Mining vs Gaming Products or something *spawn*

We don't know what's "enough". The idea is most likely to make miners lives harder on gaming lineups and I'd be surprised if they don't have a plan which would span several years already. This alone means that anything based on Turing is just a first stage. I doubt that it's a mistake that they have a gap in HX lineup between 50 and 90 for example.

 

Well you said "they don't need a lot of them" . Now " we don't know what's enoungh" .. So which one is it?

 

Which one is what? You don't know what's enough thus you can't tell how many they need.
Also I didn't say "they don't need a lot of them". Do quote what I've said in full.

 

If we had a guess on yields we could make some kind of guess on how many crypto cards nvidia can make without affecting gaming gpu supply. If yield for example was random number, 80% there potentially would be 20% additional supply without affecting gaming cards. That 20% would also be pure profit as those chips otherwise would go to garbage bin. Who knows what kind of yield nvidia has in samsung process. The crypto chips mainly really are memory hungry so pretty horribly broken chips with low clock speeds can still be used. For example there is no display out at all or tensor core or ray tracing used in crypto cards. ETH doesn't benefit from cache either so the sram inside chip could also be fairly broken and still work fine for mining. One could assume initial yields are worse as it's new chip on new process. Yields probably are better now than 6 months ago.

 

I replied to your intuitive estimation that not many mining-only GPUs would be needed initially.
I don't think I misqoted-you, no. I had the context in mind, and was addressing the shortish term future, after which other mining GPUs or new gaming GPus with mining limitations might follow (as per @manux post, which would indeed make nvidia's strategy make sense to me).
For now the competition is between those mining GPUs and the existing 30x0 series. So I've just said that no, I think that quite large quantities of mining-only gpus are needed for nvidia to address this competition.
Then you replied that we can't know how many are needed actually, right?

Thus my reply : you can't hold two conflicting positions at once in a discussion : not many gpus are needed AND we actually don't know how many are needed

 

RTX 3060 hashrates for altcoins

Spoiler

 

I read about this fascinating story from China: someone bought a mining ASIC machine (or machines?) with ~400k CNY (~US$61k). It works well at first, with heat and fans running, and actual mining profits. After a few weeks it suddenly stopped working. No mining profit at all. The buyer opened one of the machine, and find only bricks, fans, and some heating resistors. Basically a fake machine. But where did the mining profits come from? There's a small computer inside the machine, which relays the mining commands (which mining pool to use, which address to send the profits, etc.) to a remote server, where the actual mining rigs are. The seller (or more accurately, scammer) stopped the access to the remote mining rigs after a few weeks to make sure the buyer can't track back to the scammer.

 

My W5700 has finally arrived. Flashed the tweaked bios i found from igorslab. At default it was doing 49.XX MH/s at 145W. After bios flashing, it's 55.XX MH/s at 92W. Temps are not too shabby, Core temp is 50C, and Tj Memory is only 80C at 1750RPM fan speed. It's ridiculously efficient that my RTX 2060 is doing 32.XX MH/s at 95W.
I had to pull a bit mcgyver but managed to fit it in the case with some zipties. Doing 187MH/s currently, UPS's been reporting ~600W but speaker, screen, router et al is included.

Spoiler

 

Looks good ! what riser did you use ?

I went to microcenter to buy my wife a new card. Was online for 30 minutes and I was 17th in line. The only cards they got in were a 6900 which was 2k retail and a 6800 non xt and that was 950. So i decided not to buy lool.