• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

5950x and PBO overheated something or busted custom loop?

phill

Moderator
Staff member
Joined
Jun 8, 2011
Messages
13,601 (3.24/day)
Location
Somerset, UK
System Name Not so complete or overkill - There are others!! Just no room to put! :D
Processor Ryzen Threadripper 3970X
Motherboard Asus Zenith 2 Extreme Alpha
Cooling Lots!! Dual GTX 560 rads with D5 pumps for each rad. One rad for each component
Memory Viper Steel 4 x 16GB DDR4 3600MHz not sure on the timings... Probably still at 2667!! :(
Video Card(s) Asus Strix 3090 with front and rear active full cover water blocks
Storage I'm bound to forget something here - 250GB OS, 2 x 1TB NVME, 2 x 1TB SSD, 4TB SSD, 2 x 8TB HD etc...
Display(s) 3 x Dell 27" S2721DGFA @ 7680 x 1440P @ 144Hz or 165Hz - working on it!!
Case The big Thermaltake that looks like a Case Mods
Audio Device(s) Onboard
Power Supply EVGA 1600W T2
Mouse Corsair thingy
Keyboard Razer something or other....
VR HMD No headset yet
Software Windows 11 OS... Not a fan!!
Benchmark Scores I've actually never benched it!! Too busy with WCG and FAH and not gaming! :( :( Not OC'd it!! :(
I have two ideas

1) Just get a replacement board for my daily driver.

I already have quite an investment around my 011D case so getting something with that is my best option to spend the least amount of cash.
If my 5950x is toast I can still fall back to my 3950x and just wait for the AM5 platform to mature a bit and reconsider my options.

Thinking about the following options more and what fits best for my current setup.

a) - https://www.newegg.com/asrock-x570-steel-legend-wifi-ax/p/N82E16813157894
not a fan of some of the color scheme for the shrouds but some spray paint can fix that
does not have USBC header for 011D ( I didn't have one for the Master SLI/ac so not much of a loss )
this is reasonably priced and seems basically a matching replacement feature for feature (or even a bit better) for what I had plus better VRM
Still considering this

b) - https://www.newegg.com/asrock-x570-taichi/p/N82E16813157883?Item=N82E16813157883
USBC header likely unusable because of conflicting position with graphics card
older and perhaps a more proven and reliable series, but also priced a bit higher
has more layers than steel legend, higher quality PCB?
I am heavy leaning in this direction as a replacement

c) - https://www.newegg.com/asus-prime-x570-pro/p/N82E16813119196R
I might consider this Asus board.

c) - https://www.newegg.com/asrock-x570-pg-velocita/p/N82E16813157969?Item=N82E16813157969
unsure about the quality of the Killer E3100G 2.5GB LAN, I've seen some complaints, nice on paper and good a nice to have if it really works well, if junk then worst case scenario get addin card for LAN
nahimic audio, my ITX board has this and it broke with windows update so non-working feature I don't want to pay for
a bit cheaper than Tiachi, the chipset fan is unobstructed so also easy to clean too


2) Take my home server ASRock B450m Pro4 (rev1) R5-2700 and swap in the 5950x/3950x into that and repurpose that as my daily driver.

It's basically setup and all ready to go other than swapping out the CPU, I have to wait for my CPU waterblock to arrive for that.
According to https://docs.google.com/spreadsheet...qVxdCR9daIVNyMatydkpFA/htmlview#gid=639584818
this should be doable with 3950x (no PBO or overclocking of course) but I will have some all core workloads so I worry if that is really feasible with this board.
I kind of don't want to bother with this option but at the same time have been eyeing this board in the below link off and on as a home server board.
- https://www.newegg.com/asrock-rack-...eneration-series-processors/p/N82E16813140023
Get this board to possibly replace home server (b450m) this since the price is back down but it lacks of rear USB is a bit of problem.
I already have one PCIe slot adding 8 sata ports might have to get a USB3 card.


I also created a support ticket with ASRock although I have no expectations of anything happening.
Well may I suggest rather than worrying about a few pennies, grabbing something that might well be overkill for the uses (but not so overkill like it costs £1000....) but grabbing something decent and making sure it'll cope with anything you throw at it. As I mentioned, all of my Ryzen CPUs are all under top end boards and yes I know, I'm overkill Phill for a reason.

Doing a lot of crunching with the CPUs I have, 1700X's all the way to 5950X's and even a couple Threadrippers now and that's just a few new ones from AMD (quick side track... Wanna see the madness that is me?? Check the project link in the bottom of my sig... That is all!) but still one thing that I have always learnt from all of the hardware I have, do it right the first time and get something that will over do what you need it too. When I was reading up on motherboards to buy and such, I wanted the VRMs to be cool and I needed decent power delivery because I was putting the CPUs under a constant 24/7 load and yes I do undervolt/downclock but just knowing that all things being equal I've paid out for my £300 + CPU, now I have the motherboard to actually cope with it.

If you buy cheap, its kinda like buying that Lambo engine for a BMW Mini chassis... Its just stupid and a waste. I mean you have got two very nice and high end CPUs, why not spend the extra and do the same for the motherboard/s?? I'm no different when it comes to the PSUs I put in the systems I build, even the ones for my daughters (one of which is 3 and prefers bashing keys on a keyboard rather than doing anything and who can blame her at that age lol Well, sometimes lol) but use quality kit... The amount of hardware you have in there and then you buy a £50 PSU and a £100 motherboard, couldn't be a worse combo and I know @Mussels has done some amazing things when its comes to putting all the wrong kit in a build (the project log of it was flipping amazing!!) but if its not worth the extra then don't have the higher end kit and use something middle line...

I mean, when I said high end for the Ryzen CPUs I have, I mean motherboards like these -

IMG_20201224_125358.jpg

Bottom line for me, spend out the right money for the right hardware the first time, don't have to waste time and more money having to replace it and all sorts of everything else if anything was to go wrong.
 

Mussels

Freshwater Moderator
Staff member
Joined
Oct 6, 2004
Messages
55,573 (8.37/day)
Location
Oystralia
System Name Rainbow Sparkles (Power efficient, <350W gaming load)
Processor Ryzen R7 5800x3D (Undervolted, 4.45GHz all core)
Motherboard Asus x570 Gaming-F
Cooling Alphacool Apex UV - Alphacool Eisblock XPX Aurora + EK Quantum ARGB 3090 w/ active backplate
Memory 2x32GB DDR4 3600 Corsair Vengeance RGB @3866 C18-22-22-22-42 TRFC704 (1.4V Hynix MJR - SoC 1.175V)
Video Card(s) Galax RTX 3090 SG 24GB: Underclocked to 1700Mhz 0.737v (375W down to 250W))
Storage 2TB WD SN850 NVME + 1TB Sasmsung 970 Pro NVME + 1TB Intel 6000P NVME USB 3.2
Display(s) Kogan 32" 4K 70Hz + Gigabyte G32QC (1440p 165Hz) + Phillips 328m6fjrmb (1440p 144Hz)
Case Fractal Design R6
Audio Device(s) Logitech G560 | Corsair Void pro RGB |Blue Yeti mic
Power Supply Fractal Ion+ 2 860W (Platinum) (This thing is God-tier. Silent and TINY)
Mouse Logitech G Pro wireless + Steelseries Prisma XL
Keyboard Razer Huntsman TE (custom white and steel keycaps)
VR HMD Oculus Rift S + Quest 2
Software Windows 11 pro x64 (Yes, it's genuinely a good OS) OpenRGB - ditch the branded bloatware!
You want the extra power plug but you dont need it.
It's like 6 pin to 8 pin on a GPU, it helps a little with voltage stability and keeps the temperature of the wires down
by the math people use, i'd never have melted the 2x8 pin extensions on my 3090 - but they melted away at just 300W load.


