Mr. Fox

Aside from hideous thermals (especially insane hotspot) the 9070 XT was indeed a great value for budget-restricted gamers if it could be purchased for within $100 of the correct price. It was so-so at overclocking. Where it really shines brightest over all other previous Radeon GPUs is ray tracing. It jumped AMD from abject failure to 3090 ray tracing levels in a single bound. Before the 9070 XT this was AMD's most severe shortcoming in terms of image fidelity. When ray tracing first became a popular talking point I considered it a gimmick, but now my opinion is the exact opposite. I think it is the biggest visual enhancement in graphics fidelity in over a decade and I wouldn't be nearly as satisfied without it. A real game-changer (pun intended). The kind of thing where you don't know what you've got until its gone. Whenever I play games that lack ray tracing it is immediately evident and they feel very cartoonish. and similarly insane... I called this out as soon as I saw it: proprietary, fragile, modular piece of crap with no service parts. Absolute rubbish FE GPU. NVIDIA is replacing the GPU for damage control reasons, not because they care. If they cared they would not have produced a piece of garbage.

That is entirely possible if the colored plastics have a lower melting point. The conventional black plastic connectors are something close to 200°C if my understanding is accurate. Per Google Gemini: https://gemini.google.com/share/1bef51d272b8 The housing for both the conventional 12VHPWR and its minor revision, the 12V-2x6 connector, is typically made of a glass fiber-filled thermoplastic 🌡️ Material and Flammability Rating The connector plug housing is required to be made out of glass fiber-filled thermoplastic. This material is specified to meet the UL 94 V-0 classification for flame retardancy. 🔥 Melting Point The melting point of the plastic is not a commonly published specification, as the critical thermal consideration is the maximum operating temperature the plastic can withstand before deformation or failure. However, related testing provides some context: The notorious connector melting issues reported with the 12VHPWR connector were primarily caused by poor or partial seating of the connector pins, leading to high electrical resistance, excessive heat generation, and subsequent melting of the plastic. The material is designed to handle the heat generated under normal, proper operating conditions. Some third-party solutions aimed at protecting the connector have been designed to shut down power if the temperature near the connector reaches around 110°C—this suggests that temperatures significantly higher than this, caused by a fault (like an improperly seated cable), are enough to cause the plastic to melt or fail. For reference, the melting point of the lead-free solder often used in manufacturing is around 232°C, which is significantly higher than the reported plastic failure temperatures. The connector's pins heat up due to resistance, and that heat then causes the surrounding plastic to melt. The high temperature that leads to melting is generally due to thermal runaway caused by poor contact and extremely high current density in a small area, rather than an inherent weakness of the plastic under its rated operating temperature. The article below discusses the issues with the 12VHPWR connector, including contradicting specs and corner-cutting, which led to the reported melting issues. 12VHPWR is a Dumpster Fire | Investigation into Contradicting Specs & Corner Cutting

As odd as it may seem, a game that is more demanding than one might imagine is Quake II RTX. The way that game has been modded and modernized not only looks insanely good with the new textures and ray tracing, it pulls an insane amount of power. With 5090 running stock (apart from the presence of my shunted resistors) HWiNFO64 shows GPU rail powers is in the 800W range. Hard to believe a game that is so old could be so taxing on a GPU. I cannot think of a scenario where I would find a practical use for it, but the curiosity is still there. I am always curious about stuff like this regardless of whether or not I would find it useful. I may check out the trial for that reason alone. I have always preferred local storage over network storage, but my activities on PC are such that network storage is not beneficial in practical application. I have a 1TB USB SSD connected to my router with music and movies on it that are accessible from any of my computers and televisions. Other than the initial testing to confirm functionality I have never accessed any of the files on it. The only files that I actually take care to avoid losing are applications that I have purchased and those that are still functional but no longer available (or difficult to find) for public download and I have those sitting in cloud storage for safe-keeping. My interest in Linux was always driven by curiosity. It has ramped up as my growing contempt for Micro$lop and disdain for Winduhz has peaked to unprecedented levels. I'm very happy that I did not have to start from a place of total unfamiliarity and ignorance as my desire to eliminate the use of it increases.

I doubt it will happen. It might, but I would not hold my breath. If it does it will probably be priced like a 4090 and have 100°C hotspot temps.

I should just sell both of my monster computers locally for about $5K each and be done with PCs, LOL. None of these parts are worth buying at current prices. It feels almost like part of the evil plan to make us all "own nothing and be happy" whether we like it or not. @Raiderman had better snag a 5080 now and not wait for the "Super" version that will likely sell for 5090 prices, or maybe even more.

I wonder how many that had melted connectors were installed in challenging situations where space was abnormally limited and smart cable management was either very difficult or even impossible. We never know the entire story behind the photos and YouTube videos. User error includes choosing to install the GPU in an environment where its survival is unlikely. The examples of Radeon GPUs with melted connectors really dispel the notion that it was due to excessive load because those GPUs are incapable of drawing more power than the cables and connectors can handle.

