As an Amazon Associate I earn money from qualifying purchases.

Wednesday, August 3, 2022

GPU Performance Hierarchy and the Best Graphics Cards, 2022

With graphics cards finally back in stock, I've put together this comprehensive GPU performance hierarchy of the past two generations of graphics cards. These are the best graphics cards around, but it's also important to know what you're getting. I've benchmarked all of these cards, at multiple resolutions and settings. Here's how they stack up across a test suite of 15 games.

I've also listed current GPU prices, which have improved substantially in the past month, helped no doubt by Amazon Prime Day. Most AMD graphics cards are now selling at or below their official MSRPs, and Nvidia's fastest GPUs (RTX 3080 Ti, RTX 3090, and RTX 3090 Ti) are also well below their MSRPs.

GPUPerformancePrice (MSRP)Specifications
GeForce RTX 3090 Ti161.6$1,499 ($1,999)GA102, 10752 shaders, 1860 MHz, 40.0 TFLOPS, 24GB GDDR6X@21Gbps, 1008GB/s, 450W
GeForce RTX 3090153.0$1,299 ($1,499)GA102, 10496 shaders, 1695 MHz, 35.6 TFLOPS, 24GB GDDR6X@19.5Gbps, 936GB/s, 350W
GeForce RTX 3080 Ti149.0$1,049 ($1,199)GA102, 10240 shaders, 1665 MHz, 34.1 TFLOPS, 12GB GDDR6X@19Gbps, 912GB/s, 350W
GeForce RTX 3080 12GB145.6$799 ($799?)GA102, 8960 shaders, 1845 MHz, 33.1 TFLOPS, 12GB GDDR6X@19Gbps, 912GB/s, 350W
Radeon RX 6950 XT144.7$1,099 ($1,099)Navi 21, 5120 shaders, 2310 MHz, 23.7 TFLOPS, 16GB GDDR6@18Gbps, 576GB/s, 335W
GeForce RTX 3080140.3$779 ($699)GA102, 8704 shaders, 1710 MHz, 29.8 TFLOPS, 10GB GDDR6X@19Gbps, 760GB/s, 320W
Radeon RX 6900 XT134.8$859 ($999)Navi 21, 5120 shaders, 2250 MHz, 23.0 TFLOPS, 16GB GDDR6@16Gbps, 512GB/s, 300W
Radeon RX 6800 XT128.0$701 ($649)Navi 21, 4608 shaders, 2250 MHz, 20.7 TFLOPS, 16GB GDDR6@16Gbps, 512GB/s, 300W
GeForce RTX 3070 Ti121.4$690 ($599)GA104, 6144 shaders, 1770 MHz, 21.7 TFLOPS, 8GB GDDR6X@19Gbps, 608GB/s, 290W
Radeon RX 6800115.1$599 ($579)Navi 21, 3840 shaders, 2105 MHz, 16.2 TFLOPS, 16GB GDDR6@16Gbps, 512GB/s, 250W
GeForce RTX 3070115.0$575 ($499)GA104, 5888 shaders, 1725 MHz, 20.3 TFLOPS, 8GB GDDR6@14Gbps, 448GB/s, 220W
GeForce RTX 2080 Ti112.9$899 ($999)TU102, 4352 shaders, 1545 MHz, 13.4 TFLOPS, 11GB GDDR6@14Gbps, 616GB/s, 250W
GeForce RTX 3060 Ti104.9$518 ($399)GA104, 4864 shaders, 1665 MHz, 16.2 TFLOPS, 8GB GDDR6@14Gbps, 448GB/s, 200W
Radeon RX 6750 XT102.7$499 ($549)Navi 22, 2560 shaders, 2600 MHz, 13.3 TFLOPS, 12GB GDDR6@18Gbps, 432GB/s, 250W
Radeon RX 6700 XT97.0$449 ($479)Navi 22, 2560 shaders, 2581 MHz, 13.2 TFLOPS, 12GB GDDR6@16Gbps, 384GB/s, 230W
GeForce RTX 2080 Super96.6$799 ($699)TU104, 3072 shaders, 1815 MHz, 11.2 TFLOPS, 8GB GDDR6@15.5Gbps, 496GB/s, 250W
GeForce RTX 208092.7$799 ($649)TU104, 2944 shaders, 1710 MHz, 10.1 TFLOPS, 8GB GDDR6@14Gbps, 448GB/s, 215W
GeForce RTX 2070 Super85.6$579 ($499)TU104, 2560 shaders, 1770 MHz, 9.1 TFLOPS, 8GB GDDR6@14Gbps, 448GB/s, 215W
GeForce RTX 306080.9$389 ($329)GA106, 3584 shaders, 1777 MHz, 12.7 TFLOPS, 12GB GDDR6@15Gbps, 360GB/s, 170W
Radeon RX 6650 XT79.4$349 ($399)Navi 23, 2048 shaders, 2635 MHz, 10.8 TFLOPS, 8GB GDDR6@18Gbps, 280GB/s, 180W
Radeon RX 6600 XT77.6$359 ($379)Navi 23, 2048 shaders, 2589 MHz, 10.6 TFLOPS, 8GB GDDR6@16Gbps, 256GB/s, 160W
GeForce RTX 207076.0$579 ($449)TU106, 2304 shaders, 1620 MHz, 7.5 TFLOPS, 8GB GDDR6@14Gbps, 448GB/s, 175W
GeForce RTX 2060 Super72.2$409 ($399)TU106, 2176 shaders, 1650 MHz, 7.2 TFLOPS, 8GB GDDR6@14Gbps, 448GB/s, 175W
Radeon RX 660065.6$254 ($329)Navi 23, 1792 shaders, 2491 MHz, 8.9 TFLOPS, 8GB GDDR6@14Gbps, 224GB/s, 132W
Radeon RX 5700 XT61.7$399 ($399)Navi 10, 2560 shaders, 1905 MHz, 9.8 TFLOPS, 8GB GDDR6@14Gbps, 448GB/s, 225W
GeForce RTX 206060.3$340 ($299)TU106, 1920 shaders, 1680 MHz, 6.5 TFLOPS, 6GB GDDR6@14Gbps, 336GB/s, 160W
GeForce RTX 305058.8$319 ($249)GA106, 2560 shaders, 1777 MHz, 9.1 TFLOPS, 8GB GDDR6@14Gbps, 224GB/s, 130W
Radeon RX 570054.4$399 ($349)Navi 10, 2304 shaders, 1725 MHz, 7.9 TFLOPS, 8GB GDDR6@14Gbps, 448GB/s, 180W
Radeon RX 5600 XT48.5$299 ($279)Navi 10, 2304 shaders, 1750 MHz, 8.1 TFLOPS, 8GB GDDR6@14Gbps, 336GB/s, 160W
GeForce GTX 1660 Super36.5$288 ($229)TU116, 1408 shaders, 1785 MHz, 5.0 TFLOPS, 6GB GDDR6@14Gbps, 336GB/s, 125W
Radeon RX 5500 XT 8GB33.3$249 ($199)Navi 14, 1408 shaders, 1845 MHz, 5.2 TFLOPS, 8GB GDDR6@14Gbps, 224GB/s, 130W
GeForce GTX 166032.9$249 ($209)TU116, 1408 shaders, 1785 MHz, 5.0 TFLOPS, 6GB GDDR5@8Gbps, 192GB/s, 120W
Radeon RX 6500 XT28.6$179 ($199)Navi 24, 1024 shaders, 2815 MHz, 5.8 TFLOPS, 4GB GDDR6@18Gbps, 144GB/s, 107W
Radeon RX 5500 XT 4GB27.9$199 ($169)Navi 14, 1408 shaders, 1845 MHz, 5.2 TFLOPS, 4GB GDDR6@14Gbps, 224GB/s, 130W
GeForce GTX 1650 Super26.5$239 ($169)TU116, 1280 shaders, 1725 MHz, 4.4 TFLOPS, 4GB GDDR6@12Gbps, 192GB/s, 100W
GeForce GTX 1650 GDDR622.1$199 ($159)TU117, 896 shaders, 1590 MHz, 2.8 TFLOPS, 4GB GDDR6@12Gbps, 192GB/s, 75W
Radeon RX 640022.0$159 ($149)Navi 24, 768 shaders, 2321 MHz, 3.6 TFLOPS, 4GB GDDR6@16Gbps, 128GB/s, 53W
GeForce GTX 165020.0$199 ($159)TU117, 896 shaders, 1665 MHz, 3.0 TFLOPS, 4GB GDDR5@8Gbps, 128GB/s, 75W
GeForce GTX 163012.5$199 (?)TU117, 512 shaders, 1785 MHz, 1.8 TFLOPS, 4GB GDDR6@12Gbps, 96GB/s, 75W

