|
Klyith posted:The ones that are actually good are not a big discount from Z590 boards, just because the VRMs needed to run those CPUs are so expensive. It's still easily a $50 difference.
|
# ? Nov 14, 2021 15:34 |
|
|
# ? Jun 6, 2024 17:45 |
|
Tbh going forward the recommendation for Intel boards probably will/should be the lower end chipsets now that memory oc is unlocked on them. There's no overclock headroom in basically any CPU anymore, the boost algorithms have gotten real good. Even a super overbuilt vrm is kinda silly - I think the 12900k at 150w is like 97% of the performance at 240w. Just cap the power at something not stupid and save the $50.
|
# ? Nov 14, 2021 18:26 |
|
It’s not too meaningful of a distinction in most cases but the x570 has pcie4 to the chipset instead of pcie3. That means the chipset can provide pcie4 to a chipset m2 but also to other pcie slots. Honestly you could probably say “any” cases because it probably doesn’t even matter for Bitcoin mining but it’s there. Maybe if you want to use one of those multiple nvme pcie cards?
|
# ? Nov 14, 2021 19:34 |
|
hobbesmaster posted:It’s not too meaningful of a distinction in most cases but the x570 has pcie4 to the chipset instead of pcie3. That means the chipset can provide pcie4 to a chipset m2 but also to other pcie slots. Does that mean anything besides being limited to one gen 4 NVMe drive (if you have a rocket lake) on a B560?
|
# ? Nov 14, 2021 19:56 |
|
Back in June 2020, there was a whole Thing about how certain motherboard manufacturers might have been misrepresenting their power measurements to Zen CPUs in order to try and push the boost algorithm even further than what AMD's spec might otherwise let it, and it was enough of a thing that Anandtech had to come out with an article addressing this just to say that, no, folks, that's not going to kill your CPU, and it's not going to cause "electromigration", or the tendency for silicon to "degrade" over time as you feed it excessive voltage (i.e. for however much voltage you need to maintain a particular stable overclock, that you'll need more voltage after years of keeping such settings) This came up again with the release of Zen 3, and combined with internet arguments over "how much voltage is too much?", youtuber der8auer did a rough experiment of taking three Zen 3 CPUs and put them under a 24/7 stress test for 4000+ hours (about five-and-a-half months), which is the equivalent of maybe 3 to 4 years of actual/practical use, with a user putting the CPU under a heavy load for 3-4 hours per day https://www.youtube.com/watch?v=ZAww0c2m-ks The basic set-up is that all of the CPUs (two 5600X's, and one 5800X) were set to a flat 4.50 GHz with a voltage of 1.45v, which was well in excess of what would have been necessary to maintain that stable clock speed setting. der8auer then also took note of what kind of minimum voltage was necessary to run an clock of 4.70 GHz. This turned out to be 1.23v for one, and 1.26v for the two others. The idea was that if electromigration is a thing, then you might expect that after all these hours of being run under a full load and with excess voltage, that the clock:voltage ratio would no longer hold, and you need more than 1.23/1.26v to hold 4.70 GHz. None of the parts died. Not just the CPUs, but the PSUs, the CPU air coolers, and the motherboards all survived The 5800X managed to complete three runs of Cinebench at the original 4.70 GHz clock speed with the 1.26v setting. The first 5600X would complete two runs, then crash on the third. der8auer needed to increase the voltage from 1.262 to 1.268, but CB would crash on the third run again. With another change to 1.275v, the CPU would complete all three runs The second 5600X needed to be increased from its original 1.23v setting, to 1.28v, to be stable enough to complete all three CB runs This does suggest that electromigration is possible, but with some fairly big caveats: * presumably this wouldn't be an issue at all if you just left the CPUs completely at stock * even if you were manually overclocking, the test parameters were such that the CPUs were deliberately being fed voltage that was well in excess of what was necessary, either to simulate an overclocker that's brute-forcing the voltages, or even a boost algorithm that's far too aggressive. Most people still might not run into such a scenario * the test parameters were also assuming an amount of load on the CPUs that you'd really only hit after half-a-decade, if that, at which point the CPU would be well out of warranty, and possibly quite long in the tooth in terms of the upgrade cycle. * while I'm sure some of us are still rocking an i5-2500K after all these years, or even a Core2Quad, the electromigration still didn't kill the chip. You just have to give it a little more juice, and if you were ever already running the CPU on the ragged edge of your cooling solution such that more voltage isn't an option anymore, then you can still back off the OC a little to keep going * and the 5800X still didn't show signs of electromigration anyway I wouldn't draw any hard conclusions from this per se, but I thought the discussion and results was interesting
|
# ? Nov 15, 2021 08:15 |
|
Pretty sure there were people ITT spouting nonsense about how Ryzen CPUs are dying and boards are frying RAM, etc.
