The AMD Execution Thread [2007 - 2017]

I find that incredible hard to believe, the only way that would work is if AMD stock is near non-existent.
Seems more likely that someone looked at previous quarter market share and just applied market share percentage to total sales of the new cards.

 

Newegg reviews don't work like that, you can only post a review if you buy that particular product.

Both Newegg.ca and Newegg.com have similar review results too.

So what we have is pretty much North America, is the rx 480 is very hard to find, or people buying the rx 480 aren't putting up reviews or nV performance cards are selling a lot more (which were supposed to have been supply limited for the first 2 months)

 

Here's the german e-tailer I know of how makes their rough sales numbers public:
http://www.mindfactory.de/Hardware/Grafikkarten+(VGA)/GeForce+GTX+fuer+Gaming/GTX+1060.html
http://www.mindfactory.de/Hardware/Grafikkarten+(VGA)/Radeon+RX+Serie/RX+470.html

You're looking for the line "über xx verkauft".

But always remember, it's ONE datapoint from one e-tailer in one country.

 

AMD had an initial batch that is taking longer to resupply than Nvidia did with their 1070 and 1080s, this was further exacerbated that the cheap 4GB 480 was a small number of rebadged 8gb cards and still some of the largest retailers/distributors are still waiting for that to resupply.
What is holding AMD back now is the lack of the 4GB model and lack of custom AIB models for the 480 (Sapphire and Asus are sort of available but their initial batches as well have not had a good resupply).
Sells are still strong for Nvidia because they have quite a few custom AIB models that are actually available and been so for awhile now (this keeps demand strong after the early demand for reference models) - not all IHV models but quite far ahead of AMD with 480 that only now is starting to see models appear.

Cheers

 

Yeah and that trend is also seen over here in the UK.
Cheers

 

Not saying anying regarding numbers. But crypto-miners at most are going to only leave 1 review (and unlikely to leave a review at that) even if they buy over 50 cards. Reviews are an unreliable, at best, method of ranking sales. You can get a relative feel for which is selling better, but beyond that it's worthless.

Regards,
SB

 

true, but relatively more people are buying nV performance cards, tell ya this right now, either nV being bullish on next quarters numbers or AMD being bullish, one of them is most likely smoking a really big J lol. One of them is really far off the mark.

https://www.amazon.com/Best-Sellers-Computers-Accessories-Computer-Graphics-Cards/zgbs/pc/284822

I don't see any AMD cards let alone rx4xx cards up on Amazon's top sales lists. And this is updated frequently based on what people are ordering. That will take into account a miner buying 50 card.

I don't buy the entire miner thing either, that was the same thing AMD stated with the r29x launch and yeah we saw how the marketshare numbers turned out there.

 

Well, this was interesting.

http://www.amd.com/en-us/press-releases/Pages/major-pachinko-and-pachislot-2016july25.aspx

Pachislot is huge in Japan (larger than console gaming, PC gaming, and handheld gaming combined in Japan). I wonder how many machines it'll actually make it into? Regardless, since it's unlikely that AMD will get a percentage of pachislot revenue, the impact for AMD will likely be significantly smaller than what they get from their console endeavors. Still more revenue is good for a company that is a distant second to both of their direct competitors.

I still find it interesting, however.

Regards,
SB

 

You find it hard to believe NVIDIA cards are massively outselling AMD cards like they always do? I would find it hard to believe if Polaris has changed anything in this regard. Pascal easily maintains the technology gap that began with Maxwell.

 

http://www.design-reuse.com/news/40404/q2-2016-gpu-market-share.html

 

Discrete desktop now stands at 70% NV, 30% AMD, likely influenced by NV clearing stocks.
http://jonpeddie.com/publications/add-in-board-report

 

So is this possibly the 3rd win Lisa Su mentioned in 2015

.
Comes down to interpreting whether the Neo or Scorpio are considered wins or part of a revised contract I guess, or whether one thinks that was the Nintendo NX but Pachinko is more likely to provide billions in revenue (yeah appreciate this is very loose/vague figure and is probably summed) than the NX IMO, and the NX also has more rumours-leaks tied to Nvidia for now.

Would make sense this is the 3rd win and Nvidia has the NX IMO.
Cheers
Edit:
To clarify, the Lisa Su statement in 2015 was what triggered a lot of news generation-speculation that it was the Nintendo NX.