If you're looking at those numbers and wondering where they come from, they're a composite score generated by using the geometric mean of testing conducted at 1080p, 1440p, and 4K. Ray tracing hardware and DLSS factor into the scores as well, where applicable, which is why Nvidia's RTX GPUs generally outclass their AMD counterparts.

Prices are not a factor in the above rankings, but I've listed the best current Amazon price I was able to find on each GPU. Obviously, previous generation parts tend to be far less desirable, as prices have not dropped nearly enough relative to the performance on tap. Note that I do get Amazon affiliate compensation if you use my links.

With the overall performance rankings out of the way, I've also selected my top four picks for the enthusiast, high-end, mid-range, and budget markets. Here's where price becomes a much bigger consideration.

The Best Enthusiast Graphics Card: GeForce RTX 3080 12GB ($799)

Nvidia hasn't even given an official MSRP for the 12GB variant of the RTX 3080, though I'd say it's worth about $100 more than the 10GB model. It has a few more GPU shader cores, but more importantly it gets 20% more memory bandwidth. With its superior ray tracing hardware and DLSS support, the RTX 3080 12GB comes out just ahead of AMD's latest and greatest Radeon RX 6950 XT, all while costing $200 less. It's basically as fast as the RTX 3080 Ti as well, again for quite a bit less money. Without absolutely destroying your bank account, this is the top enthusiast graphics card right now.

