|
.
sincx fucked around with this message at 05:50 on Mar 23, 2021 |
#
?
Apr 6, 2019 04:04
|
|
|
|
| # ? May 16, 2024 10:43 |
|
|
sincx posted:Big air coolers will always suffer from the issue of "you're putting a huge amount of weight and torque on a few thin layers of fiberglass glued together." AIOs have an additional moving part (the pump), but their mounting mechanisms are much better. And a damaged motherboard will cost many times the price difference between an AIO and an equivalent air cooler to replace.
|
|
#
?
Apr 6, 2019 06:29
|
|
|
sincx posted:Big air coolers will always suffer from the issue of "you're putting a huge amount of weight and torque on a few thin layers of fiberglass glued together." AIOs have an additional moving part (the pump), but their mounting mechanisms are much better. And a damaged motherboard will cost many times the price difference between an AIO and an equivalent air cooler to replace. To me the biggest difference between huge tower heatsinks compared with AIOs, is that on sustained loads the case airflow becomes a big influence on temperature & noise. A liquid cooler has a much easier time dealing with poor case airflow because they dump heat directly to the outside. I've used an Antec P180 for over a decade, which is not exactly a high-airflow case, and I've found there are balance points for CPU fan vs case fans that have changed depending on which hardware was in it over the years. AIO by comparison is pretty much a single variable.
|
|
#
?
Apr 6, 2019 07:49
|
|
|
I wouldn't, but what I *would* do is get that Instapak foam and let that expand inside the case, take up all the room inside. It's what I did the last two times I had to ship to people who were not quite as technologically-adept as myself.
|
|
#
?
Apr 6, 2019 07:57
|
|
|
https://www.sec.gov/Archives/edgar/data/1046179/000119312519097160/d870977d6k.htmquote:TSMC today announced delivery of the complete version of its 5 nanometer (nm) design infrastructure within the Open Innovation Platform® (OIP). This full release enables 5nm systems-on-chip (SoC) designs in next-generation advanced mobile and high-performance computing (HPC) applications, targeting high-growth 5G and artificial intelligence markets. Leading Electronic Design Automation (EDA) and IP vendors collaborated with TSMC to develop and validate the complete design infrastructure, including technology files, process design kits (PDKs), tools, flows and IP, through multiple silicon test vehicles.
|
|
#
?
Apr 6, 2019 10:46
|
|
|
Man, TSMC are killing it.
|
|
#
?
Apr 6, 2019 11:39
|
|
|
Neat. Is Zen 3 expected to be on 5nm?
|
|
#
?
Apr 6, 2019 11:44
|
|
|
OhFunny posted:Neat. I think the current rumor is that Zen 3 will be on TSMC 7nm+, which I think is their current 7nm process but with some of the DUV steps replaced with EUV steps. From what I understand TSMC's 5nm node will be mobile-focused (so targeting low power but also low clocks).
|
|
#
?
Apr 6, 2019 15:39
|
|
|
Right about the time I want to upgrade 5nm might be in new phones. Rad.
|
|
#
?
Apr 6, 2019 19:20
|
|
|
5nm and 5G wow!
|
|
#
?
Apr 6, 2019 20:02
|
|
|
I'm planning to build close to the end of the year once Zen 2 and Navi are both out. Probably gonna be the 3700x, I'd guess. I'm not planning on overclocking it and letting the "Precision Boost" or whatever do what it can. Is the stock cooler(s) AMD started including on Ryzens good enough, or am I gonna want to add some bigger air cooler like that Noctura or whatever?
|
|
#
?
Apr 6, 2019 21:37
|
|
|
If your budget is so constrained that you have to make a determination between "buy a cooler of any kind" and "put money that would have been put into cooler, into better parts in some other category", yes, it is sufficient. At minimum, they are more capable than the loving intel slice of extruded aluminum they've been using for over a decade now.
|
|
#
?
Apr 6, 2019 21:40
|
|
|
The way I overclock my 2600X is by having a giant Be Quiet! cooler on it and letting precision boost do its' thing.
|
|
#
?
Apr 6, 2019 23:42
|
|
|
Any undervolting necessary associated with "letting PBO do its thing"?
|
|
#
?
Apr 7, 2019 04:15
|
|
|
AMD [expects] to see sales increase sharply in 2H19 Intel Core i5-9300H to i9-9980HK: More Specs Leaked AMD must be making GBS threads their pants with glee.
|
|
#
?
Apr 8, 2019 18:51
|
|
|
SwissArmyDruid posted:AMD [expects] to see sales increase sharply in 2H19 The X570 chipsets are designed by AMD in house as AMD's chipset partner ASMedia will not be able to tape out its PCIe Gen 4-related solutions until the end of 2019; however, ASMedia will still be responsible for designing AMD's chipsets for the mainstream sector. B550 coming late.
|
|
#
?
Apr 8, 2019 19:11
|
|
|
There's no way AMD lets this window of opportunity slip, not when they can force-feed OEMs where Intel can't right now. A-series with PCIe 3.0 are likely shipping to OEMs ASAP to fill their beige boxes, B-series with PCIe 4 to go into mass production around holiday? SwissArmyDruid fucked around with this message at 19:31 on Apr 8, 2019 |
|
#
?
