I Killed My 7800X3D... 

Did you even watch the video? He compared two different units. Absolutely horrible testing.

i will rob you of your components next wednesday

Did you use the noctua spacer kit?

How did you mount the Noctua??? No one is doing delids with air coolers😢. Any information would be greatly appreciated. Thank you!

@@3DxPOD You can use spacers to get the height correct, its sits niceley afterwards, loads of posts on the internet showing how this is done.......

@@FTLN Thanks I'll keep looking.

there's no point in doing any overclocking or tinkering, gains are abbyssmal

Less waittimes in rendering, more frames in Games is not the same as replacing solder with liquid Metal g​@@snowpuddle9622

If they'd maybe reduce the IHS by 50% thickness that'd be great.

the solder is almost always not evenly spread out and even if it is, liquid metal is better at conducting heat

This is a perfect example of someone repeating what they heard and not testing it themselves

Yep. Intel keeps making amazing CPUs

I mean $450 isn't hard to scrounge up

people just return them as not working lol

@@zihechen3111 you cant reduce latency, by moving farther away from the core

@@zihechen3111 The cost of the cache is wildly over-stated This is the same silicon used in the RX 6500XT which was selling for $100 The cache is rougly 30mm² doesnt include memory, doesnt include fans, doesnt include power delivery or licencing fees for the HDMI and DP ports. The 6500XT on the other hand does include all of those things, is 107mm² instead of ~30, and often comes with 2 free games, for $100, this cache is dirt cheap, and so are the CPUs, that 5950X was is cheaper to manufacture than the RX 6500XT, yet it was selling for what, 5-8x the price? On the subject of cache, You can get at least 256MB of L4 on the 5800X3D and 7800X3D in addition to the cache on the CCD, if you take the same cache thats used on the 5800X3D, and instead of cutting it to this tiny little thing a few mm wide and long and slapping it on top of the reletively small CPU cache, you instead cover the entire IO die. This can do 2 things. If you no longer place cache on top of the CCD, the CCD can now run at higher frequencies, and 2, by placing the cache on top of the much cooler IO die, the cache should be able to be more dense due to the lower thermals I mathed it out, taking the same cache density found on the 3D chip of the 5800X3D(64MB at roughly 30mm²) then instead of chopping the wafer up and using inactive silicon to cover the rest of the die like on the current X3d chips, they could just make a cache chip the same size as the entire 122mm² IO die for at least 256MB Ideally for efficiency you should still keep the 64MB of cache on top of the CPU cache, but by having this 256MB of cache on the IO die you're reducing the needs to go to system memory, reducing latency, and reducing power consumption. Just look at how efficient the 5800X3D and 7800X3D are, better performance than their non3D counterparts, while using significantly less power. Sorry if i'm a bit rambly, its late and i have been up late for the past 4 days

@@zihechen3111 The Apple M1 does use its cache smartly, but one of the reasons the M1 is so efficient is not because it has a huge cache, but because it has the power of a small dGPU, and instead of a seperate package+RAM for the CPU, and GPU, they both share the same pool of RAM and are on the same die. this means they dont have to pass through the motherboard to talk to each other, they technically dont even need to pass through the package if designed right, it could all go through silicon. The AMD 2400G, 3400G, 4750G, 5700G, 6800U and upcoming 7840U/Z1Extreme are also designed in this way(apple was not first at this, AMD has been doing APUs for years as most consoles are built on AMD APUs) The 5700G is crazy efficient, with the right RAM, you can get similar performance to a 5800X plus a GTX 1050, with the whole system using less power than the 5800X by itself, what really held the 5700G back was the vega architecture used in the GPU, which was basically 4 years old at the time the 5700G launched, yes you could close those 8 vega cores fast puwshing the 5700G to around 80w, but just cant make up for the old architecture. For reference, the steamdeck APU uses just 20w PPT, has the same number of GPU compute units as the 5700G, but offers better GPU performance than the 5700G (at 60w) thanks to the use of RDNA2 They're even both on the same process node, so it cant be argued that its only faster because one is on 7nm, and the other is 4nm

The problem with dat is Thas alot of not experienced person are gonna still buy it and end up crackling the die etc etc and start bunch rma etc to deal with it

@@HosakaBlood I was thinking that the direct die heat spreader was already attached. No need to remove it and crack the die.

@@BackToTheStart47 You can't have both a heat-spreader, and be direct die simultaneously. The whole point of direct die is that there is no heat-spreader.

@@Bayonet1809 What I am suggesting is that the direct die is already installed from Amd or Intel. No need to remove the IHS.

7800x3d is overclocked to 5.4Ghz already two weeks ago and its boost up to 5.6GHz, all without delid. ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-90UBUq1mLGY.html

You could use PBO and negative CO offset to gain performance if their is thermal headroom with undervolting and jsut let the chip keep boosting til it hits the limit

I got a similar outcome to you. -30 seems to be doable for most of these CPU’s

_back side power delivery has entered the chat_

what are your PC specs

Derbauer found that some in the previous gens had a 2nd ccd that was nuked. But physically there sometimes

it's a gaming cpu for a reason, sadly amd don't have universal cpus for gaming and production

yep, you are completely right. The voltages are very very tightly controlled, the most (and easiest) gains can be from curve negative voltage offset settings.

Yeah the Intel chips come crazy overvolted from the factory but they can be power limited and tuned to a great level where you can get great performance and reduce the heat and power consumption Exponentially.

@@NBWDOUGHBOY It's the motherboard, not the cpu.

Hwinfo

this CPU does not need 2 connectors, it never pulls above 90W. One connector is good for 300W. Why would you need 2 connectors for 600W when you only consume 90W? Do you know numbers? 90 is much smaller than 300.

Yeah, direct pressure to 4-5nm delicate sand structure. No wonder they break.

How? The setting maxes out at - 30 im BIOS? Doesn't sound like math.

@@djchotus1 I dunno, that's what the dude said... 35. I never tried over 30 cause 30 was unstable so I went back up to 25.

depends on what u consider stable? prime95 small fft with avx workload on? ye -20 CO is max on good samples. Just gamestable good cpus can make -35 CO.

at an allcore basis. u can push it further on an per core basis.

@@dwight2310 what would be the VID in full load prime95 small fft with avx and -30 CO? I'm asking because i set -30 CO and VID goes up to 1.1V....doesn't seem right or doesn't seem like the CO is working

You don't even have a 7800x don't lie please

@@nathancachia1419 ha, you are 100% right it was a typo, I have changed it to -7700x. Thanks

Nice :)

plus swapping cpu's and not doing a re test of programs and showing differences doesn't work. Your running a asus motherboard show the SP rating of the chip so we know how good the silicon of the cpu is. that could account for differences in voltage/temps/boost clock

Did it work

@@smoofwah3552 went smooth as butter.

Balls