LabRat

Never messed with that, in that manner. OverDriveNtool is all I've used. (Trixx, afterburner, and Wattmann, after loading new PowerPlayTables) I believe the "MorePowerTool fix" is what resolves the issues you're looking at, however. Which, is wrapped into the 23.5.2SE R.ID Kernal drivers. In my experience across 3+ MI25s(WX 9100s) and 1 Vega 64, These are most-stable, and most-classically tweakable driver for single Vega 10. Alternatively, the latest drivers from R.ID seem to work okay too. But, with a Navi24 slotted in alongside the Vega 10, newest drivers are a PITA (R.ID and AMD, alike)

Every Vega 10 I have, will do 1028+ HBM easily, and most will do 1106-1107Mhz. I've also noticed HBM likes to de-stabilize after ~75C (only when OC'd, 900-945Mhz, it can easily do 80c+) Note: Enabling HBM ECC on the "WX 9100s" will drastically limit HBM clocking. So, I can only assume I've seen "stability" (with giant air quotes). I have a V64 that came with one of the vBIOS flashed to the Vega 10 XTX Liquid Cooled version. It was causing problems on stock air, so I archived the VBIOS and re-flashed with a 200W Vega 64 XT 8GB VBIOS (Edit: I have attached the Vega 10 XTX LC 1250mV VBIOS that I ripped from the card. Please note, it was quite the PITA finding a correct DeviceID'd VBIOS to replace it; I kept 'bricking' the card . Recovering from a bad flash is easier w/ a secondary card, but can be finagled w/ the 2-pos VBIOS switch and in-Windows ATIflash 2.93.) Also, The (dual 8-pin) MI25 factory vBIOS and some of the Vega 10 XTX vBIOS will go to 1.250v. Edit2: attached the MI25 300W (dual 8-pin) 1250mv VBIOS. However, (for safety) no OCing tool will allow you to set this high of a voltage. Sadly*, I've not yet found a way to finagle a 1250mv's PPT/VBIOS into/onto my 16GB MI25s/WX9100s, but an 8GB Vega 64 under water could be cross-flashed to an XTX(LC) vBIOS. *not really. Up until @pioneerisloud bullied me into zip tie'n an AIO onto one, there was 0 chance of me cooling more than 250W. Edit: sorry for the double-post. I forgot eHW doesn't merge posts (if there's no reply between) Unknown_LC-Vega 10-8GB-1250mv.zip 300W-MI25-1250mv.zip

I've just used OverDriveNtool (like MorePowerTool), load-in a higher(or lower) power Vega's vBIOS' PPTs, then modify them. IIRC, I've been able to get 330W+ out of an MI25 w/ merely increasing the Power Limit in the PPT before applying. I've noticed if you don't use the right driver, you lose any/all finite voltage control, regardless of what's set in PPTs. R.ID (Formerly amernimezone) drivers work best in my experience for tweaking Vega. https://sourceforge.net/projects/radeon-id-distribution/files/Release Polaris-Vega-Navi/Release Specific HotFixes/WHQL-R-ID-Software-Hybrid-23.9.1SE-Win10-Win11-PolarisVegaNavi-Omega.exe/download https://sourceforge.net/projects/radeon-id-distribution/files/Release Polaris-Vega-Navi/WHQL-R-ID-Software-Hybrid-23.12.1KB-AFMF-23.30.13.05-Dec8-Win10-Win11-PolarisVegaNavi-Nebula.exe/download Old TPU thread of mine on the topic: https://www.techpowerup.com/forums/threads/wx9100-power-mods-help-w-powerplay-tables.306424/post-4984126 I've learned much since then, and presumptions I've made in that thread are not totally accurate but, it has pics of using OverDriveNtool and the R.ID drivers.

