News Intel GPUs - Falcon Shores cancelled

[beginner99]

My point is, for an example, consider the possibility of a mining consortium purchasing a company like EVGA, or at least their AIB section, and then using the AIB business unit to custom design cards and bios programs that exploit existing gpu silicon specifically for mining. This could entail disabling unneeded sections of the chips in BIOS, changing the memory timings to favor mining, and designing the cards for 24/7/365 maximum throughput with no extra resources put towards video out, flashy board designs, etc.

Way too costly. Mining isn't a high margin business because you need to constantly sell so you are not taking in the huge gains the coins make, only maybe this year. Plus it's on the way out. bitcoin has moved to ASICs a long time ago, ethereum will move to PoS soon, what is left? Not much for GPU mining.

 

[LightningZ71]

Its a highly speculative market with a proven, though limited, track record of moving on to the "next great coin". Etherium will likely be forked when it goes POS. There are also other coins that have practical use out there that existing cards can mine. They aren't being mined by anyone at the moment because they aren't nearly as profitable as Etherium to mine due to its present value. The cards that are out there exist, the mines exist, if there is any way for them to be put to use mining something, they will be. Remember, Etherium was a nothing coin for years and years because bitcoin could still be somewhat mined with GPUs, even though there were ASICs out there that did it much better. When Bitcoin became effectively impossible to profitably mine with a GPU, Etherium was the most profitable coin available for everyone to target. There will be another coin.

 

[DrMrLordX]

Etherium will likely be forked when it goes POS.

Ethereum has already been forked multiple times. Ethereum Classic, Ethereum MAX, etc.

 

[LightningZ71]

Of course, as has bitcoin. However, there has always been a more profitable coin to mine that was actually attainable. Mining is a "for profit" activity. Miners will mine what they can make the most money on, and Bitcoin for ASIC miners and Etherium for GPU miners have been the two biggest targets for profitability, with XMR filling in the gap for CPU miners.

When Etherium goes POS, 99% of the mining out there will be left with two choices, sell off, or mine the next most profitable coin until it does the same thing that Etherium and Bitcoin did, become unattainable.

 

[beginner99]

Ethereum has already been forked multiple times. Ethereum Classic, Ethereum MAX, etc.

Etehreum MAX is not a ethereum fork but a shitcoin scam (pump and dump). it has nothing to do with ethereum really, they juts use the name so it sounds legit.

 

[NTMBK]

Etehreum MAX is not a ethereum fork but a shitcoin scam (pump and dump). it has nothing to do with ethereum really, they juts use the name so it sounds legit.

They're all pump and dump scams...

 

[DrMrLordX]

Etehreum MAX is not a ethereum fork but a shitcoin scam (pump and dump). it has nothing to do with ethereum really, they juts use the name so it sounds legit.

Turns out you're right. Ethereum MAX is actually an ERC-20. That would make Ethereum Classic the only actual fork of Ethereum of which I am aware (Bitcoin has many more forks). Which is funny, since copying an older version of Ethereum is pretty easy for people who have experience with rolling out blockchain projects. Maybe not as easy as forking Bitcoin or Litecoin, but easy enough.

 

[moonbogg]

I'll be the first to guess at Intel's GPU ebay price for the top model: $1500.

 

[ultimatebob]

I'll be the first to guess at Intel's GPU ebay price for the top model: $1500.

It would only sell for $1500 if it mines crypto well. Give the Ethereum mining developers a few months to write optimized mining software first.

 

[moonbogg]

It would only sell for $1500 if it mines crypto well. Give the Ethereum mining developers a few months to write optimized mining software first.

I think $1500 is what gamers would be happy to pay on ebay. Miners would pay $2000+ for a card like that if it mines anywhere near a 3080. Gamers don't mind paying a few more bucks for a quality product. I think Intel should price their new cards at $1000 and up at least. $1000 is cheap these days. That's barely 3060 money.

 

[Leeea]

Lets take a moment to think about how much Intel might actually effect the market here:

So Intel is buying wafers from TSMC for their graphics solutions. Apparently their GPUs will not be fabed in house at all.

So if we go off the rumor* mill, for 2021 TSMC:
Apple 25%
AMD 9%
Mediatek 8%
Broadcom 8%
Qualcomm 8%
Intel 7%
Nvidia 6%

Now, if we look at the Intel's numbers year on year:
2019: Intel - 5%
2020: Intel - 6%
2021: Intel - 7%
It seems like Intel has been buying chips from TSMC for years now. To cover their GPU production at TSMC, they are only increasing their buy to be an additional 1% of TSMC's revenue.

