|
I sure hope the current rumors of an 7950X3D are gonna turn out true.
|
#
?
Aug 4, 2022 19:05
|
|
|
|
| # ? Jun 7, 2024 04:34 |
|
|
As expected, with AMD continuing to milk this launch as long as possible, the board partners showed a ton of their premium X670E boards today. Looks like MSI is the only one to pull 10Gbe down a tier from the halo boards, the ACE will have it as well as the Godlike. Kinda of a bummer, was hoping 10Gbe would also find its way on to stuff like the Hero. By the way all of these boards are insanely overbuilt. The hero is 18+2 with 110A power stages, with a top SKU having a 170w TDP, lol. I guess who knows, maybe this socket will last a while like AM4 and some future monster TDP part will exist. https://videocardz.com/newz/amd-shows-off-premium-x670e-motherboards
|
|
#
?
Aug 4, 2022 20:05
|
|
|
Combat Pretzel posted:I sure hope the current rumors of an 7950X3D are gonna turn out true. It makes more sense than a 7900X3D. Why waste effort putting 3d vcache on chips that are cut down? AFAIK there's nothing that would prevent AMD from validating chiplets before stacking, so it would make sense to only do it with top-binned chiplets. That said, for those reasons and more, expect the 7800X3D and potential 7950X3D to be very expensive. AMD has no reason to go cheap on those CPUs, they are halo products for a very small market. Their existence is more important than their sales.
|
|
#
?
Aug 4, 2022 20:08
|
|
|
I do enjoy that some board makers are apparently willing to pay Intel to use the Thunderbolt name for their AMD boards, and some are not. Love my compatibility with "TBT" devices.
|
|
#
?
Aug 4, 2022 20:29
|
|
|
Cygni posted:As expected, with AMD continuing to milk this launch as long as possible, the board partners showed a ton of their premium X670E boards today. Looks like MSI is the only one to pull 10Gbe down a tier from the halo boards, the ACE will have it as well as the Godlike. Kinda of a bummer, was hoping 10Gbe would also find its way on to stuff like the Hero.
|
|
#
?
Aug 4, 2022 20:29
|
|
|
Khorne posted:I stopped caring about 10gbe on boards because you can get 10gbe & 40gbe cards for $10 each on ebay. And they have the added benefit of not being rj45 which lets you get cheaper switches. They also run cooler/use less power, and if you really need rj45 for something you can get an adapter. True, but I would prefer not to add any card at all next to a 3.5+ slot GPU if I can avoid it. Guess i could get creative and put the NIC somewhere else with a ribbon cable, lol
|
|
#
?
Aug 4, 2022 20:45
|
|
|
K8.0 posted:That said, for those reasons and more, expect the 7800X3D and potential 7950X3D to be very expensive. AMD has no reason to go cheap on those CPUs, they are halo products for a very small market. Their existence is more important than their sales.  Khorne posted:I stopped caring about 10gbe on boards because you can get 10gbe & 40gbe cards for $10 each on ebay. And they have the added benefit of not being rj45 which lets you get cheaper switches. They also run cooler/use less power, and if you really need rj45 for something you can get an adapter. Combat Pretzel fucked around with this message at 21:02 on Aug 4, 2022 |
|
#
?
Aug 4, 2022 21:00
|
|
|
Combat Pretzel posted:Current system is a Threadripper. I put a Chelsio T6 in my TR box because what else is there to do with all those PCIe lanes? PCIe 5.0 in the long run will result in some hilarious densities of having 16+ x1 links that can all run at 32 GT/s.
|
|
#
?
Aug 6, 2022 23:38
|
|
|
K8.0 posted:It makes more sense than a 7900X3D. Why waste effort putting 3d vcache on chips that are cut down?. Honestly I have no idea but the current epyc lineup has 3D cache CCDs with 8, 4, 3, and 2 cores enabled. 96MB of L3 cache for 2 cores feels silly but the reduced density means the 7673x has a base clock of 3.05GHz and a boost clock of 3.8GHz which seems solid next to the full populated 8x8 ccd 7773x needing to have a base clock of 2.2GHz. Maybe it’s that simple.
|
|
#
?
Aug 7, 2022 04:27
|
|
|
I figure they'll be able to make higher clocked V-Cache desktop CPUs, because it's implied that buyers of these will be applying bigger and better heatsinks than you'd typically find in datacenters. That said, I'd expect lower boost frequencies.
|
|
#
?
Aug 7, 2022 09:01
|
|
|
hobbesmaster posted:2 cores enabled. Athlon 3000X3D   
|
|
#
?