The Best High-End Graphics Card: Radeon RX 6750 XT 12GB ($499)

Picking the best high-end GPU is a bit more difficult. The RTX 3080, 3070, and 3060 Ti are all worth a thought, but overall I give the edge to the Radeon RX 6750 XT. It has more memory, it performs nearly as well as the Nvidia cards — better if you're not worried about ray tracing effects — and it's reasonably priced at $499. You can also opt for the RX 6700 XT, which is only a few percent slower and costs about $50 less.

The Best Mainstream Graphics Card: Radeon RX 6600 8GB ($254)

There's little question about this one, as the RX 6600 delivers excellent performance and only costs $254. Such prices were practically unheard of during the past two years for a GPU of this quality, but with the massive drop off in GPU mining profitability, there appears to be a growing surplus of AMD's Navi 23 parts. Our overall performance ranking puts it quite a bit behind the RTX 3060, but that card costs over $100 more. Online prices are lower than even the RTX 3050, a card that's easily eclipsed by the RX 6600. This is arguably the best graphics card value right now.

The Best Budget Graphics Card: Radeon RX 6500 XT 4GB ($179)

This is a tough one, as the Radeon RX 6500 XT has a lot of limitations. It doesn't have much memory, it's limited to a 4-lane PCIe interface, it only has two display outputs, and it doesn't even have video encoding hardware. The problem is that the next step up basically puts us right back at the $300 midrange price point. Given you can find the RX 6500 XT for just $179, and it competes with the GTX 1650 Super that costs $60 more, there just aren't many other worthwhile options for under $200.

Looking Forward to Ada and RDNA 3

I'd be remiss if I didn't point out that Nvidia's next generation Ada architecture (RTX 4000 most likely) and AMD's next generation RDNA 3 architecture (RX 7000 most likely) are slated to arrive before the end of the year, possibly as early as September. While I could argue that you should just wait for the new parts to arrive, there's little reason to expect great prices. Both AMD and Nvidia will almost certainly launch with their fastest and most expensive GPUs first, leaving high-end, mainstream, and budget offerings for next year.

If you already have a decent graphics card, there's no rush to upgrade today. I wouldn't be surprised to see an "RTX 4080" that performs better than the RTX 3090 Ti while costing closer to $1,000. That's still a $300 jump over the current RTX 3080, and very likely such cards will sell out and go for higher prices until early 2023. AMD will probably do the same thing with an "RX 7800 XT" that's faster than the RX 6950 XT but costs at least $1,000.

Keep in mind that the wafer costs on the upcoming architectures are going to be significantly higher, and prices were paid in advance, many months ago — when GPU prices were still far higher. 5nm class GPUs will simply be inherently more expensive to make, and AMD and Nvidia will want to keep decent profit margins. If the cards have a total manufacturing cost of $500, which seems likely, I'd expect retail prices to be double that.

Monday, May 3, 2021

Chia Coin Trading Is Live, Mass Storage Shortages Could Follow (Or Not)


After a month of waiting, newcomer cryptocurrency Chia has finally hit it's trading start date. It's not listed at many exchanges yet, and there are lots of concerns and caveats with jumping on any new cryptocoin. But Chia is going for a novel appoach with a new proof of space and time algorithm. It's supposed to be far more environmentally friendly than all the proof of work coins like Bitcoin and Ethereum, but it could also shake up the storage industry in a big way.

Instead of hashing, Chia uses large plots of numbers that effectively amount to a massive Bingo card. The network determines the winning criteria every 20 seconds or so (18.75s, theoretically?), and that plot gets 2 XCH, the abbreviation for Chia's coin. The blockchain and block reward structure ends up similar to other cryptocurrencies, but the creation of the plot is the only really difficult part. After that, it's just a matter of letting the Chia plots sit around on your storage devices and hoping one of them gets a 'hit' every now and then.

I hesitate to even mention the current pricing, as it's extremely volatile right now. Early guesstimates put the starting price of XCH at $20, but IOUs for when trading commenced already hit the $1,000 mark in April. So far, we've seen a high of $1,935 today and a low of $1,116. Given the current statistics, neither price is remotely sustainable. Unless...

Yeah, unless. If you asked me years ago what Bitcoin would be worth, I would have never landed at it hitting $50,000 per BTC. Others were more prescient. People asked me about Ethereum when it launched, and I had reached the jaded point of seeing too many scam coins and ICOs and felt it was nothing special. Now, those who took the risk on the ICO have seen its value increase by 10,000X. There are a ton of other coins that haven't done nearly so well, and lots of scams as well, but Chia has some well-known people behind it. As in, Bram "BitTorrent" Cohen. Yeah, the guy that helped create BitTorrent is deeply involved in Chia. Which, I don't know ... BitTorrent isn't exactly a warm fuzzy term. It's useful for a lot of things, but also has been used for a ton of illegal downloads. Whatever.

