has intel done anything besides add more GHz and shrink the die since the release of the first core i series?
AVX and HD X000 GPU
>add more GHz
>shrink the die
Those aren't noteworthy at all ?
>>51395906
they made it impossible for consumers to overclock their machines for a 1-5% increase in performance.
unless you pay more shekels, at over 30% increase of price.
>>51395906
Isnt that the point?
>>51395966
>avx
I can agree with that, updated instruction set important for performance and compatibility.
>hd x000 gpu
I never use my igpu, and i don't know anyone that does, so that's not much of an issue. Most people would rather waste money on a gt610 because nvidia. And those that know better will get a decent gpu.
>>51395977
No, they're not. The die shrink is nice for lower temps and energy consumption, but you can always use a better cooler.
And as for more ghz? These aren't locked down like the newer gens.
>>51396068
>No, they're not. The die shrink is nice for lower temps and energy consumption, but you can always use a better cooler.
Yes, fuck technological advancement, I'll just use a fucking turbine on my P4.
I hope you're just shitposting.
>>51395906
Arch from Sandy Bridge to Haswell alone saw significant changes, the average user just can't see the result.
Serial integer performance isn't going to be making any big leaps any time soon. Pluck all the low hanging fruit and performance gains are harder and harder to come by.
>>51396150
>fuck technological advancements
Yes. If 2% increase from increasing the GHz is what you consider technological advances then fuck that. I can do it myself without paying $300.
>>51395906
Obviously they changed a lot more than that, or how else do you imagine did they increase the performance of their CPUs by that much? The clock hasn't change that much, it's not 1998 after all, die shrinks do not impact performance.
>>51396231
>die shrinks do not impact performance.
but doesn't a smaller die mean that the electrons have less distance to travel, hence resulting in a slight performance gain?
>>51396254
This is called gate delay, and it is a vital metric of gate performance, but you'd never see its practical effect in the software you're running.
>>51396254
No, not really. Electrons hardly move inside a CPU. Imagine it more like a sound wave or something, the actual current is minimal compared to electron charges. The operation latency due to electromagnetic retardation is negligible. The main objective of die shrinks is to decrease power consumption. Usually you can increase CPU clock consequently, but that doesn't work at about >5 GHz for various reasons. To get more performance nevertheless you need to increase registers, thus increasing transistor count. The CPU designs need to get more complex.
>>51396382
I thought so, electrons already move so fast.
Also, doesn't a smaller die mean less overclocking potential since its easier for elections to jump?
>>51395980
30% overclock on mine, but it's under water. Needs exponential more voltage after this point, and it's perfectly stable where it is.
>>51395980
this.
they locked their multiplies with sandybridge, forcing consumers to pay a huge hike in price for barely any performance increase.
disgusting marketing tactics.. shitty fucking corporation with sacks of shit people working there.
>>51396421
When dealing with planar gates specifically the smaller a device the more pronounced a series of issues become, and they're collectively known as the short channel effect.
You lose the ability to have a true off state, leakage current progressively increases, isolation falls off.
The reason why FinFETs exist is to address these issues. The gate structure itself actually grows in surface area despite area scaling improving. The increased gating surface area gives you better control over the channel which reduces leakage current.
The maximum achievable clocks in a given chip are controlled by hundreds of different things, you can't pinpoint any one thing as being responsible for hindering OC headroom. Different transistor libraries have different performance curves, and the IC designer could choose to intentionally sacrifice OC headroom to save further power, or vice versa.
>>51396443
moar gigahurtz does not mean better performance.
even if you went from 4 to 4.5 the performance gain is still only like 3%
>>51396469
could someone elaborate more about how Intel ruined clocking of their cpus? i. tech illiterate and don't really know what this all means yet
>>51397020
they locked their multipliers...
forcing customers to pay a huge hike in price just for the ability to lower or increase the core multiplier..
if you overclock you'll know what this means.
if you don't then you shouldn't buy the "k" varient anyway.
>>51397095
>>51397020
Best part is if you get the CPU with the unlocked multiplier so you can overclock, they remove other features you may need, so you need to choose between one or the other when the CPUs are capable of both
>>51397107
>VT-d
Around jews, you lose.
>>51397095
A8 8600 has a locked multiplier but the fsb can still be overclocked iirc.
Is this not possible with Intel too?
>>51397189
a8 7600 you fuck.
