LGA vs PGA

You don't know everyones use case. Someone might need some very specific functions on a motherboard that say only SuperMicro offers on specific boards. Stop being a gatekeeping faggot.

why wouldn't you replace the processor with a faster one when the prices of the high-end stuff drops?

(((You)))

Not him but post picture of you, your dick and your whole system along with speccy to proof legit

if he's posting proof of his AMD rig he better include his circumcision papers too

You kinda have to be retarded to bend the pins

>Intel motherboards
|||||\|||\|||||

>then
yeah, fixed alright

I like LGA more.
I was pretty happy AMD choose LGA for their high end desktop and server cpus.

I don't get why AMD just don't go all the way with LGA even on their normal ryzen chips, it's just a better design.

AMDrones shill defense force will defend this

Motherboards are cheaper than CPUs.

You have to be retarded to bend the pins. That said, LGA is better.

See

>Watch AMDfags defend this shit
you fucking useless cunt.
PGA has more contact surface.
It's better for any applications out there,
it's safer, because it locks into the socket and it doesn't create sparkles like pic related when it gets more current/voltage
Pic related is a common problem with intel cpus(the one in pic is an i7 870) when overclocked.
Same shit happened with the skylake-X overclocking twitter.com/HardwareUnboxed/status/888225519141093376

Attached: burnedpadS.jpg (550x223, 90K)

Then why AMD is using LGA for their ultra high end shit that consumes way more power than your shitty ryzen apu?

AMD should name their next architecture Bendy Ridge :)

Attached: R7 1700 jQ2WMbS.jpg (4032x2268, 2.32M)

Pins is kinda 90s tier tech. they should move to bus type slot like gpus

Attached: LPT_Dongle_-_Male_View.jpg (2990x1756, 1.88M)

AMD should not sell cpus to Inteltards.

simple math plus physics
TR and EPYC sockets have 4096 pins,
the supposed 250Watts are spread over a greater number of pads, than what intel has,
even if you manage to reach a 1kW of power consumption on TR or EPYC, this would mean that each pins sucks 1000W/4000padss which is 0.25 watts per pin.
a 1000W Intel CPU with the 2011 socket gets 1000W/2000pads, which is 0.5 watts per pin.
each one of those has greater resistance than the PGA pins, due to less surface.... so drawing even more power while having less pins, you enable more that resistance due to the contact to suck energy and get hot.
Intel had to change socket on their 2015 and later high-end CPU including the fact that they had to go over a certain consumption while having a guarantee that the socket can handle the load. They are still 400 pins bellow the TR4, they have the lga3647, but nonetheless more load is put on each pin in comparison to epyc and TR

I have done nothing so wrong to deserve an AMD CPU :(

and to push it a little further
let's say that every goldplated pad gets 1Ω resistance for both intel and AMD

2 EPYC pads having to deliver 1 watt of power.
we would need 4 pads, as we have 0.25watts per pad.
On the intel 2011 socket we would need 2 pins

Using the P=V^2*R formula for power we get for each pin
AMD:
Ptotal = Pad1 + Pad2 + Pad3 + Pad4
Ptotal = 1watt
Padx = R*V^2
R = 1Ω
Thus Ptotal=1 Watt = 4*(V^2) V^2=1/4 V=0.5 Volts

Intel
Ptotal = Pad1 + Pad2
Ptotal= 1Watt
Padx = R*V^2
R= 1Ω
Thus Ptotal=1 Watt= 2*(V^2) V^2=1/2 V=0.6944 Volts

and that's how you prove that less pads on the same power consumption need greater Voltage to deliver the that power.
The more power you push in each pin, the more it gets wasted by the resistance the touching connection makes.
I hope I didn't make any mistakes, any corrections are welcomed.

>not buying a top quality motherboard and overclocking the living shit out of a cheap CPU so it runs faster than a "high-end" CPU at a fraction of the cost.

You seem to be the stupid one.

>2 EPYC pads having to deliver 1 watt of power.
>we would need 4 pads, as we have 0.25watts per pad.
>On the intel 2011 socket we would need 2 pins
let me rephrase that, because I made correction to the problem but didn't change that part.