Let's talk some theoreticals, then. The current netspace used by Chia — which is basically the Chia equivalent of the network hashrate — is already at an astounding 1.77 EiB. That's exbibytes, as in binary exabytes, or 2^60. It's a huge number, in other words. You'd need 177,000 10TB HDDs to store all that data. Even more alarming: Chia's netspace has basically doubled in the past week, and doubled again during the week before that.

Now that XCH is a tradable commodity on a cryptocurrency exchange, the price will help determine the eventual resting point for how much netspace it will use. Right now, even if we go with $1,000 per XCH, you could theoretically earn $1,500 per month from a single 10TB hard drive (#affiliate)! But that's only part of the story.

Creating Chia plots requires about 2.7TB of data to be written. It takes a decent amount of time on a hard drive, but a fast SSD can work on multiple plots concurrently. Except, 2.7TB of writes will burn through the life expectancy of most SSDs! This is where an enterprise class SSD could make a lot of sense. Anyway, if you want to do a lot of Chia plots quickly, you'll need a decent CPU with sufficient RAM, and then an ultra-fast SSD with a high endurance rating. Intel's Optane 905p 960GB for example would be a great choice. Or get an enterprise Intel DC P3600 1600GB drive for $1250. That should allow for the creation of eight concurrent plots in perhaps five hours, or 40 plots per day.  Let's just call it 20 plots per day to be conservative — and Chia will move those plots from temporary creation storage (ie, your fast SSD) to slower long-term storage (ie, a cheap HDD) automatically.

Here's the catch: Right now, Chia farming is basically all done solo. There are mining guilds of a sort, but they're not something I'd recommend — they require you to give away the keys to the kingdom (your private wallet keys). Pooled mining (farming) of Chia is supposed to be coming, but it's not here yet. So in order to win let's just say one plot per week, you'll need ... wait for it ... about 50TB worth of stored Chia plots. Ouch. And that's before the inflation in Chia's netspace, which will drop your mining rewards proportionately.

Now we're looking at $1250 worth of hard drives, plus a $1250 fast enterprise SSD, to do around 20 Chia plots per day. It will take over a week to fill up the HDDs with plots, at which time the netspace has potentially doubled again. That's a very ugly cycle!

Ultimately, the price or netspace (difficulty, sort of) have to reach equilibrium. An RTX 3090 graphics card selling at $3000 (twice its official launch price) can each about $10 per day, so 300 days to break even. Chia farming at $1000 per XCH will potentially get you $1000 per month per 10TB HDD, with a cost of $250 or so per drive. What's more, the power required to keep those drives online is nominal compared to GPU or ASIC farming — about 10W max per 10TB, vs. 250W for a high-end GPU.

Where will Chia and XCH stabilize, and where will they be in a year? Without a lot of storage, it's going to be difficult to get started right now due to the solo mining approach, but when pooled mining becomes available things will likely get a lot better. 10TB of storage could easily net you over $1000 by next year, possibly much more. Or it could end up being wasted space and wasted write cycles on your SSDs.

One final item of note is that Chia has a premine — or in this eco-friendly coin market, a pre-farm. Chia calls this the "Strategic Reserve" and it's a whopping 21 million XCH. That's right: the creators already have 21 million Chia waiting to be utilized. Now, a premine isn't the end of the world, as the funds from such can be used for marketing, development, etc. Lots of successful coins have had a premine or ICO. But it does mean that at the current price of $1000+, a brand new coin would have a theoretical market cap of $21 billion.

Outside of the pre-farm, to date only 451K XCH have been farmed. So of all XCH currently in existence, 98% came via the pre-farm. At a rate of 64 XCH per 10 minutes from farming, for the first three years (the block rewards will halve every three years, until reaching a steady rate of 4 XCH per ten minutes after 12 years), after the first three years the farmed XCH will only account for 32.5% of all XCH in existence. In another three years, farmed XCH will represent 41.9% of all XCH. In fact, it will take approximately 21 years for the farmed XCH to surpass the total size of the initial pre-farm.

In other words, no matter how much money Chia farmers make off of XCH in the coming years, it will be dwarfed by the amount that the Chia Network 'company' makes. Keep that in mind. The people who stand to gain the most from Chia doing well already have control of millions of XCH.

And let's end where I started. The price of XCH now, just an hour or two after I began, has now dropped to just over $1000. It could fall a lot further!

Addendum: The next morning, it looks like prices have reached relative stability at around $750.

Saturday, February 20, 2021

Ethereum Breaks $2,000, Bitcoin at $57,000: History in the Making


Thar she blows! That's right: Ethereum just broke $2,000 for the first — and almost certainly not last! — time. Bitcoin meanwhile is in the high $50K range. With companies like Master Card, Tesla, and various banks investing potentially billions of dollars into blockchain technologies, none of this is going away any time soon. Which raises an interesting question: How much money does the world spend daily on cryptocoin mining?

