|
redeyes posted:I've never been so happy to have an Intel that only runs at 100c. hahaha Ryzen 5000 stays winning
|
# ? Apr 30, 2023 17:08 |
|
|
# ? May 31, 2024 21:09 |
|
FuturePastNow posted:Congrats to Asus for making overcurrent protection that doesn't work at all Doesn't look like they collaborated on anything at all
|
# ? Apr 30, 2023 17:41 |
|
Zedsdeadbaby posted:Doesn't look like they collaborated on anything at all That is definitely a failure too
|
# ? Apr 30, 2023 18:05 |
|
I'm sorry if this has been addressed already, but I just found this topic. I got one of the Micro Center bundles a few weeks ago, so I've got a 7900X and an Asus B650E-F. I'm out of my return window for everything (and I also stupidly broke a USB 3 header pin off of the motherboard while I was assembling everything), so what I've got is what I've got. I had expo and able to get the RAM up to 6000 for a little while, but it wasn't long before all this started coming out, so I turned that off. With it off, my SOC voltage topped out at about 1.07, and with it on, about 1.27. All this voltage stuff kind of goes over my head, as I've never manually overclocked or undervolted anything, but from everything I've read, I should be safe, it sounds like. Can anybody confirm this for me? I know most of the problems were with the X3D chips, and Gamers Nexus did say that they actively had to try for a catastrophic failure, but I'm all paranoid now. (Also, I've been updating the BIOS to every non-beta version that gets released.) pedrovay2003 fucked around with this message at 18:12 on Apr 30, 2023 |
# ? Apr 30, 2023 18:09 |
|
Think they were blowing up at 1.39 or so. But that still seems uncomfortably high. It should be 1.20 on a B650
|
# ? Apr 30, 2023 18:13 |
|
1.2 or 1.25 seem to be the norm, and there's only talk of slow degradation when it goes beyond 1.3. 1.27 is likely pretty safe, but I'd manually set the SOC voltage at 1.2 just to be safe. I've seen people do 1.1 even, but I'm not sure that's guaranteed to work with DDR5-6000 on all chips.
|
# ? Apr 30, 2023 18:19 |
|
Dr. Video Games 0031 posted:1.2 or 1.25 seem to be the norm, and there's only talk of slow degradation when it goes beyond 1.3. 1.27 is likely pretty safe, but I'd manually set the SOC voltage at 1.2 just to be safe. I've seen people do 1.1 even, but I'm not sure that's guaranteed to work with DDR5-6000 on all chips. I'll definitely look into this. Do I have to change any other settings in the BIOS, or should just finding the SOC setting and setting it to 1.2 be enough?
|
# ? Apr 30, 2023 18:23 |
|
pedrovay2003 posted:I'll definitely look into this. Do I have to change any other settings in the BIOS, or should just finding the SOC setting and setting it to 1.2 be enough? Just setting it should be fine. Assuming it's the same layout as the x570 bios, you want to switch to Advanced Mode -> AI Tweaker -> arrow down to DIGI+ VRM section -> "VDDCR SOC Voltage". OTOH there's a chance that your PC won't be stable running the memory at 6000 with a lower SOC voltage. At that point you'd have to make some decisions between running the memory slower with manual settings, leaving the voltage higher, or further experimentation. edit: lmao I found that video just as an example of the asus bios layout... but man is that a great example of the thing I was talking about with extreme OCers versus the needs and safties of a consumer motherboard. That's buildzoid complaining about a board that not "having a useful voltage range" as he tries to set the memory to 2V. Which would kill or quickly degrade many types of memory. Only a few examples can stand up to that much juice. Klyith fucked around with this message at 18:57 on Apr 30, 2023 |
# ? Apr 30, 2023 18:50 |
|
pedrovay2003 posted:I'll definitely look into this. Do I have to change any other settings in the BIOS, or should just finding the SOC setting and setting it to 1.2 be enough? Use HWINFO64 to check what the actual voltage is once booted. The GET voltage not meeting the SET voltage was one of the many problems that GN found. So long as it’s below 1.3 in windows under load I wouldn’t worry about it for the moment.