>Let's say that we want to deliver 1Watt to the CPU
>We would need 4 EPYC pads, since we can load them up to 0.25W/pad (as I showed above)
>and we would need 2 Intel pads on the 2011 socket to deliver the same amount of 1Watt of power, as each pad gets to 0.5Watt/pad

Yes why don't they do that?

RAM and GPU slots are near indestructible yet able to transfer huge amounts of data, but for CPU's we need something different? why?

Because you can't make a PCIe CPU?

(facepalm)
RAM : max 2 Watts per DIMM
PCIE: max 75 Watts per Slot.

geeh, I wonder why they didn't make a 2000-pin long card to plug my CPU.

We didn't say it has to be PCIe

Just solder the CPU to a PCB with a power connector and a data connector.
Then have a special CPU slot on the motherboard, probably similar to M.2 but wider and shorter.

it's not that big of an issue considering pga is easier to fix ;)

Not PCIe but the Pentium II would like a word with you.

Attached: intel-pentium-ii-350mhz-slot-1-cpu.jpg (1280x960, 158K)

tiny thunderbolt port easily pulls 105w

It would have the power connector on the CPU side, not on the motherboard side.
Just like graphics cards.

Insisting on pumping huge amounts of power through a socket style connector is exactly the dumb design flaw I'm trying to fix.

PGA has more parasitic inductance, I'm sure both companies have their reasons for their designs.
CPU bandwidth is pretty insane, i5 8500 is listed to have 41GB/s memory bandwith which is higher than PCIe v4 but lower than the future v5, and I believe memory latency through PCI is quite a bit slower

and usb standard has power delivery up to 100W
it's totally different when you try to push 100watts with constant 5 or 3.3 volts, because your amps can go as high as 20
and different when you try to push 100W when your max voltage is 1.2Volts and your amps can jump to 84.

Same problem exists with mutt and european electric grid. 110v vs 220v... it's harder for mutts to drive the same power under 110v....
imagine how harder is to push the same amount of power with 5 times less Volts.

>USB can do 20 amps
i doubt it

I could imagine certain case scenarios where you might need a lot of PCIe slots/SATA ports and maybe m.2 slots but not much CPU power, i.e like building a NAS of sorts.

USB3.1 Type C can be rated at 100W, which is why you can run decent laptops with a single USB port for charging.

Well user this works for intel because they needlessly switch sockets all the time so if you break say, your trusty ol lga 1155 socket it would not be feasible to buy another board and you would for the sake of cost effectiveness have to buy a brand new board with a new socket it's just a jewish trick.

imagine running 20 amps thru a chinese cable though

>worth more then motherboards
Are you retarded?

Yeah, there are tighter regulations for certified cables, and some high current devices will refuse to push current through a bad cable if they detect it.

Attached: disaster-girl.jpg (600x450, 42K)

LGA is electrically superior to PGA. It is the real reason why high power draw chips switched over. There's simply less resistance.

You clearly saw this when comparing Prescott Socket 478 chips versus Prescott Socket 775 chips.

It made a noticeable difference in power consumption obviously not enough to overcome the flaws in the architecture.

>Processors are worth more than motherboards so isn't PGA inherently more flawed
Perhaps, but let me remind you of a bigger flaw: TIM under the IHS.

I'm sure we'll get to a point where if you want a new high end CPU it's going to be on BGA

fixing pins in PGA CPU is easy. Not for the end user tho.

why can't they just go back to use slots?

On Intel you need to replace both anyway so it doesn't matter on which the pins are.

1) PGA is far more robust. the pins are easy to fix, and you can insert and remove the chip endlessly without wearing out anything.
2) LGA is only good for 3 or 4 insertions. after that pins get bent/warped, and lga pins are basically impossible to repair.
3) Intel claims they have better voltage and temp control with LGA, this may or may not be true, I've seen nothing to prove this is fact. One thing that is true, is you can squeeze more pins in per square inch on LGA.

Intel clearly switched to LGA in the middle of their P4 debacle in an attempt to placate MB manufacturers who were getting miffed at intel's multiple generations of P4s which all worked on the same fucking motherboard. They introduced LGA to end the upgrade path and force more motherboard sales.

PGA makes for better contact than LGA. LGA is for when you need the pin density.

Attached: 1457421791992.png (1024x1387, 1.28M)