I've run the numbers, and best-case, the Ethereum miners are using well over a billion watts of power, and Bitcoin mining is about five times that amount. That's going off how much power would be required if every miner was using the most efficient hardware possible, which obviously they're not, and so real-world power use is about three times as high as what I just estimated.

That's three billion watts, every hour of every day, just to keep Ethereum humming along, and roughly 15 billion watts for all of those Bitcoin ASICs. It's nuts! It's also completely unsustainable to keep growing at this rate, so if you're looking at mining profitability right now and wondering how long it will stay this high, I'd estimate another month or two at most, and in three to six months it will likely be less than half as profitable. But what do I know? I thought $40,000 per Bitcoin was the limit for now, and clearly that was wrong.

Anyway, back to power, at $0.10 per kWh, Ethereum would cost about $7.2 million dollars to keep running, each and every day — not including IT infrastructure and cooling, which probably makes it closer to $10 million per day. With a price of $2,000, around 6,500 ETH blocks mined every day, and 2 new ETH minted per block, that's $26 million created out of thin air. So, it's definitely got room to grow at this price. Also interesting is that there are around 1.2 million transactions per day on the Ethereum blockchain, which means each one costs something like $20 in theory — that's looking at the average of around 2 ETH per block in transaction fees.

Bitcoin is, if anything, worse. The Bitcoin network runs at around 160 million terahash per second. The most efficient ASICs do about 25TH/s using 1000W. Using only such ASICs, that would mean we'd have 6.4 million ASICs running Bitcoin, using 6.4 billion watts. But lots of older ASICs are in use, so triple that and we can estimate roughly 19.2 billion watts. Each day, every day!

Total cost at $0.10 per kWh would be about $46 million per day in power costs. Lots of big miners are getting lower cost power than that, but lost of places in the world also cost a lot more, so let's just stick with 10 cents per kWh. Bitcoin is a slow block time of around 10 minutes, so there are only 144 block per day (give or take), but the block reward is 6.25 BTC newly minted, plus around 160 BTC per day in transaction fees, so roughly 1.1 BTC in tx fees per block.

Given those stats, Bitcoin creates 900 new BTC per day, with a value of around $50 million. Notice how much closer that is to the power costs? What's more, at 160 BTC per day in transaction fees, and around 300,000 transactions per day, that means the average BTC transaction costs around $30. Fewer transactions take place, and a lot of places like mining pools and Coinbase seem to be doing transactions largely off the books — using internal mechanisms to track who has what BTC — but the net result is a lot of power and heat used for blockchain technologies.

I'm curious to see where this goes, and if we end up with a lot more regulation and at the same time more uptake of cryptocurrencies in the coming years. The potential influx of cash into the network via banks and other large companies helps ensure things will continue for a while, but we know banks and others could pull out just as quickly as they joined — manipulating the market on a large scale and trying to make a buck in the meantime.

Don't get caught holding the bag, in other words. We saw massive amounts of hype in 2017 and 2018 for blockchain, and much of that went nowhere. Then again, here we are in 2021, with BTC at more than double the previous high and closing in on triple the value. If you had and held BTC for three years and triples your value, that's a nice improvement. All the 'weak' hands that folded? Too bad for them!

Personally, I'm trying to only sell enough of what I earn to pay for power these days. Because it feels like at some point in the coming years, we'll see $100,000 per BTC, and potentially even $1 million. Also potentially $10,000 or less again, so who knows?

Monday, January 11, 2021

GPU Cryptocurrency Mining Profitability, January 2020

Bitcoin is up, and mining is making a comback. Again. At least for the time being. If you're looking for advice on the best GPUs for mining, right now, here's my breakdown and analysis, based on testing over the past few days. These are the best GPUs for mining cryptocurrency, which currently means Ethereum. I'm putting prices in as MSRP, which you probably won't find anywhere for many months. But these are theoretically prices you could pay if supply was keeping up with demand. Adjust your expectations accordingly!

All of my values are based off mining via Nicehash. That's because, frankly, I want either Bitcoin or Ethereum, not one of the other potentially garbage coins. Plus, you can easily trade BTC or ETH for a different coin if that's what you're after. You can probably earn 10% more via mining ETH directly, but then you have to manage things more and it's frankly a pain I can't be bothered to deal with. I use a standard cost of $0.10 per kWh for power, with my own power measurements taken at the outlet, using an 850W 80 Plus Platinum PSU.