|
# ? Nov 15, 2021 16:58 |
|
CoolCab posted:a b550 is a very cost effective way to get a decently performing modern motherboard. it also uses passively cooled VRMs instead of the typical actively cooled x570s which is really only mandatory for beefier chips and overclocks and comes with some drawbacks. a 5600x does not require much, iirc it's a 65w tdp. You also go X570 if you want more than like 4 usb ports in the back
|
# ? Nov 15, 2021 17:39 |
|
Yeah B550 is great and all but please remember when Ryzen 3000 series launched it. did. not. exist. Not all of us going into those CPUs had the luxury of waiting on AMD for like a whole year either.
|
# ? Nov 15, 2021 22:22 |
|
Stanley Pain posted:Pretty sure there were people ITT spouting nonsense about how Ryzen CPUs are dying and boards are frying RAM, etc. i think that was about memory controllers if you use the XMP settings
|
# ? Nov 19, 2021 14:12 |
|
Truga posted:i think that was about memory controllers if you use the XMP settings Yeah, I don't remember any widespread CPU death or DIMMs being murdered. I do remember (and experienced) 1000 and 2000 series Ryzens becoming less stable over time when XMP was enabled. For me it was something that started happening after around a year of 24/7 computing, and the fix was just to turn XMP off, while leaving the RAM speed at its rated value (or in some instances, dropping it a bit until stability was found). Dunno about 3000 or later, because I quit enabling XMP.
|
# ? Nov 19, 2021 17:44 |
|
First gen Ryzen had a not-great memory controller, so you'd have to juice the voltage quite a bit to get decent memory speeds and that probably contributed to the premature aging and instability over time.
|
# ? Nov 19, 2021 19:29 |
|
SamDabbers posted:First gen Ryzen had a not-great memory controller, so you'd have to juice the voltage quite a bit to get decent memory speeds and that probably contributed to the premature aging and instability over time. In particular a last-resort thing for getting stable memory OCs with Ryzen 1 & 2 was to boost voltage to the SOC. Not a ton, but the SOC seemed to have less headroom. And it's also constant voltage. There were absolutely some internet stories of people ruining their Ryzens by thinking the SOC could take the same +.2v that the main CPU could.
|
# ? Nov 20, 2021 03:02 |
|
The goddamn chipset fan on my x570 gigabyte elite is wearing down, loud noises all around. I can't believe such a tiny thing is the problem. What do? My thinking is: update bios so I can control the curve, it may not actually be necessary but it keeps slightly spinning up. @BIOS is the way or do I need a usb pendrive for qflash? If not fixed lube up the fan. Never done it, but I can find a youtube vid and do it. Replace the fan? This poo poo looks proprietary, where do I get a replacement? I don't want to gently caress around with jerryrigging a noctua fan. RMA the mobo? Gonna take weeks if not months probably. Any ideas? I wanna punch the guy who thought this was good
|
# ? Nov 20, 2021 12:09 |
|
Truga posted:i think that was about memory controllers if you use the XMP settings at the time I was telling you all that, my 9900k was dying due to stock settings except XMP. Guess the motherboard went a bit too high on its voltage trying to get its memory qualification
|
# ? Nov 20, 2021 12:26 |
|
I would definitely try to set a fan curve to see if you can reduce the fan speeds and noise. Gonna look for those "lol you silly goons why do you even care about X570 chipset fans just lmao" posts we occasionally had, though (sorry, I know that won't help you )
|
# ? Nov 20, 2021 12:27 |
|
orcane posted:I would definitely try to set a fan curve to see if you can reduce the fan speeds and noise. Yeah I remember those posts I'll try the curve but my BIOS is so old I don't think I have that. I'm looking for PCH_FAN from what google tells me, yeah?
