Ashtrix

Alex shows how the wiggling BS done by all the YT content creators from GN to everyone to ensure it's right is disproved by MSI power connector and shows the blatant design flaw of this garbage connector..

Seasonic Prime TX series uses all 12v pins for the PCIe outputs and label them as CPU / PCIe on the PSU side you can have all the watts you can ofc not exceed the maximum capacity. Ofc the cables vary for CPU EPS line vs GPU lines. But PSU has option not that all PSUs can do this but still a good PSU as you said should suffice. Clearly the existing 8Pins could handle it day and night 24/7 without any issue. But Nvidia wanted to savefor redesigning their stupidity of heatsink the denser PCB design too since Ampere V shaped PCBs due to the HS/Shroud design, all in all they just banked on the smaller connector and called it a day and forced the PCI SIG / Intel to push for it. Esp the fact on how Ampere lineup uses a 1/2 16Pin design with only 12 Pins, ironically this one had a latch while the 16 Pin does not. Plus none of the 3090s got burnt, and no 3090Ti too, a moot point. Also the HS not like mega tank, for instance if you scrape the Founders Edition Heat Sink Fins edges with a plastic pry tool or even your nails not even scrape just touch, you will lose the black paint, that's how awful their $1500 Heat Sink Fins are. Meanwhile Alienware 17 from decade ago heatsink fins never do that kind of BS on losing the black coating...Well It happened to me accidentally for a small imperfection on a 3090FE. Nvidia also got away with poor cooling on VRAM for 3090FE on the backside. Claiming how Micron GDD6X is fine for 118C, that's scorching hot and no matter what the heat cycling and the opposite side of VRAM having active cooling that heat will bleed everywhere and cause a mess in the long run. Garbage company end to end, esp the new VRAM fiasco and BS Fake frames. Cablemod support is really good, also their cables are really good quality not that they are way superior but they are just good as I noticed stock Seasonic cables have gold coating for the pins which go into the PSU while Cablemod does not. And they did a lot of design even though some of the QC issues exist like PSU pinout do not match for some custom orders etc and the 12VHPWR pin alignment like I showed, 8Pin connectors also do not align always for even Stock PSU cables but they do not have this BS issue because simply surface area of contact on 8 Pins is more than 12/16Pin BS. There's simply more Metal Mass, so much less reduced area of effect for any mishap. The 12VHPWR needs a v2 overhaul not just extending Sense pins and Nvidia needs to compensate every single owner who bought GPUs with this garbage connector with some money, I know both are not going to happen but I hope a Class Action Lawsuit like GTX970 VRAM crap.

Hyper X Alloy Origins is a good choice, its entirely made of AL chassis, removable USB C cable. Downside is UWP app for managing KB features. BUT the KB itself has 3 on-board profiles (which is very rare for many KBs including Corsair which do not have that), you do not need the software once you are done customizing. Another downside depending on user, it does not have media keys. https://hyperx.com/products/hyperx-alloy-origins-mechanical-gaming-keyboard?variant=42330451148957

Intel won't go anywhere. And if Intel dies that means DIY is dead, because x86 will be dead. Chipzilla status is not done over the night. It took them years of technology advancement and then fumbling as well under Brian Krzanich. Intel had world's most leading edge nodes, first 22nm, first 14nm and yet they faltered. They now have TSMC contracts up for Silicon fabrication. But they are not going to maintain that status so long, ARM is closing in fast. Ampere announced brand new ARM Server processors, and AMD already has Bergamo ready but Sapphire Rapids is DOA, vs EYPC Genoa. Their DC business posted loss last quarter. Which is why Intel is speeding up Emerald Rapids (More cores now vs Sapphire Rapids) and Sierra Forest (E-Core XEON to fight vs Bergamo and ARM). https://www.tomshardware.com/news/amd-will-hold-20-of-server-cpu-market-in-2023-analysts-say L4 is a given, I do not think they can make a chip with just 6P cores and beat the 7800X3D and upcoming RPL-Refresh. This is not a rumor or speculation, Adamantine cache (Intel L4 branding) patches were done in Linux kernel too. Intel's new MCM is chiplet tile based, they are putting the Compute tile + Xe GPU + SOC tile which will have P+E+LP-E cores and L4 access across the board like i7 5775C. L4 is slower than AMD's L3 X3D stack but given the new design of non monolithic design from Intel it may depend on how they execute.

