-iSynthesis

Why do you want to replace the caps? What are you trying to achieve? The reliability of the components used is more than adequate for normal usage, let alone as a pure backup, so is the efficiency. I would recommend against trying to mod a PSU to achieve better efficiency. It's not as simple as replacing a single component to achieve a measurable effect on efficiency and larger modifications won't be worth the time and money spent (and may well be dangerous unless you are very certain of what you are doing). Let's assume you manage to achieve the same idle power draw as the corsair (which you won't) you're saving 5W. Let's assume you manage to achieve the same load efficiency (which you definitely won't) you're saving max 30W at 1000W load. Both completely negligible in relation to the amount of research, cost of components and work you would put in. And for standby power draw, just turn it off completely.

The card severely reduces Clocks to save power when there's little load. My 6600XT right now is at 5MHz just browsing. The power consumption on the 6600s is quite low, it's showing around 80W in AMDs stress test for me. Also, temps will be very low due to the low power consumption, so I'd say what you are seeing is expected As the others suggested, check the games without VSync/Freesync to see if it clocks higher then. Good Luck

I'd be more concerned about the brand you bought rather than the rating they put on it...

Not that I know of... The USB ports on a motherboard aren't meant for charging, that's what chargers are for. They're used for data transfer to and from your PC. The only fast charging ports I've seen on PCs were front panel connectors that were directly connected via SATA Power connectors. USB 3.2 has to support 5V charging with 900mA or 1.5A, which is nothing for modern phones.

So I'm usually a quiet reader of this thread. I don't have anything to do directly with the topic but support the movement and anyone who opens up and expresses themselves no matter what. Anyways, I love the stories you post here and make sure to keep us updated. It's kind of beautiful reading these and definitely a nice breath of fresh air between the (certainly necessary) discussions that are otherwise going on. Much love to you all and I'll go into silent mode again

The OS is called Raspbian (well, apparently they renamed it to Raspberry Pi OS): https://www.raspberrypi.org/software/

Did you just drop the windows media installer file onto the USB stick or did you start the program windows media installer and select the usb drive as the target drive?

This would be pretty sick and would keep up the name of FineWine I guess^^ Looking forward to be using my Vega for the next 5 years to come lol

To answer this question, unless you know your way around circuits you'll probably have trouble fixing anything there, could be anything really... From your description it sounds like a short somewhere causing the PSU to trigger over-current protection and instantly shutting down. If it were a blown fuse or something it shouldn't trip any protection circuit in the PSU causing it to shut down, so it's likely something a bit more complex... I agree with you in that it sounds like a dead GPU, though I guess it COULD be a motherboard issue with the PCIe slot. If I understand correctly you've tried the GPU in an entirely different system though, right?

What are the rest of your specs? Especially PSU etc. It would seem odd to me that a GPU dies on a power outage somehow, though depends on what it was caused by I guess. Does it boot up for longer than .02 seconds if you run the PC without the graphics card? If you have a Post Code readout or beeper on your motherboard, does it say anything or does it turn off too quickly?

If temps don't get a lot better with a repaste I'd advise to undervolt, if anything... Might be able to get a few more MHz that way too, and keep temps down a bit.

I have the Ikea Bekant desk, I have my PC (though it is more like a mid - mATX tower) on my desk. I don't think weight will be an issue with most desks as the weight of the PC will usually be on the legs of the desk. For size, I have a somewhat small case on a rather large desk. It does kind of take place, especially with 2 monitors but I like the looks of it mostly^^

Combined High- and Lowside FETs are pretty much standard in every high-end board by now. They often offer more monitoring features (temperature, current and voltage monitoring etc.). A well designed separate FET design can be excellent (and in theory has larger surface area) but by now it's mostly used in cheaper boards as Power Stages tend to be more expensive. To OPs question, it would be helpful to know which motherboard it is about. Generally, the first number is the number of Phases the CPU gets and the second number is for the memory. Motherboard makers often bullshit about phase count. It's almost impossible to tell a good from a bad VRM simply by looking at the numbers in front.

"safest": A cheaper CPU, GPU and RAM will not be less safe. They'll just be slower. So if that's the standard, I guess those? The motherboard is another good one. People are overspending on 14 Phase VRMs while never even overclocking. I don't want to dictate what people spend their money on - but many motherboards are a waste and for the majority of people a cheap one will be more than enough. This is where you can save a ton of money without losing any performance really. Also CPU coolers I guess. The stock cooler on AMD will do fine if it comes with one, everything else is acoustics or temps for overclocking. Just don't skimp out on the PSU.

To add a bit more, storage is almost always measured to the base of 2. One byte is 2^3 bit, 2 byte, 16 bit, is 2^4 and so forth. Kibibyte, Mebibyte etc. are also to the base of 2. One Kibibyte is 2^10, Mebibyte is 2^20. With 64 bit systems, the word length of a storage element is 64 bit or 2^6 bit. This means, that the logic unit of a processor can process a 64 bit long piece of information in one step. To answer your question, we use multiples of 8 because 8 bits, or a byte, is the amount of information needed to encode a single ASCII letter. Our system is based on that pretty much.

