Ground

yeah I've by now found 2 45nm chips doing 4.5/1.2V and 4.83/1.33V, though by now I've also found a couple similarly crazy 32nm chips, though they are out of a far larger sample size (about 20kg of chips). Two of them do 5Ghz at around 1.4V.

yeah I've by now found 2 45nm chips doing 4.5/1.2V and 4.83/1.33V, though by now I've also found a couple similarly crazy 32nm chips, though they are out of a far larger sample size (about 20kg of chips). Two of them do 5Ghz at around 1.4V.

yeah I've by now found 2 45nm chips doing 4.5/1.2V and 4.83/1.33V, though by now I've also found a couple similarly crazy 32nm chips, though they are out of a far larger sample size (about 20kg of chips). Two of them do 5Ghz at around 1.4V.

45nm 4 core R15 record
W3520 on dice

E5620 rerun on dice

What is off the shelf about the iPhone’s proprietary logic boards containing Apple designed chips or parts made according to Apple’s exact specifications that were not being sold prior to Apple ordering them?
 
You could try looking at individual chips, but the spare parts being sold for iPhone repair or being used to fabricate iPhones by third parties for the most part do not touch the individual chips.
 
Just look at an ifixit teardown:
 
https://www.ifixit.com/News/iphone-11-teardown
 
There is very little that looks off the shelf. Even the parts that ought to be generic (like the battery) look like made to order versions. Apple did not just grab a battery that was being sold generically, but instead ordered one made to custom dimensions with its own markings plus other features like the chip that allows the battery to authenticate itself with the iPhone. At the very least, I have not heard of batteries with such functionality being used in anything other than an iPhone, so they are clearly not off the shelf.
 
To give another example, the camera should have a third party sensor that is sold to everyone, but the way the module is built is custom for the iPhone’s internals. The sapphire crystal used for the lens was a world first in a cellphone as far as I know. Apple clearly did not design the internals around an off the shelf camera module, such that it is a made to order part at the very least.

have changed 2Gbit dies to 8Gbit dies on an HD 7870 before, but off course you have to stay with the same technology. 

X5687. Multi of 27

AS long as you are careful while swapping the CPUs nothing should happen. 

So yeah, the voltage under load is usually the bigger issue (since thats when you see high current at the same time), so that shouldn't be as bad. Still not gonna last forever but I guess replacements are cheap enough... the testing at 1.55V was with LLC on on a Rampage II Extreme with an E5640 at 4.8 GHz (reject chip), I started losing stability after about 3 days with that setup. The chip I actually used for daily for a while didn't ever see that kind of voltage on ambient, it does 4.5/1.36V with LLC on. These chips are nothing compared to the strongest 45nm chips I've found so far though which have significantly stronger cores and sometimes insanely strong uncores too. 

x58a-ud9 and asus P6T7. The later one is especially cool since its normal height, just a little wider...

(NB is under the big chunky heatsink - low res pic without heatsinks, enough to see whats going on under them...

x58a-ud9 and asus P6T7. The later one is especially cool since its normal height, just a little wider...

(NB is under the big chunky heatsink - low res pic without heatsinks, enough to see whats going on under them...

I have tons of old parts, but for today I'm only sharing a GPU I recently got gifted

x1800XT watercooled

are people still paying that amount of money for standard 1366 boards? In Germany low-endish boards (basic P6T, ud3r, msi x58 pro, foxconn flaming blade) go for ~50-60€ now, mid tier boards maybe 90€, older high end boards (R2E, Ud7) for 120€ ish and the very very best boards for 150ish (x58a-oc, R3E, though even these can be found for less)... Zen 2 and dropping mem pricing put a serious hit in the value of x58 as a platform...

the very best 45nm chips handily beat the best 32nm chips. I have a W3520 and i7 920 that run Superpi 32M at 4.5/1.2V, none of my 32nm chips (out of around 100 by now) manage that.

eh not really. Its good for 32nm multicore stuff, but for anything else (in my experience) even the Rampage II beats it easily. I've by now had several different x58a-ocs cold (still own a working and a dead one), and while they work well the main thing they have going for them is the great VRM they have. For 45nm benching Rampage II Extreme/Rampage III Black or worst case Rampage III Extreme are much easier to run and can usually get better results. I've done this score on a Rampage II Extreme on Cascade, can't get anywhere close to that with an x58a-oc with an otherwise identical setup. 
I've recently picked up a Foxconn Bloodrage to compare, haven't gotten around to test it properly. 
I did manage finding a really nice i7 920 on ambient, really hard to control on LN2, need to run it on cascade to see what it can actually do. 4.5/~1.2V seems to be my baseline for really nice 45nm chips since my W3520 does about the same. 
 
x58a-oc you can usually sell for 200+ on ebay btw. I usually get mine for far less on forums or classifieds.
 

