHz to 255MHz. Because the computer is behind a desk and muffled, and because the fan speed is thermally controlled with SpeedFan, the computer is inaudible at idle no matter what the clock speed of the processor. When the system is used for gaming, the sounds of the game drown out any increase in fan noise.
I've had good experience with overclocking. I ran the Celeron 300A at 450MHz (50% overclock), a Duron 600 at 750MHz (25% overclock), and a Celeron 1.2 at 1.45GHz (20% overclock). Certain programs, like MP3 encoders, video encoders, or distributed-computing programs like Folding@Home, are almost entirely CPU-dependent, so they speed up appreciably. Games also depend on the video card, so they might speed up at half the rate of the CPU. A few programs (WinRAR?) are also heavily dependent on memory speed (bandwidth, latency, or both).
Let's say a 25% CPU overclock buys you 17% better frame-rates in Half-Life 2. If you have a fast video card, this is about what you should expect, according to another Xbit-labs article on CPU performance in HL2. That brings you from 70FPS to 82FPS, or 30FPS to 35FPS. That 30->35 boost can be noticeable, especially if you have vertical sync enabled on your screen. V-sync keeps the screen from tearing, but it also rounds down the FPS to the next-greatest factor of your screen refresh rate (ex: 100->50->33.3->25->20...), so going from 30 to 35 in the benchmark is like going from 25 to 33 when you are actually playing, and that is definitely noticeable.
You don't always have to sacrifice quietness for overclocking. I overclocked a 1.8GHz Athlon 64 90nm to 2.3GHz with the stock heatsink. This was doable even with PC2700 RAM, by telling the system to run the RAM as if it was PC2100, and then bringing the bus speed up from 200MHz to 255MHz. Because the computer is behind a desk and muffled, and because the fan speed is thermally controlled with SpeedFan, the computer is inaudible at idle no matter what the clock speed of the processor. When the system is used for gaming, the sounds of the game drown out any increase in fan noise.
SometimesWarrior, thanks for the illustration. I am a little disappointed with HL2 performance (more specifically, Counter-Strike), but I just began tinkering with overclocking last night. I have a feeling a gig of RAM would help more than overclocking, but I'll try it out soon.
The build in my sig is my "gaming" system, and honest to god it is inaudible when I sit at my desk. Only at night, and with concerted effort, can I hear a faint hum. I couldn't resist seeing what it can do in terms of overclocking since Asus makes it so easy to do. I have it running at 3.36GHz, which is a 20% overclock, 3:2 ratio, and my Corsair timings are 2-2-2-8. The Zalman cooler is at its slowest speed 1450 rpms. My processor gets to 47° after 30 minutes of Quake 3. Without the overclock, I can play this game for 3 hours straight and the CPU temp never goes past 44°. This system is stable thus far, and I cannot hear it at all unless I get down under the desk and put an ear to the side, within a foot.
My experience with Quake 3 and Doom 3 indicate that overclocking really do not matter. 2.8GHz is probably enough for Doom 3, and it is definitely enough for Quake 3. Both games are more dependant on the graphics card anyway. I don't even think the low memory timings have an effect. Both games seemed fine to me when the timings were 2.5-4-4-8. I guess it all depends on the application or game. I'm sure mp3 compression would be faster (noticably faster perhaps?), but I don't encode mp3s on this particular PC.
Not to hijack this thread, but I want to add a question. I have read that the FSB:Memory ratio is best when left at 1:1, but the Pentium "C" processors do well at overclocking when this ratio is mixed. Does this type of overclocking negate the benefit of doing so? For instance, is my 3.36GHz at 3:2 really performing better than the stock 2.8GHz with a ratio of 1:1? I If a mixed ratio slows the memory in order to speed up the rest of the PC, are gains really being made? A little enlightenment on this would be appreciated.
The build in my sig is my "gaming" system, and honest to god it is inaudible when I sit at my desk. Only at night, and with concerted effort, can I hear a faint hum. I couldn't resist seeing what it can do in terms of overclocking since Asus makes it so easy to do. I have it running at 3.36GHz, which is a 20% overclock, 3:2 ratio, and my Corsair timings are 2-2-2-8. The Zalman cooler is at its slowest speed 1450 rpms. My processor gets to 47° after 30 minutes of Quake 3. Without the overclock, I can play this game for 3 hours straight and the CPU temp never goes past 44°. This system is stable thus far, and I cannot hear it at all unless I get down under the desk and put an ear to the side, within a foot.
My experience with Quake 3 and Doom 3 indicate that overclocking really do not matter. 2.8GHz is probably enough for Doom 3, and it is definitely enough for Quake 3. Both games are more dependant on the graphics card anyway. I don't even think the low memory timings have an effect. Both games seemed fine to me when the timings were 2.5-4-4-8. I guess it all depends on the application or game. I'm sure mp3 compression would be faster (noticably faster perhaps?), but I don't encode mp3s on this particular PC.
Not to hijack this thread, but I want to add a question. I have read that the FSB:Memory ratio is best when left at 1:1, but the Pentium "C" processors do well at overclocking when this ratio is mixed. Does this type of overclocking negate the benefit of doing so? For instance, is my 3.36GHz at 3:2 really performing better than the stock 2.8GHz with a ratio of 1:1? I If a mixed ratio slows the memory in order to speed up the rest of the PC, are gains really being made? A little enlightenment on this would be appreciated.