What is overclocking?
 

What is overclocking? :question:

Sorry for the double-post. The site automatically told me I couldn't post more than twice every 60 seconds, and I posted to another thread. I assumed it didn't create this new thread.

(feel free to correct or add to the following)

Overclocking means the gain of performance by running certain hardware like processors, memory etc. above the manufacturer's specification. The maximum gain of performance depends on the hardware and the software you run on it. Although some manufacturers give moderate specifications and the hardware stays stable even when heavily overclocked, (especially excessive) overclocking can cause system failures and shorten the life of certain components or the system as a whole. Be aware that usually you lose warranty by overclocking!

Benjamin

Can you force your computer to overclock?

Normally, yes. Either by settings in BIOS or by setting switches on the motherboard.
Not really recommended unless you enjoy spending a lot of time fiddling with hardware setup, and is prepared to buy the occasional new CPU when you burn out the old one by accident.
That is, for enthusiasts only :)

[QUOTE=mephisto]Normally, yes. Either by settings in BIOS or by setting switches on the motherboard.
Not really recommended unless you enjoy spending a lot of time fiddling with hardware setup, and is prepared to buy the occasional new CPU when you burn out the old one by accident.
That is, for enthusiasts only :)[/QUOTE]

Is there a how to guide available?

Citrix
:cool: :cool: :cool:

Sort of. There are plenty of details that differ between CPU's and motherboards, and some systems cannot be overclocked at all.
Still, these links may serve as a starting point:
[url]http://cahtech.beigetower.org/howto/misc.php[/url]
[url]http://arstechnica.com/tweak/oc_cooling.html[/url]
[url]http://www.tomshardware.com/howto/20000808/[/url]

Overclocking is like running a car at speeds exceeding the proposed default speeds of its manufacturer. Substitute car for CPU and speed for frequency (MHz).

Overclocking in modern personal computers (introductory text)
 

Overclocking in modern personal computers (introductory text)

Written June 2004. Public domain. May be redistributed and modified without restrictions.

Overclocking is like running a car at speeds exceeding the proposed default speeds of its manufacturer. Substitute car for CPU and speed for frequency (MHz). It is also possible to overclock the main memory (RAM) and the video card's GPU.

A modern personal computer has a a northbridge chipset, a memory controller, a central processing unit (CPU), some amount of main memory (RAM) and a video card which contains a graphics processing unit (GPU). The memory controller may be either in the northbridge chipset (this is the case of 32bit Intel CPUs) or inside the CPU (this is the case of AMD 64bit CPUs).

The memory controller communicates with the main memory (RAM) by the means of a front side bus (FSB). The FSB has a frequency, usually about 200MHz in modern PCs. The internal frequency of a CPU is defined by the FSB frequency multiplied by a number, called multiplier, which is usually around 10 or 12 in modern PCs. The FSB frequency advertised by the marketing departments of hardware manufacturers is usually more than the real FSB frequency because of some special technologies employed to increase the performance of the FSB, like DDR (double data rate). Common advertised FSB frequencies are 400MHz, 533MHz, 800MHz and 1GHz, however the real hardware frequency is usually in the 133MHz-200MHz range.

In many systemboards it is possible to change the FSB frequency and the multiplier of a PC. This ability is employed by overclockers to change the frequency of their CPUs. There are two common ways to do this: Either by increasing the multiplier or the FSB speed, or both. Some also increase the electrical voltage applied to the CPU.

Overclocking can lead to increased temperatures which may lead your CPU, main memory, systemboard or video card to die or become unstable (more system crashes).

Many CPU manufacturers lock their CPUs to work only with certain multipliers. This is the case of both AMD and Intel CPUs, but some of their models marketed to enthusiasts don't have a multiplier lock.

Overclocking may make you be in the need of more powerful system cooling solutions, such as watercooling. It may also cancel your computer's warranty.

I was thinking that since overclocking barely helps you would lose more cycles posting about it and running overclocking software than gain by the tiny speed improvement.

12% is nothing to sneeze at. 2 more L-L's a year for a machine.

[QUOTE=clowns789]I was thinking that since overclocking barely helps you would lose more cycles posting about it and running overclocking software than gain by the tiny speed improvement.[/QUOTE]
It depends how long you run your machine for. A case in point is my venerable Pentium Pro system. It is rated as a PPro-200 and was shipped as such in 1996.

I overclocked it to 233MHz, where it has been ever since. It's still giving good service eight years later and is about 5-8% faster than the standard 200MHz system. As it has been running very nearly 24/7 for those eight years I have, in effect, got an extra six months computation for no additional cost.