I think the worst part of the taichis was their really really slow updates, but if they've got the final 1.2.0.7 agesa i dont see much happening any time soon on that front


Dont get budget from any brand. Get something that will last you another 10 years, because a 5950x is going to age like nehalem - it'll be viable for a long, long time.
 
Joined
Jul 30, 2019
Messages
635 (0.52/day)
System Name Not a thread ripper but pretty good.
Processor Ryzen 9 5950x
Motherboard ASRock X570 Tiachi (revision 1.06, BIOS/UEFI version P4.80)
Cooling EK-Quantum Velocity - Nickel + Plexi, EK-Quantum Reflection PC-O11 D5 PWM, EK-CoolStream PE 360
Memory Nemix DDR4-3200 PC4-25600 2Rx8 ECC Unbuffered Memory (2 sticks, 64GB, micron)
Video Card(s) XFX Radeon RX 5700 & EK-Quantum Vector Radeon RX 5700 +XT & Backplate
Storage Samsung 970 EVO M.2 2280 500GB PCIe Gen3x4 NVMe
Display(s) 2 x 4K LG 27UL600-W (and HUANUO Dual Monitor Mount)
Case Lian Li PC-O11 Dynamic Black (original model)
Power Supply Corsair RM750x
Mouse Logitech M575
Keyboard Corsair Strafe RGB MK.2
Software Windows 10 Professional (64bit)
Benchmark Scores Typical for non-overclocked CPU.
I think I've pretty much settled on getting the x570 Tiachi.


for about $67 more dollars over the Steel Legend (WiFi) I get
+ Buttons: Pwr, Rst, Clr, and Flashback
+ Better VRM
+ Better PCB
+ Debug LED
+ 3rd M.2 4x4 (when not using PCIE5)

+ Thunderbolt capable (on PCIE5) via additional ASRock Thunderbolt AIC Card (this card may no longer be available)
+ Internal front panel USB-C connector
(should come with a special cable, or upon request, for revisions of port blocked by GPU)
(there is a board revision that solves the USB-C front panel connector problem but maybe luck of the draw trying to get that)

Missing out for not getting the Steel Legend
+ PCIe x16/x16 in dual might be more useful vs. Tiachi in dual at x8/x8 where GPU would have less than full x16
+ saving about 67 dollars

$223 seems to be fair price for the Tiachi x570.

Thoughts?

Well may I suggest rather than worrying about a few pennies, grabbing something that might well be overkill for the uses (but not so overkill like it costs £1000....) but grabbing something decent and making sure it'll cope with anything you throw at it. As I mentioned, all of my Ryzen CPUs are all under top end boards and yes I know, I'm overkill Phill for a reason.

Doing a lot of crunching with the CPUs I have, 1700X's all the way to 5950X's and even a couple Threadrippers now and that's just a few new ones from AMD (quick side track... Wanna see the madness that is me?? Check the project link in the bottom of my sig... That is all!) but still one thing that I have always learnt from all of the hardware I have, do it right the first time and get something that will over do what you need it too. When I was reading up on motherboards to buy and such, I wanted the VRMs to be cool and I needed decent power delivery because I was putting the CPUs under a constant 24/7 load and yes I do undervolt/downclock but just knowing that all things being equal I've paid out for my £300 + CPU, now I have the motherboard to actually cope with it.

If you buy cheap, its kinda like buying that Lambo engine for a BMW Mini chassis... Its just stupid and a waste. I mean you have got two very nice and high end CPUs, why not spend the extra and do the same for the motherboard/s?? I'm no different when it comes to the PSUs I put in the systems I build, even the ones for my daughters (one of which is 3 and prefers bashing keys on a keyboard rather than doing anything and who can blame her at that age lol Well, sometimes lol) but use quality kit... The amount of hardware you have in there and then you buy a £50 PSU and a £100 motherboard, couldn't be a worse combo and I know @Mussels has done some amazing things when its comes to putting all the wrong kit in a build (the project log of it was flipping amazing!!) but if its not worth the extra then don't have the higher end kit and use something middle line...

I mean, when I said high end for the Ryzen CPUs I have, I mean motherboards like these -

View attachment 256604
When I got the Master SLI/ac I had a R5 2600. It was my return to using AMD (after maybe 10 years) and it was fairly good experience and a lot of things to try. I never really expected Zen2/3 to become available for my MB but when it did I ended up playing the upgrade game (testing the reality of AMD's ecosystem on upgradability) but also passing on equipment that worked well to family for birthdays, Christmas, etc... according to their needs as I continued to explore. I'm on my B550 right now which is actually my emergency backup system (when not used for gaming) where I had planned to retire my 3950x with the 5950x being on my daily.
Bottom line for me, spend out the right money for the right hardware the first time, don't have to waste time and more money having to replace it and all sorts of everything else if anything was to go wrong.
I agree, normally try to match my parts with each other. AM4's unique position with extended backward compatibly is interesting but also full of problems in the long run encouraging users to potentially put suboptimal or just plain bad combos together. Add overclocking with a bad combo from playing the upgrade game and boom you get exactly what I did. I think it was a mistake to open 300/400 series boards to 16core when the consistent quality needed for the best user experience really didn't arrive until x570/B550.
 

tabascosauz

Moderator
Supporter
Staff member
Joined
Jun 24, 2015
Messages
5,030 (1.85/day)
Location
Western Canada
Missing out for not getting the Steel Legend
+ PCIe x16/x16 in dual might be more useful vs. Tiachi in dual at x8/x8 where GPU would have less than full x16
+ saving about 67 dollars

Double x16 doesn't exist on consumer platforms (maybe only with some PLX wizardry from years past). Steel Legend is 4.0 x16 on the main slot, and 4.0 x4 on the lower "x16" slot. All you have to do is check the backside of the board; it's wired for x4. Also telltale sign is the Crossfire certification only - SLI requires x8/x8, which is why the Taichi has it.

The lanes still come from the CPU - it only has 24 x 4.0 lanes total, not nearly enough for dual x16. If you have 2 GPUs or other bifurcation needs, 4.0 x8 is still 3.0 x16 bandwidth - plenty enough.

I'm a big proponent for the POST code, BIOS flashback, and rear IO clear CMOS button, even if you don't OC. AMD's penchant for providing continuous feature updates through BIOS makes those three features pretty much worth their weight in gold.

it helps a little with voltage stability and keeps the temperature of the wires down
by the math people use, i'd never have melted the 2x8 pin extensions on my 3090 - but they melted away at just 300W load.

EPS =! PCIe,

Solution to that would be to not use sketchy extensions? With thick solid pin connectors there is no reason why they'd get hot under normal use or ambient OC, but extensions don't necessarily guarantee the same thickness of wire. eg. the SF750 uses 16AWG on its 8-pin EPS and PCIe, but 18 gauge for the rest.

Extra 4-pin ATX12V or 8-pin EPS doesn't give Zen extra magic dust. There is no OC benefit. Board makers aren't extra to think ahead like that, not when the CPUs won't even touch 300W ever on air.
 