|
# ? Nov 20, 2021 12:29 |
|
Like if it was the standard fans I'd be fine with it, or gigabyte could unfuck a little pocket change and install the tiny noctuas, but no it has to be a proprietary fan with noneuclidean geometry. gently caress you
|
# ? Nov 20, 2021 12:30 |
|
dex_sda posted:If not fixed lube up the fan. Never done it, but I can find a youtube vid and do it. This is not a thing that actually works on 99% of PC fans. All types of sleeve / rifle / FDB bearings are sealed -- they have to be to work -- so if you can somehow get lube in that means it will come right back out. The small number that still use ball bearings (not a thing on a tiny chipset fan) may benefit but by the time the fan is making bad noises it's generally too late. If the bearing is going bad you're kinda hosed, but you should first take the computer out and clean it well. Maaaaybe it's just jammed with dust and gunk. dex_sda posted:Replace the fan? This poo poo looks proprietary, where do I get a replacement? I don't want to gently caress around with jerryrigging a noctua fan. The fan isn't going to be proprietary. It's not standard or common, but they're never 100% proprietary. If you are able to get it out of that heatsink thing and read the model #, you can get a replacement. However that may require buying from ali express and shipping from china. Could take just as long as the RMA.
|
# ? Nov 20, 2021 13:50 |
|
Klyith posted:This is not a thing that actually works on 99% of PC fans. All types of sleeve / rifle / FDB bearings are sealed -- they have to be to work -- so if you can somehow get lube in that means it will come right back out. The small number that still use ball bearings (not a thing on a tiny chipset fan) may benefit but by the time the fan is making bad noises it's generally too late. Klyith posted:The fan isn't going to be proprietary. It's not standard or common, but they're never 100% proprietary. If you are able to get it out of that heatsink thing and read the model #, you can get a replacement. However that may require buying from ali express and shipping from china. Could take just as long as the RMA. Yeah it sucks. I'll check to see if it's visible maybe.
|
# ? Nov 20, 2021 13:53 |
|
You can absolutely relube cheap sleeve bearings and usually get years more out of them. They also do not need to be sealed to work, and especially not sealed well. Also, LOADS of them are never sealed to begin with. Most bearings in general are not sealed. Many fan bearings are "sealed" by a sticker, and as long as you take it half off, use something like a toothpick to put a light grease into the little hole while turning and wiggling the fan to help distribute it, it will work fine in most cases. In other cases there is no hole, but if the fan is already hosed, just make a hole and try - it's not like you have anything to lose, and I've had it work. orcane posted:Gonna look for those "lol you silly goons why do you even care about X570 chipset fans just lmao" posts we occasionally had, though (sorry, I know that won't help you ) Yeah, there was a lot of this and it was embarrassingly stupid. The smaller a fan is, the shittier it is, and no one was going to put a high-end fan to cool a 13w max load. K8.0 fucked around with this message at 14:25 on Nov 20, 2021 |
# ? Nov 20, 2021 14:22 |
|
K8.0 posted:Many fan bearings are "sealed" by a sticker, and as long as you take it half off, use something like a toothpick to put a light grease into the little hole while turning and wiggling the fan to help distribute it, it will work fine in most cases. In other cases there is no hole, but if the fan is already hosed, just make a hole and try - it's not like you have anything to lose, and I've had it work. Yeah that's what I've seen on the youtubes
|
# ? Nov 20, 2021 14:47 |
|
up until the point that they came out with the x570S, my plan was always to wait until the fan starts dying, make an adapter bracket for a 40mm Noctua, snip the connector and splice if necessary.
|
# ? Nov 20, 2021 15:04 |
|
can you just unplug the silly little fan?
|
# ? Nov 20, 2021 15:23 |
|
CoolCab posted:can you just unplug the silly little fan? This imo
|
# ? Nov 20, 2021 16:06 |
|
Unplug the fan, check the chipset temp with something like HWinfo64 and if it doesn't go over like 90C and the system is stable I wouldn't worry about it.
|
# ? Nov 20, 2021 17:57 |
|
I'm so happy I got a X470 with my 3900X.