Paul

Most overclockers buy their hardware with the intention of overclocking. Some do it to save money and some do it for fun(i fall in here). Most OCers dont buy according to the model numbers, they buy according to the core used in a processor.

Most processors are exactly the same(accept for their clock speed). For Example, the AMD Athlon 2500+, 2600+, 3000+, and 3200+ are all build on the Barton core. Barton is build on the 130nm process and features 512K of L2 cashe. Barton will overclock to 2.3ghz on average (my barton does 2.5ghz and it isnt rare for a moble barton to hit 2.8ghz). there is a sizable difference in price between the 2500+ ($80) and the 3200+($200'ish? It was $400 when i bought my 2500+), but they both OC roughly the same.

Here is another example for the P4. When the northwood core came out some time ago, many Overclockers rushed out and bought the cheapest northwood P4 available, the 1.8ghz version. Because this processor had the same core as its faster (and more expensive) brothers, it wan't uncommon for it to reach 2.8ghz. That's 156% of the rated performance!

[QUOTE=xilman]in effect, got an extra six months computation for no additional cost.[/QUOTE]
Well...eclectricity...

:wink:

well overclocked CPU doesn't take much more power than normal CPU

i suppose

How different are the Prime95 run times on not overclocked and clocked PC's?

Citrix
:cool: :cool: :cool:

Also, do one of you who know how to get your processor to OC happen to be my neighbor ?;o) I see some links but when it comes to messing with my hardware you can equate me with any combination of the following smileys: :no: :ernst: :censored: :huh: :help:

The simplest form of overclock your cpu is to adjust some bios
parameters. Most MB allow you to change FSB in a range. You
can inch the FSB up till the system becomes unstable(watch
your way out. make sure your MB has a botton to reset setting).
Your memory must be able to take the frequency or it must be
run async to the FSB. Note memory, AGP and PCI are all run on
the same FSB although to a ratio. OC friendly MB locks AGP/PCI
so they are not affected by FSB.

Pumping up multiple cant not be done on P4. They are all locked.
Early AMD XP are unlocked but not any more. I heard mobile XP
and opteron are unlocked but I dont own any of them. For multiple
unlocked CPU, you can change it by bios setting or jumpers.

I am lucky to have an early TB XP1700+ which is multiple unlocked.
Too bad my GByte MB is not very oc friendly. I ended up increasing
the multiple by 1 and run it as xp1900. I had to up its vcore by .25v.
It has been 18 months.

Recently I bought a $69 XP2500 Barton and ECS N2U400 combo from
Fry's. With stock voltage and cooling, I was able to up its FSB from
133 to 200. So it runs like a XP3200. Too bad my memory is PC2700
(333MHz). Cant move it up. I am overjoyed with the result.

Some MB come with software to change FSB. That wont cause your
board lock but the setting doesnt survive power cycle.

XP2500+ overclocked to XP3200 bench:

AMD Athlon(tm) XP 3200+
CPU speed: 2205.04 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE
L1 cache size: 64 KB
L2 cache size: 512 KB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
L1 TLBS: 32
L2 TLBS: 256
Prime95 version 23.8, RdtscTiming=1
Best time for 384K FFT length: 28.553 ms.
Best time for 448K FFT length: 32.939 ms.
Best time for 512K FFT length: 35.188 ms.
Best time for 640K FFT length: 46.606 ms.
Best time for 768K FFT length: 55.702 ms.
Best time for 896K FFT length: 66.663 ms.
Best time for 1024K FFT length: 74.374 ms.
Best time for 1280K FFT length: 99.364 ms.
Best time for 1536K FFT length: 117.043 ms.
Best time for 1792K FFT length: 143.913 ms.
Best time for 2048K FFT length: 157.219 ms.

[QUOTE=MrHappy]Well...eclectricity...

:wink:[/QUOTE]
I have not been able to detect any significant change in the machine's power consumption. I guess there must be some but it's lost in the noise.

Paul

[QUOTE=xilman]I have not been able to detect any significant change in the machine's power consumption. I guess there must be some but it's lost in the noise.
Paul[/QUOTE]

With a P6-200@233 the increase in power consumption should not exceed 8 Watts.

Your 8/365/24/7 use translates that to the cost of about 25 pints of good english beer served in a cozy pub.

Maybe you should reconsider overclocking next time :wink:

[QUOTE=TauCeti]With a P6-200@233 the increase in power consumption should not exceed 8 Watts.

Your 8/365/24/7 use translates that to the cost of about 25 pints of good english beer served in a cozy pub.

Maybe you should reconsider overclocking next time :wink:[/QUOTE]
I think I can afford the cost of an extra three pints per annum. As I said earlier, it's lost in the noise.