You are taking a Grand Tour video and applying it to usps. It just doesn't work. A usps route is pretty much the perfect scenario for an electric. Literally. Lots of stop and go means you can get a good amount of energy back. Short routes that should be well within current electric capabilities. This combined means that a single charge a day would probably get you through most routes. The charging infrastructure at the depot wouldn't have to be insanely fast charging, it would just have to charge them over night. This is neither expensive in comparison to other spendings usps has nor does it take a long time to do.

It depends on the power supply, but most ones reach peak efficiency somewhere around half of it's max power draw. However, from that point up until full power draw the decrease in efficiency is pretty small, usually only a couple %. Power supplies operate at least efficiency at low loads. This added together means that your peak input power will be at full load and with a slightly decreased efficiency from what it's max rating is, for an 80+ gold PSU you'll be above 80% close to the 90% mark. Refer to the chart in this wikipedia article to estimate efficiency at different levels of power draw of 80+ PSUs: https://en.wikipedia.org/wiki/80_Plus Taking the estimate of 80%, which is far below spec but calculating it safely, you're looking at a max power draw of around 815 watts.

So the temps have always been trash, right? Also with the old cooler? It's an odd issue with everything you've tested, the only thing that would come to my mind is that they didn't apply the Thermal Interface Material properly when they made the CPU (not sure if those Xeons were soldered). Other than that, maybe try a full reset of the bios and check stock temps...

This strongly depends on the battery inside. Some specially designed Lithium-Ion batteries may operate a bit below -50°C even. But for normal ones you shouldn't go below 0° when charging (storing it colder is another topic). This may / will decrease it's capacity irreversibly. They will also charge substantially slower as the internal resistance rises with lower temps. I myself have used power banks at below that, they were fine afterwards but I'd recommend putting the power bank into a warm bag or something, tightly wrap it maybe (It should produce a bit of heat itself so if it's 0°C outside it should be a bit more if it's wrapped into something)

Probably a whole lot of input lag if I had to guess^^ In all honesty though, most servers you mentioned are not built for graphics rendering. They'd probably be pretty crap at it and trying to get thousands of gpus to work together would be a nightmare as 4 way SLI/Crossfire never really worked well in the first hand. And more would be a nightmare or impossible to work to create single frames together...

That's the debug tool, not sure what it shows but it's not artifacting. Frametimes maybe, no clue... Which modpack are you playing? What is your CPU / GPU usage like? (Detailed on the CPU part, per-core usage etc.) I'd recommend tweaking around the optifine settings some, some settings can have huge impact on fps. Minecraft with that many mods I'd recommend at least allocating 8GB of ram and make sure to enable multi-core rendering. There's a setting under Performance I believe about rendering chunks more frequently, I'd suggest turning that off. I'll check tomorrow but 33 chunk updates seems high if you're just standing still.

Motherboard GPUs are a relict of the past, pretty much the only application they're still used for is server motherboards for basic Displays. They have many disadvantages; They can't be powerful as that would require active cooling. They would increase the cost of the motherboard as it would be another processor that would have to be soldered onto the motherboard. Putting it onto the CPU is an easy way out; The GPU is just another addition to the existing silicon. It's connected to the memory anyways and it's actively cooled. It just needs a few extra phases power delivery on the board. They're a viable option nowadays for servers as basic display outputs. A very weak GPU is enough to do so. Also, it wouldn't be viable to try to integrate an iGPU onto a massive die such as an Epyc or Xeon with as many pins as they have already anyways. Also many servers don't require any display output at all so it would be a waste to implement one.

Yes, that's the exact instructions I followed. I did not let it get hot before, I found it not really being necessary but if you're willing to go the extra step it surely can't be bad... Just don't burn yourself^^ If you need any help understanding steps taken I can try and translate the important stuff from the article (or you're german yourself?) Good luck!

MX4 is just fine, that's what I used too. The STRIX Vega has one row of VRMs, I ended up getting a thermal pad replacement for it (3mm Thermal Grizzly Minus 8). It was not needed but got the VRM temps down quite a bit. It kind of depends what your temps are there, ASUS really f-ed up their design so they ran hot as hell but you might not have to...

To find out if heat is the issue try turning up your fans to 100%, this should get your hotspot temp lower. If it still crashes it's something else. My vega is running on 1x8pin (PSU side) to 2x8pin (GPU side) without a problem since I repasted. Even overclocked shooting over 330W. Repaste, use a lot of Thermal Paste on Vegas. Start off by tightening the two screws further away from the PCIe slot FIRST, only when they are bottomed out screw in the other two till the end. This should drop your hotspot temp, which may be the reason for the crashes you're experiencing (This happened to me). To find out if it's the issue or the PSU try turning up your fans to 100%, this should get your hotspot temp lower. I agree with the others, if you're going for a 3080 you'll want to get a new PSU.