Also have an x58a-GD65, Rampage II Extreme and Rampage III Extreme each with blown up VRMs in various boxes that I have to get fixed, spent a couple hours yesterday trying to get the second x58a-oc working again. Bloodrage, signed x58a-oc and R3E at least fully work. 

eh not really. Its good for 32nm multicore stuff, but for anything else (in my experience) even the Rampage II beats it easily. I've by now had several different x58a-ocs cold (still own a working and a dead one), and while they work well the main thing they have going for them is the great VRM they have. For 45nm benching Rampage II Extreme/Rampage III Black or worst case Rampage III Extreme are much easier to run and can usually get better results. I've done this score on a Rampage II Extreme on Cascade, can't get anywhere close to that with an x58a-oc with an otherwise identical setup. 
I've recently picked up a Foxconn Bloodrage to compare, haven't gotten around to test it properly. 
I did manage finding a really nice i7 920 on ambient, really hard to control on LN2, need to run it on cascade to see what it can actually do. 4.5/~1.2V seems to be my baseline for really nice 45nm chips since my W3520 does about the same. 
 
x58a-oc you can usually sell for 200+ on ebay btw. I usually get mine for far less on forums or classifieds.
 

Also have an x58a-GD65, Rampage II Extreme and Rampage III Extreme each with blown up VRMs in various boxes that I have to get fixed, spent a couple hours yesterday trying to get the second x58a-oc working again. Bloodrage, signed x58a-oc and R3E at least fully work. 

eh not really. Its good for 32nm multicore stuff, but for anything else (in my experience) even the Rampage II beats it easily. I've by now had several different x58a-ocs cold (still own a working and a dead one), and while they work well the main thing they have going for them is the great VRM they have. For 45nm benching Rampage II Extreme/Rampage III Black or worst case Rampage III Extreme are much easier to run and can usually get better results. I've done this score on a Rampage II Extreme on Cascade, can't get anywhere close to that with an x58a-oc with an otherwise identical setup. 
I've recently picked up a Foxconn Bloodrage to compare, haven't gotten around to test it properly. 
I did manage finding a really nice i7 920 on ambient, really hard to control on LN2, need to run it on cascade to see what it can actually do. 4.5/~1.2V seems to be my baseline for really nice 45nm chips since my W3520 does about the same. 
 
x58a-oc you can usually sell for 200+ on ebay btw. I usually get mine for far less on forums or classifieds.
 

Also have an x58a-GD65, Rampage II Extreme and Rampage III Extreme each with blown up VRMs in various boxes that I have to get fixed, spent a couple hours yesterday trying to get the second x58a-oc working again. Bloodrage, signed x58a-oc and R3E at least fully work. 

Some 1156 Dice action
i3 540 - chip does 4.8 at 1.4V on air

i3 530 - chip does 4.6 at 1.4V on air

wasn't alone on the session, but as hwbot is currently down I can't add my friends scores to this. Gonna add those later!
 
Friends scores:
G6950 - chip does 4.8 at 1.4V on air

i5 760 - chip does 4.4 at 1.4V on air

x99 with xeons works fairly well with Reg ECC. Pretty sure all asus boards at least, I've heard some got it to work on pretty much any board even.

Picking up 1366 for a daily hasn't been a good idea for at least half a year now, basically since DDR4 prices started dropping. Ryzen 1xxx series has been quite cheap for quite a while now...
 
I'm not quite sure yet, but I could imagine x99 with Reg ECC and Xeons could get interesting once the first gen of DDR4 servers get decommissioned.

Sammy A is a 16Gbit IC, I’m almost certain its not gonna get close to bdie. I’m rather expecting a new 8Gbit IC (perhaps 8GB D-Die) to be the next interesting Samsung IC. 
Micron Rev E (the new micron stuff) seems neat for sure, sounds like it’ll be easier to run then bdie, especially when benching, which makes it quite exciting to me. 

unlikely for this one - bdie is at its limit (and apparently a lot harder to run). The new run is using micron

i7 920 at 5.09 
 

i7 920 at 5.09 
 

i7 920 at 5.09