Threadripper Pays for Itself Using Cryptocurrency Mining @ [H]

Games like GTA V selling like hotcakes is killing gaming. There's absolutely no fun to that game after the 1st hour, just a reskin of previous GTA Games so now everyone is looking for that magical open world without any decent story line.

No People are being stupid and spending $$$$ for DLC for GTA Online-and Rockstar is making a killing on it.

I enjoyed the story that GTA:V had and the online portion was fun, but after a while it was just a grind with no real payback and Rockstar wanting you to open your checkbook to make any meaningful progress with it-I tuned it out and haven't played it in ages.
 
Mind you, even an old AMD FX can be a 3€/day CPU if your electricity cost is as cheap as in Argentina or Brasil, 0.05cent/kWh

I believe there are some misconceptions about Brazil's energy rates.
Expect to pay R$ 0.337 to R$0.599 for each kWh before taxes. Source.
But said values ( somewhere around USD 0.9 to 0.19/ kWh) only apply to the first 200kWh consumed in a month. the next 200kWh would cost double that and every kWh consumed above 400kWh costs triple. Taxes add 40-60% over that figure. I live in a house of with wife and 2 kids: energy bill ~$165/month, without mining.

So Brazil's miners usually only have profitable energy rates when mining in the colder months in the south of the country, using the mining rigs as heaters for their houses. And their ROI is longer than that of Arizona/Texas, because hardware here costs at least 60% as much from import taxes.
Been there, done that. The grass is always greener on the other side of fence.
 
I believe there are some misconceptions about Brazil's energy rates.
Expect to pay R$ 0.337 to R$0.599 for each kWh before taxes. Source.
But said values ( somewhere around USD 0.9 to 0.19/ kWh) only apply to the first 200kWh consumed in a month. the next 200kWh would cost double that and every kWh consumed above 400kWh costs triple. Taxes add 40-60% over that figure. I live in a house of with wife and 2 kids: energy bill ~$165/month, without mining.

So Brazil's miners usually only have profitable energy rates when mining in the colder months in the south of the country, using the mining rigs as heaters for their houses. And their ROI is longer than that of Arizona/Texas, because hardware here costs at least 60% as much from import taxes.
Been there, done that. The grass is always greener on the other side of fence.

Hmm, I have a pal in another forum and we often chat, I might have mixed hos country up, could be Argentina. He says he pays 0.06 US_cent/kWh. I took that as true.Maybe rates differ between regioons of the country too. His source of energy iirc are nearby Hydroplants.
 
Paraguay is the cheapest. Final rate is $52 to $63 for MWh (unlimited ), around your quoted $0.05/kWh. Source.
Argentina costs ~$0.08/kWh up to 300kWh/month. Source.
 
Odd that when you have CPU and GPU mining going the system is unusable. I have my 4670K and 1080Ti both going and my system is plenty usable, youtube, surfing, Outlook/Excel/Word/etc, no issues.


Also, you look like Ron Hextall. LOL
Say what? I tried turning on XMR on the CPU in my Nicehash client after watching the video, and I was surprised to find that it had no obvious effect on the system's usability at all. Mining Eth on the GPU has a noticeable effect, but at least using the Nicehash legacy client, it doesn't seem to make any difference, other than the obvious extra heat.
 
With crypto crashing, what's payback time now, 3 years?
Impossible to say. The price of the major cryptos is lower *right now,* but it goes up and down all the time anyway. Based on the current price, payback time looks like it'd probably be in the 18-24 month range for the 1950X.

That said, it looks to me like you could use the $450 TR CPU and make roughly the same money, but spend half as much (not counting the mobo and memory).
 
Not as volatile? Maybe.
Riskier? Absolutely.

CPU-mined crypto tends to be newer, riskier crypto. It can, and does, dissappear overnight sometimes.

I must say that I'm surprised at the tone of the article and video, when it comes to the click-bait-ey "Pay for Your Threadripper By Mining" statements.

If you calculate the ROI today, there is a *100% chance* that calculation will be wrong. And due to the volatility of the space, we're not even talking about a chance of being "off by a day or week or so". It could be that you never ROI the hardware, or you could possibly earn enough magic internet money to ROI it in 2 months. But guessing about when that is based on a SPOT calculation of the last price someone paid for Monero, the current difficulty, hashpower targeting the coin is a fool's errand. That doesn't even take into account the price change of power over time, nor does it consider the power "tier" structure, whereby the power company may charge more per Kwh based on your usage (i.e., the more you use, the more you pay per Kwh). Also, not enough people do this while understanding that they do need to account for the money required for a higher power bill. You may have calculated it in the equation to guess profitability, but it often surprises folks that start mining when they didn't prepare themselves for $X greater power bill the next month.