Apr 8, 2019 19:24
|
|
|
SwissArmyDruid posted:
lol @ b550 having pcie4. """mainstream""" machines don't even make pcie3 sweat
|
|
#
?
Apr 8, 2019 21:00
|
|
|
The objective is to make intel marketing dept sweat, not the motherboards.
|
|
#
?
Apr 8, 2019 21:36
|
|
|
Malcolm XML posted:lol @ b550 having pcie4. """mainstream""" machines don't even make pcie3 sweat This is my assumption as well, 300/400 should support Ryzen 3000 and in the instances where high perf/enthusiasts are involved they can afford the premium for the higher end x570 chipset. From a marketing perspective they can say PCI-e 4 support on all their processors which is an easy number for consumers to understand. AMD 4, Intel 3? ok AMD is 1 betterer. More bandwidth will be good for I/O though on the mainstream chipsets.
|
|
#
?
Apr 8, 2019 21:50
|
|
|
incoherent posted:The objective is to make intel marketing dept sweat, not the motherboards. pcie4 actually needs $ pumped into the boards for signal integrity reasons so they are gonna say AMD RYZEN 3000 with PCIe 4.0* * only on X570
|
|
#
?
Apr 8, 2019 23:50
|
|
|
Malcolm XML posted:pcie4 actually needs $ pumped into the boards for signal integrity reasons so they are gonna say AMD RYZEN 3000 with PCIe 4.0* ** only on slot 1
|
|
#
?
Apr 9, 2019 00:11
|
|
|
.
sincx fucked around with this message at 05:50 on Mar 23, 2021 |
|
#
?
Apr 9, 2019 00:18
|
|
|
sincx posted:Does the chipset also get PCIE 4.0 to the CPU? That's a nice boost if so. Well if it was Intel, they'd immediately chop it from four lanes of PCIe 3 to only three lanes of PCIe 4, and still hang TB 4 off of it instead of piping it straight into the CPU.
|
|
#
?
Apr 9, 2019 00:27
|
|
|
WHAT THE gently caress IS THAT?!
|
|
#
?
Apr 9, 2019 02:44
|
|
|
....why does it have BGA instead of contact points?
|
|
#
?
Apr 9, 2019 03:15
|
|
|
priznat posted:** only on slot 1 They clearly need to put in some kind of pcie3->pcie4 bridge next to the slot, so they can say all the slots are pcie4***. *** performance may be limited by pcie3
|
|
#
?
Apr 9, 2019 03:16
|
|
|
Why 56 cores? is this supposed to compete with the 64 core Epyc?
|
|
#
?
Apr 9, 2019 03:18
|
|
|
But remember guys to only use a pea-sized amount of paste.
|
|
#
?
Apr 9, 2019 03:45
|
|
|
CommieGIR posted:....why does it have BGA instead of contact points? love 2 throw my $MONEY cpu in the trash when my motherboard dies. I had to double-check the date to make sure it wasn't an april fool's joke, with Cutress's gormless grin there. edit: Seriously, though. I know that SP3/TR4 is all sorts of jank when it comes to the torque wrenches, and the carrier, and the two-stage mounting system, and having to remount the chip a few times to make sure that you get contact on every pin, but holy poo poo, I do not think this is the way to solve that problem! edit edit: I'm sure that there will be some kind of program to ship the dead boards back to Intel or the board OEM to desolder and reball the processor before putting it onto a new board, but jesus. Are rework stations going to become a standard part of IT going forward? I'd better loving brush up on my hot air reflow skills. SwissArmyDruid fucked around with this message at 04:34 on Apr 9, 2019 |
|
#
?
Apr 9, 2019 04:02
|
|
|
Craptacular! posted:But remember guys to only use a pea-sized amount of paste. GC's got you covered! 
|
|
#
?
Apr 9, 2019 04:54
|
|
|
Am I supposed to not eat that? ^ It looks like a tub of ice cream.
|
|
#
?
Apr 9, 2019 05:06
|
|
|
Ice cream is kind of a "heat sink" right? E: At least they're not trying to run it at 5GHz like last year! Stickman fucked around with this message at 05:28 on Apr 9, 2019 |
|
#
?
Apr 9, 2019 05:26
|
|
|
...
|
|
#
?
Apr 9, 2019 06:19
|
|
|
CommieGIR posted:Why 56 cores? is this supposed to compete with the 64 core Epyc? Two 28 cores (Intel's largest) glued together, then soldered to the motherboard.
|
|
#
?
Apr 9, 2019 06:54
|
|
|
.
sincx fucked around with this message at 05:50 on Mar 23, 2021 |
|
#
?
Apr 9, 2019 07:00
|
|
|
I'm still waiting to see how they try to spin this at Computex.
|
|
#
?
Apr 9, 2019 08:01
|
|
|
SwissArmyDruid posted:I'm still waiting to see how they try to spin this at Computex.
|
|
#
?
Apr 9, 2019 08:37
|
|
|
Budzilla posted:They could say that they are glued together like their competitor's CPUs. Instead of calling it "glued together", Intel settled for calling the process "foveros".
|
|
#
?
Apr 9, 2019 08:39
|
|
|
|
| # ? May 16, 2024 10:43 |
|
|
I found something in the attic of my parents I guess that's from the late 90s? 
|
|
#
?
Apr 9, 2019 13:13
|
|