|
# ? Nov 20, 2021 19:46 |
|
CoolCab posted:can you just unplug the silly little fan? i might if i can find the connector, it must be hidden by the gfx card lol. either that or it's hidden inside the goddamn sick 1337 gamer cover on the lovely rear end fan in which case full rip i'll try doing it tomorrow
|
# ? Nov 20, 2021 19:49 |
|
I cleaned it up some extra and I noticed it gets better with each hot -> cold cycle so I restarted it a bunch of times and I think it loosened up whatever lube is inside cos it went quieter again. I'll open up the sticker, clean out the gunk and lube it up the next time this happens which will probably be soon. Thanks for the suggestions, gently caress the engineer who decided this
|
# ? Nov 21, 2021 10:22 |
|
Ah yes it's the engineers fault again that AMD rushed pcie 4.0 to the consumer market, on a crappy old GloFo node chipset and firmware nowhere near ready even for dev boards.
|
# ? Nov 21, 2021 11:34 |
|
your first mistake was buying gigabyte
|
# ? Nov 21, 2021 11:41 |
|
ijyt posted:your first mistake was buying gigabyte it really was man, although at the same time two of my close 'into building pcs' friends who made a ryzen platform got different stuff and one took 10 months to have a working PC with no issues because all the x570 mobos from other manufacturers kept loving up, and the other had the fan start loving up sooner. it was not ready for release dex_sda fucked around with this message at 12:06 on Nov 21, 2021 |
# ? Nov 21, 2021 12:04 |
|
5800X down to $340 on Amazon. 5900X $490.
|
# ? Nov 21, 2021 12:56 |
|
ijyt posted:your first mistake was buying gigabyte every Brand sucks, and the epic disaster that gigabyte has turned into recently wasn't foreseeable then. personally if I had a x570 with a mobo fan, I'd have jumped on every bios update to look for new options for the fan control. so if any mistakes were made, that would by my choice.
|
# ? Nov 21, 2021 14:14 |
|
i've had zero issues with my asus x570, but i'm on a 3950x and only since spring 2020. dunno if they had more issues with 5000 series or earlier on it does have a tiny fan on it, but i don't think it ever starts, and also is completely covered by the reference radeon heatsink lmfao
|
# ? Nov 21, 2021 14:36 |
|
Truga posted:i've had zero issues with my asus x570, but i'm on a 3950x and only since spring 2020. dunno if they had more issues with 5000 series or earlier on I got my x570 when the 3 series basically dropped, and I hear a lot of issues from those earlier boards
|
# ? Nov 21, 2021 14:38 |
|
Got a Gigabyte X570 Aorus pro wifi. No issues at all. Love it.
|
# ? Nov 21, 2021 17:13 |
|
Klyith posted:the epic disaster that gigabyte has turned into recently wasn't foreseeable then. Eh... we've known Gigabyte had questionable practices ever since they were caught several years ago making motherboard revisions over time where a previously released/advertised model, would see features dropped on the new revisions but weren't indicated as such. They'd also then send these out as RMAs so the people getting warranty replacements were effectively getting inferior boards. It's why if I were going to buy Gigabyte, it'd only be for an nVidia card since they have to meet nVidia's minimum requirements at least.
|
# ? Nov 21, 2021 17:44 |
|
I'd buy select Gigabyte monitors, because most monitor problems are on arrival and as long as you buy from a store with good return policies it doesn't matter who the manufacturer is. Everything else I'd stay away from. I've seen allegations of them refusing RMAs on GPUs that have clear PCB defects.
|
# ? Nov 21, 2021 22:13 |
|
K8.0 posted:I'd buy select Gigabyte monitors, because most monitor problems are on arrival and as long as you buy from a store with good return policies it doesn't matter who the manufacturer is. Everything else I'd stay away from. I've seen allegations of them refusing RMAs on GPUs that have clear PCB defects. They deny warranties on GPUs with signs of stress near the PCIe connectors even if the stress has nothing to do with the issue, all while making cards they know will naturally exhibit said stress because GPUs are so loving big these days.
|
# ? Nov 21, 2021 22:37 |
|
|
# ? Jun 6, 2024 17:45 |
|
Well I just bought a Z590 from Gigabyte. Thought it was just exploding PSUs that we had to worry about.
|
# ? Nov 21, 2021 22:53 |