Miners might target Nvidia Maxwell next

[antihelten]

With all this talk about Jon peddie and what not, it is important to note that the current spike in the prices of AMD card in the US and the lack of stock, largely coincided with the rise in price of litecoin.

Now assuming that AMD increased their supply of cards as a response, the effect of this wouldn't be apparent until Q1 of 2014, for which we don't currently have any data.

Thus the fact that Jon Peddie data for Q4 2013 doesn't show any significant increase in AMD sales is hardly surprising, as the effect of mining wouldn't become apparent until Q1 2014.

 

[Keysplayr]

But wait. How long have people known about and have been mining with AMD cards?

 

[antihelten]

But wait. How long have people known about and have been mining with AMD cards?

For a long time obviously, however there was a bit of a lull in GPU mining of bitcoin due to the arrival of ASICs, and it arguably wasn't until the explosive rise in Litecoin price last november that GPU mining became big again.

This can also be seen from the fact that there actually was a (short) period were the new hawaii cards were priced more or less at MRSP in the US not to mention the cheap 280x's, which wouldn't have been the case if the viability of GPU mining had been constant up till now.

 

[KingFatty]

Yeah what really makes me think is the disparity/contradiction between the data that is collected, and the behavior of the after-market pricing for AMD cards.

If the crazy prices were merely from low supply of new cards, that would only affect the new cards being released and used cards would be at the same price.

But because used cards are *also* at ridiculous prices, then there is probably a parallel situation between new and used cards, where demand is really driving things.

So the Peddie data is just not up-to-date enough to reflect this? It seems to show the opposite of this?

 

[daxzy]

But wait. How long have people known about and have been mining with AMD cards?

Since 2011, when the Radeon 5xxx series were the first monster hashers for bitcoins.

 

[Keysplayr]

Since 2011, when the Radeon 5xxx series were the first monster hashers for bitcoins.

That's a really long time now.

 

[ocre]

With all this talk about Jon peddie and what not, it is important to note that the current spike in the prices of AMD card in the US and the lack of stock, largely coincided with the rise in price of litecoin.

Now assuming that AMD increased their supply of cards as a response, the effect of this wouldn't be apparent until Q1 of 2014, for which we don't currently have any data.

Thus the fact that Jon Peddie data for Q4 2013 doesn't show any significant increase in AMD sales is hardly surprising, as the effect of mining wouldn't become apparent until Q1 2014.

Absolutely I am interested in the Q1 data. But the latest data we got from Q4 is all we got. The AMD prices did go up in this time period though, so its not completely irrelevant. But i agree that the Q1 data is very important.

What is interesting about the Q4 results is that AMD lost marketshare and volume. This cannot be ignored. If there was a huge boost in mining card sales towards the end of the quarter, AMD should have had stock to spare as they ended the quarter slipping. Yet across the product stack, across the board prices went up on AMD GPUs and they didnt even sale as many cards as last Q. This is Q4 which usually is up from Q3. So no matter how you slice it, something is a miss.

My gut says that there could have been a supply issue in play. Could have been related to the console obligation, maybe or maybe not. Also I believe that their was an increased interest in Mining during that period. I believe these too things caused the prices to go up.

But Q4 is not enough data by itself. I would love to see Q1 but that data is not available yet. Without it, the conclusion is premature I think. But it is all we got. I really didnt expect it things to be like this. We started seeing articles suggesting that the AMD price hikes where due to supply and not massive demand. The Q4 results seem to support this. It will take Q1 results to completely confirm or debunk this.

 

[3DVagabond]

They were likely predicting a down trend. They also had old stocks to clear. If I had to guess they were caught with their pants down by the mining surge and didn't have the inventory to take advantage.

I thought that I had read somewhere that TSMC had excess manufacturing capability though. They were somewhere in the 60% range on capacity. I don't remember the article though and would like to look back again to be certain, but I think I recall that figure.

When the shortage first hit Dave Baumann tweeted he was going to China to see if he could increase supplies. I never heard anymore on that. IF he was successful it takes something like 10 weeks to produce wafers (someone correct me if I'm wrong), then ~1mos in manufacturing, and if sent by boat, ~2-4wks, depending on stop overs, to ship. That would put us to somewhere around now for increased supply if there is any relief coming from Dave's visit.