Component $ per Day Power per Day Profit Break Even
GeForce RTX 3090 ($1499) $7.50 $0.96 $6.54 ~229 days
GeForce RTX 3080 ($699) $6.50 $0.86 $5.64 ~124 days
GeForce RTX 3070 ($499) $4.50 $0.62 $3.88 ~129 days
GeForce RTX 3060 Ti ($399) $4.40 $0.58 $3.82 ~104 days
GeForce RTX 2080 Ti ($1199) $4.40 $0.72 $3.68 ~326 days
GeForce RTX 2080 Super ($699) $3.40 $0.61 $2.70 ~259 days
GeForce RTX 2070 Super ($499) $3.30 $0.52 $2.69 ~186 days
GeForce RTX 2060 Super ($399) $3.10 $0.48 $1.82 ~154 days
GeForce RTX 2060 ($299) $2.30 $0.48 $1.82 ~164 days
Radeon RX 6900 XT ($999) $6.00 $0.84 $5.16 ~194 days
Radeon RX 6800 XT ($649) $5.25 $0.84 $4.41 ~147 days
Radeon RX 6800 ($579) $4.50 $0.72 $3.78 ~153 days
Radeon RX 5700 XT ($399) $4.00 $0.66 $3.34 ~119 days
Radeon RX 5700 ($349) $3.75 $0.55 $3.20 ~109 days
Radeon RX 5600 XT ($279) $2.70 $0.48 $2.22 ~126 days

So, there you have it. If you could buy the GPUs, the best options right now are the RTX 3060 Ti and the RX 5700. Neither one is readily available at the pricese listed, and we could see prices more than double the MSRP in the coming months, depending on how things develop. But if you get the right hardware, assuming you already have a PC, breaking even on the GPU in just four months still looks possible. At the outside, you might break even in a year. Assuming of course that Bitcoin doesn't go tits up again.

Sunday, January 10, 2021

The Rise and Fall of Bitcoin, Round Four

The only thing predictable about cryptocurrencies at this point is that they're unpredictable. Sure, there are larger patterns that repeat, but knowing when a new swing up is starting, or a new trend down is on its way, is tricky business. The past month has seen Bitcoin eclipse the previous high of 2017, followed by a surge to more than $40,000. That appears (maybe!) to have been the peak, as prices dropped over 20 percent just today (Jan 10, 2020). Will Bitcoin and the other cryptocurrencies rebound and head back up? Will they collapse back to early 2020 prices? I could guess, but then I'm no better than the rest of the pundits. All I know for certain is that cryptocurrencies are extremely volatile, the past month or two being a prime example of that volatility.

I've spoken with a few people over the past week who basically wanted to know what I thought about Bitcoin. One is an investment advisor, and naturally people he advises right now are asking about Bitcoin. Others were computer enthusiasts wondering if now is the time to start mining. My answers to both were somewhat nebulous, but I'll share them here.

Should you invest in Bitcoin or any other cryptocurrency, right now? No, I don't think so. We're already far into the surge upward, and the inevitable collapse will eventually follow. As fast as Bitcoin rocketed to $40,000, it can drop back to $10,000. Since we're still at about $33,000 (update: now past $50K), I wouldn't want to get caught holding the bag. Maybe we'll see $50K before a crash, maybe even $100K! Maybe not.

Longer term, you could still buy in today and very likely recoup your investment in the coming years, but history suggests we'll see $10,000 long before we see $100,000. Wait for the next drop, and if you still want to buy into Bitcoin or some other cryptocurrency, that's the time to make your move. Play the long game, buy when it's low, sell when it's high, but don't buy everything and don't sell everthing — I'd sell 25, 50, or even 75 percent right now, then set a target where you'll buy back in with a similar 25, 50, or 75 percent of what you put aside.

If you have 1 BTC today, then, sell half of it for $20,000 (give or take) right now and keep the other half. Then plan on buying $10,000 worth of BTC when it reaches your desired threshhold, like maybe $10,000. That theoretically gets you $10,000 of profits, plus increases your BTC holdings to 1.5 BTC. Do that a few times over the coming years and you could end up with $50,000 in the bank and 4 BTC or whatever.

As for mining, if you have the hardware available (meaning, a graphics card), I figure any time the net gains from mining are at least double the cost of the power you put into it, that's an okay time to mine. Right now, a GPU that uses perhaps 350W of power (including the rest of the PC) can generate about $5-$7 per day of BTC, using Nicehash. The power cost for running such a PC 24 hours a day is going to be 8.4 kWh, so if you pay $0.12 per kWh, that's $1 per day in power, which means it's a good time to mine.

If you pay $0.30 per kWh, that's $2.52 in power per day, which means it's still a viable time to mine ... but you really ought to consider lower power locations rather than mining in California or Europe or wherever. Cheap power can get as low as $0.05 per kWh, which means it's usually far easier to turn a decent profit.

But more on that in my next post...

Wednesday, December 23, 2020

How to Set Up Network File Sharing on Your Home Windows Network

If you're wondering how to copy files from PCs over the network on Windows, it's not too difficult. I do this for every PC I build, which has been quite a few over the years.

First, you need to know the other PC's network name. You can find this by checking your system information. Opening up the Control Panel, then System and Security, and then System will get you the appropriate page. Far easier: Press Win+X, release both keys, then press Y for System (Win being the Windows key on your keyboard).

