|
It's something you never ever really "git gud" at unless you do it often enough for it to become mechanical. Also, as CPU packages become more chiplet-heavy the "X" and "Big Pea" methods won't be valid anymore over a thin uniform application over the entire spreader. The last time I did it was with Kryonaut and I just got exasperated and redid it the first time. I remember AS5 being so loving simple by comparison and the only paste I flat out hated was AS Ceramique because I did such a good job in applying it that getting the HSF off my Tualatin P3 required my twisting the sink to finally break the super tight seal that'd been made.
|
#
?
Dec 4, 2021 05:55
|
|
|
|
| # ? May 9, 2024 16:41 |
|
|
Rinkles posted:I hosed up and ended up repasting three times, because I noticed that the cooler was loose, but only by the second attempt did I realize it was because I forgot to screw in the mounting brackets. At least this was before I moved on to do anything else. You learn by doing, it's the best way
|
|
#
?
Dec 4, 2021 11:56
|
|
|
Let it also be said that the penalty for using too much paste is far less of a problem than the penalty for not using enough. Too much paste makes a mess and squeezes out of the sides but otherwise performs fine, too little and there will be air gaps between the surfaces and the performance will suffer significantly. Always error on the side of too much.
|
|
#
?
Dec 4, 2021 13:03
|
|
|
Also, both with regard to "should I repaste if I take apart fresh assembly" and "I'm worried about using too much", unless you're test benching or building an incredible number of systems, you're usually going to wind up throwing away most of the paste you buy. Just Paste It.
|
|
#
?
Dec 4, 2021 14:01
|
|
|
https://www.intel.com/content/www/us/en/support/articles/000088261/processors/intel-core-processors.html Good to see they're on top of the DRM issues, there's already only three games left to fix
|
|
#
?
Dec 4, 2021 15:11
|
|
|
BIG HEADLINE posted:It's something you never ever really "git gud" at unless you do it often enough for it to become mechanical. Also, as CPU packages become more chiplet-heavy the "X" and "Big Pea" methods won't be valid anymore over a thin uniform application over the entire spreader.
|
|
#
?
Dec 4, 2021 15:27
|
|
|
BlankSystemDaemon posted:Servers have had big heat-spreaders for decades, and I don't think I've ever seen the "X"/"Big Pea" paste pattern recommended there ever. It's almost always 6 dots equidistantly separated from each other and the edges, like rolling 6 on a 6-sided die. Yeah well maybe if server scrubs used a better method, they wouldn't have such trouble breaking 3 GHz 
|
|
#
?
Dec 4, 2021 17:57
|
|
|
ConanTheLibrarian posted:Yeah well maybe if server scrubs used a better method, they wouldn't have such trouble breaking 3 GHz Of course, whether its a 7nm, 14nm, or any other value of process doesn't really reflect reality since there's no way you could fit a gate into 7 or 14 nanometers, when the Van der Waals radius of a single silicon atom is 210picometer - which would mean that the gate would have to be only 35 or 70 atoms thickness. 
|
|
#
?
Dec 4, 2021 18:30
|
|
|
Newest parts are Ice Lake / Intel 7 now. But I think we are overanalyzing the joke here!!
|
|
#
?
Dec 4, 2021 21:31
|
|
|
Cygni posted:Newest parts are Ice Lake / Intel 7 now. But I think we are overanalyzing the joke here!! For reference, I think 7nm equates to about 22nm of actual gate material.
|
|
#
?
Dec 4, 2021 21:47
|
|
|
Intel 7 is a "10nm" technology, not 7nm. Their supposedly 7nm process is "Intel 4"
|
|
#
?
Dec 4, 2021 22:18
|
|
|
Dr. Video Games 0031 posted:Intel 7 is a "10nm" technology, not 7nm. Their supposedly 7nm process is "Intel 4" AMD, ARM, RISC-V ODMs, IBM and every other architecture under the sun are doing the same - it's not just Intel.
|
|
#
?
Dec 4, 2021 22:26
|
|
|
Gate lengths can be close to the fake “nm” marketing name, but gate pitch for all of em (TSMC 7/6, Samsung 7, Intel 7) is 54nm I think, with similar max density too.
|
|
#
?
Dec 4, 2021 23:01
|
|
|
And since finFET we're in 3D structure territory, making just gate pitch or length technically even less meaningful. At least by doing the nm unit in the name it's being honest that it is just a node label and marketing term.
|
|
#
?
Dec 5, 2021 11:03
|
|
|
IIRC the nanometer number they quote these days means "if we were still using planar transistors, this is how small they'd have to be to match the process's density".
|
|
#
?
Dec 5, 2021 15:27
|
|
|
Beef posted:And since finFET we're in 3D structure territory, making just gate pitch or length technically even less meaningful. Though I find it very funny that in the same breath they talked about nanometers being a useless marketing term now, they announced that they'll be returning to measurement-based node labels in a few years anyway with "20A"
|
|
#
?
Dec 5, 2021 21:34
|
|
|
ConanTheLibrarian posted:IIRC the nanometer number they quote these days means "if we were still using planar transistors, this is how small they'd have to be to match the process's density". TSMC straight up says its just marketing now. TSMC VP of Research posted:“Today, these numbers are just numbers. They’re like models in a car – it’s like BMW 5-series or Mazda 6. It doesn’t matter what the number is, it’s just a destination of the next technology, the name for it. So, let’s not confuse ourselves with the name of the node with what the technology actually offers.” I like that they say that while simultaneously pushing "half step" nodes like N6, lol.
|
|
#
?
Dec 5, 2021 21:37
|
|
|
Once they get to 1, they should just start counting up again. They'll probably be close enough to a billion transistors per mm2 that they can just pretend that's what it means.
|
|
#
?