Last edited:

Mussels

Freshwater Moderator
Staff member
Joined
Oct 6, 2004
Messages
55,573 (8.37/day)
Location
Oystralia
System Name Rainbow Sparkles (Power efficient, <350W gaming load)
Processor Ryzen R7 5800x3D (Undervolted, 4.45GHz all core)
Motherboard Asus x570 Gaming-F
Cooling Alphacool Apex UV - Alphacool Eisblock XPX Aurora + EK Quantum ARGB 3090 w/ active backplate
Memory 2x32GB DDR4 3600 Corsair Vengeance RGB @3866 C18-22-22-22-42 TRFC704 (1.4V Hynix MJR - SoC 1.175V)
Video Card(s) Galax RTX 3090 SG 24GB: Underclocked to 1700Mhz 0.737v (375W down to 250W))
Storage 2TB WD SN850 NVME + 1TB Sasmsung 970 Pro NVME + 1TB Intel 6000P NVME USB 3.2
Display(s) Kogan 32" 4K 70Hz + Gigabyte G32QC (1440p 165Hz) + Phillips 328m6fjrmb (1440p 144Hz)
Case Fractal Design R6
Audio Device(s) Logitech G560 | Corsair Void pro RGB |Blue Yeti mic
Power Supply Fractal Ion+ 2 860W (Platinum) (This thing is God-tier. Silent and TINY)
Mouse Logitech G Pro wireless + Steelseries Prisma XL
Keyboard Razer Huntsman TE (custom white and steel keycaps)
VR HMD Oculus Rift S + Quest 2
Software Windows 11 pro x64 (Yes, it's genuinely a good OS) OpenRGB - ditch the branded bloatware!
"PCIe x16/x16 in dual"
Doesnt exist on AM4, its always 8x/8x

tabasco: these were branded XPG cables, not cheap crap.
I'm not saying its neccesary, but if you have the option, plug em in.


The problem isn't with the chipsets or generations, it's with badly designed budget boards.
In theory as long as people keep PBO off and dont all core OC, a 5950x wont use any more power than previous gen chips

Tried to find a graph with as many zen generations as possible - notice the the 15W difference from a 2700x to a 5950x?
1659325330204.png


I get the feeling something went wrong with your loop or monoblock and caused the bang and damage, and you should have been fine with nothing more than VRM throttling at worst
 
Joined
Jul 30, 2019
Messages
635 (0.52/day)
System Name Not a thread ripper but pretty good.
Processor Ryzen 9 5950x
Motherboard ASRock X570 Tiachi (revision 1.06, BIOS/UEFI version P4.80)
Cooling EK-Quantum Velocity - Nickel + Plexi, EK-Quantum Reflection PC-O11 D5 PWM, EK-CoolStream PE 360
Memory Nemix DDR4-3200 PC4-25600 2Rx8 ECC Unbuffered Memory (2 sticks, 64GB, micron)
Video Card(s) XFX Radeon RX 5700 & EK-Quantum Vector Radeon RX 5700 +XT & Backplate
Storage Samsung 970 EVO M.2 2280 500GB PCIe Gen3x4 NVMe
Display(s) 2 x 4K LG 27UL600-W (and HUANUO Dual Monitor Mount)
Case Lian Li PC-O11 Dynamic Black (original model)
Power Supply Corsair RM750x
Mouse Logitech M575
Keyboard Corsair Strafe RGB MK.2
Software Windows 10 Professional (64bit)
Benchmark Scores Typical for non-overclocked CPU.
I think the worst part of the taichis was their really really slow updates, but if they've got the final 1.2.0.7 agesa i dont see much happening any time soon on that front
I remember seeing a lot of complaints about ASRock slow bios updates. AMD's backward compatible upgrade game I don't think helped matters as vendors had to keep re-updating their older product lines which eventually bit MSI users in the butt pretty hard.

I see the Tiachi x570 looks to have AGESA Combo V2 PI 1.2.0.6b and 1.2.07 in beta. Normally I don't run a beta bio unless I need a fix because I never had bios flashback. I was running a beta AGESA Combo V2 PI 1.2.0.7 on my Master SLI/ac for a month or so already due to a problem I found ZenTimings was broken and couldn't read some values. They ended up fixing it in BIOS\UEFI and giving me a new beta it in a couple of days turnaround. Similar thing early on when I switched to ECC ram (again exploring the claims of ECC support or lack thereof) having problems they patched me up pretty quickly and got it working so that's kinda why I'm also sticking with ASRock for now.

Dont get budget from any brand. Get something that will last you another 10 years, because a 5950x is going to age like nehalem - it'll be viable for a long, long time.
The only reason I still have the B450m (other than perhaps of being a tech horder) is because it was originally a gift for my mother with a 2200g but win10 18 to 19 upgrade totally messed up her system. I replaced her board with MSI Tomahawk while I tried to investigate wtf happened. I ended up keeping the board as a light home server and stuck my 2700 in it as the final stop for that board planning to pass my 2600 down as an upgrade to her later.

"PCIe x16/x16 in dual"
Doesnt exist on AM4, its always 8x/8x

tabasco: these were branded XPG cables, not cheap crap.
I'm not saying its neccesary, but if you have the option, plug em in.
I see now. I read the spec sheet wrong.

AMD Ryzen series CPUs (Vermeer, Matisse)
- 2 x PCI Express 4.0 x16 Slots (single at x16 (PCIE1); dual at x16 (PCIE1) / x4 (PCIE4))*
 

Mussels

Freshwater Moderator
Staff member
Joined
Oct 6, 2004
Messages
55,573 (8.37/day)
Location
Oystralia
System Name Rainbow Sparkles (Power efficient, <350W gaming load)
Processor Ryzen R7 5800x3D (Undervolted, 4.45GHz all core)
Motherboard Asus x570 Gaming-F
Cooling Alphacool Apex UV - Alphacool Eisblock XPX Aurora + EK Quantum ARGB 3090 w/ active backplate
Memory 2x32GB DDR4 3600 Corsair Vengeance RGB @3866 C18-22-22-22-42 TRFC704 (1.4V Hynix MJR - SoC 1.175V)
Video Card(s) Galax RTX 3090 SG 24GB: Underclocked to 1700Mhz 0.737v (375W down to 250W))
Storage 2TB WD SN850 NVME + 1TB Sasmsung 970 Pro NVME + 1TB Intel 6000P NVME USB 3.2
Display(s) Kogan 32" 4K 70Hz + Gigabyte G32QC (1440p 165Hz) + Phillips 328m6fjrmb (1440p 144Hz)
Case Fractal Design R6
Audio Device(s) Logitech G560 | Corsair Void pro RGB |Blue Yeti mic
Power Supply Fractal Ion+ 2 860W (Platinum) (This thing is God-tier. Silent and TINY)
Mouse Logitech G Pro wireless + Steelseries Prisma XL
Keyboard Razer Huntsman TE (custom white and steel keycaps)
VR HMD Oculus Rift S + Quest 2
Software Windows 11 pro x64 (Yes, it's genuinely a good OS) OpenRGB - ditch the branded bloatware!
I see now. I read the spec sheet wrong.

