Cracking code and curiosities, I turn caffeine-fueled late nights into bytes of wisdom. Welcome aboard! I’m Alex, with 20+ years of mastering the art of coding and now, the craft of content creation. Subscribe to my channel and join me in unraveling software enigmas with in-depth tech reviews tailored for developers, with a dash of humor. Emojis suck 🚀
i'm reaaaaaaaaally sceptical about these benchmarks, too good to be true, specially since windows has been struggling a lot to get windows 11 running decently on ANY computer
These Signal 65 guys are not to be trusted. As others have noted Ryan Shrout was the former head of client CPU marketing for Intel for years and was responsible for HIGHLY manipulative marketing campaigns that bordered on fabrication and outright lies. His reputation is not good, and it’s best to assume that anything Signal 67 is putting out is at best highly cherry picked. And right off the bat it’s VERY easy to see that they’ve done exactly that with the choice of comparison systems and the choice of benchmarks that they’ve run. Lots of battery tests going against shitty Intel CPU’s that aren’t competitive. What’s missing? Oh yeah, anything from AMD made on a modern processor node. Also, in terms of tests that stress the GPU, again product selection matters. The competition is Apple (their GPU is capable but nothing to brag about) and more Intel - specifically Arc (which most OEM’s won’t even touch because of how bad it is!) and Xe integrated graphics. Not a single benchmark featuring nVidia OR AMD graphics. I have a feeling these test results are going to be like the fabled Principaled Technologies fiasco from about 5 years ago. Technically correct, but the conclusions will be the exact opposite from ‘real world’ test comparisons that ordinary people will experience. Ryan Shrout has been a paid shill his entire career. Who’s paying for these early benchmarks, and how has he manipulated the tests to give his paymaster the results they want? Until that’s answered its best to toss that entire report in the trash where it belongs!
I needed to get into Apple silicon so I could run LLMs locally, but after rumors of M4 Max by early next year didn’t want to buy an M3 now, So I found a killer deal on a fully loaded M1 Max: 64GB/8TB (2x the SSD I need, but it was too good to pass up) for $2,300. It’s a massive upgrade from my late 2020 i7 MBP, which I can probably get ~$1k for as it has a 4TB SSD, and then I’m sure I’ll get more than $1K for this M1 Max when I want to step up to a M4 machine in 10-12 months or so. I love the used Mac market! 👍😁
Qualcomm have a team of 8 engineers working on Linux support for the SoC - including Vulcan drivers for the GPU. Most of their drivers have already been merged into the mainline Linux kernel, but things like the GPU have not yet been merged. They can't guarantee that OEMs will support Linux, but the SoC itself will. The Laptops boot off standard UEFI, so no need for any fancy image flashing like the Pi - should just be a case of throwing in a Fedora usb and installing Linux like you would on any normal x86_64 machine. Hopefully the OLED screen on the Surface laptop has a high refresh rate - the MBP has a high refresh rate but very high latency (noticeable while gaming, for what little games are supported on macOS+Crossover). Cannot wait to see how games perform with FEX and Proton! Gnome has fancy trackpad gestures like MacOS, really want to see if the ergonomics of the Surface laptop trackpad with its haptic trackpad rivals the MBP. I'll replace my MBP with a Surface for work and play if Linux support is great and the laptop is within 90% the feel of the MBP.
Given these scores, it'll be interesting to see what market the reviewers think the OEMs are targeting with these new devices. Are the OEMs going to address the good, not great, single core performance? Or are they just going to pretend Apple doesn't exist? (The same way Apple likes to pretend that other companies don't exist.)
All of these 15” laptops with fans can barely just, not beat the MacBook Air with no fan, from last year, M4 is already out at this point, which is significantly faster than M3😂
Signal 65 looks like a return of PCperspective which lost credibility when we discovered Ryan was running a side business getting paid by Intel. He then left pcper and landed a job at intels marketing division where he again just produced more slanted graphs. I'm not seriously looking at any graph or testing he has been involved in producing as he has proven he can be bought.
I returned the M3 Pro laptop after a few days using it because the screen had a defect from the factory. Got my money back and it’s been a few months now.
The “benchmarks” seem all over the place, more than usual for the tech industry. And until normal people can test these out, I don’t buy the hype. That said, the price of these laptops looks competitive and making 16 GB RAM industry standard is a welcome move, although a bit late.
Maybe in 10 years they'll all be adopting RiskV instead of ARM. And adding other *PU types to cope with other specific uses that are now still managed by NPU, GPU, CPU
Performance ARM processors may not be as powerful as other processors and may have lower performance at the same clock speed. This limits their ability to handle more demanding applications, such as high-end gaming and intensive computational tasks. Multitasking ARM processors may not be as efficient at multitasking as other processors.
Limited speeds: Some ARM processors have limited speeds Scheduling instructions: Scheduling instructions can be difficult with ARM processors Software disadvantage: Windows on ARM has a software disadvantage compared to traditional Windows because most applications and games are not Arm-native Strict load/store architecture: ARM is a strict load/store architecture where every instruction is fixed-width and either accesses memory or performs a computation on registers Performance: ARM-based chips from Qualcomm fall short in terms of performance
More generally, x86 processors have a higher raw performance than ARM processors. This means you can "plug and play" your software into an x86 CPU and expect it to perform well, regardless of how much power the device uses.
Anyone who thinks ARM does not have trade-offs is mistaken, esp. since Intel is rewriting the book on x86, totally reengineering it, and will move from silicon to glass first.
My understanding is that Intel Core Ultra 7 is manufactured with Intel 4 node. M3 is TSMC N3B. It is claimed that both are comparable in terms of die size, power, and performance. X Elite is manufactured with TSMC 4 nm, which should be cheaper to make. Hopefully we customers can see the saving soon.