Looks like efficiency is up a good 5% too :up:Quote:
Originally Posted by Titan7171
Printable View
Looks like efficiency is up a good 5% too :up:Quote:
Originally Posted by Titan7171
It's actually the other way around.Quote:
Originally Posted by Opteron146
Some impressive numbers indeed. I still think low end bloomfield aka i7 930 would edge it out though. I'm on the fence about building a dedicated media encoding rig and it's between the x6 or the i7 930. It'll all boil down to whichever of these is $200 or lower and how well it competes against the bloomfield. I have a Microcenter near me so I'll be factoring in the fact that I could pick up a i7 930 for $199.99.
Exactly :). The low-k dielectrics prove to really help quite a bit in low rc wire delay and less leakage power(a strong point for Thuban turbo mode implementation!). Practically what AMD achieved with Thuban (same power envelope or better-140W 965BE- while adding 50% more cores along with Turbo Core ability) almost equals intels 45->32nm transition since intel added 50% more cores and more cache and stayed roughly in the same power envelope of 130W the older QC top of the line bloomfiled chip at the same clock had! The difference is that intel expanded L3 cache in Westmere design ,but their cache is inclusive and they just didn't want to go lower with cache/core ratio when compared to Bloomfield. AMD's L3 cache is exclusive spill over cache and increasing it over 6MBs for 6 core chip wouldn't make much difference.Quote:
Originally Posted by Opteron146
So chips with ultra-low-k will run hotter :ROTF: ?Quote:
Originally Posted by OhNoes!
http://en.wikipedia.org/wiki/Low-KQuote:
Replacing the silicon dioxide with a low-κ dielectric of the same thickness reduces parasitic capacitance, enabling faster switching speeds and lower heat dissipation.
dont waste your time Opteron146, this is the same guy that said these chips would be better off with no turbo
One thing it doesn't allow is higher voltages though...
You're talking about two different things. A high leakage chip (leaks power) and therefore runs cooler, whereas the opposite is true for a low leakage chip. It is about efficiency.Quote:
Originally Posted by Opteron146
Also, low heat dissipation = means a hotter chip; no? The heat generated from current has to go somewhere.
Edit: Did you realize I highlighted a particular portion of your post? Don't try to throw blinds here.
That's one argument, the other argument is that the Nehalem's L3 cache is kind of a L2.5 Cache.Quote:
Originally Posted by informal
Each core has its 2MB L3 slice, which could be accessed faster than the other 2MB slices. That's also the reason why overall L3 cache latency is higher for Gulftown, I guess the cache slices are connected through a ring bus. With 6 cores you'll get 2 more stops -> higher latency.
Why is everyone trying to do cpu technology 101 here?..........what is it thats trying to be explained here? This is the same technology as Deneb basically. Please take the Intel tech talk/ somewhere else
High leakage means more power is turned into heat.Quote:
Originally Posted by OhNoes!
High leakage chips are more suited for LN2 because they are leaky transistors and like to use more voltage.
Low leakage is more efficient but less tolerant to voltage and puts out less heat in general.
Will you guys stop argueing and babbling like idiots and acting like you know everything now? I mean everyone.
You seem to be a funny guy .... did YOU see what I quote above and which part was highlighted ?Quote:
Originally Posted by OhNoes!
Quote:
lower heat dissipation.
LowER means less heat dissipation than without low-k.
If you do not know what's heat-dissipation is ... please google it ;)
@crazydiamond:
I guess you are right ...
@BeepBeep2:
Why not ?Quote:
One thing it doesn't allow is higher voltages though...
@ Opteron146
It's pointless to argue with him since he actually believes (shock ensues!) that a high leakage chip runs ,wait for it,wait for it, cooler .The dude just redefined physics laws :p: .
Oh BTW, I guess he didn't hear of AMD Tweaker Edition either :p:. Those were some great clocking high leakage chips,but they ran hot as hell and consumed A LOT of power/heat. That's why AMD never intended to sell them in the first place,they were bad for retail since they were actually a lot less efficient than regular Denebs...
Its gonna be one long intel visit in the AMD section.........Dont they have another new expensive socket coming out that they can talk about,lol.
It probably just won't scale.Quote:
Originally Posted by Opteron146
Core i7 uses high-k and is pretty power efficient, and on 45nm they use no more than 1.3v on avg.
AMD's TWKR 42 CPU was a Deneb CPU that did not meet any power requirement or thermal requirement for their lineup. It runs really hot at the same voltages as normal CPU's and sucks way more voltage (want) and power (need). This made it suitable for LN2 cooling where an air or water cooler won't be a limitation and you can feed it insanely high voltages. C2 TWKR's ran almost as fast as good C3 retail CPU's under LN2.