 

As per the press release it's using a pre-existing carrizo die APU, not semi-custom silicon.

 

Well you had the financial 2014 announcement that was subtly different to the Lisa Su 2015 announcement that were subtly different to this announcement, all pointing to 2016 gaming win, so not sure if the announcements could be deemed to be completely accurate from a projection perspective.
Yeah thinking about it does seem more likely the Pachinko deal did come from left field and was a very recent thing and nothing to do with any previous announcements *shrug*.
Just me over thinking it

Cheers

Edit:
Just remembered and checked.
And again Lisa Su April 2016 announcement is subtly different to the 2014 and 2015 one.
This time it mentioned 3 semi-custom design wins, stressing it is not the existing consoles; so that looks to be the XBox One Slim, Neo, Scorpio, especially as that announcement mentioned one would be 2017 and we know Scorpio is coming out then.
So there is possible leeway with the projections in the earlier announcements, although even allowing for such leeway as most of you say the 2015 announcement would probably still not be the Pachinko.

 

I know for a fact that no casino game machines never use semi-custom hardware because there's just no need for it. AMD Steamroller/Excavator embedded APUs are actually pretty popular within western casino gaming machines because of their decent GPU performance and price compared to similar offerings from Intel, and power consumption isn't that important either. Having them going into Pachinko machines is most probably just an extension of what's already happening in the western solutions.

 

http://techreport.com/news/30551/amd-claws-back-some-discrete-gpu-market-share

 

A relevant comment from Tech Report gives an interesting contrarian data point to these Mercury and Peddie reports by looking at monthly Steam survey results. These aren't quarterly sales, these are user system reports, so these numbers tend to have a much smoother trend, essentially low-pass filtered. They also only report from Steam, which is self-selected to be gamers: performance oriented users. And finally, there's one month lag in these reports, so the July report doesn't even include any RX 480 shipments, and only a fraction of a percent of systems report GTX 1070/1080 (which had been shipping for less than one month). These Steam numbers will be much more interesting when we see the September or October report, which will show the first rush of RX 480 and GTX 1060 adoptions.
The 2016 popularity results:
:: : AMD : NVDA
JAN: 26.2% 54.8%
FEB: 26.0% 55.8%
MAR: 25.5% 56.4%
APR: 25.4% 56.6%
MAY: 25.5% 56.6%
JUN: 25.1% 56.7%
JUL: 24.7% 57.0%

 

Steam hasn't been properly detecting AMD graphics cards for years, besides:

And a whopping 20% of these performance oriented users are using Intel graphics? And 30% of the whole GPUs cake isn't even DX12 capable?
4th most popular GPU is Ivy Bridge's HD 4000, followed by "Intel Haswell" (it's not even detecting the GPU), followed by a Geforce GT 720M which is a Fermi GPU that somehow Steam says it's DX12 capable.

And then we see 3% less DX12-capable GPUs in July than there were in May, even though there are practically zero pre-DX12 GPUs in the market nowadays?


Perhaps it's time to acknowledge what the Steam Survey in the GPU part really is: a giant non-reliable mess.
Valve certainly doesn't care about solving that crap so I don't know why anyone would waste time with those results.

 

That doesn't tell the entire story, there was a drop of 20% in volume sales across the entire quarter, which the brunt of that was taken by nV. AMD hasn't sold more cards, they just maintained or a slight uplift of sales, combine with the loss of sales (volume) nV covers it. * the reason for total sales loss is quite simple, Pascal and left over stock of Maxwell, if you start going through the line items of AMD and nV quarterly reports, its very clear that nV's margins improved and that covered the loss of sales and then some. But AMD on the other hand... well ya get what I'm seeing. No change in margins, not that much change in the line items, but of course having all computing as one line item, its all mucky.

I would like to ask this question not sure if its too blunt but here it is.... How smart is it for a CEO to mix and match numbers from two different independent sources that use two totally different methods to calculate market share of GPU and use those numbers to show share holders? Then when we get the numbers we can look at their financial sheets, and then figure out what is really going on.

One is shipped and the other is AIB cards sold.

I find the reporting on this a bit lackadaisical.

I have only read one article thus far that has mentioned the drop in total AIB sales, and that is Hexus.

 

Q2 is always the weakest quarter.