You can see the network ID of the PC on the System page under Device Specifications, then Device name. For this particular PC, the network name is "XPS15" (it's an old Dell XPS 15, I'm not that creative with my network names). If you've never specified a name for your PC, it's probably something lame like "Desktop-A8LDK09" or something you'd never remember without looking it up. Go ahead and change that if you want, but you'll need to restart the PC for it to take effect.

You're only half way through accessing PC's storage over your network, sadly. Next, you need to share a folder on the PC. I fly fast and loose and just share the whole damn drive. It's a home network, I don't invite people over and give them the network password unless I trust them, and I also require authentication to access shared folders. You should, too!

So, open Windows Explorer (shortcut: Win+E). Find "This PC" on the left and click that. Then right-click the drive you want to share (ie, C:). Choose Properties, then click the Sharing tab, then click the Advanced Sharing button. Give the share a name (C is default), then press the Permissions button. Select "Everyone" and click the Remove button. Then click the Add button. For the name, type "Authenticated Users" and press the Check Name button. You should now have the share with access permissions set to Read for Authenticated Users. Again, I fly fast and loose so I give Authenticated Users the Full Control option. If your PC has multiple drives, you can do this for each drive you want to share.

You're still not quite done. Your user account needs a password, and you need to know the user account name. You can use the full Microsoft account email address (eg, and password, if your PC is set up with a Microsoft account. Otherwise, open your C:\Users folder to find out the abbreviation for your user name (which you'll need to use for local accounts). You also must have an account password. It can bet a dumb password, but give it something you'll remember that isn't too easy to guess, so not "password" hopefully. "ThisIsMyPassword" should be fine (except it's not now that I've publicly suggested it).

Repeat all of these steps for each PC on your network. Congratulations! You're starting to become an IT professional! (I'm not really joking. This is SysAdmin 101 stuff, but it's very useful!)

Now you're ready to access another PC over the network. Open Windows Explorer (Win+E), and in the address bar (which should say "Quick access" by default), type in the network name of the other PC, but with two backslash characters before the name. So in my case, I type:


Alternatively, on the left side of Explorer, scroll down to Network and click that. Hopefully the other PCs on your network show up, but without the extra step of sharing folders you likely won't be able to see anything on those PCs.

That's it. The first time, you might get a prompt for the user name and password. You can save it if you don't think any hackers or other 'bad' people will get on your PC on your home network, which is a reasonably safe bet. You should now see the "C" folder you shared earlier, which is the entire drive of your other PC. You can even map the folder to a network drive if you want it to always be accessible. Either way, if you browse to the Steam folder (or whatever other folder you want), you can now copy, paste, and delete files.

Simple! Okay, not really that simple, but I do this stuff all the time and it quickly becomes second nature.

How to Transfer Steam Game Downloads Between PCs

Anyone with multiple PCs has probably ended up in a situation where they want to transfer a Steam game from one PC to another. Steam is perhaps the easiest of all digital gaming platforms when it comes to moving games around—better than GoG Galaxy, Ubisoft Connect, and certainly much better (for now at least) than the Epic Games Store. The same process that can be used to move Steam files between two different drives on the same PC extends to copying game files between different PCs.

First, close down Steam on both PCs. Then open the Steamapps folder on the source PC, and the Steamapps folder on the destination PC. By default, this will be:

C:\Program Files (x86)\Steam\Steamapps

If you're like me, it will be C:\Games\Steam\Steamapps, but whatever. You should see a few subfolders, plus some .ACF files. You'll need to copy over both the ACF file for whatever game you want to transfer, plus the actual game files, which are located in the Common subfolder. 

But which ACF file(s) do you need? You'll have to open them up with a text file viewer like Notepad++ (seriously, no one should still be using Notepad!) to figure out which game each file corresponds with. Alternatively, browse to and search for the game you want to copy. Open the appropriate page and you'll fine the Steam app number in the URL. For example, Cyberpunk 2077 is app number 1091500. Copy the ACF file to the other computer, then go to the Common subfolder and copy the game folder (which can be quite large).

Once the copy is finished, launch Steam on the second PC and it should automatically detect the new game files. You're done! You might need to wait for Steam to sync your cloud saves, but otherwise that's it.

Of course, the above assumes you already know some basic networking stuff. If not, check out my How to Set Up Network File Sharing on Your Home Windows Network guide.

Why wouldn't you just download the game on both PCs, though? Simple: Data caps. They suck, but many of us still deal with them because we have no other alternative. Comcast / Xfinity has a 1229GB per month data cap, which is 205GB more than the unofficial data cap that's been in place for nearly 20 years. My Internet connection (via Comcast the entire time!) has gone from 50Mbps down/6Mbps up to 350Mbps down/12Mbps up over that same time.

Put it this way. Back in 2005, with a 75Mbps connection at the time, it would have required 30.3 hours of continuous downloading at the maximum speed possible to hit my old 1TB data cap. Today? I could chew through my entire monthly 1.2TB limit in just 7.8 hours! Plus, game sizes have massively increased during that same time. 2-5GB games used to be the norm, then 10GB, 15GB, and 20GB. Now, 50-75GB minimum is typical of any major game release, and there are multiple games that weigh in at far more than 100GB. Downloading Cyberpunk 2077 twice would use up 12 percent of my data in less than an hour.