AMD Ryzen series CPUs (Vermeer, Matisse)
- 2 x PCI Express 4.0 x16 Slots (single at x16 (PCIE1); dual at x16 (PCIE1) / x4 (PCIE4))*
those spec sheets are often copy pastes from the manual and lose formatting, i've seen a lot of screwed things there

Like there's meant to be a 1. 2. 3. for the different CPU generations and PCI-E lanes/generation, but they dont copy over and you end up with a gibberish mess

its 16/0/4 or 8/8/4, at 3.0 or 4.0 unless it's one of the older APU's when its 8/0/4 i think
 
Joined
Jul 30, 2019
Messages
635 (0.52/day)
System Name Not a thread ripper but pretty good.
Processor Ryzen 9 5950x
Motherboard ASRock X570 Tiachi (revision 1.06, BIOS/UEFI version P4.80)
Cooling EK-Quantum Velocity - Nickel + Plexi, EK-Quantum Reflection PC-O11 D5 PWM, EK-CoolStream PE 360
Memory Nemix DDR4-3200 PC4-25600 2Rx8 ECC Unbuffered Memory (2 sticks, 64GB, micron)
Video Card(s) XFX Radeon RX 5700 & EK-Quantum Vector Radeon RX 5700 +XT & Backplate
Storage Samsung 970 EVO M.2 2280 500GB PCIe Gen3x4 NVMe
Display(s) 2 x 4K LG 27UL600-W (and HUANUO Dual Monitor Mount)
Case Lian Li PC-O11 Dynamic Black (original model)
Power Supply Corsair RM750x
Mouse Logitech M575
Keyboard Corsair Strafe RGB MK.2
Software Windows 10 Professional (64bit)
Benchmark Scores Typical for non-overclocked CPU.
I get the feeling something went wrong with your loop or monoblock and caused the bang and damage, and you should have been fine with nothing more than VRM throttling at worst
I went back to inspect the monoblock because of your comment earlier but I think it's a solid piece on the backside with no opportunity to leak from that position. I'm not sure how they joined the VRM and CPU plates to the larger block but whenever I disassembled the block I don't recall a way to remove those portions of the block. EK fluid has a particular annoying odor and I didn't smell any around the gaskets along the outter edges of the monoblock. If there was a leak it would have had to come from the rad above in the 1/3 of the fin stack area above the CPU. The rad ports are too far out of the way for a leak there to have gotten under the monoblock. There were no signs of fluid loss from the maintenance I did about last month so while not impossible to rule out I think it's unlikely a leak occurred unless it was from the top gasket of the monoblock but again no evidence of fluid there.

In the past few weeks I was doing all core workloads (not overclocked) for hours at a time so maybe there is a possibility incremental damage was occurring on the board without me knowing it and like a pimple I popped it with PBO.
 

tabascosauz

Moderator
Supporter
Staff member
Joined
Jun 24, 2015
Messages
5,030 (1.85/day)
Location
Western Canada
tabasco: these were branded XPG cables, not cheap crap.
I'm not saying its neccesary, but if you have the option, plug em in.

The problem isn't with the chipsets or generations, it's with badly designed budget boards.
In theory as long as people keep PBO off and dont all core OC, a 5950x wont use any more power than previous gen chips

I get the feeling something went wrong with your loop or monoblock and caused the bang and damage, and you should have been fine with nothing more than VRM throttling at worst

Fair enough. 16AWG too, what a letdown from ADATA.

Stock 2700X isn't far from stock 5950X but 2700X also maxes out/doesn't clock further at like what, 200W on air? Even PBO takes you way past 200W on 5000. OP was all-core testing north of 200W when doodoo hit the fan - add to that the bad thermal pad contact. Besides, VRM lists have been around before Ryzen 3000 and back then it was no secret that awful boards aren't a good choice for 2700X OC.

Stock 5950X would have been [probably] alive, though not exactly ideal.

I remember seeing a lot of complaints about ASRock slow bios updates. AMD's backward compatible upgrade game I don't think helped matters as vendors had to keep re-updating their older product lines which eventually bit MSI users in the butt pretty hard.

I see the Tiachi x570 looks to have AGESA Combo V2 PI 1.2.0.6b and 1.2.07 in beta. Normally I don't run a beta bio unless I need a fix because I never had bios flashback. I was running a beta AGESA Combo V2 PI 1.2.0.7 on my Master SLI/ac for a month or so already due to a problem I found ZenTimings was broken and couldn't read some values. They ended up fixing it in BIOS\UEFI and giving me a new beta it in a couple of days turnaround. Similar thing early on when I switched to ECC ram (again exploring the claims of ECC support or lack thereof) having problems they patched me up pretty quickly and got it working so that's kinda why I'm also sticking with ASRock for now.

Oh hell no. If there's one thing ASRock is half-decent at, it's acceptably-finished BIOS updates. You don't know the meaning of nightmare until you've been with Asus and Gigabyte. And MSI's gimped "slim" 16MB BIOS.

Newest =! best. AGESA 1200 to 1203c was a winning tradition. It's been all downhill from there with AMD breaking performance/breaking PBO in preparation for 5800X3D. Latest 1207 is finally okay again for general performance but the 145A+ EDC problem is still there I think (but not a problem unless pushing PBO and hard).

I think the L3 "fix" for Windows 11 is in 1206 or something, but if you're on Win 10 then you can just sit pretty on 1201/1202/1203. 1207 has the TPM stuttering fix iirc.
 
Joined
Jul 30, 2019
Messages
635 (0.52/day)
System Name Not a thread ripper but pretty good.
Processor Ryzen 9 5950x
Motherboard ASRock X570 Tiachi (revision 1.06, BIOS/UEFI version P4.80)
Cooling EK-Quantum Velocity - Nickel + Plexi, EK-Quantum Reflection PC-O11 D5 PWM, EK-CoolStream PE 360
Memory Nemix DDR4-3200 PC4-25600 2Rx8 ECC Unbuffered Memory (2 sticks, 64GB, micron)
Video Card(s) XFX Radeon RX 5700 & EK-Quantum Vector Radeon RX 5700 +XT & Backplate
Storage Samsung 970 EVO M.2 2280 500GB PCIe Gen3x4 NVMe
Display(s) 2 x 4K LG 27UL600-W (and HUANUO Dual Monitor Mount)
Case Lian Li PC-O11 Dynamic Black (original model)
Power Supply Corsair RM750x
Mouse Logitech M575
Keyboard Corsair Strafe RGB MK.2
Software Windows 10 Professional (64bit)
Benchmark Scores Typical for non-overclocked CPU.
Also 90C is the MAX operating limit and its probably throttling at that point. You shouldn't push it to hard.
Too high EDC + (on) high(est) temp can kill your CPU overtime easily. PPT(w) comes second.
I found a 3rd video (2018) on this MB buildzoid discovered the VRM temperature sensor is broken on this boards non wifi counterpart.
 
Joined
Sep 3, 2019
Messages
2,700 (2.27/day)
Location
Thessaloniki, Greece
System Name PC on since Aug 18th 2019, 1st CPU R5 3600
Processor Ryzen 9 5900X, ECO mode 88W PPT
Motherboard Gigabyte X570 Aorus Pro (Rev1.0), BIOS F36d, V2 1.2.0.7
Cooling Corsair H110i 280mm (w/ Contuctonaut since Jan 2020)
Memory 2x16GB G.Skill Trident Z Neo (GTZN) 3600MHz 1.40V CL16-16-16-16-32-48 1T, tRFC:274, B-die
Video Card(s) MSI RX 5700XT Gaming X, GPU 2100MHz, VRAM 1750MHz
Storage Samsung NVMe: 980Pro 1TB (OS 2022), 970Pro 512GB(2019) / SATA-III: 850Pro 1TB(2015) 860Evo 1TB(2020)
Display(s) 2007 24" EIZO FlexScan S2411W 1920x1200 16:10 60Hz Samsung S-PVA 14-bit (16.7M/1.06B colors) 6ms G2G
Case None... naked on desk
Audio Device(s) Sound Blaster Z <--optical link--> Logitech Z5500 5.1 500W, Logitech G935 headset
Power Supply Corsair HX750i
Mouse Logitech MX Master (Gen1)
Keyboard Logitech G15 (Gen2) w/ LCDSirReal applet
Software Windows 10 Home 64bit (21H2)
"PCIe x16/x16 in dual"
Doesnt exist on AM4, its always 8x/8x

tabasco: these were branded XPG cables, not cheap crap.
I'm not saying its neccesary, but if you have the option, plug em in.


