Team overclocks Core 2 Quad to 5.1GHz, claims world record -- too bad it's not
By Nilay Patel 
posted July 31st 2008 5:02PM
posted July 31st 2008 5:02PM
[Via PC World]
posted July 31st 2008 5:02PM
A look back on popular stories from today in a specific year.
Now that we've thrown 'em off the trail, use the form below to get in touch with the people at Engadget. Please fill in all of the required fields because they're required.
Not bad for a fail.
overclock the world!
zzZZZz
that's ok, they still have the biggest e-peen even if they're wrong about everything
clock speed means nothing engadget, we thought you would know that by now. Obviously this quad core processor at 5.1GHz is faster than a P4 at 8.18GHz
Precisely.
Yes, but I think Nilay was referring to the highest clock speed record instead of the fastest CPU record.
"Obviously"...
It isn't faster. It's more efficient.
While the "wheels" concept is somewhat true, it is more accurate to use the analogy of trying to move from one apartment to another with one U-Haul supercharged to go 100 MPH or 4 U-Hauls that can only go 60MPH each.
If you only have stuff to fill one truck, then the 100MPH truck will get you moved faster because of less travel time. But if you have enough crap to fill 4 tricks, then you'll get done faster with the 60MPH trucks because you only have to make 1 trip as opposed to 4.
IE. In real world application of an OS that can take advantage of multiple cores, you will see improved performance in the 5.1GHz Q6600 over the 8Ghz P4. But only under those conditions.
Correct me if I am wrong(which i could be very wrong) but the P4 is a single core at 8.18. the core 2 extreme is a dual core each at 6. this is a quad each at 5.1. so isnt this processor faster in the sense that you could say it is 20.4 ghz?
Again I am probably wrong.
no. car moves at 10 mph, each wheel goes at 10mph, you dont ad up the speed of the wheels and say its going at 40mph... in the same way, the chip is still operating at 5.1ghz, there are just 4 cores, all operating at that speed.
but if a thread(process) could be divided into 4 parts, would not this processor process faster than the P4?
No.
It isn't one core at 20.4GHz.
It's 4 cores at 5.1GHz.
If it can't run a single threaded program at faster than 8.18GHz then it's not faster than 8.18GHz. Single threadeds work on 1 core, i.e. one of these 5.1GHz cores, so no matter how many individual cores they have it's not faster than 5.1GHz.
It's like wheels on a car. Cars have 4 wheels, not 1 big wheel. You can use them together to make a single job easier, but at the end of the day it's still 4 average wheels, not one big one.
ok, thankyou for the clarification, i knew there would be people on this site with some good insight. So basically multicore processors are better for multi threaded programs, like 3d work.
Hamidxa below is probably right in saying that it's the fastest overall performance, but in terms of single core speed it's not.
You *can* divide the process into 4 threads and then run it faster (although 4 5.1GHz won't be the same as a single 20.4GHz because of how it splits it and puts it back together), but like h0rk above said, you can only compare like with like. 4 wheels at 10mph aren't the same as one wheel at 40.
Even if that is what they were claiming (which is silly), the Skulltrail platform also uses Quad-core processors, not dual core.
Both of the earlier responses are totally true...
But if you had RTFA, you would know that they disabled two of the cores because that apparently allows them to attain higher clock frequencies...
The 9775 is a Quad Core.
Kamokazi: ergh, you're right, scrap my comment about Hamixda.
BlowURmindBowel: haven't RTFA, but if that's true.... Yeah, not quite what it's made out to be then!
Lol. Ive been waiting for someone to start this arguement on engadget for a while to start to laugh an noob responses. The car with 4 wheels thing dosen't work. Simply because, if you have a 4 wheel drive car, then the engine would still output the same as a 2 wheel drive car. So the same amount of energy is transformed into forward energy (or flops in terms of computers).
Also, IMO, benchmarking on the basis of clock speed is silly. Especially comparing a P4. A P4 extreme (used to hit 8.18GHz) is 4GHz by stock. So thats about a 110% increase. This overclock on a Q6600 is more like 140%.