They are first launching the new processors on Intel 4, aka Intel 7nm. Which Supersedes Intel 7, aka 10nm. Also note TSMC 7N is inferior to Intel 7, but the issue is Intel is 5 years late. Intel 4 is now denser than TSMC 5N as well. Intel launched Tigerlake 11th gen on Intel 7 aka 10nm for BGA trash. And relegated Rocket Lake to suffer on 14nm while it was designed for 10nm in mind, and cut the 2C4T and IMC downgraded, they knew that LGA1700 will sell a lot due to PCIe4 Chipset and new I/O and with a disaster like RKL the graphs will look good vs 10/11th gen. And BGA will get sales regardless because Intel 11th gen BGA was on 10nm and sold a ton at that time AMD was not that competitive on BGA land. Plus ADL was on 10nmESF, which is 10nm+ now aka Intel 7. So Intel is repeating the same, they are experimenting their Intel 4 node on BGA trash and give them the most important thing - L4 cache this time on Multi Tile design and with the new Baby core LP-E added to mix they can mint out new BGA chassis and use them for 1-2 years again, so OEMs will also be onboard. Plus smaller silicon die tiles. I think TSMC is involved too for their Xe GPU tile. As for Desktop LGA1800 socket, they will launch them as a stop gap. The MTL-S Intel 4 desktop processors will beat the RPL/R and Zen 4X3D in games, so they will advertise "GAMING" a lot so they will get sales volume and wont bother with SMT / MT workloads. Remember most computer builds are now just pathetic gaming boxes only. And once Zen 5 launches with 16C32T, Intel may launch Arrow Lake-S with 8P and 16E to fight Zen 5. But I do not get one thing here, Intel launching 2 CPU refreshes in a single year ? Maybe they know AMD Zen 6 will not use AM5 socket, so they cant use LGA19xx or whatever for Lunar Lake / Nova Lake in the future. But if AMD uses Zen 6 also on AM5, then it's toast lol. But this is all LGA. BGA will be selling boatloads for their new MTL-S processors. Intel management team is absolute BS. They had everything, i7 5775C, Broadwell L4 eDRAM CPU which beats 7700K and approaches 10th gen i5 performance in Games. Imagine if they kept on using it since 5th gen to 10th, 12+ the refinements would be insane. But they did not care as it costs money and also because Intel management and Investors kept on eating that Data center Market money once Brian left after Spectre Meltdown and 10nm disasters, the investors jumped ship to Apple and Management is rotting it away plus a ton of talent bleed. They even let Jim Keller leave in middle of the design of his project Now AMD and ARM are destroying their DC / HPC marketshare finally and also x86 share is ceding away to AMD, Intel responded by axing Optane / 3D-XPoint (World's best Flash storage technology and it will never be replaced) and SSD NAND business and 5G Modem and a lot and now adding the EDRAM back to their processors. Their previous quarter is a disaster in the entire history of Intel corp (HPC is losing money...only Client got them cash)

