SkyHound0202

In CMYK printing process, the lighter the color, the less ink/toner the printer will use. By switching to a light color, it will save you a few microliter/microgram per page, which translate to about 0.0005 to 0.0025 EUR per page (depend on print technology and coverage), which is neglectable in small print volume, but does add up in the long run (saves up to 25 EUR with 10k prints). Compare to switching to lighter colors to save ink I would argue that using third party supply is more economic since it will save around 0.001 to 0.005 EUR per page (up to 50 EUR per 10k).

I won't necessarily call it a bad thing if a "starting at" price is low enough that makes a previously unobtainable tech (in terms of price and/or performance) available and accessible to more people, while also informing the customers what caveat ("catch") it may have. For some (if not most) people, their biggest concern is not performance, but price. To them, solving the problem of owning a device take precedence over the actual performance of the device.

GPU die is the part that is least likely to fail on a graphic card. Fan fails over time, power delivery degrades after extensive use, solder joints break from stress, connectors break due to user error. So if done correctly, finding used GPU dies elsewhere and remanufacturing them with new components on a new board is a good way to go. ("Kinology", etc.) Heck, even professionally refurbishing a used a card by giving it new paste and new fans plus a thorough check is a valid option. (rest of the tested) Not only does it reduce e-waste, but it also bring the price down in general. But the problem is that Nvidia and their partners doesn't want these, since someone buying one of these remanufactured, unlicensed cards means one less die Nvidia will sell to their partner and one less card their partners will sell to you. So they have been clamping on it pretty hard. AMD on the other hand doesn't have such limitation. A remanufactured RX 6600M card (mobile Navi 23 die on PCIe card, mostly used for mining) functions as expected on desktop, while Nvidia mobile GPU require either extensive hacks or will be eventually blocked from newer driver updates.

At some point they should just adopt the 8-pin EPS12V (i.e., the CPU power connector rated for 336 watts maximum) for consumer graphic cards as well. It's already being used in server graphic accelerators (Tesla) and professional graphic cards (Quadro) anyway. 8-pin PCIe power connector (150 watts max) is nearly two decades old and won't cut it in the current hardware landscape.

What's next? The eBay exclusive Ryzen 9 5950X3D? We knew there's at least one somewhere in an AMD lab...

Of course you don't need a new PC, if it's a) recent enough, and b) upgradable. A Kaby Lake i5 is not even old, despite being listed as "Discontiuned" on Intel ARK. True while it won't received any more driver update for its iGPU, but the CPU itself still sports all the modern features (AVX2, PCIe 3.0, NVMe, Optane, etc.) and four capable cores at a reasonable clock speed. And thank goodness that Acer uses standard form factors for their system. Had it been a custom Dell or those 12VO board, a power supply upgrade would be impossible. (I will skip the RAM, storage and GPU part since they are generally swappable and readily available, unless soldered. Also skipping OS as it's software and up to user preference, but modded Windows or Linux should be fine)

My hot take under the 9900KS video in 2019: Intel's foundries are their liability. Moving to a smaller-but-immature 10 nm node now means losing out the current older-but-mature 14 nm node capacity, which Intel clearly would not until they have sorted out their 10 nm problems. It's not about microarchitecture, but how their foundries (by extension, their manufacturing business model) works. And a bunch of people immediately jumped in without even knowing what I was talking about. Someone brought up the broken Tick-Tock model and be like "130 nm FTW". Then there's people keep mixing architecture advantage and node advantage and came up with "1x nm Zen1/+ competes with 22 nm Haswell or 14 nm Broadwell". Anyway, I was mocked and humiliated that day for merely understanding Intel's broken business model of holding onto their fabs and let node supersedes architecture. Years later, lo and behold, Intel's 10th and 11th gen is still utilizing 14 nm. I was right.

The SSD type doesn't matter since they will all bottlenecked by the SATA2 interface found on the PS4 Slim anyway, but ANY SSD is an upgrade when compared to the mechanical hard drive that comes with the PS4 Slim. You will see some improvements in terms of boot & load time and texture streaming in game. There's no compatibility issue. Reason that few people uses 870 series for PS4 upgrade is that it's still relatively new (released two years ago, general availability at MSRP one year ago) and the fact that older 860 series (also other brands) can be have for cheap. 870 QVO is a decent choice since it uses QLC (4-bit) NAND to offer large capacity on the cheap. While it has less write endurance compares to Samsung EVO and PRO line (QVO 1 TB only has a rated 360 TBW, 60% of EVO of the same size), the gaming environment is less write intensive (unless you ) so it shouldn't be a problem. Do keep you saves backed up just in case.

Short answer: Unlimited tests in 3DMark and Offscreen testes in GFXBench. Long answer:

Guess I missed the point. Thought you were doing transcoding primarily & constantly. If you just want the CPU to be efficient, you can try dropping the Package Power Tracking ("PPT", W), Thermal Design Current ("TDC", A) and Electrical Design Current ("EDC", A). Default PPT/TDC/EDC for the 5900X should be 142/95/140. You should be able to safely drop it to 128/80/125 to run the chip at 95W TDP, then further to 88/60/90 and effectively turn your Ryzen 9 5900X (105W TDP) into an OEM-only Ryzen 9 5900 (65W TDP). This would result in slight performance loss but huge efficiency gain. See this for more info. (if you want even lower, try 45W TDP at 61/45/65).

Maybe get one of those Tesla P4 for less than 100 USD and pass it through unraid to take advantage of NVENC?

You need to go to Control Panel\Network and Internet\Network Connections and find your adapter (in your case it should say something like Realtek 2.5 GbE family controller) then right click for Status and check Connection-Speed. This denotes your link speed, i.e., the speed of the physical link over the cable. For gigabit ethernet it should say 1 Gbps (note: despite the controller on motherboard being a 2.5 GbE one, the fritzbox only have a gigabit port.) Any test yield online through website like speedtest net only indicate the speed of your internet subscription package, and it not indicative of the speed achieve in a local network (LAN). Also, I am afraid there's no real need to get an expensive cat 8 cable when a cheap cat 6 cable (even a cheaper cat 5e cable when shorter than 55 meter) can do 2.5 GbE in a home environment.

You can manually enable "write caching" to use some system memory (RAM) as a write to improve performance. It should be enabled by default. If not you can go to Device Manager, select the drive from Disk Drive section, right click and select Properties then Policy tab. Tick the check box "Enable write caching on the disk". Keep in mind this method may only improve the performance slightly as DRAM-less SSD are inherently a bit slower than their DRAM-equipped counterpart, but in day to day operation there should be any noticeable differences except extreme edge cases, and any SSD is far superior than a traditional mechanical drive.

I would love if they release an "extreme" edition with a better and more efficient processor (Snapdragon 782G, 860 or 870 comes to mind) that can not only function as a streaming device but also a capable android gaming handheld in its own right.

They could've just use the T-series 35W processors, or even the 65W standard ones. Though they will still draw much greater power (generating more heat) at peak turbo.