i5-3570k : not bad but a little disappointed...

[LagunaX]

Looks likes there is another U.K. review mirroring my results:
http://forums.aria.co.uk/showthread....w*-56K-WARNING

[OneMoar]

Originally Posted by Aquinus

Everyone seemed to talk down to me when I said to everyone that IVB will generate more heat than SB and was going to be a major issue for over-clocking. Wait, what is that? The IVB i5 runs hotter than my SB-E at a lower overclock? I was hoping I was going to be wrong, but physics doesn't lie. Smaller circuitry = more heat even if the CPU uses less power.

that depends on the voltage you are at Smaller circuitry = less Resistance = less voltage/amperage needed = lower heat

[de.das.dude]

One of my Friends has an i5 2500K which died twice in 6months. he was running @4Ghz, stock volts

[LagunaX]

Been running 24/7 4.8ghz he must of had a bad psu overvolting the chip or inadequate stock cooling.

[de.das.dude]

Originally Posted by LagunaX

Been running 24/7 4.8ghz he must of had a bad psu overvolting the chip or inadequate stock cooling.

nope. he is a reviewer, so cant accuse him of those stuff. he checked it himself. this will be the second time he will be RMAing

[OneMoar]

Originally Posted by de.das.dude

nope. he is a reviewer, so cant accuse him of those stuff. he checked it himself. this will be the second time he will be RMAing

hes doing something wrong or his board is defective
dirty power could also be to blame

[Aquinus]

Originally Posted by OneMoar

that depends on the voltage you are at Smaller circuitry = less Resistance = less voltage/amperage needed = lower heat

Actually smaller circuitry has more resistance because there is less area for the charge to move, therefore increasing current density, which determines how much heat is generated. According to Ohm's Law, I = V / R, where I is current, V is voltage, and R is resistance, in a circuit where resistance doesn't change (ideal fully loaded circuit,) R becomes a constant (which isn't true, because hot metals conduct electricity less efficiently than cold metals) so any increases to voltage will increase amperage.

So lets say you have a CPU that runs stock at 1.30v (for an easy number,) with some static resistance R, that has a TDP of 90 watts. The calculated current for the CPU would be P = I * V, since we're solving for current, we get P / V = I, 90-watts / 1.3 volts = 69.2 Amps. Using the amperage we can calculate the theoretical loaded resistance of the CPU (this is all very theoretical, a lot doesn't work this way, but it will give an idea for numbers.)

Since I = V / R, we can reformulate that to be R = V / I which would be R = 1.30v / 69.2 Amps = 1.88x10^-2 Ohms. Using this resistance we can calculate the current of an over-clock at 1.42 volts.

I = 1.4v / 1.88x10^-2 = 75.5 Amps which is 6.3 amps more, a 10% increase in current.

According to Joule's first law, Q is directly proportional to (I^2)*R. So heat increases exponentially as current increases. So if you double the current, heat quadruples. You triple the current, heat increases by a factor of 9.

So increasing the voltage on our pretend CPU by 9% just increased your heat generated by 19%. Smaller circuitry has the benefit of using less power, but has the adding issue of generating more heat because of the smaller circuitry. The exponential heat problem starts earlier for IVB because it is already running hot because of the smaller circuitry.

IVB's lower stock voltage (and in turn lower current,) is what makes IVB use less electricity, not the smaller circuitry (which actually diminished some of that benefit.)

[radusorin]

Didin't this hole small die less heat less power thing worked till now ? Or is it from now on ?

starting to get confused

[Vulpesveritas]

@radusorin It is still beneficial due to the voltage decrease, and it is projected down into the single nm for silicon I believe. After that they may have to turn to other mediums such as carbon nanotubes.
The 3D transistors may also be contributing to heat, although that is merely speculation on my part.

[OneMoar]

Originally Posted by Aquinus

Actually smaller circuitry has more resistance because there is less area for the charge to move, therefore increasing current density, which determines how much heat is generated. According to Ohm's Law, I = V / R, where I is current, V is voltage, and R is resistance, in a circuit where resistance doesn't change (ideal fully loaded circuit,) R becomes a constant (which isn't true, because hot metals conduct electricity less efficiently than cold metals) so any increases to voltage will increase amperage.

So lets say you have a CPU that runs stock at 1.30v (for an easy number,) with some static resistance R, that has a TDP of 90 watts. The calculated current for the CPU would be P = I * V, since we're solving for current, we get P / V = I, 90-watts / 1.3 volts = 69.2 Amps. Using the amperage we can calculate the theoretical loaded resistance of the CPU (this is all very theoretical, a lot doesn't work this way, but it will give an idea for numbers.)