Thanks for this post bro papu. I have finally got the Cablemod Kit before last week, I was debating a lot of things regarding NVMe SSD which I will try to mention in a later post. In short Samsung, WD, all NVMe drives are a joke because M.2 standard is not use in Enterprise and they use all drives which are non 2280 M.2 so the quality is garbage (Controllers, Firmware, NAND Flash, BIOS issues, SLC caching) Anyways so the 90 Degree adapter I got from Cablemod I noticed that the bottom female connector is basically not flush, so I emailed to them and they kindly shipped me an RMA without asking anything. Now that I got another one, and I started inspecting it. Here are the photos, basically if you see the connectors the female pins on the 90 degree adapter the newer one are not perfectly straight. I also have the Cablemod 12VHPWR direct PSU cable too, it's unopened yet so I took a look and those are nicely straight. I will send another message to Cablemod for this and tell them to improve their manufacturing process. Below right is the new adapter look at the first row... Second thing, is I found this video who mentions the difference with the Nvidia stock adapter. Notice how Nvidia adapter is perfectly straight while BeQuiet is not. Looking at Igor's NTK (Tulip design Intel ATX3.0 revision also puts this in mandatory) vs Astron (Dimple outdated) Also for some reason Cablemod's 12VHPWR cables sense pins run 2 unsheated naked wires. Not sure why it is the case. Not sure if it's my OCD or not but it does not look streamlined like Nvidia adapter, I bet all the custom cables are having this flaw. Why is this important ? because as per the whole zoomed images that Igor's Lab provided the new NTK connector Tulip design latches to the male pins inside the female socket for the GPU connector and imagine if you have such variances and over the time what may happen ? I suppose the Northridge videos show what can happen, since most of those lot are from Cablemod directly. 8 Pin connectors are way thicker and are robust and never had this issue and won't have even if the pins are not perfect due to simple commonsense thinking - the 8 Pins have more metal contact mass vs 16 pins, so they are really feeble and the amount of current flow is also higher on the 16 pin vs a single 8 pin connector. Period. I already mentioned how AMD's R9 295X2 using dual 8 Pin handle 500W but mentioning it again and we all know Intel SKL-X HEDT i9 10980XE draws over 600W of power from 8 Pin CPU cables without issue when OCed to 5.1GHz all core, why does this BS company Nvidia has to do with this garbage design ?? Respect for that Northridge Fix on telling how this is not some "User Error" as quoted otherwise by GN but rather a serious design failure. GN still does not think this is a flawed product, look at their new fancy site which has some failures mentioned and conveniently says Nvidia's issue is "fixed as per GN Standards" but the expert repair guy is telling NO. His video catalog on Youtube clearly shows the dude knows about his stuff. Buildzoid also mentioned how the design is garbage vs traditional 8Pin. Really a mess.

Huge thanks to brother @Tenoroon for his good gesture on letting me borrow his GPU, the 860M 2GB card came packed well and I have been running it since last Sunday with same loads and one day I left it overnight no issue. And today I just played 6 hours of Dota2 with friends, 70W max power consumption on the GPU. So far not a single power off shutdown issue. Only a single day on iGPU the machine shutdown this happened 2 weeks back, perhaps the VRMs got overloaded and heated up after loading it on Dota2 for both GPU and CPU, It never happened again though on the same load. However now the same workload on the GPU like before and no issue. While 980M would just simply shutdown the computer at random playing music, watching movie etc. So far so good. I hope this means MXM slot is fine and Motherboard as well.