Interesting video. Brother @electrosoft asked a rhetorical question not long ago about why nobody we know has melted connectors. Nobody could give a definitive answer. This makes me even more curious. 1600W for 15 minutes did not melt the connector. I think the answer might be "defective Chinese trash" after watching this. More accurately, "expensive defective Chinese crap." People are getting worried about exceeding 60-70°C and twice that much heat didn't melt the plastic connector. Thank you. I think I can still do better. Just need to get my core temps to stay down in the 20°C range or lower.

The curse of Frank Azor. Congratulations AMD. You really found yourself a winner.

You explanation piques my curiosity so I may have to investigate it. Even if it turns out I do not have a need for it, I am still curious. I would say at least 1300W. The Lian Li Edge seems like a solid and affordable option. I have one in my 4090 build. I have a Thermaltake GF3 1650W PSU in the 5090 build. I like both of them. The GF3 is hard to find. It has dual 12VHPWR sockets and like 6 PCIe 8-pin sockets in addition. I may have to drop down to 0.001 Ohm shunt resistors to bump my power limit. I backed off my core clock a bit and the scores went up. I ran the benchmark again with GPU-Z and HWiNFO64 running on my second monitor so I could watch and it is still showing power limit perf cap reason, so... hmmm. I'm pulling 1350W from the wall already. HWiNFO64 shows like 959W on GPU power rails. Apparently that's not enough. My core temp is still hitting 41°C with 9°C water, so that's not helping. I guess I am going to have to think about using liquid metal on the GPU. I don't want to, but 41°C is definitely not helping https://www.3dmark.com/3dm/144631382 | https://hwbot.org/benchmarks/3dmark_-_steel_nomad_dx12/submissions/5918232

It is definitely a specification engineering flaw and inherently defective design and not an NVIDIA defect. Any company stupid enough to use it is going to have burnt and melted wiring and connectors just because it's trash. The only thing we can blame NVIDIA for is being the stupid bastards that made this defective piece of crap the new normal.

I think the temptation will be easier for me because I have worked from home as a "virtual office employee" since 2002. I've never worked from an office in all of those years, although I used to have to travel extensively and excessively. Thankfully that ended about 7 years ago. Now I just stay home all the time. I never go anywhere and I don't want to. My compute is walking from my breakfast stool in the kitchen to my desk in the office. About 30 feet one way. There is no "stopping by" anywhere on the way home. It's both a blessing and a curse. I'm so used to staying home all the time I never want to go anywhere, LOL. The hour drive to Micro Center will be a massive pain in the butt. There won't be any impulsive shopping as long as I don't just go there for the heck of it. After about 13 years of 75% overnight travel and only being home one or two days a week, I don't like driving even 30 minutes to go do something anymore. (Mrs. Fox finds that pretty annoying and boring, but she hasn't forgotten my absence 5 to 6 days a week for many years ) All that being said, I'm sure if I have something specific I want to buy I will be more than happy to waste 2 hours of my life driving back and forth to get better pricing and a better warranty, no shipping and the overall better Micro Center experience. I'm going to like not having to purchase almost everything online anymore. That always sucks.

I would love it if this works and we could make our own drivers for Windows 7 on whatever modern hardware we choose to run.

Interesting....

Off-Topic: The Phoenix Micro Center grand opening for VIP members is 11/5. My wife better hide my wallet. 🤣 It's an hour each direction from where I live, but at least it's not the 6-hour drive each direction to Tustin, CA.

Another reason to disable it. Windows Updates help keep the PC repair industry working. "Please be patient, Windows is currently bricking your computer..." From the selected start time...

Happy to help. I have not been able to beat any of my Strix high scores with the Apex using my best 9950X. https://hwbot.org/benchmarks/cinebench_-_r23_multi_core_with_benchmate/submissions/5898291 https://hwbot.org/benchmarks/y-cruncher_-_pi-1b/submissions/5904464 I hate cutting aluminum and you are right about using a dremel or grinder. It loads up the griding disk with metal and resists the process. Using a hacksaw or a jigsaw is the easiest way to cut aluminum, but you can't use a hacksaw for some things. My MSI X870E Carbon and the Z790i Edge both had the rear I/O heatsink made about 1/16" too long and had contact interference with the GPU backplate. I was able to install the GPU in both motherboards but it was jammed against the backplate hard enough to damage the anodized finish on the GPU backplate. The NVMe heatsink was also touching the backplate on the Z790i Edge, but not jammed against it super hard. I had to install the GPU first, then the NVMe heatsink.