Since I = V / R, we can reformulate that to be R = V / I which would be R = 1.30v / 69.2 Amps = 1.88x10^-2 Ohms. Using this resistance we can calculate the current of an over-clock at 1.42 volts.

I = 1.4v / 1.88x10^-2 = 75.5 Amps which is 6.3 amps more, a 10% increase in current.

According to Joule's first law, Q is directly proportional to (I^2)*R. So heat increases exponentially as current increases. So if you double the current, heat quadruples. You triple the current, heat increases by a factor of 9.

So increasing the voltage on our pretend CPU by 9% just increased your heat generated by 19%. Smaller circuitry has the benefit of using less power, but has the adding issue of generating more heat because of the smaller circuitry. The exponential heat problem starts earlier for IVB because it is already running hot because of the smaller circuitry.

IVB's lower stock voltage (and in turn lower current,) is what makes IVB use less electricity, not the smaller circuitry (which actually diminished some of that benefit.)

BUT
all the goes right out the window when we start talking about super conductive materials
won't be to far off down strange haxoring starts to occur when you get down below 10nm

[theoneandonlymrk]

Originally Posted by Vulpesveritas

The 3D transistors may also be contributing to heat, although that is merely speculation on my part.

that would make a bit of sense as there is more surface area to the gate so it should dissipate heat quicker, but the shorter interconnect lentgh and lower gate operateing power normally mitigates this ,its a new tech and its not like its worse

[Aquinus]

Originally Posted by OneMoar

all the goes right out the window when we start talking about super conductive materials
won't be to far off down strange haxoring starts to occur when you get down below 10nm

Except...
A: CPUs aren't made of super conducting material.
B: 10nm has bigger issues than this, chip manufacturers will have to figure out how to get around the quantum tunneling problem before they get past 16nm.

[radusorin]

Originally Posted by Aquinus

Except...
A: CPUs aren't made of super conducting material.
B: 10nm has bigger issues than this, chip manufacturers will have to figure out how to get around the quantum tunneling problem before they get past 16nm.

So for now, as long as the materials remain the same the maximum size will be 16nm?

@Vulpesveritas the same thought haunts my mind. Well at list SB (no 3D) less heat than IVB (whit 3D) and till now at list it seems to be a performance increase as well. I mean apparently a 4.8 IV can keep up whit a 5.0-5.2 SB.

[Sinzia]

Sounds like hardcore overclockers might want to wait for the next stepping then, I had high hopes for OCing out of the gate on this one.

[Vulpesveritas]

It would be funny if this will give AMD an enthusiast edge for OC this year.... eh fanboys can hope, right? Then again the resonant clock mesh may cause a similar issue, and we won't know till Trinity.
On topic though, yeah it's seeming about a 10% IPC increase for Ivy. So it's not all bad news, though I wonder how it does under higher volts.

[Random Murderer]

Originally Posted by radusorin

So for now, as long as the materials remain the same the maximum size will be 16nm?

@Vulpesveritas the same thought haunts my mind. Well at list SB (no 3D) less heat than IVB (whit 3D) and till now at list it seems to be a performance increase as well. I mean apparently a 4.8 IV can keep up whit a 5.0-5.2 SB.

except that clock for clock SB-E is outpacing IVB. IVB-E is going to be a real wild card in terms of performance, heat, and overclocking, but we'll have to wait until they're released to know for sure how they stack up to IVB, SB, and SB-E in those aspects.

[OneMoar]

It won't be long before they START using Super conductors

[Vulpesveritas]

@onemoar: Or at least get creative. I believe Intel is trying germanium alloy and patonting it for use in processors, while I remember something with IBM and carbon nanotubes.
I don't believe either is a superconductor at high tempratures. So eh.

[Aquinus]

Originally Posted by Vulpesveritas

@onemoar: Or at least get creative. I believe Intel is trying germanium alloy and patonting it for use in processors, while I remember something with IBM and carbon nanotubes.
I don't believe either is a superconductor at high tempratures. So eh.

Carbon nanotubes is to combat the quantum tunneling problem at 16nm and smaller.

[OneMoar]

http://en.wikipedia.org/wiki/Room-te...superconductor
^^^

[Vulpesveritas]

Originally Posted by OneMoar

http://en.wikipedia.org/wiki/Room-te...superconductor
^^^

Right well... I don't see carbon nanotubes or germanium on that list. lol

[BrooksyX]

Hmm glad I just went ahead and went the sandybridge. Yes ivybridge seems nice but it doesnt seem like its panning out to be that much better.