I was about to post this.,..saw it yest. Glad to see it. All these Techtubers are just monetizing heavily on the AMD - ASUS situation to max and fooling people enmasse esp GN throwing a bunch of Electron Microscope zoom pics and look at comments everyone is just simply salivating and no one is really into trying to find the truth nor even think about the inconclusive outcome. Basically It's known knowledge to all of us here. Socket Sense vs Die Sense. ASUS APEX on Intel allows Die Sense readings from Z590 series exclusively using a hardware switch on mobo enabling that feature vlatch (bro @Papusan should know it..) and the ROG engineer in that above video knows how the AMD boards are made, basically unlike Intel there's no physical switch at-least on C8H, they just incorporated it onto the top end stack of AM5 boards so any HW sensor data polling will result in accurate reading (perhaps related to how Intel vs AMD Mobo design works) in our case HWInfo. Now look at Igors lab, and HWbusters YT channel, both use same tool to monitor the Motherboard read out points and show it on the same UI claiming how the 0.05 voltage is extra over the AMD 1.3v fixed cap. As such reputed testers fail to capture another data point. Both of them have Gigabyte board.. use same tool and miss same thing. GB posted a video too check below highlighting how the HW voltage readouts are not accurate because of the Die Sense integration (they are dumb as rock though why not just speak like ASUS lol just muted and showing it normal person won't even understand even 1% let alone grasp all that) It's really insane how they all are using a single data point. GN used the readout point from near PCIe slot if we see in their video snapshot below..Yea it's all cool lol but why just completely ignore the HWinfo.. These lads failed to be called as Techtubers. Just another YT reviewer crowd capitalizing on the sensationalism. Anandtech captured data on ASRock X670E, and they used HWInfo which is why their data is shown accurate but they also missed the point on Die Sense. However their testing is interesting as they test the total Power draw and Current plus Temperatures as well across all AGESA revisions. Here's more... from the actual real deal in HW Builzdoid (he already stated how 12VPHWR did not have good ceiling vs 8Pin) - https://nitter.snopyta.org/Buildzoid1/status/1658131479275216896#m clearly mentions how the the testing is not accurate. Second if you go all the way below that tweet responses, some fellow Nords of bro Papu checked it themselves too by following the ASUS engineer and found out same result.. They used HWInfo + ASUS engineer mentioned points to read actual Die Sense VSoC. AMD's new AGESA really locks it down is the conclusion. Even with increased LLC. AMD has MCM design I was not aware of anything how VSoC is related to die sense, if it was Intel it would have struck me in a second on die sense, as we do only check VCore there's no MCM design to read another voltage point.

Don't want to critic him too much but here's my thought. All the video entire run of it has basically - Acoustic Scans, X Ray imaging, Cross section Microscope and then another deep Electron Microscope. Ton of Technical terminology thrown at the user, and what we are looking at is a Nanometer scale transistor packaging from world's advanced lithography of TSMC 5N under a Microscope, it will look same for older processors or any node of Semiconductors wafers. Then proceeds with their FA Lab contact to have even more ton of Technical terminology thrown like 20 items or more list of cause of failure. Ultimately there's no root cause conclusion which was actually the one being hinted at and discussed in their prev piece. This just makes Steve look super smart although his script is basically from the Lab team, credit is its looking interesting under microscope well, that's all. I expected the outcome, but I wanted to wait and see. It's simple, the problem's nature is pure Electrical Engineering standpoint and we cannot really see anything as 99.99% of the people do not know this high tech stuff. People do not even know how TXAA (Nvidia technology) blurs their game to garbage due to Resolution of Textures being downgraded as a result of AA blending which is the default on all AAA and we cannot disable it then they upscale that using DLSS lower quality image apply sharpening filter instead of using DLAA which is real Super Sampling unlike DLSS !! let alone the VLSI engineering. Fact is if we see anything under the Electron Microscope be it our skin, any piece of PCB, CPU things always look different and interesting due to simple fact of Naked eye vs Microscopic scale of perception. Just shows how over current, over voltage flow disrupts the electrons thus moving the chemical composition of the PCB and the CPU to changes, entirely expected to see how Copper and others flow inside out other elements... For Example lets look at these beautiful die shots NexGen Nx586 440nm RISC processor (Tried to rival Intel x86 Pentium at that time, tried but could not..) Baikal M1000 28nm (Russian ARM designed processor made on TSMC 28nm) However.... ASUS blaming... I won't say no to that. ASUS is a jackass company, look at their forum they just killed the forums, insanely worst QC dept of BIOS literally putting garbage that does not work (to recall beta BIOS fix for RTX 40 PCIe 4.0 fix on Z590 boards works and stable does not, then it has Anti Roll Back for Stable, nasty new practice by ASUS on ARB) Then the boards that I had to return back.. was a hell. Go to AMD forum entire ASUS forum us run by tons of issues. Same for Intel esp their Z690 series is a massive POS. Shame how their BIOS is top in UI but their BIOS itself is now not good add the other QC problems and their Software.