For the GPU, I would say select whichever one costs less, especially if you are putting a waterblock on it. My Zotac 5090 Solid OC was a good buy compared to the other more expensive options that deliver nothing for the extra money. It is an excellent GPU. The air cooler on it was fantastic (unlike some of the other affordable brands/models). It ran freakishly cool for an air cooled GPU. The only 50-series GPU I would recommend avoiding like a plague is an FE model. For overclocking potential probably the best GPU silicon quality most consistently will be an AORUS Master, but the cost vs benefit isn't justified. I love overclocking more than anything else I do with a computer, and really the only reason computers matter to me at this point, but paying a WHOLE LOT more for a very small gain in GPU benchmark scores is just not a very intelligent decision. I have owned the following X870E motherboards and I list them in my order of preference: X870E AORUS Master (best overall - only flaw is no way to disable WiFi/BT in BIOS) X870E-E Strix (replacement for second AORUS Master that arrived with shipping damage) X870E Apex (returned first for refund, second was junk, I am using #3) X870E Carbon (returned for refund - good mobo, but no async BCLK and weird glitches) X879E Taichi (my least favorite out of all AMD motherboards I have owned - hated it) I had a X870E Taichi and hated it. The PCIe bifurcation was garbage and I did not care for the firmware. I have only owned two ASRock motherboards and did not like either one. I had a B850 AORUS Elite that I used in a build for my granddaughters and it was excellent. The only criticism I had was the PCIe slots below the GPU slots were X1, but using them did not drop the GPU to X8. This is unavoidable with anything below X870E dual chipset due to a lack of PCIe lanes with an non "E" AMD dual chipset motherboard. PCIe X1 dramatically reduces NVMe speed... makes NVMe speed like SATA SSD. If you plan to insert anything in other PCIe slots in addition to your GPU in an AMD motherboard the "E" version is an absolute must have. The only complaint I have with the Gigabyte boards is no way to disable WiFi/BT in the BIOS. Super stupid flaw they could fix effortlessly if they cared. If you use WiFi/BT and use Windoze 11 as your main OS (I do not do either one) this truly is a non-issue. It really pissed me off that Gigabyte did not provide that option in the BIOS. I asked twice and both times they said no... "you're the only person complaining about it" (essentially we don't care what you want and you are not worth the minimal effort needed to make a BIOS as good as our competitors). Gigabyte is the only brand I know of that omits this essential basic BIOS option. The Strix was an accidental blessing. I purchased a second AORUS Master from Central Computers on sale for less than what I paid for the first. The big and heavy NVMe heatsink under the GPU was not latched. Apparently shipped from the factory without being latched. It flopped around inside of the box and broke several things and scratched up things that did not get broken. I asked them to open the box and inspect before shipping a replacement. They had quite a few in stock and ended up opening all of the boxes and all were damaged in the same way. They offered the Strix for no difference in price. I accepted. The Strix is better than the AORUS Master in terms of firmware. A close second only because I could not install both of my Sabrent quad NVMe X4 cards like I could in the Master. It only has one extra PCIe slot. The AORUS had two, both usable at X4 without dropping the GPU from X16 to X8. The AORUS Master allowed me to install 10 NVMe SSDs and 4 SATA drives while maintaining the GPU at X16. The Apex is a great motherboard with a glaring engineering defect entirely due to an idiotic PCIe slot arrangement. I can only use the X4 PCIe slot above the GPU. The Sabrent quad NVMe card's heat sink touches the GPU backplate. The Strix performs as well as the Apex in terms of the CPU overclocking. It has asych BCLK and I can use the Sabrent card in the bottom slots without the GPU dropping to X8 like it does in the Apex. If I knew everything I know now before buying my first I probably would have purchased two X870E-E Strix Gaming WiFI. If I were going to recommend one, it would be the X870E-E Strix as the best all-around X870E motherboard with the fewest flaws and compromises. Hope this helps. https://www.newegg.com/asus-rog-strix-x870e-e-gaming-wifi-atx-motherboard-amd-x870e-am5/p/N82E16813119682

I have never been exposed to Unraid. I have obviously heard of it from those that build and use home servers, just never had any reason to use it myself. It is a full OS with a GUI or a CLI only? Linux? Did you (or do you plan to) get replacements from EVGA for the PSUs?

Recently discovered this channel. This guys is great. Pulls no punches and calls the balls and strikes regardless of who is up to bat. Go Alberta, go...

Interesting... #1 and #3 are the same reasons for me, too.

Maybe if I can buy one cheap enough. I wouldn't mind getting my hands on an AORUS Master 5090 and adding a second 12V-2x6 connector to it. If not then I might shunt the 4090 Suprim and add an EVC2 to it to see how close I can get to 5090 performance with a vasectomy reversal. I think whatever I buy going forward, whether it is made by Nvidia or AMD is going to include an EVC2, unless another K|NGP|N option with unlocked voltage surfaces. Shunt mods are going to also be as routine as delid and bare die have become for me. If I can't do that then I don't have any use for it and there is no point in buying it. Stock is no more exciting to me than running a Celeron.

From selected start time...

People that are willing to enable this dung so they can play the latest Battlefield and Call of Duty need to watch this video. It should be terrifying to anyone that has a brain that still functions correctly. Hard to believe that any person of average intelligence could actually believe that Micro$lop is an above-board company that has the interest of consumers at heart. They do not care if people pirate Windows because they are data pirates. And YouTube is kissing their butt.

YouTube and the Micro$lop Mafia. Screw 'em both. May their nasty executives go broke and burn in hell for eternity. I prefer Rumble over YouTube anyhow. I hope a bunch of YouTube content creators get ticked off, leave and move to Rumble.