The "sweet spot" model is the 280X, and that was the chips he was trying to get more of. Maybe??? the reason for the recent 280 nonX announcement is that they upped production of Tahiti chips and they have the stocks of the less than perfect chips to use, as well? Or, maybe it's just that through regular production that there's finally enough culled chips to release another model.

This is all speculation on my part, but if the 280 nonX drops and we have supplies of 280X's as well, that might be what has happened.

 

[ocre]

regardless. It would indicate that the surge in GPU mining isnt that big. Amd that GPU mining is not bigger than the traditional GPU markets. That is all i set to prove. And i think i have done that

 

[TestKing123]

regardless. It would indicate that the surge in GPU mining isnt that big. Amd that GPU mining is not bigger than the traditional GPU markets. That is all i set to prove. And i think i have done that

No you have not. This very website (anandtech.com) even acknowledges the pricing pressure miners (at the individual and group level) have put on AMD cards, except you for some reason. Misquoting, selectively quoting irrelevant figures or ignoring the proper metrics that obviously shows this trend isn't going to disprove this.

 

[KingFatty]

Do you have another lead on any data sources, besides the Peddie data? I even tried searching a form of LexisNexis, but kept getting results pointing to articles writing about the recent Peddie data.

I guess Peddie has a monopoly on that kind of data analysis, and it's got some holes in it to where it's impossible to quantify what share of the market that miners have, and what they've purchased as a percentage of the discrete video card market?

 

[ocre]

No you have not. This very website (anandtech.com) even acknowledges the pricing pressure miners (at the individual and group level) have put on AMD cards, except you for some reason. Misquoting, selectively quoting irrelevant figures or ignoring the proper metrics that obviously shows this trend isn't going to disprove this.

no one is arguing that miners haven't driven the prices up. the argument is that that there hasnt been such a large surge in mining to the point that miners out number gamers. Perhaps you should pay closer attention. also want to address the irrelevant figures claim. that's so funny. How is posting data about cards sold and marketshare irrelevant when we are discussing a supposed massive surge in miner gpus sold? Can't get anymore relevant, sorry.

 

[antihelten]

Also want to address the irrelevant figures claim. that's so funny. How is posting data about cards sold and marketshare irrelevant when we are discussing a supposed massive surge in miner gpus sold? Can't get anymore relevant, sorry.

Because the theoretical massive surge in miner GPUs couldn't really have happened any earlier than Q1 2014, which is about the earliest AMD would have been able to increase the supply to accommodate the increased demand (considering that they would have to order extra wafers, and then go through the whole production of the extra GPUs which doesn't happen overnight).

The only way the surge in gpu mining could have had a significant effect on GPUs sold in Q4 is if AMD by chance had a large number of GPUs in stock in anticipation of a rise in sale. However to my knowledge no semiconductor company stockpiles products on the off chance that a sudden surge in sales might happen, the only case in which a company is sitting with extra stock is usually if a product doesn't sell as well as expected (like what happened with Llano).

So the only thing the Jon Peddie data for Q4 2013 really tells us that AMD didn't have a large stockpile of GPUs sitting around, but we pretty much knew that already.

The interesting question currently is whether or not AMD have increased their supply in reaction to the demand from mining, since there are a number of reports talking about problems with securing certain components necessary for the GPUs, but we won't really know until the data for Q1 2014 comes out.

 

[Will Robinson]

Right now I imagine AMD is shipping every fast mining chip it can produce.
That's an enviable position to be in.
If they weren't selling out the prices would fall dramatically.

 

[KingFatty]

the argument is that that there hasnt been such a large surge in mining to the point that miners out number gamers. Perhaps you should pay closer attention.

And just for a bit of a clarification, it's about the number of GPUs bought by miners vs. gamers. One miner could buy hundreds of GPUs, even though he's just one person. But one gamer is unlikely to buy GPUs by the hundreds, and usually buys a single GPU and sometimes 2.

 

[tweakboy]

How about SkyLake gl