I think the same, assuming it has gone bad since the black screening and causing some kind of trip off. I ran the machine for 48 hours now without getting any shut off issue. I loaded the CPU to max, by playing Dota2 with iGPU Intel HD 4600 + the usual 3.3GHz clockrate with undervolt that has been stable for years. Surprised the game engine Source is so optimized even after all these years it runs on the Intel HD outdated chip from 2013. So 70C-75 max on CPU package with that kind of load. And did not have shut off, it has some hiccups while loading the game initially before a match, but it's lack of the GPU and CPU esp the MQ chip's VRM limitations, not the powercut issue. And ofc ran the machine again for whole night for p2p download. This confirms that CPU VRMs are not at fault at the very least, and the power delivery for the CPU as a whole from DC Jack line in. As for the 980M GPU... If you recall, I created that 980M Blackscreen thread back at NBR and had over 100 pages of discussion on how poorly these were made, unfortunately I did not get the MOSFETs added to the GPU. As for Shuntmod, nope. The card just had Prema vBIOS but it was hot at higher voltage so I use Svl7+Johnksss version. So maybe the whole Blackscreening got worse and fried something on the VRMs that is doing this kind of issue, fingers crossed all the way hoping its not the MXM slot power delivery. For the Sparks thing, I remember when I plugged a new battery the port was having sparks but the machine and battery worked for long time, hope this AW 17 R1 is as resilient as other Rangers that still live.

Update. Some good news on AW17 R1 Ranger. I removed the MXM altogether and ran the machine for 24Hrs+ and yest night I did not shut the PC and kept it powered on. So no shutdown or powercut or throttling. It went on, downloading p2p for whole night. Still did not shut down and kept it running, now going to see for next 24hours. If that is the case going forward then - I think it's MXM GPU failing in requesting power draw through MXM PCIe slot ? Is that possible, because irrespective of Power draw it fails, like starting a Heaven Bench does not cause it to shut down immediately nor playing Dota2 can cause, but rather can occur anytime on idle or downloading something or watching movie, and even happens if the GPU is not used but just connected to MXM slot itself on SG Mode using IGPU only. So I suspect there's some electronic failure on the MXM VRMs over the time which causes it to fail, any ideas..? I think brother @Tenoroon's generous offer would help me in testing it out for one final test, to use his 860M and check if it's the MXM Slot or 980M MXM GPU. And if it's MXM GPU, then I'll be very happy as no mobo replacement required and can just try to get either Dell Quadro P5000 or HP Quadro P5200 MXM as replacement. On an another note, finally the Cablemod order is complete and got the email for FedEx it is right now in China.

Thanks a lot for suggestions, yea that MSI option seems really damn good compelling option like Precision 7720 but these all laptops (X170SM/KM, MSI, Precisions and older Clevo DTRs) are hard to find in good well maintained condition. I'm also trying to see if I can get this AW17 R1 repaired as like the machine and do not want to simply dump it. P775 is kinda too old so I did not put any research into them, P870 is a good choice but then again, I thought X170 series is somewhat newer with more recent LGA desktop processors and works better for a new purchase prospect. Esp given the age of those P775 and P870 platforms and my old machine too. So it has to be pampered which kinda are unicorns. I saw Framework thought it's too new for a gen 1 launch and as you add modules it gets more expensive and the HW performance is not that much for the Price esp when you see the price to performance of a Desktop. Only reason why I was into Socketed Clevo and others being I feel like the Mobos are built better rather than the welded versions. The Initial idea of BGA Lenovo Legion 7i I dropped it. $2000 of BGAware and for that cash I could just pick up a brand new X670E AMD AM5 board mATX box that would smoke out even the next gen due to AM5 socket being Zen 5 ready forget the BGA junk. While the BGA one might live as long as warranty exist but it's going to be expensive later on and replacement parts will be insanely expensive. I see AW17 parts being very cheap (including mobo), I see the BGA AW parts by side and they cost 5x more lol for which a P5000 MXM could be bought or even a P4000 choice being an option makes it outlast a lot of machines that came after it in welded format. Premamod is a mega bonus perk and an incentive but BGA is really a massive let down you mean there are Laptops with Premamod + BGA ? But the more I think the more old NBR BGA-Hate threads I recall haha,