So, this thing could run 4 different processes at 5G each, where the single core 8G P4 would be stuck prioritizing the four on one core and achieve less than the slower clocked quad in the same amount of time? Or something like that?
Not to get off track but some of us are forgetting that the P4 was a different platform. The P4 has longer piplines and required those electrons to move faster(ie 4.5Ghz). Intel has gotten smarter since AMD was kicking their ass and they decided to shorten up the pipline and other tings.
this is nice. over 5Ghz is awsome.
i would love to see some benchmarks to this.
I also can wait for an OS system that can fully utilize multicore cpus and use them in a way that they can be split or segregated, As well as a fully 64bit programs. The hard ware is here people and can be upgraded, lets hope the programmers can keep up.
People commonly make the mistake of thinking multi-core systems only benefit from multi-threaded apps. This is incorrect. While it is true that a single threaded application will run the same regardless of the number of cores the system, that is only relevant if that is the ONLY application running. Say you run an intensive application and it executes on one of the cores then you launch 2 or 3 other apps, those are balanced by the OS and executed on the other cores, leaving your original application to suck all the juice from the forst core untouched. There is still a HUGE advantage to run multi-core even if you use only single-threaded apps. Anyone can run multiple instances of CPU Burn-in to test this out for themselves.
With liquid nitrogen anything is possible!
Nilay you moron,
A C2Q at 5.1 Ghz is clock for clock and overall much much faster than a P4 @ 8 GHz.
So in terms of pure frequency this may not be the fastest "clocked" per se, but in terms of relative performance, this is a world record.
The point being of course, all clock frequencies are not created equal.
The amount of cycles per second depend on how the CPU architecture as a whole handles instruction sets, and the complexity of those instruction sets.
To that end, the way a P4 attains its frequency is different than the way a C2Q attaints its frequency, and as such, the argument boils down to one of apples versus oranges.
On that note, this can still easily be argued as the fastest "clocked" processor in the world.
I mean, hell, would you prefer seeing a 386 ramped up to 10 Ghz and then argue otherwise?
Of course not, that would be ludicrous.
It might be a world record for the Q6600, but it's not overall because of the QX9775.
ijyt,
Certainly.
However, Engadget's (Nilay in this case) point of drawing comparisons to a P4 @ 8 Ghz as a means of discrediting this achievement is spurious at best.
Engadget editors should know better, that slapping a "GHz" rating on a processor is meaningless in the grand scheme of things when the speed of a processor really depends on how the architecture as a whole handles instructions/sets, rather than just what frequency its clocked at alone.
It was really inane to reference a P4 8Ghz once again, because anybody who knows anything about the performance of the C2Q line when compared to any of the P4 line of processors will be able to tell you that even in single-threaded apps, that C2Q @ 5.1 Ghz will trump the 8 GHz P4 due to numerous variables associated with the design of the chip, and not solely the frequency itself.
But clock speed is one of the problems for overclocking; getting more throughput isn't. You don't run into problems because of too much data, you run into problems because your device overheats, or because somewhere, some transistor doesn't switch quite fast enough, the next transistor thinks it 'said' one not zero, and it reads/writes wrong data, reads/writes to the wrong place, does the wrong instruction, or just plain fails (some internal failures don't map cleanly to external characteristics).
So while it's quite possibly true that you get more throughput with this rig, that's not the best metric for overclocking success.
P4's come stock at about 4ghz, so of course they will be higher.
I agree, it may be a world achievement for a Q6600 which Tom's Hardware should be recognized for and what they probably were GOING FOR NILAY...
Engadget writters need better writting prose. In any case, yes each processor is different. You can compare a P4 to a Q or QX processor, they just arn't the same. What about cache, CPU instructions, architecture, bus speed, and the like? Its like comparing a Geo Tracker that has a supercharger with 400 horsepower to a mustang with the same horsepower. You just can't hahaha. Nilay...XD
What is LNO2? Are you trying to say Liquid Nitrogen? Liquid nitrogen is just N2; it's not a dioxide. Maybe you meant LN2?
Ok, so I understand that liquid nitrogen would dissipate any heat instantly, but wouldn't this also slow the electrons flowing through the board to a near crawl?