The problem isn't with the chipsets or generations, it's with badly designed budget boards.
In theory as long as people keep PBO off and dont all core OC, a 5950x wont use any more power than previous gen chips

Tried to find a graph with as many zen generations as possible - notice the the 15W difference from a 2700x to a 5950x?
View attachment 256607

I get the feeling something went wrong with your loop or monoblock and caused the bang and damage, and you should have been fine with nothing more than VRM throttling at worst
The thing is that when you leave everything on their "luck" it could go crazy in no time...
The OP was "playing" with PBO and CurveOptimizer, but left the limits to the board.

And here is the situation you can be in, without knowing it, or be prepared for...
CB R23 MT

1. This is PBO on (my) Board limits (PPT:900W, TDC:480A, EDC:215A) with no Curve Optimizer enabled.
PBO_test_02 enabled_900_480_215_no_CO_SC21060.png

Wait, it gets better...

2. PBO on board limits, with CO from -5 (to best CCD) to -10 (to worst CCD)
PBO_test_03 enabled_MB_900_480_215_CO_5_10_SC21583.png

3. PBO on board limits, with CO from -10 to -20
PBO_test_04 enabled_MB_900_480_215_CO_10_20_SC22000.png

This can go even further on EDC if I released it beyond 215A and give more negative curve, but I wouldn't dare as this was too much already.

And this is controlled limits

4. PBO on custom limits (PPT:200W, TDC:130A, EDC:160A) , with CO from -5 to -10
PBO_test_05 enabled_Manual_200_130_160_CO_5_10_SC21875.png

5. PBO on custom limits (PPT:200W, TDC:130A, EDC:160A) , with CO from -10 to -20
PBO_test_05 enabled_Manual_200_130_160_CO_10_20_SC22176.png

6. PBO on custom limits (PPT:160W, TDC:115A, EDC:135A) , with CO from -10 to -20
PBO_test_07 enabled_Manual_160_115_135_CO_10_20_SC22176.png

Anyone interested to know which one gave the best score in R23?

----------------------------------------------

I found a 3rd video (2018) on this MB buildzoid discovered the VRM temperature sensor is broken on this boards non wifi counterpart.
Well this can explain the burnout
I'm really sorry man... Hope the 5950X is ok
 
Joined
Jul 30, 2019
Messages
635 (0.52/day)
System Name Not a thread ripper but pretty good.
Processor Ryzen 9 5950x
Motherboard ASRock X570 Tiachi (revision 1.06, BIOS/UEFI version P4.80)
Cooling EK-Quantum Velocity - Nickel + Plexi, EK-Quantum Reflection PC-O11 D5 PWM, EK-CoolStream PE 360
Memory Nemix DDR4-3200 PC4-25600 2Rx8 ECC Unbuffered Memory (2 sticks, 64GB, micron)
Video Card(s) XFX Radeon RX 5700 & EK-Quantum Vector Radeon RX 5700 +XT & Backplate
Storage Samsung 970 EVO M.2 2280 500GB PCIe Gen3x4 NVMe
Display(s) 2 x 4K LG 27UL600-W (and HUANUO Dual Monitor Mount)
Case Lian Li PC-O11 Dynamic Black (original model)
Power Supply Corsair RM750x
Mouse Logitech M575
Keyboard Corsair Strafe RGB MK.2
Software Windows 10 Professional (64bit)
Benchmark Scores Typical for non-overclocked CPU.
The thing is that when you leave everything on their "luck" it could go crazy in no time...
The OP was "playing" with PBO and CurveOptimizer, but left the limits to the board.
Just to clarify I did not use Curve Optimizer. My intent was to raise the limits but keep the CPU in control of cores and voltage so not to bypass it's safety features for current and temperature. I hope I said that correctly. My CPU never did get past 89c from what I observed but defiantly on the 2nd attempt to get some VRM readings the VRM spiked past 80c before the board blew out.
 
Last edited:
Joined
Sep 3, 2019
Messages
2,700 (2.27/day)
Location
Thessaloniki, Greece
System Name PC on since Aug 18th 2019, 1st CPU R5 3600
Processor Ryzen 9 5900X, ECO mode 88W PPT
Motherboard Gigabyte X570 Aorus Pro (Rev1.0), BIOS F36d, V2 1.2.0.7
Cooling Corsair H110i 280mm (w/ Contuctonaut since Jan 2020)
Memory 2x16GB G.Skill Trident Z Neo (GTZN) 3600MHz 1.40V CL16-16-16-16-32-48 1T, tRFC:274, B-die
Video Card(s) MSI RX 5700XT Gaming X, GPU 2100MHz, VRAM 1750MHz
Storage Samsung NVMe: 980Pro 1TB (OS 2022), 970Pro 512GB(2019) / SATA-III: 850Pro 1TB(2015) 860Evo 1TB(2020)
Display(s) 2007 24" EIZO FlexScan S2411W 1920x1200 16:10 60Hz Samsung S-PVA 14-bit (16.7M/1.06B colors) 6ms G2G
Case None... naked on desk
Audio Device(s) Sound Blaster Z <--optical link--> Logitech Z5500 5.1 500W, Logitech G935 headset
Power Supply Corsair HX750i
Mouse Logitech MX Master (Gen1)
Keyboard Logitech G15 (Gen2) w/ LCDSirReal applet
Software Windows 10 Home 64bit (21H2)
Just to clarify I did not use Curve Optimizer. My intent was to raise the limits but keep the CPU in control of cores and voltage so not to bypass it's safety features for current and temperature. I hope I said that correctly. My CPU never did get past 89c from what I observed.
Ok, my bad

Though as you can see on first screenshot, even without the curve, but with PBO enabled(motherboard) things are getting too high. Around 200A EDC in my case
 
Joined
Jul 30, 2019
Messages
635 (0.52/day)
System Name Not a thread ripper but pretty good.
Processor Ryzen 9 5950x
Motherboard ASRock X570 Tiachi (revision 1.06, BIOS/UEFI version P4.80)
Cooling EK-Quantum Velocity - Nickel + Plexi, EK-Quantum Reflection PC-O11 D5 PWM, EK-CoolStream PE 360
Memory Nemix DDR4-3200 PC4-25600 2Rx8 ECC Unbuffered Memory (2 sticks, 64GB, micron)
Video Card(s) XFX Radeon RX 5700 & EK-Quantum Vector Radeon RX 5700 +XT & Backplate
Storage Samsung 970 EVO M.2 2280 500GB PCIe Gen3x4 NVMe
Display(s) 2 x 4K LG 27UL600-W (and HUANUO Dual Monitor Mount)
Case Lian Li PC-O11 Dynamic Black (original model)
Power Supply Corsair RM750x
Mouse Logitech M575
Keyboard Corsair Strafe RGB MK.2
Software Windows 10 Professional (64bit)
Benchmark Scores Typical for non-overclocked CPU.
Ok, my bad

Though as you can see on first screenshot, even without the curve, but with PBO enabled(motherboard) things are getting too high. Around 200A EDC in my case
Ok thank you for this, I think I am beginning to understand this better now from your example to discuss it.

