Mr. Fox

This is the original already delidded/bare die SP118 CPU back in the Strix. (I still need to delid the new one and put it in the Master, along with installing the EVC2 on the 5090.) https://hwbot.org/benchmarks/y-cruncher_-_pi-1b/submissions/5904464 https://hwbot.org/benchmarks/y-cruncher_-_pi-2.5b/submissions/5904468

So the 6000C26 kit is fine. Modest improvement. Tighter timings and lower voltage, but not a remarkable change in overall performance. Now I need to decide if it is worth keeping considering my loss of 16GB in memory capacity.

Well, the new 9950X is close to an identical match to the better of my existing two. So, definitely not a lottery winner but good enough to keep. Look how close they are. Crazy. (#1 is the older CPU and #2 the one that arrived today.) The newer one might be just a hair better, but not enough difference to matter. So, now it is delid and bare die and then test the 6000C26 32GB kit (which is already installed and running the same custom profile as the old 48GB kit).

Don't care about the game, but it has always been too expensive to buy just for the benchmark. Now it's worth getting. Less than an hour until the sale ends. @Papusan - for $2.99 definitely worth buying for the benchmark. https://store.steampowered.com/app/507490/Ashes_of_the_Singularity_Escalation/ https://www.ashesofthesingularity.com/escalation/metaverse#/ladders/benchmark/overall/Crazy_4k

I've never been interested in this game because I don't care for the genre, but I always wanted the benchmark. It has always been too expensive, but now Steam has it on sale for $2.99 and I grabbed it. Hurry... less than an hour before the sale ends. @Papusan Here is a stock run on the 4090. https://www.ashesofthesingularity.com/escalation/metaverse#/ladders/benchmark/overall/Crazy_4k

I really question the intelligence of continuing this hobby sometimes. I mean, I love it and probably always will, but I despise how most of what is for sale today is grossly overpriced rubbish. It is frustrating, infuriating and inexcusable. It really pisses me off when I spend money (sometimes lots of it) on something that I expect to be exceptional and the only thing about it that is exceptional is how exceptionally lousy it turned out to be. There should be a punishment for selling trash, but there isn't and that is why it is so common.

It can run 8200 just fine and do most things, but will not pass memory tests without errors. It could be the CPU or the memory modules as well. It has the better of my two CPUs in it, but the memory controller might be a bit weaker. Because I am running both systems bare die and liquid metal I don't want to just move them around for the sake of curiosity. The memory performance at 8000 and the same timings as 8200 on the Master are almost the same (obvious in the screenshots above) except that the CPU in the Master maxes out at 2167 FCLK. It will not even boot at 2200. I suspect that is why they are nearly the same memory performance and latency in spite of the 200MT/s higher speed on the Master. I've got another 9950X and a 6000C26 32GB kit arriving late today for testing. If they are better than what I already have I will keep one or both, delid the CPU if it is better, etc. If not, will return both of them for a refund tomorrow. What is really goofy and hard to explain (as the case is with most AMD quirks and weirdness) is that the memory overclock is dialed in nicely and totally stable at 8200 on the Master with the crappier CPU, but if I lower it to 8000 the same memory clock becomes unstable. https://hwbot.org/benchmarks/y-cruncher_-_pi-2.5b/submissions/5904049

A little tighter. I ordered a 6000C26 32GB kit (will arrive this evening) that I am going to try on the Strix to see if it runs better with those. It's falls a bit short on memory overclocking compared with the AORUS Master. Even with the better CPU installed in it the Strix struggles with 8200 compared to the Master. It doesn't seem that Nitropath does anything to make it better, in spite of the marketing rhetoric from A$$zeus. The ROG firmware is better, (stating the obvious,) but the AORUS board is better. Wish I could crossflash the ROG BIOS to the Master. I will see if the Strix does better with the 6000C26 kit. Getting the tPHYRDL to match on the Strix has been elusive. The idiots at Gigabyte don't even give an option in the BIOS to disable the WiFi/BT module that I don't ever use (and lacks W10 drivers).

Check this out. This could do a lot to help avoid melted connectors. Active load management per pin on the 12VHPWR socket. https://www.igorslab.de/en/aqua-computer-ampinel-exclusive-test-the-smart-plaster-for-nvidias-open-12v2x6-wound-and-the-ultimate-lifesaver-for-any-affected-graphics-card/ Property / Function Description / Value Current balancing Active distribution and reduction of currents on all 6 lines of the 12VHPWR plug Monitoring Continuous measurement of current, voltage and temperature at plug connections and electronics Alarm configuration Multi-level: visual, audible, software termination, PC shutdown, emergency shutdown of the GPU power supply Operation OLED display (128×64 px), one-button control, integrated buzzer (approx. 85 dB) RGBpx lighting 8 integrated LEDs, effects for operating states, alarms and user configuration Software support Supported by aquasuite (Windows): Data logger, overview pages, virtual sensors, profiles Profiles Manual or automatic switching, up to four profiles Playground Virtual software sensors, output actions, global profiles, hotkeys aquasuite web Data export, import and publication via Aqua Computer Server Compatibility Support of HWiNFO and AIDA64 via shared memory Firmware Update and language selection via aquasuite Security functions Integrated protection mechanism up to complete shutdown of the GPU power supply Materials Aluminum, silicone, PMMA, epoxy resin Dimensions 72 × 75 × 30 mm Weight approx. 100 g Supply voltage 12 V DC ±5 %, max. 0.5 A Maximum output power 650 W Maximum power loss 10 W Ambient temperature 10-40 °C (non-condensing) Audible alarm Buzzer, approx. 85 dB Lighting 8 digitally controlled RGB LEDs Interfaces USB 2.0, 2× signal output, 1× temperature input

Yup, also shows all for the Strix. But, tPHYRDL still shows a mismatch. I think it is accurate on the Strix. I'm not sure that tPHYRDL mismatch is actually an issue though. Seems like a lot of noise around that topic to me. I don't think it actually matters, but getting them to match is difficult on the Strix. Maybe it is the "special" A$$zeus NitroPath gimmick that slightly diminishes the memory overclocking on the ROG crap. Edit: Nope, almost. VDDIO is still not being read on the Strix. Oh well. At least the Master has all of the readings now.

Yes, that is Red Fox. He is a smart boy, but has a sassy mouth. @electrosoft, @Raiderman and @jaybee83 there is a new version of ZenTimings available that shows all of the sensors for voltage now. I will test to confirm that is true on the Strix as well, but it shows all sensors on the Master. It also shows tPHYRDL is matched whereas the last beta showed them mismatched (even though it was not). Also shows the Nitro settings in the lower right corner.

Free and safe versus not free and not safe malware OSes from crApple and Macro$lop... https://youtu.be/Zt9gH34Zw2Q

Nice to see something decent going down in price. Total opposite of what we normally see. Thermalrite products are generally respectable quality.

So far no problem but I do not think I will ever have any level of confidence in the durability or resilience of the petite arson connector. The risk willl always be there and requires monitoring. It is an unfortunate and unnecessary situation that we find ourselves in. And the hits just keep on coming...

Change that occurs purely for the sake of effecting change is never good or welcomed by anyone other than the inventors. There was zero legitimate need for this. It provides no benefit whatsoever to any consumer. The only logical explanation for it is that it was a very carefully executed strategy to benefit NVIDIA and the PSU makers and create new revenue streams through deliberately fabricated obsolescence. People buy PSUs with 10+ year warranties and continue using them after the warranty expires. What better way to force consumers into buying something they don't need?