|
# ? Apr 30, 2023 18:52 |
|
I really appreciate the insight, everyone. I think I've given myself a headache with all of this research, but I'm glad that it sounds like I didn't waste my money with my upgrades. I just need to learn techniques to be more careful than usual.
|
# ? Apr 30, 2023 19:03 |
|
Nah, your component selection is fine. It's important to know that this has only been confirmed to have happened to a small handful of people, making it an exceedingly rare issue. Just make sure the voltages are sane, and it's all good.
|
# ? Apr 30, 2023 19:28 |
|
Dr. Video Games 0031 posted:Nah, your component selection is fine. It's important to know that this has only been confirmed to have happened to a small handful of people, making it an exceedingly rare issue. Just make sure the voltages are sane, and it's all good. That does make me feel better. I know that the non-3D chips are also safer, so I'm taking solace in that.
|
# ? Apr 30, 2023 19:50 |
|
On my 7950X3D, my current settings are VSOC to 1.15V and VDDIO/MVDD/MVDDQ (all sorts of memory related voltages) to 1.2V, for overclocking DDR5-4800 UDIMMs to 6000. Seems to work fine so far, i.e. 0 WHEA errors (ECC modules), no BSODs under load. I suppose I could attempt to go lower, but seems like there's enough headroom, to make it unworthwhile to potentially create drama on my system.
|
# ? Apr 30, 2023 20:37 |
|
Dr. Video Games 0031 posted:Disable EXPO immediately and update to a newer BIOS. Yeah I turned EXPO off as soon as I saw those voltages and just installed 1401, and now seeing SOC at 1.25V. Maybe will tune that down later on but for now this seems Safe Enough.
|
# ? Apr 30, 2023 20:54 |
|
Combat Pretzel posted:On my 7950X3D, my current settings are VSOC to 1.15V and VDDIO/MVDD/MVDDQ (all sorts of memory related voltages) to 1.2V, for overclocking DDR5-4800 UDIMMs to 6000. Seems to work fine so far, i.e. 0 WHEA errors (ECC modules), no BSODs under load. I suppose I could attempt to go lower, but seems like there's enough headroom, to make it unworthwhile to potentially create drama on my system. Might I ask which DDR5 ECC UDIMMs you bought?
|
# ? Apr 30, 2023 22:33 |
|
phongn posted:Might I ask which DDR5 ECC UDIMMs you bought?
|
# ? Apr 30, 2023 23:19 |
|
Zedsdeadbaby posted:Doesn't look like they collaborated on anything at all that lone overworked BIOS engineer probably spent his entire time to get memory training that got exponentially more complicated in DD5 working than worrying about anything else
|
# ? May 1, 2023 02:16 |
|
on the up side, since I'm still too scared to play any games on my pc rn, so I'm about to do the completely insane thing that is turning on my ps5 and actually playing it.
|
# ? May 1, 2023 02:56 |
|
In AMD's defense re: ProcHOT not working... with an AIO working properly it's pretty hard to get something like a 7950X up to 95C before it hits the PPT limit (142W?), they need to hire someone whose sole purpose is to do very wrong, absurd things on purpose and destroy unholy amounts of hardware.
|
# ? May 1, 2023 04:53 |
|
New Zealand can eat me posted:In AMD's defense re: ProcHOT not working... with an AIO working properly it's pretty hard to get something like a 7950X up to 95C before it hits the PPT limit (142W?), they need to hire someone whose sole purpose is to do very wrong, absurd things on purpose and destroy unholy amounts of hardware. ah yes, AMD, the famous impoverished kid that works in his mom's basement
|
# ? May 1, 2023 04:55 |
|
New Zealand can eat me posted:In AMD's defense re: ProcHOT not working... with an AIO working properly it's pretty hard to get something like a 7950X up to 95C before it hits the PPT limit (142W?), they need to hire someone whose sole purpose is to do very wrong, absurd things on purpose and destroy unholy amounts of hardware. ...1usmus, Buildzoid, Kingpin, at least one of those guys is looking for a job.