In your example you started with your board limits (instead of Ryzen Master defaults) before gradually applying your curves until they approached the board limit of your EDC (that looks to be the first limit reached). So if someone were to blindly accept the Ryzen Mater defaults 1000, 420, and 460 they could get themselves into real trouble without understanding they should really start around their actual board limits. This was a critical step I didn't understand or was aware of when using PBO in Ryzen Master. When you combine this with the CPU safety features only protect the CPU, not the VRM, it makes sense and I think I had a misunderstanding in this regard. This misunderstanding prevented me from seeing that, not only should I not accept Ryzen master defaults, I shouldn't just plug in some numbers some person on reddit recommended to put in there. When you raise the limits the CPU will ask, the board may try it, the CPU might still be OK, but the board can now be stressed past it's limits and one would hope it finds a way to throttle or shuts down. In the worst case scenario you damage your board and/or CPU.

So I'm also reading PBO has a hard limit of allowing 105W TDP CPUs to draw ~220W. I think you were trying to tell me before basically my board could not do that. So when I raise the limits in RM too high the CPU will try to continue boosting until it reaches that limit which my board cannot supply. The attempt causes the VRM's to overheat and chaos ensues. Since my boards thermal protection doesn't work I get the worst case scenario - game over.

Please correct me if I am wrong.

I'm up late trying to piece together the correct understanding these past few days so I have to retire for the night. Thank you again for your example.

Final question for the night. Would a 7zip on Ultra all cores of 350GB worth of files be more stressful than CB R23 MT run?
 
Last edited:
Joined
Sep 3, 2019
Messages
2,700 (2.27/day)
Location
Thessaloniki, Greece
System Name PC on since Aug 18th 2019, 1st CPU R5 3600
Processor Ryzen 9 5900X, ECO mode 88W PPT
Motherboard Gigabyte X570 Aorus Pro (Rev1.0), BIOS F36d, V2 1.2.0.7
Cooling Corsair H110i 280mm (w/ Contuctonaut since Jan 2020)
Memory 2x16GB G.Skill Trident Z Neo (GTZN) 3600MHz 1.40V CL16-16-16-16-32-48 1T, tRFC:274, B-die
Video Card(s) MSI RX 5700XT Gaming X, GPU 2100MHz, VRAM 1750MHz
Storage Samsung NVMe: 980Pro 1TB (OS 2022), 970Pro 512GB(2019) / SATA-III: 850Pro 1TB(2015) 860Evo 1TB(2020)
Display(s) 2007 24" EIZO FlexScan S2411W 1920x1200 16:10 60Hz Samsung S-PVA 14-bit (16.7M/1.06B colors) 6ms G2G
Case None... naked on desk
Audio Device(s) Sound Blaster Z <--optical link--> Logitech Z5500 5.1 500W, Logitech G935 headset
Power Supply Corsair HX750i
Mouse Logitech MX Master (Gen1)
Keyboard Logitech G15 (Gen2) w/ LCDSirReal applet
Software Windows 10 Home 64bit (21H2)
Ok thank you for this, I think I am beginning to understand this better now from your example to discuss it.

In your example you started with your board limits (instead of Ryzen Master defaults) before gradually applying your curves until they approached the board limit of your EDC (that looks to be the first limit reached). So if someone were to blindly accept the Ryzen Mater defaults 1000, 420, and 460 they could get themselves into real trouble without understanding they should really start around their actual board limits. This was a critical step I didn't understand or was aware of when using PBO in Ryzen Master. When you combine this with the CPU safety features only protect the CPU, not the VRM, it makes sense and I think I had a misunderstanding in this regard. This misunderstanding prevented me from seeing that, not only should I not accept Ryzen master defaults, I shouldn't just plug in some numbers some person on reddit recommended to put in there. When you raise the limits the CPU will ask, the board may try it, the CPU might still be OK, but the board can now be stressed past it's limits and one would hope it finds a way to throttle or shuts down. In the worst case scenario you damage your board and/or CPU.

So I'm also reading PBO has a hard limit of allowing 105W TDP CPUs to draw ~220W. I think you were trying to tell me before basically my board could not do that. So when I raise the limits in RM too high the CPU will try to continue boosting until it reaches that limit which my board cannot supply. The attempt causes the VRM's to overheat and chaos ensues. Since my boards thermal protection doesn't work I get the worst case scenario - game over.

Please correct me if I am wrong.
Yeah you got it...

I should point out that all settings was done into BIOS. I do not trust any software for this.
For board limits there is actually a "Motherboard" setting for PBO.

You know what the "funny" thing is? All that Current (A) going into CPU in the first 3 screenshots (out of 6) is just a total waste that only stresses the CPU silicon and potentially VRMs (depending the board).
Allow me to state the R23 scores of each run

---------------------------------------

I dont have the screenshots of R23, you have to take my word for it.
Also those were quick settings just to show the idea of PBO limits and CO.
It wasn't a real optimization but rather a demo of it.


1. PBO disabled (PPT:142W, TDC:95A, EDC:140A) with no Curve Optimizer enabled.
Actual avg PPT: 139W, EDC: 138A
R23 MT: 20932

2. PBO on board limits (PPT:900W, TDC:480A, EDC:215A) with no Curve Optimizer enabled
Actual avg PPT: 151W, EDC: 194A
R23 MT: 21060

3. PBO on board limits (PPT:900W, TDC:480A, EDC:215A) with CO from -5 (to best CCD) to -10 (to worst CCD)
Actual avg PPT: 158W, EDC: 203A
R23 MT: 21583

4. PBO on board limits (PPT:900W, TDC:480A, EDC:215A) with CO from -10 (to best CCD) to -20 (to worst CCD)
Actual avg PPT: 165W, EDC: 209A
R23 MT: 22000

--------------------------------

5. PBO on custom limits (PPT:200W, TDC:130A, EDC:160A) , with CO from -5 (to best CCD) to -10 (to worst CCD)
Actual avg PPT: 161W, EDC: 156A
R23 MT: 21875

6. PBO on custom limits (PPT:200W, TDC:130A, EDC:160A) , with CO from -10 (to best CCD) to -20 (to worst CCD)
Actual avg PPT: 164W, EDC: 158A
R23 MT: 22021

7. PBO on custom limits (PPT:160W, TDC:115A, EDC:135A) , with CO from -10 (to best CCD) to -20 (to worst CCD)
Actual avg PPT: 159W, EDC: 134A
R23 MT: 22176

Final question for the night. Would a 7zip on Ultra all cores of 350GB worth of files be more stressful than CB R23 MT run?
Never test that so I wouldn't know
 
Last edited:

tabascosauz

Moderator
Supporter
Staff member
Joined
Jun 24, 2015
Messages
5,030 (1.85/day)
Location
Western Canada
I found a 3rd video (2018) on this MB buildzoid discovered the VRM temperature sensor is broken on this boards non wifi counterpart.

Think that's all you need to know really. Not even the fact that the temp sensor is broken (as are most wildly inaccurate software sensors), but that the board literally doesn't have working OTP on the VRM.
 
Joined
Jul 30, 2019
Messages
635 (0.52/day)
System Name Not a thread ripper but pretty good.
Processor Ryzen 9 5950x
Motherboard ASRock X570 Tiachi (revision 1.06, BIOS/UEFI version P4.80)
Cooling EK-Quantum Velocity - Nickel + Plexi, EK-Quantum Reflection PC-O11 D5 PWM, EK-CoolStream PE 360
Memory Nemix DDR4-3200 PC4-25600 2Rx8 ECC Unbuffered Memory (2 sticks, 64GB, micron)
Video Card(s) XFX Radeon RX 5700 & EK-Quantum Vector Radeon RX 5700 +XT & Backplate
Storage Samsung 970 EVO M.2 2280 500GB PCIe Gen3x4 NVMe
Display(s) 2 x 4K LG 27UL600-W (and HUANUO Dual Monitor Mount)
Case Lian Li PC-O11 Dynamic Black (original model)
Power Supply Corsair RM750x
Mouse Logitech M575
Keyboard Corsair Strafe RGB MK.2
Software Windows 10 Professional (64bit)
Benchmark Scores Typical for non-overclocked CPU.
Tiachi ordered. Now we wait.