Remember: it costs REAL money now to earn *the chance* at more magic unicorn internet money later.

The other flaw in this common misconception is that it assumes you sell the crypto the same time you "make" it. It's really more about the value of the crypto one currently holds, and when someone might choose to exchange that into something else (be it fiat currency, BTC, DASH, whatever...). And therein lies the guessing game.

But it's all certainly enough that I would tell people "You can make magic unicorn internet money, and you might be able to turn that into USD, but it ain't a picnic, and it's far from guaranteed."

Just my 0.000003 BTC on the subject.
If you don't have a Threadripper, you could have just said so.
 
Monero is the 13th largest crypto coin by market cap. It's not really some fly by night (crypto-night even) coin that's going to disappear in the morning.
 
All I can say is I am without my gaming PC as my 980Ti smoked (but still just barely under warranty thank goodness) but I have been waiting 2 weeks so far because EVGA has NO STOCK to send me a compatible replacement card. And they told me why. Cryptominers. Personally, I think they still should have kept some stock to fulfill warranty claims but I don't make the rules. Not to mention my new Ryzen 1800x build is on hold because 1KV and above (major brand) power supplies are all backordered for the same reason. So, ya. miners are not my friends right about now.
 
Not as volatile? Maybe.
Riskier? Absolutely.

CPU-mined crypto tends to be newer, riskier crypto. It can, and does, dissappear overnight sometimes.

I must say that I'm surprised at the tone of the article and video, when it comes to the click-bait-ey "Pay for Your Threadripper By Mining" statements.

If you calculate the ROI today, there is a *100% chance* that calculation will be wrong. And due to the volatility of the space, we're not even talking about a chance of being "off by a day or week or so". It could be that you never ROI the hardware, or you could possibly earn enough magic internet money to ROI it in 2 months. But guessing about when that is based on a SPOT calculation of the last price someone paid for Monero, the current difficulty, hashpower targeting the coin is a fool's errand. That doesn't even take into account the price change of power over time, nor does it consider the power "tier" structure, whereby the power company may charge more per Kwh based on your usage (i.e., the more you use, the more you pay per Kwh). Also, not enough people do this while understanding that they do need to account for the money required for a higher power bill. You may have calculated it in the equation to guess profitability, but it often surprises folks that start mining when they didn't prepare themselves for $X greater power bill the next month.

Remember: it costs REAL money now to earn *the chance* at more magic unicorn internet money later.

The other flaw in this common misconception is that it assumes you sell the crypto the same time you "make" it. It's really more about the value of the crypto one currently holds, and when someone might choose to exchange that into something else (be it fiat currency, BTC, DASH, whatever...). And therein lies the guessing game.

But it's all certainly enough that I would tell people "You can make magic unicorn internet money, and you might be able to turn that into USD, but it ain't a picnic, and it's far from guaranteed."

Just my 0.000003 BTC on the subject.
You do realize that the 1950x in Cryptonight has better hash rates than Titan V? Right? You do know that you can convert virtually any Crypto into fiat or whatever currency you want -> Help pay for whatever. Hell, you could convert virtually daily from Nicehash to Coinbase to $ with minimal amount of fees. So help paying for the 1950x or offsetting the cost for that platform is reasonable and is an option for folks to know about and as a consideration if they own one or are considering the cost of buying one. So this was a great service/read from HardOCP -> No one else even looked at this seriously. As for the ROI, it will most likely be much less than the time estimated by the review.

Hell if I add up all my current currencies I have at TODAY's rates it already payed for 2 1080Ti's, 2 Vege FE's and then some in less than 5 months. Tomorrow values of the same coins/Cyrpto will most likely will be higher meaning even more profit. Since most will have more than just the 1950x, as in a GPU or GPU's, that also mean the ROI will be shorter if you put more assets mining.

As for unicorn money, pull out some of the green stuff in your wallet -> that is the real unicorn money.
 
You do realize that the 1950x in Cryptonight has better hash rates than Titan V? Right? You do know that you can convert virtually any Crypto into fiat or whatever currency you want -> Help pay for whatever. Hell, you could convert virtually daily from Nicehash to Coinbase to $ with minimal amount of fees. So help paying for the 1950x or offsetting the cost for that platform is reasonable and is an option for folks to know about and as a consideration if they own one or are considering the cost of buying one. So this was a great service/read from HardOCP -> No one else even looked at this seriously. As for the ROI, it will most likely be much less than the time estimated by the review.