Of course it's not just gaming and downloads. Windows updates, Android and iOS app updates, game updates, and other software updates all take their pound of data. But if you've cut the TV cord and are going full streaming? One hour of 4K content on Netflix or Amazon Prime represents 10-20GB, depending on the encoding quality, and between family members it's easy to use hundreds of GB per month on streaming content. In today's world of streaming video? 2TB is a reasonable data cap, especially considering I already pay $30 extra for a higher bandwidth.

How to Transfer Epic Games Store Downloads Between PCs

If you have more than one PC that you use for gaming—like, say, a desktop PC in the office and a laptop for elsewhere—you might want to have the same game installed on both PCs. For some platforms (eg, Steam, GoG Galaxy, Ubisoft Connect) it's as easy as copying the files from one PC to the other, and sometimes importing the already downloaded folder. Epic Games? Not so simple, but at least it's possible—and it's far better than the Microsoft Store, where you have no option other than to download the full game on each PC.

For the Epic Games Store, the process is a bit more painful. First, before you do anything else, make sure the Epic Games Launcher isn't running (on either PC). Next, you need to copy the game files. These are usually located in your Epic Games folder, but can be installed elsewhere. If you've done a custom install to a different folder, you'll need to put the files into the exact same location on the second PC. (Workaround below.) This is why I create a C:\Games folder and have Steam, GOG, Ubisoft, and Epic all installed in subfolders.

This guide assumes you already know some basic Windows networking stuff. If not, check out my How to Set Up Network File Sharing on Your Home Windows Network guide. Alternatively, you could copy all of the files to a (hopefully fast!) USB drive (I have a Samsung T5 1TB SSD [#affiliatelink] that works great). Either way, step one is to copy the files over:

The next step is where things can get tricky. Go to your C:\ProgramData folder (it's hidden by default, so you might need to tell Windows Explorer to show hidden files and folders). You can try to be clever about what you copy, but I find it's easiest to just copy the entire Epic folder from the first PC where you downloaded over to the second PC (and replace any duplicate files that don't have the same time and date).

Once the above is done, launch the Epic Games Launcher on the second PC and it should pick up any games you have on both the second and first PC. There's a catch, however: If you only have a game installed on the second PC, it will disappear and you'll have to jump through extra hoops. Crap. So, your first PC should be a superset of all the games you want on the second PC. But there's another possibility, which is more work and rather a pain in the rear, but it's still better than the complete redownload of a large game.

To execute the workaround, start the download of the game you want on the second PC, pause it, and exit the Epic Games Launcher (say yes to the warning message). On the first PC, of course make sure the game is fully updated, and then exit Epic Games there as well. Now, copy all of the files from the game folder on the first PC, into the game folder subdirectory named:

[Game Folder]\.egstore\bps\Install

Once the copying of all the files is complete, launch the Epic Games Store and resume the download. It should start "verifying" all of the game files. On a fast SSD, this happens at about 400MB per second, so an 80GB game can still take 3-4 minutes to verify. If you're using a slower SSD or a hard drive, it might take 10 minutes or more.

If you see your network traffic start up again, though, the workaround ... didn't work. Sorry. I'm not sure what causes that, but some games don't seem to like doing things this way. Fortnite is a good example, and it probably has difficulties because it gets updated about every other day. As always, your experience may vary, but if neither approach worked for you, leave a comment with the game name and details of what happened.

Why wouldn't you just download the game on both PCs, though? Simple: Data caps. They suck, but many of us still deal with them because we have no other alternative. Comcast / Xfinity has a 1229GB per month data cap, which is 205GB more than the unofficial data cap that's been in place for nearly 20 years. My Internet connection (via Comcast the entire time!) has gone from 50Mbps down/6Mbps up to 350Mbps down/12Mbps up over that same time.

Xfinity, your data caps SUCK!

Put it this way. Back in 2005, with a 75Mbps connection at the time, it would have required 30.3 hours of continuous downloading at the maximum speed possible to hit my data cap. Today? I could chew through my entire monthly limit in just 7.8 hours! Plus, game sizes have massively increased during that same time. 2-5GB games used to be the norm, then 10GB, 15GB, and 20GB. Now, 50-75GB minimum is typical of any major game release, and there are multiple games that weigh in at far more than 100GB. Downloading Cyberpunk 2077 twice would use up 12 percent of my data in less than an hour.

Of course it's not just gaming and downloads. Windows updates, Android and iOS app updates, game updates, and other software updates all take their pound of data. But if you've cut the TV cord and are going full streaming? One hour of 4K content on Netflix or Amazon Prime represents 10-20GB, depending on the encoding quality, and between family members it's easy to use hundreds of GB per month on streaming content. In today's world of streaming video? 2TB is a reasonable data cap, especially considering I already pay $30 extra for a higher bandwidth.