|
# ? May 1, 2023 04:57 |
|
It's criminal that buildzoid has to pay for any of his hardware. He's such a treasure trove I'm not sure what point you're trying to make. The idea here is that all of the other limitations were working and this is the first generation where they hit those before they hit the thermal ones. E: It wasn't you're just too quick on the draw. New Zealand can eat me fucked around with this message at 05:38 on May 1, 2023 |
# ? May 1, 2023 04:58 |
|
New Zealand can eat me posted:It's criminal that buildzoid has to pay for any of his hardware. He's such a treasure trove If that's directed at me, It was an agreement and a short list of names that would love to break bioses over their knees, repeatedly, and get paid for it.
|
# ? May 1, 2023 05:27 |
|
SwissArmyDruid posted:If that's directed at me, It was an agreement and a short list of names that would love to break bioses over their knees, repeatedly, and get paid for it. a $24B/year company spending maybe like $100K/year on guys doing extreme CPU testing and generating good PR for is bad actually
|
# ? May 1, 2023 05:57 |
|
grack posted:Gamer's Nexus video on the 7800X3D "Rapid unscheduled disassembly", lmao
|
# ? May 1, 2023 07:48 |
|
On my 7900X and ASUS B650E-F, hwinfo reports the core VIDs, VDDCR_VDD, VDDCR_SOC, and VDD_MISC. Is VDDCR_SOC the only one I should be concerned about getting above 1.3V?
|
# ? May 1, 2023 08:04 |
|
One of the temperature sensors on my ASUS B650 board is constantly reporting being at 106 C, even at idle. What the gently caress
|
# ? May 1, 2023 08:15 |
|
Tbh sounds like the amd boards are still as bad as 20 years ago.
|
# ? May 1, 2023 08:54 |
|
This poo poo just solidified my decision to skip AM5 for a while yet, so I just pulled the trigger on a 5800X3D to upgrade my 3700X/X570 build while they're still available, and a 2x32GB 3200MHz kit that matches my old 4x16GB. Also swapping my 3070ti for a 3080 with a friend who doesn't need a 3080 in a secondary PC for some cash in between, and have a 2TB Kingston KC3000 drive sitting ready to go as well, to replace an a2000 1TB. At the end of this, I hope to have a (reasonably) faster computer with 96GB ram, and to use the leftover parts in a HP prebuilt that's sitting without CPU/RAM/NVMe for a new proxmox host.
|
# ? May 1, 2023 09:10 |
|
I've also now just noticed that every two seconds the RGB LEDs on my Corsair H150i will blink off for a split second. I think this machine might just be straight up haunted. Anyone here got any holy water I can borrow?
|
# ? May 1, 2023 09:21 |
Combat Pretzel posted:The 16GB 1R ones of these: https://www.kingston.com/en/memory/server-premier/ddr5-4800mts-ecc-unbuffered-dimm
|
|
# ? May 1, 2023 12:59 |
|
Wibla posted:This poo poo just solidified my decision to skip AM5 for a while yet, so I just pulled the trigger on a 5800X3D to upgrade my 3700X/X570 build while they're still available, and a 2x32GB 3200MHz kit that matches my old 4x16GB. Consider a 3600mhz kit, that's the sweet spot for 5000 series
|
# ? May 1, 2023 13:06 |
|
Josh Lyman posted:On my 7900X and ASUS B650E-F, hwinfo reports the core VIDs, VDDCR_VDD, VDDCR_SOC, and VDD_MISC. Is VDDCR_SOC the only one I should be concerned about getting above 1.3V? Yes. Kazinsal posted:One of the temperature sensors on my ASUS B650 board is constantly reporting being at 106 C, even at idle. That is normal, it's what happens when the chip doesn't have a thermistor probe hooked up to that sensor slot. They'll generally report their min or max value. So you'll have one entry in hwinfo that's below freezing or above 100C and you just ignore it because it's impossible. Or anyways it was impossible before chips started spontaneously combusting. Wibla posted:This poo poo just solidified my decision to skip AM5 for a while "Don't buy the first gen of new DDR platforms" has been pretty solid advice for the last 20 years.