Dec 6, 2021 00:45
|
|
|
Dr. Video Games 0031 posted:Though I find it very funny that in the same breath they talked about nanometers being a useless marketing term now, they announced that they'll be returning to measurement-based node labels in a few years anyway with "20A" Technically it's still not measurement based, they're using A not Å, you're just making the connection like the "nm" comes after Intel 7 or 4.
|
|
#
?
Dec 6, 2021 03:35
|
|
|
IBM is already down to 2 nm chips, so the next step down is sub-1000 picometres
|
|
#
?
Dec 6, 2021 03:50
|
|
|
Probably should update ASAP? Gamer's Nexus reported that some Gigabyte motherboards got vulnerabilities updates, and mine was among them. e:date says 11/09 for F7, but when I checked a few days ago F6 was the latest BIOS available
|
|
#
?
Dec 6, 2021 07:06
|
|
|
gradenko_2000 posted:IBM is already down to 2 nm chips, so the next step down is sub-1000 picometres Perfect, number go up
|
|
#
?
Dec 6, 2021 07:44
|
|
|
Deuce posted:Perfect, number go up
|
|
#
?
Dec 6, 2021 14:35
|
|
|
USB is so dumb No, those aren't all the same speed.
|
|
#
?
Dec 7, 2021 21:27
|
|
|
Orks taught us long ago that red makes it go faster. Obvious stuff, here.
|
|
#
?
Dec 7, 2021 21:50
|
|
|
I guess USB isn't that 'universal' anymore now that are several types.
|
|
#
?
Dec 7, 2021 23:26
|
|
|
why are mobo so stingy with the usb-c?
|
|
#
?
Dec 7, 2021 23:56
|
|
|
incoherent posted:why are mobo so stingy with the usb-c? More expensive connector with more stringent requirements.
|
|
#
?
Dec 8, 2021 00:15
|
|
|
TOOT BOOT posted:I guess USB isn't that 'universal' anymore now that are several types.
incoherent posted:why are mobo so stingy with the usb-c?
|
|
#
?
Dec 8, 2021 00:15
|
|
|
Rinkles posted:USB is so dumb My Gigabyte also has USB 2.0 ports lol. Granted they're designed for mouse/kb/DACs etc but still, having them all USB 3.2 would have been better. Thankfully it only has 2 USB 2 ports, the rest are all USB 3.1/3.2 and a Type-C 3.2 on the back as well as an internal connector that goes to the Type-C 3.2 on my case front panel along with 2x 3.2 USB-A ports.
|
|
#
?
Dec 8, 2021 00:49
|
|
|
mrk posted:My Gigabyte also has USB 2.0 ports lol. Granted they're designed for mouse/kb/DACs etc but still, having them all USB 3.2 would have been better. What is 3.1? Don't you mean USB 3.2 Gen 1/2???   And USB4 is honestly even worse, lol
|
|
#
?
Dec 8, 2021 01:14
|
|
|
Well PC Manufacturers, I made it. Despite your directions Ah, usb consortium! Welcome! I hope you're prepared for a forgettable standard!
|
|
#
?
Dec 8, 2021 01:37
|
|
|
I still remember when the usb consortium announced the new names for the 5/10/20 gbps standards. Even for some of the poorest performing standards bodies out there it is exceptionally rare one invents a naming convention so colossally bad that it makes the whole internet pause for a moment because everyone that reads it thinks it is a prank.
|
|
#
?
Dec 8, 2021 02:24
|
|
|
The naming and poo poo does suck. It does get a little tricky though because they were adding the ability to add multiple lanes to a controller instead of a single so you're left with trying to communicate something like GenX with 2 lanes compared to GenY with a single lane and them both being the same speed but I guess it's like when someone asks you a question and your brain freezes and you blurt out some gibberish and 5 minutes later you're wondering what the hell you're thinking only this time it's a group of people and a ratified document.
|
|
#
?
Dec 8, 2021 02:46
|
|
|
DrDork posted:Orks taught us long ago that red makes it go faster. Is this right, tho? I’ve got black ones and one red one. I plug my drives into the red one assuming (naturally) that it’s the best one. It’s still slow as gently caress, so who knows but usb sucks rear end nevertheless
|
|
#
?
Dec 8, 2021 02:49
|
|
|
BlankSystemDaemon posted:It was never universal, but today it's even less: This. And also, on a single host device, you can have USB C ports: that can output video, and ports that cannot that support thunderbolt and ports that cannot that can accept power and ports that can only provide power And heaven help you if the pets are not clearly labeled for any of that. Then you have USB C cables that night only be USB 2.0 or not tested for the power spec you need, or incapable of any number of other functions. Basically, USB is a hell scape of compatibility issues.
|
|
#
?
Dec 8, 2021 05:31
|
|
|
You guys aren't wrong to diss it, but I can generally plug anything into anything and it basically works. It will often tell me if it could be working better or maybe I don't even notice. Seems to me it is stupid, but good enough. At least you can plug it in in either direction and it's pretty small and fits pretty snug.
|
|
#
?
Dec 8, 2021 05:49
|
|
|
Yeah, I do not miss the bad old days of parallel ports and making sure my serial ports had a high enough UART rating to run a proper external modem.
|
|
#
?
Dec 8, 2021 06:19
|
|
|
I don't miss setting IRQs either if it comes to that, but let's not pretend that USB isn't completely hosed.
|
|
#
?
Dec 8, 2021 06:40
|
|
|
|
| # ? May 9, 2024 16:41 |
|
|
If you wish to identify your ports USB You must first answer these riddles three
|
|
#
?
Dec 8, 2021 06:40
|
|