1% of TSMCs revenue going to Intel's GPUs is not going to effect anything in this shortage market. The Intel GPUs will be even more vaporware then the AMD/Nvidia ones. Another possibility is Intel is waiting until 2022 to do its buy, but if that is the case we will not be seeing these Intel GPUs any time soon.


*the rumor mill claims this is from cnbc in March 2021, but I have not yet found the original source. Intel's large 3nm purchases at TSMC seem unlikely to effect things prior to 2024.

 

[mikk]

And the other 6%? You have to assume the wafer allocation didn't change more towards Xe HP, otherwise it's a flawed analysis. Intel shipped the iphone modem XMM 7360/XMM 7480 some years ago made on TSMC which they don't have anymore.

Overall 7% seems surprisingly high considering that AMD is barely higher with lots of CPUs and GPUs made on TSMC. Furthermore Xe HPG is launching in Q1 2022, your 2021 1% "analysis" is a result from only one quarter at best with the bigger ramp likely in 2022.

 

[eek2121]

It would only sell for $1500 if it mines crypto well. Give the Ethereum mining developers a few months to write optimized mining software first.

As I understand it, you won't be able to mine Ethereum when the Intel cards are released. ETH will be PoS instead of PoW by then.

 

[ultimatebob]

As I understand it, you won't be able to mine Ethereum when the Intel cards are released. ETH will be PoS instead of PoW by then.

Ethereum has been saying that they were going to switch over to Proof Of Stake back since 2018. At this point, I'll believe it when I see it.

 

[beginner99]

Ethereum has been saying that they were going to switch over to Proof Of Stake back since 2018. At this point, I'll believe it when I see it.

Well you can see it and you can actual stake already if you want to. In fact there are so many stakers already and a long queue by now you are probably too late to the game.

As I understand it, you won't be able to mine Ethereum when the Intel cards are released. ETH will be PoS instead of PoW by then.

From a pure technical perspective probably true but I'm pretty sure it (= The Merge = merging of ETH1 chain with PoW with beaconchain with PoS) will be delayed due to 1 major issue: client diversity or better said lack thereof. The way PoS works, no more than 1/3 of validators are allowed to be offline for the chain to still finalize*. Which means no client should have a market share above 1/3 or else if that client is faulty, the network will go down. Currently the first completed and leading client has about 70% market share. Ethereum foundation hasn't commented on this yet but I'm pretty sure they will need to and that the merge will not happen if client diversity does not increase.

* for the cryptohaters in here: no you can not use this to attack the network. First you would need a huge amount of ETH, 32 ETH for each validator and then there is a finite size of allowed validators and right now it already takes 10 days till you can enter the network. Besdies that you will lose all your funds. Currently 1/3 is over 100k validators meaning 100,000 times 32 ETH = 3.2 Mio ETH at roughly $3000 dollars. thats almost 10 billion $.

 

[NTMBK]

Can we please stop turning this into Yet Another Crypto Thread? This thread is meant to be about Intel GPUs.

 

[DrMrLordX]

Can we please stop turning this into Yet Another Crypto Thread? This thread is meant to be about Intel GPUs.

Well I think the gist of it is: Intel dGPUs probably won't see much mining duty, for better or for worse.

 

[ultimatebob]

Well I think the gist of it is: Intel dGPUs probably won't see much mining duty, for better or for worse.

Plus, there are no Intel GPU's yet, unless you're a lucky beta tester who's probably under NDA right now. That leaves us with just speculation, or talking about how awesome the forum moderators are

 

[RTX2080]

DG1 to have performance around GT1030

 

[uzzi38]

To have? Dude, DG1 has been out for a while now

It's slightly worse than TGL iGP for average framerates with much worse lows (and more driver bugs).

 

[RTX2080]

To have? Dude, DG1 has been out for a while now

It's slightly worse than TGL iGP for average framerates with much worse lows (and more driver bugs).

Sorry, I lol at myself now. I've long time not being informed in GPU segment.

 

[Shivansps]

What would be the chance of Intel not releasing GPU OCL drivers for the Xe GPUs? They have a golden oportunity to win the gamer market if they do that.

 

[DrMrLordX]

What would be the chance of Intel not releasing GPU OCL drivers for the Xe GPUs? They have a golden oportunity to win the gamer market if they do that.

Depending on whom you ask, the existing OpenCL drivers for Intel iGPUs are bad enough that they may as well not exist. At least as far as mining software is concerned.

 

[mikk]

Intel is the only graphics company who cares about OpenCL these days. It's not about the driver, the mining software devs doesn't care for iGPUs.

 

[Tup3x]

Doesn't Intel have their OneAPI thingy as CUDA competitor? I'd assume that developers would use that if they end up making a mining client for Arc GPUs