electrosoft

When the cache is your enemy.....

I was listening to a YT go on about the connectors and he did note that many black connectors could be scorched too but you just can't see it. That gave me pause.... Outside of the discoloration, many of the burned yellow connectors are not physically changed (while some clearly are). It makes me wonder how many have scorched their connectors in daily use and don't even know it? Good early detection system. Maybe in the future all makers should make the actual insertion portion white/red/yellow so you can clearly see if there is some scorching or discoloration going on in daily use.... Nah bro! Your worst decisions allow us to continue to live vicariously through your purchasing decisions! 🙂 Looking forward to your results and $800 over an Astral actually isn't terrible seeing as the XOC vBIOS hasn't shown up for us mere mortals in almost 9 months and counting.... I think I said this before, but once you exhaust that cache things can get wonky depending on the game but if the game can 100% sit in that cache it's lights out wins for AMD every time. Problem is games are getting bigger and bigger and texture active and look ahead calls can tank that cache super fast. AMD (and Intel even) need to increase that cache size even more now. But for AMD and X3D you need to tune your system just like Intel so when you do fall outside of the cache your system is ready to minimize the hit as much as possible. Take a game like Fallout 76 due to the engine age everything sits in the cache at all times and it just destroys Intel and scarily so. Ironically, their GPUs can't hold a candle to Nvidia's cards in Fallout 76. Best combo continues to be X3D + Nvidia GPUs. WoW on the other hand started off with Intel having such smooth, consistent lows that the overall gameplay was better but recently with newer updates across the board, I'm no longer getting that chunk or stuttering with AMD and I haven't in quite awhile after switching to the 9800X3D. On the other hand, that tuned 59/45/60 tuned SP109 14900KS in the wife's system and 9070xt at 1440p Ultra 10 max RT is an absolute beast and with the newest drivers from AMD and refinements from Blizzard in Tazevesh it is locked at 165fps just about everywhere now. All the previous issues with AMD cards on WoW are a thing of the past and I continue to be very impressed with the 9070xt. 275hx in my Alienware is holding it down nicely keeping that 5070ti mobile pegged to 100%.... Basically, shut up and play.... 🤣 Looking good! Is there a particular reason you decided to downsize one of your Gigabutt boards and keep smashing your noggin against the wall trying Apex boards over and over....and over? 🤣 Hopefully with next gen AMD chips really addressing the fabric issues in a major way, this will be a thing of the past and we'll get cache on both chips. While it was a pipe dream for the 9950X3D with both CCDs having X3D, logically with its latency and fabric issues, it just didn't make sense.

All I'm saying is the math isn't mathing....and luck is not at play in engineering fault tolerance testing. Plus with the user reports all we have is their subjective takes ("I swear I checked everything right before going to church and feeding the homeless!") usually rife with oversights and human error we do not know or see. All we see is the end product which is a burned cable and/or fused connection. Then people run with it..... If the connector was that flawed, those who routinely push their cards should suffer a much noticeable higher rate of failure yet amongst the enthusiasts on the forums the rate of failure is minuscule if even that. I've long thought this since the 4000 series and it continues to prove itself out over and over. I frequent the various enthusiast forums, enthusiast FB tech message groups, tech focused Discord along with clans and guilds chock full of 4090 and 5090 owners gaming 8-12hrs/day and more waiting for this influx of burned connectors amongst the various types used considering how hard the hardware is being pushed even on non shunted cards running max OC's constantly hitting 600-650w....yet......much ado about nothing. I ran my 4090 OC'd from start to finish. I am doing the same with my 5090 too. One thing I do for my cards since Ampere is I run them max Underclocked when not gaming/benching then Max OC when gaming. Look, it's a cheap part to replace for the cable. I can swap in numerous types or models if I want. That isn't the issue. Zero loyalty to a damn connector cable 🤣 The one and only time I had issues was with the CM adapter cable I was going to use and it kept black screening and crashing and it turned out to be severely problematic and was a legitimate issue even outside of human error or potential inherent problems. The replacement one they sent me is still sitting sealed in its pouch on my shelf. I am really curious about the Ampinel and am almost 100% I'll be picking one up just out of curiosity. And you're correct, many on OC.net DO spend more money on quality products than normies. That just adds even more weight to my argument. With all that being said, I still wish they had stuck with 8-pin PCIe connectors but I guess with the way power consumption is going, we would end up with cards with 4-5 of them if not more at some point on certain models. Imagine a HOF with 8 8-pin connectors..... 🤣

The gift that keeps on giving across all spectrum including PSUs, tentacles and AMD and Nvidia! Ugh.... On the flipside, has anyone in this thread experienced a burned out connector? Over on the enthusiasts forums it is indeed rare to encounter a burned connector especially at proper 600w spec. Why can you have some throw 800w+ through their connectors on the regular or bench at 1200w+ and no problems and Bobby the Basic User burns up their connectors? Even in that realm, it is a small sample across all connectors.... How is it we all ran XOC last gen and not one burned connector among any of us? Why is this?