When you put it through your engine, thats a mechanical process, the air passes through the intake into the cylinders where it is supercompressed due to its density.
But on electronics? This is close to the method they used to slowing down a lightbeam due to the electrons in the particles becoming nearly frozen.
Somebody please explain this....
hey guys, i've invented black hole cooling! you just flip this switch like this to activate the wormhole and oH SHI-----
The electrons being frozen is utter nonsense at least at those relatively high temperatures. Superconductors work only at very low temperatures and have electrons flowing through them no problem ( wait even better than at room temperature ). To see a quantum effect on the electron you need to go to temperatures way lower than this then you get to see what is called a bose-einstein condensate.
Yes it would slow them down but not nearly enough to make a difference.
Yes, its the bose-einstein condesate that was used to slow down light particles. But even the slightest change in temperature would surely have some effect on the electrons.
Its just basic science that heat excites (speeds) up electrons and cold starts to slow them down. I'm not saying they are frozen, I'm just saying at -129C or whatever the temperature was, won't that be inhibiting the electrons somewhat as opposed to an alternative like speeding them up without dropping the temp on the CPU?
Well all the things you learn in thermodynamics about molecules or atoms slowing down at lower temperatures is not quite applicable to electrons. Electrons are inherently quantique in nature and they don't move they rather tunnel. As said you need very very (a fraction of a Kelvin) low temperature to see effects on electrons. As for the bose einstein condensate being used to slow down photons it's quite the other way around since they use a technique called laser cooling where laser beams selectively slow the atoms that are moving along their axis using the doppler effect. For more information take a physical chemistry course.
@Lowest Ranked
Solid state physics is a bit more complicated than that. I actually have a test on the subject tomorrow, and the over-simplified answer is that there are two primary mechanisms that determine conductivity in semi-conductors. One is due to impurity scattering which is proportional to temperature and the other is lattice scattering which is inversely proportional to temperature. Up until a certain temperature, which depends mainly on the material used and the doping, the conductivity increases until lattice scattering is the dominant mechanism and then conductivity decreases. The impurity scattering is the simple idea everybody knows (well, part of it is anyway) that as temperature increases you have more electrons moving around and so the material can conduct electricity more efficiently. But at a certain temperature not many more electronics become ionized as the temperature goes any higher, and at this point it just causes the lattice - the grid of atomic connections making up the material - to vibrate more violently and be more likely to interact with the ionized electronics. This causes the conductivity to decrease. This model works well for silicon from around 200-500 degrees Kelvin, but at higher temperatures this model breaks down the conductivity skyrockets due to an increase in the intrinsic electron levels.
If you're really interested in learning more, Berkeley posts their lecture notes for an introductory course online. That's not the course I took, but my school requires a password. I hope this helps - it's pretty difficult to explain this stuff at all let alone without diagrams.
http://www-inst.eecs.berkeley.edu/~ee105/sp08/
So you mean "anything is [NOT] possible" with liquid nitrogen. Damn I shoulda wiki'd that crap.
This is actually a world record for the Q6600.
Team Italy doesn't hold the highest frequency record anymore too, someone beat them by 1Mhz. :)
http://valid.x86-secret.com/show_oc.php?id=218928
Light speed to slow, go straight to Ludicrous Speed. Ludicrous Speed, GO!
That was @Hamidxa, Damn you engadget comment system...
Spaceballs :D.....greatest skit movie, ever!
The word "record" can be quite vague. It's a world record for this particular chip on this particular motherboard to reach this particular speed. :)
I'd say they have the record for the quad core.
LNO2? That would mean liquid nitrogen dioxide, which would be extremely dangerous and no good for cooling. You want liquid nitrogen.
Hate to dissapoint (bah, who am I kidding, I love it), bu tthat's a Q66. It's twice the processor yours could ever be.
Agree, aside from forums their site has become useless. Boo-urns tomshardware...
This is a world record. The highest clocked Core 2 Quad. Seriously. What were you thinking when you said it wasn't a world record?
That a Core 2 Extreme had been clocked higher just several months before?