Aug 7, 2022 11:52
|
|
|
okay you jest but i am now intensely curious how that would run
|
|
#
?
Aug 7, 2022 12:00
|
|
|
hobbesmaster posted:Honestly I have no idea but the current epyc lineup has 3D cache CCDs with 8, 4, 3, and 2 cores enabled. Yeah I think it's just one form of market segmentation or another. If you have a low threadcount workload that really benefits from the cache, but will max out power/thermals on just 2 cores per CCD, paying less than half as much for 7373Xs probably makes sense. It's not the same in the consumer market, because no one is paying such a huge premium for the chips and cache.
|
|
#
?
Aug 7, 2022 15:46
|
|
|
threadripper pro 5995WX review. note that they don't throw in a 5800X3D to compare with https://www.youtube.com/watch?v=yQ-c6SmVuDY
|
|
#
?
Aug 8, 2022 15:08
|
|
|
kliras posted:threadripper pro 5995WX review. note that they don't throw in a 5800X3D to compare with cups the size of ipads
|
|
#
?
Aug 9, 2022 05:29
|
|
|
Combat Pretzel posted:I figure they'll be able to make higher clocked V-Cache desktop CPUs, because it's implied that buyers of these will be applying bigger and better heatsinks than you'd typically find in datacenters. That said, I'd expect lower boost frequencies. Note that while data center heatsinks are usually slimmer than high-end consumer ones, this doesn't mean they are necessarily any less powerful. Unlike in the consumer space, DC doesn't give a poo poo about noise, so the airflow over the heatsink can be 10x higher.
|
|
#
?
Aug 9, 2022 13:05
|
|
|
asus finally updated their remaining motherboards with an agesa 1.2.0.7 bios update btw you probably don't need to update if you're on win10 with ftpm disabled in bios though
|
|
#
?
Aug 10, 2022 17:08
|
|
|
kliras posted:asus finally updated their remaining motherboards with an agesa 1.2.0.7 bios update btw I thought using the 5800x3d was an exception but in looking for details it apparently reduces performance? It’s a pretty consistent 1% in synthetics. The rumor was that the .7 AGESA had fixed some issues for the x3d. I guess fixes may actually make things slower I guess. My x3d is running on a MSI x570 board with AGESA .7 and it’s current cinebench performance is a hair better than the original firmware from April or whatever it was released. I am running a negative voltage offset from PBOtuner now however, so I guess that made up the difference.
|
|
#
?
Aug 10, 2022 19:58
|
|
|
If you have a use for it, just PM me. I'll edit this post when it's been claimed. Claimed! mdxi fucked around with this message at 03:42 on Aug 14, 2022 |
|
#
?
Aug 13, 2022 19:53
|
|
|
Just got this error in my event viewer My processor's a Ryzen 5 3600 I got new a couple years ago. Does anyone know what could be the cause?
|
|
#
?
Aug 13, 2022 22:40
|
|
|
NotNut posted:Just got this error in my event viewer Are you undervolting or anything similar?
|
|
#
?
Aug 13, 2022 22:52
|
|
|
Kibner posted:Are you undervolting or anything similar? No. I don't even know how to do that
|
|
#
?
Aug 14, 2022 00:26
|
|
|
Until you see it happen more than once, I would just ignore it. Maybe update your bios if you feel like you need to do something.
|
|
#
?
Aug 14, 2022 01:38
|
|
|
mdxi posted:I have a Ryzen 3600 which I will ship to anyone in the US/Canada. It was gently used in my desktop for a while. It's been cleaned and packaged in the clamshell + box of the 5600G which replaced it, including the factory HSF. I'm not asking for any money, which is why I'm not putting this in SA-Mart. PMed you about the Ryzen!
|
|
#
?
Aug 14, 2022 03:06
|
|
|
https://twitter.com/VideoCardz/status/1559223593267040257 If true, that would be the same day Raptor Lake is formally announced, with its availability expected in October. Announcement for Zen 4 still expected this month.
|
|
#
?
Aug 15, 2022 19:27
|
|
|
Cygni posted:https://twitter.com/VideoCardz/status/1559223593267040257 When's the last time that we had competitive CPUs launching directly alongside each other like this? 2023 should be a good year for nerds.
|
|
#
?
Aug 15, 2022 20:33
|
|
|
It’s newegg but they have a 5600 for $140 which is a fine price point for a “last in socket” upgrade that probably won’t hold back the next gen GPUs. https://www.newegg.com/amd-ryzen-5-5600-ryzen-5-5000-series/p/N82E16819113736?Item=N82E16819113736
|
|
#
?