Found a video that suggest the extra 4 pin can be used to help supply PCIe power if your using a lot of PCIe devices.
 

tabascosauz

Moderator
Supporter
Staff member
Joined
Jun 24, 2015
Messages
5,030 (1.85/day)
Location
Western Canada
Tiachi ordered. Now we wait.

Found a video that suggest the extra 4 pin can be used to help supply PCIe power if your using a lot of PCIe devices.

Not the same thing. We're talking about an extra 4-pin ATX12V or 8-pin EPS beside the existing 8-pin EPS. They can only contribute power to the CPU, through the main VRM.

What you're referring to is an additional 6-pin PCIe sometimes located in the middle or bottom of the board (my Unify-X has one at the bottom edge). It's intended to provide additional power to the PCIe slots if using multiple power hungry GPUs, in case they exceed the spec 75W draw through their x16 slots. The 6-pin takes the load off the poor 24-pin when the GPUs exceed 75W slot draw.

Wendell is a whiz when it comes to servers, but he regularly makes a *lot* of incorrect assumptions when reviewing consumer hardware, and just rolls with it. You'll see what I mean when you watch his board reviews. Power doesn't just float across the PCB into whatever device it chooses.
 
Joined
Nov 18, 2010
Messages
6,184 (1.41/day)
Location
Rīga, Latvia
System Name HELLSTAR
Processor AMD RYZEN 9 5950X
Motherboard ASUS Strix X570-E
Cooling Custom Loop. Two 360ies + 280 rad. 8x Nidec Servo Gentle Typhoons. EK-Quantum Momentum monoblock.
Memory 4x8GB G.SKILL Trident Z RGB F4-4133C19D-16GTZR 14-16-12-30-44
Video Card(s) ASUS 1080 Ti SLI + water blocks
Storage Optane 900P + Samsung PM981 NVMe 1TB + 750 EVO 500GB+ 1TB 980PRO
Display(s) Philips PHL BDM3270 + Acer XV242Y
Case Lian Li O11 Dynamic EVO
Audio Device(s) Sound Blaster ZxR
Power Supply Fractal Design Newton R3 1000W
Mouse Razer Basilisk
Keyboard Razer BlackWidow V3 - Yellow Switch
Software Windows 11 insider
Power doesn't just float across the PCB into whatever device it chooses.

Current does flow like that actually or are you trying to defy the basic ohms laws? PCB's internal powerplanes are really low resistance and current flows towards the shortest sink path. If the consumption rises on PCIe current will naturally go there.

So please let's not be so hasty with assumptions like that. BTW I have no idea who that reviewer is or whatsoever, I selfdom use YT for PC review stuff. The additional PEG's on bottom part of the board were made because of the economy on layers as power planes needed to be split. Basically it acts like plaster, multiple return paths actually cause problems, like VRM noise, strange EMI other things.
 
Joined
May 2, 2017
Messages
7,761 (3.80/day)
Location
Back in Norway
System Name Hotbox
Processor AMD Ryzen 7 5800X, 110/95/110, PBO +150Mhz, CO -7,-7,-20(x6),
Motherboard ASRock Phantom Gaming B550 ITX/ax
Cooling LOBO + Laing DDC 1T Plus PWM + Corsair XR5 280mm + 2x Arctic P14
Memory 32GB G.Skill FlareX 3200c14 @3800c15
Video Card(s) PowerColor Radeon 6900XT Liquid Devil Ultimate, UC@2250MHz max @~200W
Storage 2TB Adata SX8200 Pro
Display(s) Dell U2711 main, AOC 24P2C secondary
Case SSUPD Meshlicious
Audio Device(s) Optoma Nuforce μDAC 3
Power Supply Corsair SF750 Platinum
Mouse Logitech G603
Keyboard Keychron K3/Cooler Master MasterKeys Pro M w/DSA profile caps
Software Windows 10 Pro
Current does flow like that actually or are you trying to defy the basic ohms laws? PCB's internal powerplanes are really low resistance and current flows towards the shortest sink path. If the consumption rises on PCIe current will naturally go there.

So please let's not be so hasty with assumptions like that. BTW I have no idea who that reviewer is or whatsoever, I selfdom use YT for PC review stuff. The additional PEG's on bottom part of the board were made because of the economy on layers as power planes needed to be split. Basically it acts like plaster, multiple return paths actually cause problems, like VRM noise, strange EMI other things.
What they're saying is that the VRM 12V powerplane isn't actually connected at all to the PCIe 12V powerplane on the motherboard. The EPS connectors feed the CPU VRM, and only the CPU VRM. PCIe slots are fed through the 24-pin, as well as any ancillary 12V connectors that might be added to the board. These are entirely physically separate power planes, so power definitely doesn't just float across the PCB - it needs actual connections through copper traces to move through. (And yes, that means that running several 75W PCIe devices on a board that only supplies PCIe power through the 24-pin is a tad sketchy.)
 
Joined
Nov 18, 2010
Messages
6,184 (1.41/day)
Location
Rīga, Latvia
System Name HELLSTAR
Processor AMD RYZEN 9 5950X
Motherboard ASUS Strix X570-E
Cooling Custom Loop. Two 360ies + 280 rad. 8x Nidec Servo Gentle Typhoons. EK-Quantum Momentum monoblock.
Memory 4x8GB G.SKILL Trident Z RGB F4-4133C19D-16GTZR 14-16-12-30-44
Video Card(s) ASUS 1080 Ti SLI + water blocks
Storage Optane 900P + Samsung PM981 NVMe 1TB + 750 EVO 500GB+ 1TB 980PRO
Display(s) Philips PHL BDM3270 + Acer XV242Y
Case Lian Li O11 Dynamic EVO
Audio Device(s) Sound Blaster ZxR
Power Supply Fractal Design Newton R3 1000W
Mouse Razer Basilisk
Keyboard Razer BlackWidow V3 - Yellow Switch
Software Windows 11 insider
What they're saying is that the VRM 12V powerplane isn't actually connected at all to the PCIe 12V powerplane on the motherboard. The EPS connectors feed the CPU VRM, and only the CPU VRM. PCIe slots are fed through the 24-pin, as well as any ancillary 12V connectors that might be added to the board. These are entirely physically separate power planes, so power definitely doesn't just float across the PCB - it needs actual connections through copper traces to move through. (And yes, that means that running several 75W PCIe devices on a board that only supplies PCIe power through the 24-pin is a tad sketchy.)

ah yes yes, but current doesn't only mean the hot wire, but ground too.

The problem occurs exactly because of ATX24PIN power and multi rail PSU's have different rails for PCIe devices and you basically connect them together in unwanted manner while using the additional PEG connector supplying more current to PCIe slots. The problem isn't really mitigated using single rail PSU as you alter return paths via the ground plane and that's same for any of those, this causes sometimes some fishy problems I described as the device combination each PC user does are wildly different.
 
