All Activity

If this is accurate it is wonderful news.

This reminds me very much about the fake glued on heatsink grills Dells design engineers added for the sake of "better cooling" for the older slimy modern Alienware Jokebooks. It was there only for the design😄

Hello everyone, first of all I am very happy to have found this forum. I have been a hard-core & die-hard user of the AW18 for 12 years now and simply refuse to buy another laptop for daily use. One of my 4 models has developed the following problem: The DIMM RAM access controller 1 seems to be faulty. I have switched all kind of RAM to all positions and found that whenever any RAM is placed in controller 1, the laptop does not POST and I get four beeps. I have tried all combinations, so the access controller seems to be faulty. There is no visible damage and I have tried cleaning it, but that didn't change anything. As the maximum RAM for my processor running Windows 7 Ultimate seems to be capped at 32 GB, I wanted to ask if it is possible, instead of changing or repairing the motherboard to use the following combination on the 4 controllers: controller 0: 16 GB controller 1: leave empty controller 2: 8 GB controller 3: 8 GB Any suggestions would be greatly appreciated. Regards, Omadon

Nice results! That board and CPU look like a great value. I'm extremely confused by Intel's decision to delay a Refreshed SKU until that time frame given Nova Lake is supposed to be out later this year. Reminds me of 11th gen and 12th gen. Either they had to heavily bin to get chips able to run the new clocks or they actually fixed some things under the hood. I'm leaning towards heavy binning, but hoping for actual latency fixes that can make the refresh something that will be more impressive over an overclocked ARL chip. I also ordered a 12 core BTL-S chip lol. Mouser had them as out of stock backorder, and I decided I wanted something else to mess around with. The order went through and according to them, they have ordered one from the factory lol. I'll be shocked if it ever arrives, more shocked if it works. 🤞🤞

Thinking out loud, the AI buildup toward 2025 feels like it triggered a delayed realization at the top of the major CPU vendors like Intel, AMD, and ARM. For decades, Intel, AMD, ARM, and others built their market share through platform control. OEM relationships, ecosystem lock-in, and incremental architectural leverage made the CPU the anchor of the system. Then, almost overnight, they found themselves paying billions to accelerate someone else’s technology, reshaping entire systems around it, and effectively handing over slices of the very market they spent decades carving out. That is a reversal of logic. Instead of defending the platform, they subsidized the erosion of their own leverage. Value migrated away from the CPU and toward an external accelerator that began defining how systems are built, priced, and scaled. At some point, executives notice when they are funding their own displacement. What seems to be happening now is a quiet correction. Rather than continuing to pour capital into accelerating a smaller player’s dominance, money appears to be flowing back inward. Into their own roadmaps. Their own architectures. Their own standards and system-level control. This is not something you see announced in press releases. You see it in behavior, hesitation, and shifting priorities. You also see it in the market. Intel and AMD are no longer acting like companies resigned to becoming feeder components. At the same time, NVIDIA’s stock is not following the straight-line hype trajectory you would expect if total platform capture were inevitable. That divergence suggests reassessment behind the scenes, not universal buy-in. If the AI story were as simple as the headlines claim, capital flow would be one-directional. It is not. This does not mean NVIDIA goes away. It means the assumption that the entire industry will continue financing its expansion indefinitely is likely wrong. Platform owners eventually defend platforms. The AI boom did not just create demand. It exposed a power imbalance that incumbents tolerated for too long. What we are likely seeing now is the early phase of that imbalance being corrected quietly, deliberately, and before it becomes irreversible. The solution is simple, but overlooked: eliminate the silo. Unified, coherent memory allows CPUs and accelerators to operate on the same data without costly copies. Once that bottleneck is gone, the perceived GPU advantage evaporates. CPUs retain their performance, and accelerators return to their intended role: augmenting the platform, not defining it.

In short, A little bit of the bottom of the palm rest (the bottom part of it which faces up when you stare at the bare GPU, would have to get trimmed to make room for the mosfets/PD of the non-maxQ version. However the non-maxq version is the same length as the p5000 that was supposed to fit the 7540/50, as far as I can tell it's just longer between the two of them. the Width is the same and only one screw mount (related to screwing it to the palmrest) is different, other than that, the palmrest + heatsink modification (flatten it out since the "dip" for the mosfets will be in the wrong place, and then either adding a shim/small bit of aluminum to connect the mosfets to the 7550 heatsink), is all that is needed, and since they both share the same BIOS there shouldn't be a whitelist conflict.

or simply a properly debloated, tweaked and tuned system.

I think you mean 38000 and 28200. Your unit should easily be capable of a 35000 score and for example Notebookcheck even got A 38500 score in their review, possibly a better binned chip: https://www.notebookcheck.com/Dell-Pro-Max-18-Plus-Test-18-Zoll-Workstation-mit-USB-C.1167532.0.html

I will check tomorrow morning.

If you can "make it fit" (and I would genuinely love to see that), then it will probably work. I have never heard of this being done, but... The physical interface is the same. I don't think there will be an issue with needing to modify the NVIDIA driver INF.