Hell if I add up all my current currencies I have at TODAY's rates it already payed for 2 1080Ti's, 2 Vege FE's and then some in less than 5 months. Tomorrow values of the same coins/Cyrpto will most likely will be higher meaning even more profit. Since most will have more than just the 1950x, as in a GPU or GPU's, that also mean the ROI will be shorter if you put more assets mining.

As for unicorn money, pull out some of the green stuff in your wallet -> that is the real unicorn money.
Are you factoring in taxes? Like it or not its a reality now, especially if you're selling for USD on coinbase.
 
Are you factoring in taxes? Like it or not its a reality now, especially if you're selling for USD on coinbase.

If you make less than $38,000 and hold your coins for more than a year, what percentage do you pay?
 
You pay 0% if you make under $38,000 because after a year it is a long term capital gain.
 
You pay 0% if you make under $38,000 because after a year it is a long term capital gain.
That said, I'm sure if you cashed out a certain percentage or more than your yearly income through something like Coinbase with a ACH transfer to your bank account, it would likely trigger an audit, or the bank might file a SAR which would trigger investigation.. lol.
 
Although the topic is interesting, I don't really agree that one should buy TR's for mining. You might as well buy overpriced Vega 64's.

Vega 64 does about 2000 H/s on Monero on less than 200W from the wall. Even overpriced, you can buy them for $900. Heck, Newegg has the Vega FE for around $950, and they get about 2200 H/s. Most importantly is that they scale better. You can get a mobo with several PCI-E slots and it will scale out. If you went the CPU route, you'd have to buy a lot of extra components.

Edit: I do find it interesting that the TR 1950X does better than a Broadwell-EP E5-2699 v4. The latter has way more L3 cache and more core/thread count. My single E5-2699 v4 does about 900 H/s (pulling 125W from wall).
 
Although the topic is interesting, I don't really agree that one should buy TR's for mining. You might as well buy overpriced Vega 64's.

Given the prices, you probably shouldn't buy a TR specifically for mining. However if you already have a TR, or plan to buy one, it's not a terrible deal.

Some people never turn off their systems. If that's the case the 108 idle draw is already a sunk cost. That being said, with this kind of thinking your E5-2699 may be more energy efficient (depends on idle value).

So what you're really looking at is 136W for 1450H/s whenever you're not doing stuff on it. So in the particular setup that HardOCP demonstrated, the CPU is getting ~10H/Ws, while the GPU component is only doing ~6.5H/Ws.
 
ZiggyDeath

If you are in need of (or currently have) a multicore system like the TR, there certainly isn't any negative to mining Monero, IMO. But I think if you're looking for profit and scalability, Vega 64/1080 Ti is still where it's at.

For reference, my E5-2699 v4 (using a MSI X99-SLI board) idles around 80W from the wall. Monero mining at 900 H/s only pulls in 125W from the wall (measured via Kill-a-Watt). With a 45W differential, its pretty stupid to not mine, given the fact that the system is already there.
 
Yea, threadripper can make a profit mining. So can most modern cpus. People have known that Ryzen excels at Cryptonight mining for a while. Even AMD FX is pretty good compared to Intel. (My FX 8350 does 360 H/s... at least it's good at something...)

But when it comes to price/performance, and price/watt, you are far better of with a BIOS modded RX 570/580 or RX Vega. Even an RX 580 for $500 would get you better price/performance than threadripper. So yea, if you have threadripper, feel free to mine on it, but it's a terrible idea to go out an buy threadripper cpus to mine on.
 
If you are in need of (or currently have) a multicore system like the TR, there certainly isn't any negative to mining Monero, IMO. But I think if you're looking for profit and scalability, Vega 64/1080 Ti is still where it's at.
Exactly, the argument is more towards utilizing underutilized resources. Also a person is is likely to have/access to CPUs rather than a whole lot of videocards.

For sure if you're in it to make money, a CPU rig doesn't make much sense.

For reference, my E5-2699 v4 (using a MSI X99-SLI board) idles around 80W from the wall. Monero mining at 900 H/s only pulls in 125W from the wall (measured via Kill-a-Watt). With a 45W differential, its pretty stupid to not mine, given the fact that the system is already there.
The power draw is impressive. I assume that the 125W figure is after the turbo drops off?
 
Monero apparently is very cache intensive, but not quite execution intensive. So IMO, its relatively little stress on the CPU.

The E5-2699 v4 numbers is *system* power draw with turbo disabled, so base clock is steady at 2.2 GHz. I saw that the TR 1950x was being clocked at 4 GHz, so that's probably why its hash rate is much higher. Too bad O/C options suck on newer Xeons because I think even a mild O/C to 2.5-2.7 GHz will greatly improve hash rate while not increasing power draw by much.
 