As soon as I saw this, I thought, "Oh, he's purposely trying to trigger @Papusan and @Mr. Fox!" 🤣 Looks good! I, too, loop my connector(s) over the top of my GPUs when able so you can see the whole card and it takes a bit of the downward pull off the connector. I've done this since Ampere as an additional reason on top of aesthetics. Hmmmm, I wish you had the other board too to see if you could reproduce this on two of them. Might need to talk to other X870e owners to try to recreate a similar test scenario to see if they can trigger a memory fail.

Love by the sword, die by the sword @Papusan!! Time to order your pair (or two!) ASAP! 🤣

It didn't even cross my mind l2c vc controller support would be an issue so I understand. Hopefully they can provide an update or if your other one is supported. ----------------------------------------------------------------------- More MSI connector woes, but again with the sheer amount of cards being sold, it does come down to potential user error or one off issues. This was on a Gaming Trio (not like that matters) This isn't like the CM debacle which was much more prevalent and seemed like a daily occurrence with a crap ton of units showing up at repair shops across the world. Lightning striking the second user TWICE is just odd...... The bulk of users over on the forums shooting 800w+ daily through their units having zero problems. The vast majority of 4090 users never having a problem. Just like my 4090 I run my 5090 full tilt. The one bonus about using the supplied OEM tentacle is if it burns, you're completely covered from the OEM/AIB unlike if you use a third party connector. I will always continue to use the supplied connector. I might pick up an Ampinel simply because it looks interesting as a protection mechanism.

I'm so glad the damage was material only and everyone is safe and sound @Papusan!! Material items can be replaced / covered. Loved ones can not. Good fortune it didn't hit the house or do even worse. By chance, was your collection of Crocs maybe caught under the tree? Maybe? 🙂

Claims of turning a 9070 into a 9070xt have been greatly exaggerated.....as always..... Agreed, the juice isn't worth the squeeze on that one. Eventually a time will come around where you might have a scenario where you can test it in transit, but with everything set up as now? Nope. So both the 9950X and memory are retail rolling the dice? Good luck! My experience with EXPO is similar with some kits. They work as rated, but if you lower the frequency, they suddenly become unstable. This was the case with a Crucial 6400 kit and G.Skill 6000 kit. I've also had several needing more voltage on the Aorus B650i than rated versus the x870e Hero. I can see why kits get returned with such unstable variability. In the end, I just end up dialing each and every kit in manually. Same as I did for years on Intel. That, in turn, shows the variance between even "same" kits which made me realize (you too obviously) that binning also exists with memory kits too to go along with CPUs, GPUs and depending on how hard you're pushing even motherboards. Ugh... 🤣

Can the Strix run 8200 looser or caps out at 8000 regardless?

Overall trend from makers like Thermalrite and ID Cooling and more along with even MSI and Phanteks offering excellent 360mm AIOs for $100 or below. Go to PCpartpicker and the sheer amount of 360mm AIOs available for sub $60 is refreshing to view. Even 360mm models with LCD displays now can be had for sub $100. This flies in the face of Asus and NZXT trying to pass off display enabled AIOs for $440 and $300 respectively. Insane and truly made for the idiots out there.

Most modern users, even over on OC, all prefer clean runs to their 5090s and below. Most are either using the supplied PSU 12vhpwr cable or custom ordering one for their PSU with the connector included. Just about every model in the Nvidia consumer stack now uses it. Who in this forum are using the tentacle and running it to 4x PCIe 8 pin connectors to the 5090 besides me? Unfortunately we're three generations in now on Nvidia and the 12vhpwr connector. Nvidia dominates the market so it makes sense that PSU makers are starting to default to having one on their PSU What I would like to see though is always having at least 4x 8 pin connectors too along side it especially on models 850w+ and up but the bean counters have run their numbers and some models are sacrificing them for $$$. For lesser wattage models I can kind of give it a pass. So far we've only seen two AMD models with it: The Taichi and the Nitro+ 9070xt models. I suspect more will adopt it for future models. 😞

🤣

What the hell man!!!! Et Tu @Raiderman?!?! 🤢 On the other hand, the system looks great as long as you remove that abomination from the foreground.... 🤣

Decent video showing once again the power of X3D but more importantly the overall futility of 6000 vs 6400 vs 8000 for AM5 along with the strength of 2DPC boards vs 1DPC with 800 series chipsets.... I'd like to see this run back on non X3D chips....