Mr. Fox

So, I think I had mentioned that I had an old G.SKILL 8000 CL40 XMP 48GB kit that I could never get to run stable on any Ryzen platform I have tried to use it on. It ran fine at 8400 C38 on Z790 and I just saved it in case I might need it. Well, I got it stable on the Strix. Not at 8000, but 6400 with 1:1 (Gear 1) mode and it runs pretty nice. I might be able to tighten things up a bit more, but the way RAM prices are so stupid right now we have to make do with what we have because we not only cannot afford to waste money on better RAM, finding it to buy it is difficult. The only thing you can buy is overpriced crappy bin quality right now.

I wonder how much "strategy" actually exists in the AI space. Maybe NVIDIA finally realized they were sinking too much into something that would provide no return ii the money spent. It may also have something to do with financial institutions starting to shy away from loaning money to cash incinerators because they realized doing so was not very intelligent. There are no assets or collateral, just concepts. None of the data is "owned" by the AI companies, merely pirated and hijacked data they have mined to use as they see fit, whether we like it or not.

Hi bro. I wish I did. I have not seen any memory for sale at a reasonable price since the AI clown posse screwed up the rest of the world with their insane agenda. Even used RAM is being sold at ludicrous scalper prices on eBay. This was the "cheapest" good bin 32GB kit I could find, which I ordered today for a new build and it was priced about 250-300% more than what it should be. I hope it works as well as my G.SKILL 6000 CL26 kit does. Crossing my fingers. V-COLOR Manta XSky DDR5 32GB (2x16GB) 6000MHz CL26 1.45V SK Hynix IC Gaming Desktop Memory Module Black, for AMD EXPO (TMXSO1660826KKK) Edit: I just texted you some info on an inexpensive DDR5 kit for sale on overclock.net.

Yeah @safedisk was saying I was the only one having issues and it was either my settings or cooling were not right. Totally not true. A few others posted that any AGESA after 1.2.7.0 has been problematic for them as well. I figured it out (what needed to be changed) and saved the profiles, but I will be using BIOS flashback to downgrade both machines to older/better firmware that does not have AGESA 1.3.0.0 or 1.3.0.0a. The 2101 version had some kind of issue with Bitlocker keys not working or something like that. That does not matter to me one iota since I disable Bitlocker, but it would be a major problem for anyone that does. The Kingbank 8400 48GB kit in the Strix will not even flash the new BIOS correctly. It hangs flashing "zone 5" and soft-bricks the mobo. It can only be flashed using the BIOS flashback method, so you know that something isn't as it should be. There is no reason for the flash process to fail with BIOS defaults set, and I tested the process twice. It failed both times and required BIOS flashback recovery both times.

I finally got the Apex and Strix stable with AGESA 1.3.0.0 and 1.3.0.0a but overall it wasn't worth the effort. I am not identifying any benefit whatsoever to having the newer firmware. Performance is slightly lower with the "latest and greatest" cancer... as usual. Cinebench scores are lower and old memory tuning profiles that were stable before were no longer stable. M_ORDERING changes according to @safedisk. ODT values also had to be tweaked for my systems to avoid Karhu and TM5 errors.

I just discovered this channel. Pretty decent heavy grind. The CGI in this one is pretty stunning.

Hey stranger. Where you been? Welcome back. You were missed. No idea on the RTX 6000. It's not a product that fits into a sane person's discretionary hobby budget so I've not given it even a glance. Curiosity killed the cat fox, so I just don't want to subject myself to temptation for nonsense like that.

As best I can tell it does not seem to be clock-blocked for me. My stock default voltage is 1.070V and I was able to use the EVC2 to set a +50mV offset to get it to about 1.125V and the core offset of +190 seems to hold. This was bumping up to just under 800W with a moderate load Vulkan stress test. The first image is with the EVC2 connected, so HWiNFO64 cannot "see" the WireView. The second image is with the EVC2 USB cable unplugged after applying the voltage offset so that the WireView sensor show data in the OSD. GPU-Z doesn't show much of anything accurately in terms of voltage, power or PerfCap reasons at this point. The second image shows WireView Pro II sensors in RTSS overlay. Edit 1: the way the new Imgur UI is made totally sucks. Changing to the old style only fixes the look and feel, but not the degraded functionality. Very lame, and the way it is nearly impossible to post direct image links that skip their ugly sucky crap landing page really is super annoying. Seems like every time you find something awesome, some idiot (or group of idiots) find a way to screw it up. I may have to start looking for an alternative that isn't a piece of trash, but I am not sure there is one. Edit 2: I'm trying Image Chest now. Hopefully it will be what Imgur used to be. Updated images and links.

I will give it a go after work.

The petite arson connector will always be the weakest link. Two is better than one, but not as good as none. Two places for disaster to strike. The only way to fix the problem is to get rid of the engineering defect and stop tempting fate. The higher the performance limit goes, the more likely it becomes that disaster will strike. MSI will always be MSI, which is less than ideal. But, I can see the logic in their concern about a $5000+ GPU becoming a financial liability under warranty. If they were intelligent they would have never released the 5090 Lightning because of the defective 12V-2x6 specification standard. They're not intelligent, just dishonest.

I think the vBIOS has to support it. If it does not that might be why there are no examples of it having been done on numerous other video cards rather than just a select few.

That's totally dishonest. They should take the 15% depreciation off of the current market price or refund 100%. If that user were smart he would provide evidence of the current price and insist that Silicon Power send a replacement kit from that retailer or any other retailer that has it in stock. If they refuse, then take them to small claims court. If the product has a lifetime warranty (which most DRAM does) it would be inappropriate to charge depreciation. If something is guaranteed for life there is no legal basis for depreciation.

About 12-13A per pin.

Quake II RTX - silky smooth on Linux at 5120x2160. Had to lower my power limit with LACT to 500W because the shunted 5090 with the 800W Matrix vBIOS was making the WireView Pro II scream like a stuck pig. My UPS also started squealing because it exceeds its 1500W capacity.