AMD: Speculation, Rumors, and Discussion (Archive)

Status
Not open for further replies.
I think there's a point to be made with your starting premise, but adjustments are required. I'm not going to rewrite your argument for you though when it is so apparently hasty.

For what it's worth I suspect, but have no evidence, that NVidia's per chip prices were more solid. And perhaps NVidia's margin on those chips has improved over the lifetime. etc. I just think you over-egged your position.

The recently posted "failure lab" had the engineer referring to, I think, "10 million parts a month". 7.5 million "VR capable" chips sold over a 2 year period implies 3% of NVidia's parts, in the absolute best case. I'll let you work that into your agument.
 
hell no its going to be around $300, lowest $200, There is no way its going to go less than that, and its all going to come down to where the 1070 is priced.

This card if it doesn't use GDDR5x, which seems likely now I don't see it going higher than $300.
 
hell no its going to be around $300, lowest $200, There is no way its going to go less than that, and its all going to come down to where the 1070 is priced.

This card if it doesn't use GDDR5x, which seems likely now I don't see it going higher than $300.
$300 is literally no cheaper than the current "VR capable" GPUs. So, that would be a complete fail at "making VR capable graphics more accessible".

Of course, one has to wonder whether the cost of VR headsets themselves obviates the value proposition of the graphics card... Maybe AMD dreampt up the mainstream VR GPU when they were under the impression that VR headsets would be $3-400 :eek:
 
Good point didn't think about that. But there is some lead time for VR games, so possible a bit lower don't see going down too much.
 
The recently posted "failure lab" had the engineer referring to, I think, "10 million parts a month". 7.5 million "VR capable" chips sold over a 2 year period implies 3% of NVidia's parts, in the absolute best case. I'll let you work that into your agument.
I did a double take at that figure. Jon Peddies number for total dGPUs for 2015 was 50 million, (with 5.9 million being $300 and up "enthusiast" level GPUs, which fits reasonably well with 7.5 million VR-capable GPUs.)
 
It's often claimed that the real money is in the mid-end, but $1.5B is nothing to sneeze at. (It's probably larger than AMD's total GPU revenue.) And probably much higher margin as well, and profit driver as well.
I think this has changed for at least 1, maybe 2 years and I would be very surprised if the market analysts at both large manufacturers didn't also think of this.

Reason as follows: Seeing what a 4-5 year old PC with a then-at-least-decent CPU can do today, there's no reason to invest in something other than the graphics card and maybe the RAM to have a fully functioning gaming machine still in your home. Take an i5-2500 or i5-3470, both absolutely non-high-end, but still very valid when paired with a fast graphics card, especiall in higher resolutions of 1080p and up.

I think an increasingly high percentage of PC gamer realize that they need to upgrade their mediocre gaming card from 4+ years back (maybe a GTX 460 or HD 6850). The people who only buy a new PC and get ripped off with high megabyte and core numbers have ventured on to consoles for gaming and use a tablet or laptop for other stuff.

That's why the GTX 970, despite their ...err... oneusly communicated memory/L2/ROP configuration was such an important product for Nvidia at the timeframe: It hit just the right price point people were willing to invest in order to keep their rigs viable for gaming for another 2-3 years at least.

So, is this going to be a $150-200 card that can do VR?
249 US-$ comes to mind as an attractive and oftentimes used price point. It leaves both room for maneuvering as well as products above and below.
 
I think people are underestimating Polaris 10 a bit here. Sure, there will probably be around $250 model, but AMD is literally trying to cover whole market with Polaris 10 and 11, which should mean that "full Polaris 10" will fit higher, maybe not the 980 Ti level suggested by some rumour, but still notably higher than GTX 970/R9 290
 
I think people are underestimating Polaris 10 a bit here. Sure, there will probably be around $250 model, but AMD is literally trying to cover whole market with Polaris 10 and 11, which should mean that "full Polaris 10" will fit higher, maybe not the 980 Ti level suggested by some rumour, but still notably higher than GTX 970/R9 290

My interpretation of their rhetoric and road map is that AMD is at least temporarily ceding the high end to Nvidia.

-Talking up perf/W instead of absolute performance.
-Positioning against 960 -970 range, TAM, VR min spec pitches.
-Vega. Amd doesn't have the HPC and professional market share foothold to justify a non-gaming chip like GP100, and 2017 is too soon to replace Polaris, so it seems like Vega is to address the high end.
 
Last edited:
My interpretation of their rhetoric and road map is that AMD is at least temporarily ceding the high end to Nvidia.

-Talking up perf/W instead of absolute performance.
-Positioning against 960 -970 range, TAM, VR min spec pitches.
-Vega. Amd doesn't have the HPC and professional market share foothold to justify a non-gaming chip like GP100, and 2017 is too soon to replace Polaris, so it seems like Vega is to address the high end.
Yes, Vega will obviously be high end, but you need to remember that the "high end" moves higher every generation, too.
My guess is Polaris 10 is Hawaii in terms of CU counts etc, but with updated architecture of course, and it will address the market from 970/290-level min to around Fury X / 980 Ti level or bit under, spreading over 3 or even 4 models, while Polaris 11 would cover anything under VR spec
 
Yes, Vega will obviously be high end, but you need to remember that the "high end" moves higher every generation, too.
My guess is Polaris 10 is Hawaii in terms of CU counts etc, but with updated architecture of course, and it will address the market from 970/290-level min to around Fury X / 980 Ti level or bit under, spreading over 3 or even 4 models, while Polaris 11 would cover anything under VR spec

Your performance projections seem reasonable to me, though I'm just speculating on which market segments I believe AMD is addressing (or rather, not addressing). I think silent guy's interpretation of Roy Taylor's quote is correct--it's a tacit admission that AMD doesn't have anything big(ish) and fast, and it wasn't spun very artfully either. What I'm imagining is a situation roughly analogous to if this had happened at 28nm:

Nvidia launches GK104, GK106, and GK100.
AMD launches Pitcairn and Cape Verde.
 
Your performance projections seem reasonable to me, though I'm just speculating on which market segments I believe AMD is addressing (or rather, not addressing). I think silent guy's interpretation of Roy Taylor's quote is correct--it's a tacit admission that AMD doesn't have anything big(ish) and fast, and it wasn't spun very artfully either. What I'm imagining is a situation roughly analogous to if this had happened at 28nm:

Nvidia launches GK104, GK106, and GK100.
AMD launches Pitcairn and Cape Verde.
Err, AMD launched Tahiti too (in fact, it was launched couple months before Pitcairn and Cape Verde)
 
Total Addressable Market, i.e. potential sales.

I think people are underestimating Polaris 10 a bit here. Sure, there will probably be around $250 model, but AMD is literally trying to cover whole market with Polaris 10 and 11, which should mean that "full Polaris 10" will fit higher, maybe not the 980 Ti level suggested by some rumour, but still notably higher than GTX 970/R9 290
Are they? Or is Polaris the new line-up replacing R9 300 series with Fiji still sitting on top, waiting to be reliefed by Vega?
 
Err, AMD launched Tahiti too (in fact, it was launched couple months before Pitcairn and Cape Verde)

That's what I'm saying. Imagine if this time, they didn't. See what I mean?

Also AFAIR GK100 was much later, that may not be true of GP100, which is why I threw it in.
 
Last edited:
Total Addressable Market, i.e. potential sales.


Are they? Or is Polaris the new line-up replacing R9 300 series with Fiji still sitting on top, waiting to be reliefed by Vega?


Yeah this is most reasonable because if they get the performance up to the 980ti, the Fury line is dead. There would be no need to still be making them.
 
Status
Not open for further replies.
Back
Top