Paul

[QUOTE=TauCeti] good english beer [/QUOTE]
That's a contradiction. :wink:

[QUOTE=MrHappy]That's a contradiction. :wink:[/QUOTE]

If you need something to forget the endlessly pouring rain in england, it is "good" :)

Seriously: I loved the "bitter beer" there. There seems to be no such thing here in germany. Maybe it's under embargo like the blue pringles.

[QUOTE=Citrix]How different are the Prime95 run times on not overclocked and clocked PC's?

Citrix
:cool: :cool: :cool:[/QUOTE]

My Athlon's timings are 136% of its stock speed.
My older Athlon's timings are 76% of its stock speed(old motherboard, 100mhz FSB limitation)
My P4's timings are 120% of its stock speed.
My P2's timings are 150% of its stock speed.
My Via/Cyrix's (C3/EDEN) timings are 137% of its stock speed(which is slow to begin with).

Not all computers are created equal.

Just wondering:

Why are personal computer speeds measured in Hz, while supercomputer speeds are measured in FLOPS? Why the diffferent units?

Hz vs. FLOPS
 

A personal computer usually is a uniprocessor machine, that is, it has only one processor (CPU).

A supercomputer is usually made by connecting together thousands of CPUs (usually using standard Gigabit Ethernet or some other networking technology like Infinibad or Myrinet - [url]http://www.myri.com/[/url] ).

Many times the CPUs used for supercomputers do not exceed frequencies of 500MHz or 1GHz. This is because they want stability, low temperature and low electricity usage. However now many supercomputers use standard server CPUs, like AMD Opteron, with speeds starting at 1.4GHz.

Hz is frequency while FLOPS is performance. 1 MFLOPS is one million floating point operations per second. 1 GFLOPS is one billion. One TFLOPS is one trillion.

It would be innapropriate to use Hz for supercomputers. You could say that this supercomputer is 500MHz because it uses 10,000 500MHz CPUs, or you could say it is 5,000,000MHz (5 THz - terahertz) because 10,000*500=5,000,000. In either expression, the information is not very useful for a scientist who wants to run some scientific software on it (mprime? :banana:). The scientist needs to know how fast he/she will get the results. If we express the computer power in terms of FLOPS, that is, in how many mathematical operations it can complete every second, the scientist, who presumably knows how many mathematical operations his/her software needs, can calculate when he/she will get the desired result.

Also, usually many scientists share the same supercomputer at the same time. For example, a supercomputer may serve a hundred scientists at the same time (i.e. running 100 programs at once). If I am not mistaken this is called time sharing. It is much more scientific to measure how much mathematical power each scientist gets, rather than how much frequency.

Remember: Frequency is just one factor in performance. An AMD Athlon at 2.2GHz has the same performance as an Intel Pentium4 at 3-3.2GHz.

I think we should use FLOPS for personal computers too. One might argue that most people do not know about mathematics ("floating point operations per second? what's that man?":w00t: ). However, I am sure that most people also do not know about physics (hertz). :banana: Apple recently used FLOPS to advertise its new PowerPCs. I think the PC industry may start using FLOPS after every desktop-laptop PC is based on a dual-core or dual-processor setup...

So if FLOPS is a more useful measure, why not use it for PCs too? Also, how can I figure out how many FLOPS my PC has?

Thanks for the information

measuring FLOPS for a PC
 

You can measure the FLOPS of your PC by benchmarking. I think SiSoft Sandra can measure the FLOPS.

United Deviced Power Calculator
 

You may try this:

Go to [url]http://www.ud.com/solutions/[/url] and click the POWER CALCULATOR image down the menu at the left.

It says an AMD Athlon at 2GHz has 3 gigaflops (3 GFLOPS) of performance, when used with the United Devices grid software at 100% utilisation.

It gives 3 measurements:

Dhrystone ALU: 8161 MIPS
Whetstone FPU: 2250 MFLOPS
Whetstone iSSE2: 4110 MFLOPS

Don't know what it means...

Hz is just speed, while FLOPS is power.

analogy
the water comes out the pipe at 15 fps vs. the water comes out the pipe at 5 MGD

One may seem fast 15 feet a second, but 5 Million Gallons a Day is a huge amount of flow.

ALU is Areithmetic-Logic Unit inside the CPU, it processes integer mathematical operations and boolean logic. MIPS is million instuctions per second. An instuction may be the addition of two numbers or a logical operation (AND, OR, XOR etc).

FPU is a unit inside the CPU for floating point arithmetic, that is, non-integer numbers. It measures in FLOPS.

Some CPUs, including Intel Pentium 4 and AMD Athlon64, have SSE2 capability. The SSE2 capabilities of a CPU can also be measured in FLOPS, because they handle floating point arithmetic at higher speed than FPU, but they are highly specific (i.e. FPU is general-purpose, SSE2 is for special operations).