Aug 16, 2022 02:57
|
|
|
are there any settings i can look to reducing power draw for a Threadripper 3960X? total system draw is 180W at "idle" but shoots up to 280W on even a light load (e.g., YouTube) i've tried stuff like disabling PBO but it didn't do anything. oddly enough, changing screen refresh from 144 to 100hz reduced power draw by 30W i'm OK with high draw when doing compute or gaming but would like to reduce it for light loads ideally
|
|
#
?
Aug 16, 2022 03:24
|
|
|
Could use some help identifying probable cause here: - Desktop (5950x, RTX 3080, MSI Unify x570) wouldn't power on this morning. Last used last night about 10 hours earlier, clean shutdown. - Power switch on front was unresponsive. Switch lights up when started normally, even that light wouldn't come on. - No evidence that any power was flowing to the motherboard. - Switched between power ports on strip, switched between power strips, switched between wall sockets. No dice. All other devices on those strips worked with no problems. - Tried a few times over the first hour and decided to go buy a new power supply as a first check when I had time later. - Randomly tried power switch on front roughly 3.5 hours later and the machine booted like it had never had a problem. Any ideas as to what this could be? Should I be looking at a new power supply? v1ld fucked around with this message at 17:08 on Aug 16, 2022 |
|
#
?
Aug 16, 2022 17:04
|
|
|
shrike82 posted:are there any settings i can look to reducing power draw for a Threadripper 3960X? Don't know anything about Threadripper but: are you using the AMD power plan? Check with any of the utilities that show you hardware info if your cores are actually spending time in idle states. Though that power draw could well be your GPU as evidenced by your monitor change.
|
|
#
?
Aug 16, 2022 17:11
|
|
|
v1ld posted:Could use some help identifying probable cause here: What's the make and model of your power supply?
|
|
#
?
Aug 16, 2022 17:50
|
|
|
Sidesaddle Cavalry posted:What's the make and model of your power supply? CORSAIR - RMx Series RM850x 80 PLUS Gold Purchased from Best Buy on Aug 22nd 2021. Wonder if it's still under warranty and I should ask for a replacement just to be on the safe side. E: There's a 2 year Corsair manufacturer warranty on it. v1ld fucked around with this message at 18:26 on Aug 16, 2022 |
|
#
?
Aug 16, 2022 18:14
|
|
|
That seems like a problem in the power delivery chain. Could be a loose cable somewhere in the chain from the wall (including internally), could be a bad random bitflip in the bios eprom, could be a shaky power button. You may laugh, but i have literally seen that happen. Could also be that the computer ghosts were mad at you today. Modern computers are fuckin weird sometimes and the chain of power protection and sensing features in the PSU and board sometimes gets off sync, misreads a variable, gets something stuck in memory, or just wigs the gently caress out. If you've reseated all the cables and eliminated the obvious failure points, and it still happens again or you are doing something absolutely uptime critical, the PSU is probably the first place to start swapping. If it happens again, I would also swap the DIMMs around (and try booting with just 1) before i bought anything. I know that sounds dumb for a probable power issue, but I can't tell you how many strange failure modes I've fixed by swapping the DIMMs and forcing the board to retrain the timings, which sometimes stick even after BIOS clears. Again, computer ghosts at play.
|
|
#
?
Aug 16, 2022 18:46
|
|
|
Thanks, that's all very sound advice. I'll keep an eye on things and reseat stuff internally if it happens again. The complete lack of response to power on followed a few hours later by "what, you called earlier? didn't hear you" like it was no problem at all is just super weird enough that it probably has a very mundane cause. Knowing I have another year of warranty is nice.
|
|
#
?
Aug 16, 2022 18:53
|
|
|
v1ld posted:- Switched between power ports on strip, switched between power strips, switched between wall sockets. No dice. All other devices on those strips worked with no problems. So, this does sound a lot like the PSU was responsible, but I think more like the lock-out from a short or other protection rather than failure. When you trip the protections they typically won't turn on until you unplug or flip the rear switch for a bit. If the switching strips & sockets you did was quick each time I suppose it's possible that you never de-powered it long enough to reset the lock-out. Alternately, maybe you have a intermittent short. I'd open up the PC and review your cables & plugs, particularly if you do fancy cable management routing (which can pinch wires). Also maybe tilt the case around to listen for a loose screw or whatnot. Aside from that, anything that happens once is coincidence. Until it happens a second time I'd note it and move on. Though as a totally separate thing, IMO it's always nice to have a second PSU around. PSU problems can be pretty whack to diagnose, so swapping a PSU is high on my list of shotgun troubleshooting for persistent mystery problems. A cheaper model is fine since it's just the backup / spare used to repair your friend's PC. v1ld posted:E: There's a 2 year Corsair manufacturer warranty on it. That PSU should have a 10 year warranty.
|
|
#
?