Lower leakage means low heat and more efficiency but less scaling with voltage. Of course, if it runs faster anyway it won't matter. Some CPU's on air/water won't scale much over 1.48v, some 1.55v. It's all different and there is a sweet-spot. My CPU runs hot under stability testing and loves to take 1.55v all day and 1.6v+ for validations. While some people hit my speeds at ~1.45v
Contradictions galore! Since you insist on me spelling it out in layman terms, let me oblige.
A high leakage chip is very inefficient at utilizing power (because of the leaks) so at say 1.20v, a high leakage chip will generally run cooler (than the more efficient, low-leakage chip). The low-leakage chip, because it utilizes power more efficiently also generates more heat because you have to realize that: input power = output heat. What you're describing is a process improvement, not a high leakage chip characteristic. The reason why high lekage chips are suitable for ln2 is simply because they can take more voltage; more voltage=more power=more heat <- needs to be cooled by ln2. Makes sense?
Benches that show greater improvements from HT are also the same ones that show greater improvements from extra cores.Quote:
Originally Posted by ajaidev
It won't run cooler... A high leakage chip will run hot as hell,but it WILL scale with voltage and you WILL need to cool it to the extreme since the leakage current will then SKYROCKET with more voltage you add. You can see all this with TWKR edition chips,these are great examples. You need to understand that less leakage means less wasted power and less heat,period. Leakage current is a bad thing,a big problem in process tech. business.
How much do I have to dumb this downQuote:
Originally Posted by OhNoes!
A fluorescent light bulb is like a low leakage transistor. Light represents work being done and heat equals the heat from a transistor.
An incandescent bulb represents a high leakage part, in which more energy is converted to heat and work per watt efficiency goes down.
You don't make sense and you contradict everything anyone has ever tested or seen or proven. :ROTF:
no, leaky chips are hotter, hence they are less efficient ( electric wise )Quote:
Originally Posted by OhNoes!
Let's take two chips; one with leaky transistors and one with not so leaky transistors.
the chip with not so leaky transistors consumes 125W under load. That 125w is converted to heat.
the chip with leaky transistors consumes the 125W plus more (let's say 140W) because some power is being lost due to the leaking. that 140W is being converted to heat.
The leaky chip is outputting 140W of heat compared to the 125W of the not so leaky chip. It runs hotter.
OhNoes! has just self-destructed in a bad way... One thing is being wrong,whole other is digging your self even deeper with even bigger nonsense.
Now back to Thuban 1055T OCing fun :). We need imamage awake and testing that bad boy ASAP :)
Maybe he is Bizzaro Superman where everything is opposite:rofl::ROTF:Quote:
Originally Posted by BeepBeep2
I guess I'm lost here
I thought this thread was about the Phenom II x6 1055T
oh well....
@OhNoes
Watts = Voltage times Amps. In a computer, the voltage that the cpu gets stays pretty much the same, it draws on the 12v rail. Given that at least one variable in that equation is a constant, the power that a chip draws is proportional to the amps, the amount of current. In a high leakage and low leakage chip voltage is applied and stays the same, but the current is different for each chip. Let me explain this by the following example. You and your friend are in an open flat meadow standing 30 yards apart. If you try to whisper to your friend of course he cannot hear you, if you yell he can hear you, but he can hear you best if you cup your hands around your mouth and yell. In this example, voltage is the distance the sound has to travel, it is constant, amps is how loud you have to yell to be heard, and watts is how loud you sound to your friend. If you just yell at your friend, he will hear you (high leakage chip), but if you cup your hands around your mouth and yell, he will hear you better and louder since you are directing the sound at him instead of just in his general direction (low leakage chip). So the more and more you have to yell, the hoarser your voice will get due to strain. When you have to give more current, you produce more watts. In a cpu this extra current that is being applied to make the chip work, alot of it is wasted energy, and that is heat. Heat is a byproduct of wasted energy since the foundation of physics is based on the fact that matter cannot be created or destroyed. For the same voltage, if you have to put in less amps(current) into the chip to make it work, it will of course have lower watts than a chip that requires more amps to run. Leakage is the term being used for requiring more current and thus also creating much more heat. So when someone says high leakage chip that means that particular chip requires more amps to run than a low leakage chip. When applied to the formula W = V x A where in computers V is a constant and doesnt change, the more amps, the more Watts that will be produced.
The reason the rev C2 Phenom II X4 965 were 140w and the rev C2 Phenom II X4 955 were 125w is due to the fact that the 965s required more amps to operate at their stock speeds, and thus drew more watts. It isnt rocket science, but pure and simple Physics.
^all i see is blah blah blah, ohnoes needs to stay out of the amd section blah blah blah