So your CPU can compute up to 8161 MIPS when used for integer arithmetic and boolean logic, 2250 MFLOPS when used for general purpose floating point arithmetic, and 4110 MFLOPS when used for SSE2 arithmetic.

Also note than SSE2 accepts 128-bit instuctions, while ALU and FPU are only 32bit, If I recall correctly.

Prime95/Mprime can utilise SSE2 when available. However, it is faster in Intel P4 than AMD Athlon64. AthlonXP doesnt have SSE2.

[QUOTE=Citrix]How different are the Prime95 run times on not overclocked and clocked PC's?[/QUOTE]For Prime95 it is almost only the CPU speed that matters. This means that it can be better to lower the memory/FSB ratio (if memory fails first) so that you can get a higher CPU speed.

And don't forget to run a lot of double-checks to see that your overclock is prime-stable (not all errors are detected by Prime95). On new computers I keep lowering the FSB until I have around 20 matching double-checks.

[QUOTE=jinydu]Just wondering:

Why are personal computer speeds measured in Hz, while supercomputer speeds are measured in FLOPS? Why the diffferent units?[/QUOTE]

Because supercomputers usually have a lot of CPU while pc has 1. And when
you have thousands of cpus, it is not uncommon one or two of them arent
functioning.

[QUOTE=optim]ALU is Areithmetic-Logic Unit inside the CPU, it processes integer mathematical operations and boolean logic. MIPS is million instuctions per second. An instuction may be the addition of two numbers or a logical operation (AND, OR, XOR etc).

FPU is a unit inside the CPU for floating point arithmetic, that is, non-integer numbers. It measures in FLOPS.

Some CPUs, including Intel Pentium 4 and AMD Athlon64, have SSE2 capability. The SSE2 capabilities of a CPU can also be measured in FLOPS, because they handle floating point arithmetic at higher speed than FPU, but they are highly specific (i.e. FPU is general-purpose, SSE2 is for special operations).

So your CPU can compute up to 8161 MIPS when used for integer arithmetic and boolean logic, 2250 MFLOPS when used for general purpose floating point arithmetic, and 4110 MFLOPS when used for SSE2 arithmetic.

Also note than SSE2 accepts 128-bit instuctions, while ALU and FPU are only 32bit, If I recall correctly.

Prime95/Mprime can utilise SSE2 when available. However, it is faster in Intel P4 than AMD Athlon64. AthlonXP doesnt have SSE2.[/QUOTE]

So the 4.110 Gigaflops reading is the most appropriate one to compare with Primenet's 15.145 Teraflops?

[QUOTE=jinydu]So if FLOPS is a more useful measure, why not use it for PCs too? [/QUOTE]
FLOPS means less for a PC than for a supercomputer.
A supercomputer's primary purpose is to crunch (floating point) numbers, so the number of floating point operations it can do per second is of interest.
PC's are used for lots of things, and numbercrunching is typically just a small part of it. A typical PC program does a little computing, reads or writes a little (to disk or database), and does a little screen updating. Hence, FLOPS says less about 'how quick will my program run', since the variation in what PC programs spend their time on are so much greater.
Basically, FLOPS is not important for general PC users - common programs like Word and Internet Explorer do not use floating point operations at all.

But the emphasis on GHz instead is probably mostly the effect of Intel marketing strategy.

A weird thing about my computer is it seems to gain/lose more time than the speed differences would suggest. My computer seems to be running about even when it runs at 80 ms per iteration. But when I get it to run at 76.1 ms per iteratioin for 20 hours, my expected completion date goes down by 4 to 5 hours. That's 24 to 25 hours of work in 20 hours, so a 20 to 25% improvement. I should only expect about a 15 minute gain. Kinda interesting...

[QUOTE=jinydu]So the 4.110 Gigaflops reading is the most appropriate one to compare with Primenet's 15.145 Teraflops?[/QUOTE]
Yes.

The reduced completion time ( 4 or 5 hours ) is the amount of time saved
over the full time left for the test ( weeks/months ) using the slightly faster
time per iteration, not the gain for that one day.

A time savings over long period not per day.

[QUOTE=dsouza123]The reduced completion time ( 4 or 5 hours ) is the amount of time saved
over the full time left for the test ( weeks/months ) using the slightly faster
time per iteration, not the gain for that one day.

A time savings over long period not per day.[/QUOTE]

I don't think so. The remaining time has not gone back up, and it doesn't fluctuate during slower days at all. It seems to behave more like if it assumes you will be running the program at some constant speed, which on my computer seems to be about 78-80 ms per iteration.