|
# ? May 1, 2023 13:34 |
|
To be clear, there will usually be two places where SOC voltage is reported, one by your CPU and another by your motherboard. VDDCR_SOC is the CPU reading, and there's probably a separate voltage reading under your motherboard in hwinfo. For gigabyte, it's reported as "CPU VCORE SoC," but i'm not sure about Asus. I'm also not sure which is the "real" one, but the Gigabyte motherboard reading likes to hover around 50mV higher than the CPU reading.
Dr. Video Games 0031 fucked around with this message at 14:04 on May 1, 2023 |
# ? May 1, 2023 13:45 |
|
Kazinsal posted:I've also now just noticed that every two seconds the RGB LEDs on my Corsair H150i will blink off for a split second. Was that with a HWInfo sensor window open? Mine was doing the same thing, stopped as soon as I turned off HWInfo.
|
# ? May 1, 2023 14:22 |
|
Enos Cabell posted:Was that with a HWInfo sensor window open? Mine was doing the same thing, stopped as soon as I turned off HWInfo. I had the problem the entire time I owned my last pc, gonna be slightly salty if it turns out this whole time it was because I always had hwinfo running so I can have temps on my task bar
|
# ? May 1, 2023 14:29 |
|
Dr. Video Games 0031 posted:To be clear, there will usually be two places where SOC voltage is reported, one by your CPU and another by your motherboard. VDDCR_SOC is the CPU reading, and there's probably a separate voltage reading under your motherboard in hwinfo. For gigabyte, it's reported as "CPU VCORE SoC," but i'm not sure about Asus. I'm also not sure which is the "real" one, but the Gigabyte motherboard reading likes to hover around 50mV higher than the CPU reading. Do I need to worry about the mobo's "CPU VDDIO / MC" sensor? Some of the posts earlier were discussing that as well.
|
# ? May 1, 2023 14:38 |
|
Josh Lyman posted:Scrolling down to the mobo, my Asus does have a "CPU SOC" which is about 30mV higher than the CPU sensor. same cept for me it's 20 instead of 30, strix x670e-e
|
# ? May 1, 2023 14:47 |
|
Dr. Video Games 0031 posted:I'm also not sure which is the "real" one, but the Gigabyte motherboard reading likes to hover around 50mV higher than the CPU reading. Josh Lyman posted:Scrolling down to the mobo, my Asus does have a "CPU SOC" which is about 30mV higher than the CPU sensor. They're both real, just measuring at different locations! And voltage on the mobo sensor being higher than the CPU sensor is also totally normal. The mobo is measuring it somewhere on the VRM side. Then the voltage drops a bit from resistive loss as it goes through the power plane, the pins, and into the CPU. So the CPU sensor on the inside (which also isn't a super-accurate sensor) sees less volts. This chunk of video is a good explanation of the many different sensors for voltage, and how the most accurate one you can't even see because it's connected directly between the CPU die and the VRM controller. (And also an explanation of what LLC is.) Josh Lyman posted:Do I need to worry about the mobo's "CPU VDDIO / MC" sensor? Some of the posts earlier were discussing that as well. No, as far as anyone knows that's not being mucked with by mobo bios.
|
# ? May 1, 2023 15:27 |
|
|
# ? May 31, 2024 21:09 |
|
Klyith posted:"Don't buy the first gen of new DDR platforms" has been pretty solid advice for the last 20 years. So true. The second revision/generation of mobos/ram/stuff is cheaper, faster and simpler to use
|
# ? May 1, 2023 15:34 |