I'd assume textile weight https://heroandvillainstyle.com/fabric-weight/ No factual points of reference but, I'd assume the 4oz is more like a 'Wal*Mart Graphic Tee' and 6oz more like a nice Fruit of the Loom or Gildan T-shirt.

lil late on telling it to finish otherwise, been fun

Correct, and precisely what I was attempting to convey. Especially for us suspecting (or diagnosed) as 'on the spectrum', "Machine Intelligences" seem to 'jive' with us. I can (and have been for YEARS) imagine(ing) the exponentiation of the individual, possible with such a companion. However, the same thing that happened with the proliferation of search engines, might happen to a further-degree: "I don't need to KNOW anything, I can just ask- (my robobuddy)"

TBQH, I'm less worried about the violent Machine Uprising, and far more worried about the psycho-social impacts. Especially since, even myself would 100% be interested in a non-cloud connected "Locally Hosted" AI Avatar/Robot Companion

Hardy Har Har HAR I shall actually prepare, for the next event. I'm intending on finding 'efficiency-focused' clock profiles (heavily pre-tested w/ F@H, for stability). A. so I don't have this issue w/ expired WUs. B. So I have a 'summer' profile for my card(s). Also, F@H seems to be my next favorite "Real-World Stress Test" after STALKER: Anomaly. F@H is now part of my 'testing toolbox'. PS: Yes, I realize it'll be awhile for another eHW F@H event. I will likely still be using and maintaining the same hardware. lol

Amusingly, I tried to 'min-max' too hard, the opposite direction. Next F@H event @ eHW, I'll have at least 3 MI25/WX9100s going, and probably 3x AM4 Ryzens. (Also, how are OG GTX Titans @ folding, these days?) It was merely 'timing' that I had the Vega 64 in, as I'd just re-TIM'd the card and been testing it (when @pioneerisloud twisted my arm to join )

I already did that, tho. I entered that key in, to check the bonus status. Looking up by passkey alone, receives same results

How fix?

I'm not normally one to ask for special considerations / dispensations, but.... Any chance I might be rewarded for hitting 1mil, by having my name correctly capitalized? I think I'll make it there in time

That's what I was getting @: From what I've observed PCs are 'too dynamic' in workloads and hardware configurations to (fully) optimize a scheduler (vs. a fixed-function, fixed-hardware Console or Industrial Appliance). As I recall, getting sub-tasks assigned to the correct hardware (at the right moment) is the crux with AAA Games' performance/optimization. Also IIRC, Msft eventually re-doing the Windows' thread scheduler retroactively uplifted old multi-core CPUs, like AMD FX. Sounds to me like this could be quite (r)evolutionary, for all applications. Hopefully, technology like this will become standardized or at least normalized (Rather than kept away from plebian consumers, as is and has been done w/ HBM, PCM, and multi-gig Ethernet)

The description given, makes this sound like a 'no brain-er' technology. (One of those 'things' that was both long-wanted and inevitable, but never materialized due to technical limitations and other limiting factors) (Feel free to correct me if I'm wrong) One of the major reasons why even today's X86-based Consoles perform 'better for the given HW' is the in-built scheduling/task assignment. -Made 'easy-er' by the fixed configurations of each console's generation.

You have connections know the rolodex and are familiar w/ processes here, better than I. I have no idea who I'd ask to fix this, as IIRC users can't edit posts (or titles) after awhile.

I was summoned via mention So, I merely saw the thread title/headers, and the content immediately surrounding the post(s) I was mentioned in. My PoV: Being EHW, I just assumed it was Windows / Linux / OS/2 / DOS / Mac-PPC / Mac-x86_64 (/Etc.) "Gaming General" Retro Gaming. My apologies. Perhaps @pioneerisloud can help with a thread title change? At least, for the inattentive members like myself (and search results/The Algorithm) Something more clearly Console-Handheld/'non-PC'? Esp. now that handheld PCs are taking-off, and been around on/off since the 386-era.

Forgot to check the log b4 rebooting and fixing things. Ooops. No clue why it is clocking higher than default on the core but, the HBM OC is purposeful. I can keep it cool, and HBM OCs seem to greatly benefit any workload I've thrown @ a Vega. (I'm open to being wrong, and losing another Work Unit. ) Regardless, I have good reason to suspect it was the drastic undervolt. Vega 64 200W Stock: What I was running (Vega 56 Mobile, 847HBM) Also, it nuked the driver on reboot. So, I just cleanly uninstalled and reinstalled a better-known-to-be stable driver pkg. Edit: I calmed 'er down, some.