Chris Titus: 🆕 The Linux Book The book is now feature complete, but I'm waiting for publication. I want to create and update a few video tutorials for certain sections. This project will continue to receive updates well into the future as the Linux ecosystem evolves. Visit thelinuxbook.com

Not if you're afraid to make a little modification to the palmrest and the existing heatsink :3 I will make this fit and prove it :3

Asus have now increased the prices on their US webstore. Vanilla black Astral is now $3510 USD. Nice, only $50 higher than previous premium above MSRP. And the Astral LC is now $3870. Or +150 USD. Anyone have seen Nvidia have increased MSRP for their FE models? https://www.asus.com/us/motherboards-components/graphics-cards/all-series/filter?Category=NVIDIA&Spec=354488 Yup, custom made firmware only for the very few. Theres no magic in matrix bios, they just needed to up the xbar freq bit to match good effi. Also bios can have same name, for example ending .50 but still have multiple revisions The white Astral 5080 is now $1950 for those that prefer a downgrade. Or for new buyers. Yup, 5080 close to old 5090 MSRP. ASUS TUF Gaming GeForce RTX™ 5070 Ti 16GB is $1170 USD. For xx70 tier card. No one is able to stop this man. Greed and incompetence is the main driver that let Jensen get what he want. Money over the lands own needs and security. This is just sad. Or better say a joke that should never have happened. Nvidia CEO Jensen Huang to visit China as company prepares to start H200 shipments to the country And no warnings will help... Nothing! What next... Sell AI chips also to North Korea? The outcome is still the same. More cash in and more AI chips out. Anthropic's Dario Amodei says allowing Nvidia H200 sales to China is like "selling nuclear weapons to North Korea" "I think this is crazy"

How will it be when Microsoft increase AI usage from 30% and up to 50% of the coding? Patch Tuesday nightmare continues: Certain Windows apps will freeze during use Microsoft's latest Patch Tuesday update (KB5074109) continues to wreak havoc, by now breaking applications like Outlook that interact with cloud storage services such as OneDrive.

Forgot to mention it’s Cinebench R23.

Which CineBench exactly?

This evenings efforts, iout Gain was stuck at 300 so effectively I was able to push the card again. Around 650-700w, I wasnt able to change the Iout gain from 300 so ran as is. Next time I'll try 250-275. Core 3475 Mem 2764 Voltage 1.158 Card is stuck at 683 again after one of the failed attempts so I guess thats all for tonight. Wasnt able to breach 3500 Core. :(

Have any of you with the Ultra9 285HX chip done any Cinebench tests? Results from online reviews, like Notebookcheck, get MultiThread scores around 3800 although I'm only getting 2820. I'd like to see how my results compare to other users' results.

💯 ‼️ Very well said and highly accurate.

Yup. Your AI-powered girlfriend will arrive in 5 years, according to Microsoft's AI CEO We knew you'd been wondering when this day would come, and according to Microsoft's AI CEO, Mustafa Suleyman, your AI-powered girlfriend or boyfriend will arrive in five years. Microsoft is already starting to advertise that AI is more than just a chatbot...

i3 8100 will work, buy cheapest second hand one

Does anyone know if a i3-8100 would work as a test cpu? They're super cheap and would help me diagnose it without taking it to the shop, which would cost more than the 2nd hand i3-8100

someone knows?: the cpu fan connector, its a "JST 1.25 - 4-Pin"? and any possible way to set the TDP limit higher for the CPU? the max is 61W, then the mainboard cut it, thats a bit annoying, even with vcore offset from -0.050V n cache voltage offset from -0.015V running at 4100MHz allcore/cache into the powerlimit. maybe shunt mod possible on the mainboard?

just will order another CPU fan, fits perfect in the secondary HDD/SSD bay n can blow fresh air straight to the GPU heatsink, need to make a fan cable splitter bcs will put it straight to the CPU fan connector(should handle 0.6A in total). so few holes in the bottom cover, connect the cable, put the fan in n fresh air to the mem + GPU heatsink(will solder 2mm copper heatsinks on it for maximum performance) g).

I think the issue is deeper than AI specifically. When you enter a building, do you take the stairs or the elevator by default? When you travel somewhere unfamiliar, do you use a paper map or Google Maps? When you need to remember a phone number or an address, do you commit it to memory or store it on your phone? No matter how we frame it, our cognitive and physical habits have been eroding for decades. It happens gradually and mostly without us noticing. I grew up in a time when elevators were rare. You took the stairs because there was no alternative. When you needed a phone number or an address, you used the yellow pages or asked around. Your brain was the address book. I carried hundreds of numbers and locations in my head and could recall them instantly. Navigation worked the same way. You learned your surroundings by observing them. You placed mental markers along the way. Physical maps mattered. An atlas was a normal household item. Many kids today have never even seen one. AI is not the beginning of this shift. It is just the latest step in a long process where convenience replaces effort and external tools quietly take over functions we used to exercise ourselves. The problem is not using technology. The problem is forgetting what we give up when we stop exercising those abilities altogether.