Aug 16, 2022 19:00
|
|
|
shrike82 posted:are there any settings i can look to reducing power draw for a Threadripper 3960X? 2950X without PBO, Geforce 3080, two 4K120 displays. Power draw is like 270ish watts on the current meter on light browsing workload. That said, 105-110W on idle with displays in sleep mode. That said, I guess subtract around 20W, because I'm still getting 24W on the meter due to the modem, RPi and UPS on said socket. --edit: I'm anticipating the drop in power draw when moving to Zen 4 on AM5.
|
|
#
?
Aug 16, 2022 19:14
|
|
|
Klyith posted:That PSU should have a 10 year warranty. It doesn't because I think the platform is from CWT. I think only their Seasonic platforms come with 10 year warranties iirc. This is hard information to find now... Pouring one out for JonnyGuru.
|
|
#
?
Aug 16, 2022 19:48
|
|
|
Klyith posted:So, this does sound a lot like the PSU was responsible, but I think more like the lock-out from a short or other protection rather than failure. When you trip the protections they typically won't turn on until you unplug or flip the rear switch for a bit. This could be it. Both the strips have their protection lights still on but it's still possible the PSU tripped. While I flipped the rear switch, I may not have waited long enough after that for a lock-out to expire. Didn't know PSU did any form of lock out, that's good to know too. quote:Though as a totally separate thing, IMO it's always nice to have a second PSU around. PSU problems can be pretty whack to diagnose, so swapping a PSU is high on my list of shotgun troubleshooting for persistent mystery problems. A cheaper model is fine since it's just the backup / spare used to repair your friend's PC. My last PSU fried a couple years out of warranty but the nearby Best Buy has Corsairs in stock so I haven't been too worried about having a replacement on hand. But it's a good idea. E: That last PSU lasted 8 years and was rated for 7. Just saying since I haven't shown very strong warranty reading skills so far in this thread. It imploded when it decided to leave us, with some impressively heavy grey smoke that was heavier than air and flowed across the floor. Didn't damage anything within the case. quote:That PSU should have a 10 year warranty. It does, my mistake. Found the box it came in and that does say 10 years. I may have been looking at the wrong PSU on their support site. Thanks for pointing this out. v1ld fucked around with this message at 20:02 on Aug 16, 2022 |
|
#
?
Aug 16, 2022 19:58
|
|
|
Canine Blues Arooo posted:It doesn't because I think the platform is from CWT. I think only their Seasonic platforms come with 10 year warranties iirc. This is easily googlable. The RMX has had a 10 year warranty since 2016, and the plain RM have had 10 since 2019. Corsair uses a number of different OEMs, but any particular one is pretty ignorable because Corsair does way more work on their PSUs than the average Gigabyte or whatever brand. The RM(x), TX, & CX lines they input on design and specify the components. IMO whichever OEM did the manufacturing is kinda irrelevant in their case. Canine Blues Arooo posted:This is hard information to find now... Pouring one out for JonnyGuru. RIP. (He works for Corsair now.) Techpowerup is a very good source these days. And GN would be if their had a higher volume of reviews.
|
|
#
?
Aug 16, 2022 20:39
|
|
|
|
| # ? Jun 7, 2024 04:34 |
|
|
I got a 'TITAN WORKSTATION' PC build a guy bought direct from Titan PC's website. This thing is the biggest joke i've ever seen. It has some decent parts, Asus X570 ACE workstation board, 5950x CPU WITH a tiny 120mm noctua single fan cooler, and some 120mm noctua fans. The case has Zero airflow, as in a 1/2" gap on the front side panel and no top fans or vents, one 120mm rear vent/fan and a GTX 3090 FTW3.. absolute beast of a setup with zero airflow. And the best part a no-name chinese 1250w PSU.. that is turning itself off immediately. I decided not to risk it after one shut off in front of me after about 2 minutes. $4000+ bux and I have to spend another about $600 to get a working safe system with a good PSU and some airflow. Jesus
|
|
#
?
Aug 17, 2022 00:13
|
|