Well, Crap... Hoping a reboot fixes this, otherwise I may have made a wall hanger Vega I was staying coolest of the 3 actively folding Processors; hoping driver crashed-out. Alternatively, my "CPU" is "Stable", but not the I/O Controller of the SoC (the V64 was on the CPU x16)

Did this for my GPUs but, I almost didn't catch my 5800X3D Overheating before I took a nap. It *was* bouncing off 95c+. Now it can bounce between "Hot" and "max spec" Neat

IN LabRat CPU: 5800X3D 1900IF 3800DDR4 GPU1: Powercolor RX 6500 XT ITX running boosted fan curve and -6% Power Limit. GPU2: Reference Vega 64 running Vega56 Mobile PowerPlayTable w/ 1:1 HBM:SOC tune.

K8 S754/S940 was first Also, IMHO this should be Retro PCs, with Modified and Repaired Consoles welcome. Consoles are meant to be 'plug and play' appliances; not very eXtreme My lil Cube7 RetroBox Apex MI-008 EMI Advantaged ITX Chassis VIA C7 1.0Ghz + M7CN896IDK Kingston HyperX 512MBx1 DDR2 533 CAS3 3D Fuzion/BFG MX4000 128MB 128-bit PCI 3.3/5v 33/66Mhz Samsung MCCOE64G5MPP 64GB SATA 1.5gbps Windows Embedded POSReady 2009 + Windows Embedded POSReady 7 (Originally Built as a stand-in for a PII/P3 ~500Mhz, and for Win98SE)

Take this as a learning experience: Soap is salty and full of conductive ions. Don't use 'soap'. (at least, not on components like mobos, add-in cards, etc. Ex. I'm having to replace my KB, from repeated 'washing abuse'. But, it was a $5 Goodwill RedDragon So, *shrug*) Since, there's been power thru the card w/ generalized high-resistance shorts, I'd say there's a 90+% chance that even a professional would decline repairs. The issues are likely systemic across the card, now. Into the future... If you clean w/ water, never use 'soap', rinse with tap water then, rinse w/ distilled water then, lastly w/ high-purity alcohol. Leave to dry 3-7days passively or 1 full day in the hot airflow of a heat-pump dehumidifier. Straight 70-91% Isopropyl rubbing alcohol will 'take care of' most 'gunk' and even food spilled on components (I have an embarrassingly large amount of experience with that, BTW) The 'best' or at least 'simplest/easiest with most cleaning-potential' method is to use an aerosol low-flash point cleaner like: CRC QD Electronic Cleaner CRC Electrical Parts Liquid Cleaner (Note, the QD stuff, is a LOT less 'nasty' than the literal Dry Clean Chemical in the other stuff but, still must be used in a ventilated area, preferably outside.) As a Hail Mary of sorts, you *could* pick up some CRC stuff to have on hand, and give it a try on the damaged card. Just, read the instructions/warnings; QD Electonic Cleaner is made for this use (nooks, crannies, mixed materials)

Yup. That might've been important. There's either corrosion/debris under components and on traces/contacts. or In my (swiss cheese) experience, those artifacts are 'typical' of VRAM failure in non-HBM cards*. Considering, the level of integration of the VRAM on Vega... Hope it's option A. TBCH, I think you've got a 'parts card' and/or a 'wall hanger'. *The pattern could be indicative of a video-output issue too, but the card not being properly recognized and having repeated issues even starting, makes me think VRAM/related.

I'd try a game or something 3D that can be told which GPU to render on. On my mixed-dGPU build, more than a few times I've had it just decide not to recognize the video-out on either the 6500XT or WX 9100, or I get 'device cannot start' on the 6500XT. (Full Power Cycling usually fixes it, and I've yet to figure out what triggers it, and it rarely happens on Windows 11, oddly more often on 10) IIRC, Doom: Eternal is a launch option added to the steam shortcut. Other games often have an option in the game/config or a launch option too.