Joined
May 2, 2017
Messages
7,761 (3.80/day)
Location
Back in Norway
System Name Hotbox
Processor AMD Ryzen 7 5800X, 110/95/110, PBO +150Mhz, CO -7,-7,-20(x6),
Motherboard ASRock Phantom Gaming B550 ITX/ax
Cooling LOBO + Laing DDC 1T Plus PWM + Corsair XR5 280mm + 2x Arctic P14
Memory 32GB G.Skill FlareX 3200c14 @3800c15
Video Card(s) PowerColor Radeon 6900XT Liquid Devil Ultimate, UC@2250MHz max @~200W
Storage 2TB Adata SX8200 Pro
Display(s) Dell U2711 main, AOC 24P2C secondary
Case SSUPD Meshlicious
Audio Device(s) Optoma Nuforce μDAC 3
Power Supply Corsair SF750 Platinum
Mouse Logitech G603
Keyboard Keychron K3/Cooler Master MasterKeys Pro M w/DSA profile caps
Software Windows 10 Pro
ah yes yes, but current doesn't only mean the hot wire, but ground too.

The problem occurs exactly because of ATX24PIN power and multi rail PSU's have different rails for PCIe devices and you basically connect them together in unwanted manner while using the additional PEG connector supplying more current to PCIe slots. The problem isn't really mitigated using single rail PSU as you alter return paths via the ground plane and that's same for any of those, this causes sometimes some fishy problems I described as the device combination each PC user does are wildly different.
No, the problem here is that you took a single quote out of context, misunderstood it, and responded to it as if it was saying something other than what it was saying, and are now failing to recognize that fact.

Also, who uses (or makes) multi-rail PSUs these days? And won't current take the shortest, lowest resistance path, whereever that may be? And in a single rail PSU, won't all those paths, no matter through which cables, return to the same ground rail in the PSU? Bridging rails in a multi-rail PSU is obviously to be avoided, but I can't imagine there's much overlap between owners of multi-rail PSUs and owners of motherboards with ancillary PCIe power connectors, seeing how one is hardly made any longer, and the other is rare and mostly reserved for HEDT/workstation boards.

As for this:
The additional PEG's on bottom part of the board were made because of the economy on layers as power planes needed to be split.
That is simply not true. The additional PEG ports were added because without them, the board wouldn't be able to deliver 75W of power to each PCIe slot without risking burning out the 24-pin connector or cable, or worse, the board. There are only two 12V lines on a 24-pin, after all, and with the EPS12V rail not connected to the rest of the board, there's no safe way of providing significantly more than 75W to the PCIe slots without adding further power connectors.

This can absolutely lead to noise and weird resonances, but that's why you have electrical engineers working at both PSU and motherboard manufacturers that can work to minimize this.
 
Joined
Jul 30, 2019
Messages
635 (0.52/day)
System Name Not a thread ripper but pretty good.
Processor Ryzen 9 5950x
Motherboard ASRock X570 Tiachi (revision 1.06, BIOS/UEFI version P4.80)
Cooling EK-Quantum Velocity - Nickel + Plexi, EK-Quantum Reflection PC-O11 D5 PWM, EK-CoolStream PE 360
Memory Nemix DDR4-3200 PC4-25600 2Rx8 ECC Unbuffered Memory (2 sticks, 64GB, micron)
Video Card(s) XFX Radeon RX 5700 & EK-Quantum Vector Radeon RX 5700 +XT & Backplate
Storage Samsung 970 EVO M.2 2280 500GB PCIe Gen3x4 NVMe
Display(s) 2 x 4K LG 27UL600-W (and HUANUO Dual Monitor Mount)
Case Lian Li PC-O11 Dynamic Black (original model)
Power Supply Corsair RM750x
Mouse Logitech M575
Keyboard Corsair Strafe RGB MK.2
Software Windows 10 Professional (64bit)
Benchmark Scores Typical for non-overclocked CPU.

Mussels

Freshwater Moderator
Staff member
Joined
Oct 6, 2004
Messages
55,573 (8.37/day)
Location
Oystralia
System Name Rainbow Sparkles (Power efficient, <350W gaming load)
Processor Ryzen R7 5800x3D (Undervolted, 4.45GHz all core)
Motherboard Asus x570 Gaming-F
Cooling Alphacool Apex UV - Alphacool Eisblock XPX Aurora + EK Quantum ARGB 3090 w/ active backplate
Memory 2x32GB DDR4 3600 Corsair Vengeance RGB @3866 C18-22-22-22-42 TRFC704 (1.4V Hynix MJR - SoC 1.175V)
Video Card(s) Galax RTX 3090 SG 24GB: Underclocked to 1700Mhz 0.737v (375W down to 250W))
Storage 2TB WD SN850 NVME + 1TB Sasmsung 970 Pro NVME + 1TB Intel 6000P NVME USB 3.2
Display(s) Kogan 32" 4K 70Hz + Gigabyte G32QC (1440p 165Hz) + Phillips 328m6fjrmb (1440p 144Hz)
Case Fractal Design R6
Audio Device(s) Logitech G560 | Corsair Void pro RGB |Blue Yeti mic
Power Supply Fractal Ion+ 2 860W (Platinum) (This thing is God-tier. Silent and TINY)
Mouse Logitech G Pro wireless + Steelseries Prisma XL
Keyboard Razer Huntsman TE (custom white and steel keycaps)
VR HMD Oculus Rift S + Quest 2
Software Windows 11 pro x64 (Yes, it's genuinely a good OS) OpenRGB - ditch the branded bloatware!
Tiachi ordered. Now we wait.

Found a video that suggest the extra 4 pin can be used to help supply PCIe power if your using a lot of PCIe devices.
That makes sense

You know, I wonder if this ties into the issues some people had with USB device disconnects due to the PCI-E bus resetting (AGESA updates reduced it, but it's still semi common - i've only ever caused it with a damaged PCI-E riser cable and a crappy asmedia USB-C port)
High power draw PCI-E/USB devices, combined with someone skipping an extra power connector...
Even if they're seperated, It's given me something to think about next time someone brings that issue up (are they using all the power connectors, extensions, etc. I have a 4 ->8+4 dubious splitter here in the spare parts bin)
 
Joined
Jul 30, 2019
Messages
635 (0.52/day)
System Name Not a thread ripper but pretty good.
Processor Ryzen 9 5950x
Motherboard ASRock X570 Tiachi (revision 1.06, BIOS/UEFI version P4.80)
Cooling EK-Quantum Velocity - Nickel + Plexi, EK-Quantum Reflection PC-O11 D5 PWM, EK-CoolStream PE 360
Memory Nemix DDR4-3200 PC4-25600 2Rx8 ECC Unbuffered Memory (2 sticks, 64GB, micron)
Video Card(s) XFX Radeon RX 5700 & EK-Quantum Vector Radeon RX 5700 +XT & Backplate
Storage Samsung 970 EVO M.2 2280 500GB PCIe Gen3x4 NVMe
Display(s) 2 x 4K LG 27UL600-W (and HUANUO Dual Monitor Mount)
Case Lian Li PC-O11 Dynamic Black (original model)
Power Supply Corsair RM750x
Mouse Logitech M575
Keyboard Corsair Strafe RGB MK.2
Software Windows 10 Professional (64bit)
Benchmark Scores Typical for non-overclocked CPU.
That makes sense

You know, I wonder if this ties into the issues some people had with USB device disconnects due to the PCI-E bus resetting (AGESA updates reduced it, but it's still semi common - i've only ever caused it with a damaged PCI-E riser cable and a crappy asmedia USB-C port)
High power draw PCI-E/USB devices, combined with someone skipping an extra power connector...
Even if they're seperated, It's given me something to think about next time someone brings that issue up (are they using all the power connectors, extensions, etc. I have a 4 ->8+4 dubious splitter here in the spare parts bin)
There seems to be some debate here about this. I wonder if it can be easily settled by some testing using a multi-meter.
 
Top