Too bad O/C options suck on newer Xeons because I think even a mild O/C to 2.5-2.7 GHz will greatly improve hash rate while not increasing power draw by much.

Not familiar with Xeons, but could you use something like XTU to override the boost timer? Or perhaps some other utility?

Just screwing with the turbo timers allows you to squeeze a lot more out of a Intel CPU without the messiness of voltage changes.
 
Not familiar with Xeons, but could you use something like XTU to override the boost timer? Or perhaps some other utility?

Just screwing with the turbo timers allows you to squeeze a lot more out of a Intel CPU without the messiness of voltage changes.

To my knowledge, Intel 'fixed' the loopholes regarding E5 overclocking either by unlocked multiplier (v1, v2, some v3) or forcing turbo modes (most v3). You can overclock BLCK, but that only yields small gains.

Of course if I'm wrong about it, someone point me to the light!
 
I've been wanting to build a TR system since they came out, but just couldn't justify the cost when my Ryzen 7 does everything I throw at it. But this little press release from AMD gave me the excuse I needed to go ahead and build a TR system. Combine great pricing on the 1950x and MBs at Microcenter and I'm for $1100. Unfortunately, I needed to buy a new PSU to support the 2nd CPU power socket required by the ASRock Taichi board.

Currently hashing 1420 H/s drawing 160W using Nicehash, so yeah, it will take me 18 months to recover my costs but I'm thrilled and any kind of cost recovery. For me, this makes more sense than overpaying for a bunch of GPUs, even if they recover their costs faster.
 
I've been wanting to build a TR system since they came out, but just couldn't justify the cost when my Ryzen 7 does everything I throw at it. But this little press release from AMD gave me the excuse I needed to go ahead and build a TR system. Combine great pricing on the 1950x and MBs at Microcenter and I'm for $1100. Unfortunately, I needed to buy a new PSU to support the 2nd CPU power socket required by the ASRock Taichi board.

Currently hashing 1420 H/s drawing 160W using Nicehash, so yeah, it will take me 18 months to recover my costs but I'm thrilled and any kind of cost recovery. For me, this makes more sense than overpaying for a bunch of GPUs, even if they recover their costs faster.
And that is exactly what the story is about, it is NOT about building TR mining rigs, it is about using what you already have to recoup some cost. Looking forward to your build!
 
I would love to build a TR system as my plex/file/VM server & video encoder box. Silly FX8320 on that thing gets under 10 fps on Handbrake on SHQ. Then again, that's more than double what my old P2-955 was doing, it averaged 5 fps or less. I have no real interest in mining, but if the box is going to be on 24/7 anyway and idle 90% of the time I'm not doing vid work... I guess I would look into that.

Stupid ram prices still keeping me from making the leap though... that and I just paid my taxes and that was an unexpected nasty surprise.
 
I've been seeing this topic pop up all over the place, but this is the first place that started talking about it by mentioning that it was AMD marketing that sent the idea out there. Are older chips Intel chips just too inefficient to do this with or could you still come out ahead even with say, a 4930K? That's how Bitcoin was many years ago - you didn't buy hardware to mine it - you used what you already had in the spare time when you weren't using it.

Edit: Looked into it, and it seems you can indeed do it with older CPUs like the 4930K - and get about 300 instead of 1400 like Threadripper, but with power I think it would have made a LOT more sense 3 months ago in the brutal winter we had here where any power used on mining was basically free heating than it does now with it starting to warm up and you having to pay for both system usage and air conditioning usage.
 
Last edited:
I just sourced a Ryzen 1700 and board to "upgrade" my Plex server from an old X58 setup. Mining in the background of course. Figured it was more efficient (should be after all it's 6-7 years newer) and would run cooler.
 
I've been seeing this topic pop up all over the place, but this is the first place that started talking about it by mentioning that it was AMD marketing that sent the idea out there. Are older chips Intel chips just too inefficient to do this with or could you still come out ahead even with say, a 4930K? That's how Bitcoin was many years ago - you didn't buy hardware to mine it - you used what you already had in the spare time when you weren't using it

I guess it was more of an interview than a press release:

https://www.forbes.com/sites/jasone...e-mining-profit-than-a-gtx-1080/#116d3d265add

One key to getting good performance is fitting the work into L3 cache, so it doesn’t have to fetch it from memory. With Cryptonight you need 2MB L3 cache per thread to make it efficient, which is why the Zen CPUs perform better than intel.
 
Don't forget that any profit must be claimed on income tax, so that is likely another 30-40% off, making payback rate longer.
true but I doubt tax man is gonna come at you for 200 dollars or invest in tracking you that will likely cost them over 1000. They got bigger whales to fry.
 
Back
Top