Greetings folks, I was thinking about this machine as a secondary PC for myself and this being last LGA socket vs the BGA land. (yea MXM is dead and the GPUs are not even close to desktop parts but it's at-least not a thin and light jokebook i guess)., If you don't mind can anyone clearly mention if the shutdown problems that were mentioned are they applicable to the entire X170-KM or only a few machines exhibit that behavior or just 11th gen on this machine ? Prema is not there sadly for this machine I guess, and Russian folks unlocked BIOS option exists (edit - sorry I did to check properly, ViktorV's the contributor for the Unlocked BIOS), does it provide any important features ? Also using 10900K in the same machine, can it get back 3200MHz DDR4 speed vs 2933MHz max for 11900K ? Anything important, I mean is this machine really worth getting or is it plagued by some EC / Firmware / BIOS bugs which ruin the ownership experience. Heatsink fitment is that good or not that good ? Sorry if I'm being straight forward in these, I hope you guys understand as spending money and time is valuable. Please feel free to mention any other important aspects if I missed. Thanks in advance.

I guess the time has come sadly. Yest it got shutdown on iGPU, SG mode. I disabled the Nvidia GPU too from Device Manager. Maybe one last test is to remove MXM and see.... @Vasudev, I did not update the machine to any new Microcode, the BIOS is also old A14+ unfortunate that your Echo is failing too. Yea I agree there's not much on the BGA land now, however I checked most of the BGA laptops and some videos yesterday. Perhaps there's some VRM component failure or PCH related issue I don't know if CPU failure can cause such things too.., cannot be sure at all hard to find out. Now I'm just thinking about 3 options of new machine - Clevo X170 KM with a lot of baggage attached.., 1/2 BGA Precision 7720 Refurbished for cheap. A BGA Lenovo Legion 7i 2022. Unfortunate.

That is a good idea, it never struck me. Thank you !! I have the 120Hz panel I was saving it for Pascal 1070 MSI or P5200. And the eDP cable too. But as of now its 60Hz LVDS panel is only installed. So I did what you suggested did the MUX key combo fn+F5 and it asked me to reboot for switch and then the Dell ePSA Diagnostics worked (before when it was running on PEG only mode the ePSA won't boot, perhaps due to Custom vBIOS on 980M). As @Mr. Fox asked about ePSA, yea it passed. The error was Battery not installed and so it says "One or more errors found" and displayed the failure, last pass was in 2016. But as I checked the results tab everything passed except the Battery so ePSA needs to pass every single test then only it may pass the machine as 100% pass. And this is not a thorough test but rather normal one I think in the deep test it runs deeper Memory tests. So after booting into Windows 8.1 Enterprise 64Bit, when I right click as expected NVCP does not exist. But I tried to install the Intel HD 4600 Graphics drivers they were failing to start installation, I do not remember how to do this, I booted to the BIOS and put the Display Graphics to IGFX from SG... But in Windows it did not change anything. Tried older ones from Dell and they worked BUT it black screened and the OS display got corrupted lucky we have Unlocked BIOS so I flipped Video Adapter back to PEG mode and restored the OS which removed the bad Intel HD drivers. Later ran DDU and removed the Nvidia drivers and now the Intel HD driver successfully installed. 980M does not show up as it says Microsoft Display Adapter and it shows. That's expected I believe. Basically had to run DDU before, anyways no more 980M for now. I will run the machine on iGPU exclusively and see how it does. Thanks.