Hmm my guessSo Nvidia will leave the $150(or below) market to AMD, you also said that the 460(4GB) is DOA or something to that effect? Will be interesting to see how people react to a ~180$ 1050, assuming there's no 1050Ti.
4GB is future proof but 3GB is not
Hmm my guess
1050Ti (1060 3GB-1 SMX) for $200
1050 further cut down Pascal for $160
GTX950 $120-140
And then the sub $100 Nvidia rebrand nightmare begins with
GT940= rebranded Maxwell 384 cuda cores for $100
GT930=rebranded Kepler 384 cuda cores GT 740(GTX 650) for $80
You seem to put forth a mocking tone on this but to me it's a huge difference. 4GB is the RAM amount of every recent mid range product (290, 970, 980) as well as many lower-end products (380X, 4GB 960s) which means it is not out of consideration that developers target that amount. 3GB is the amount of older cards (280X, 780, etc), so 3GB is basically a generation behind 4GB. That is a big deal.
AMD has a difficult road ahead as Polaris has been disappointing.Whether the problems with Polaris are design or process issues related remains to be seen.
Maybe you're disappointed, though I don't know why. The 470 and 480 top the charts on performance/$.
Add in massive savings on Freesync over Gsync and the fact you can mine (should you choose) and have the card pay for itself and the AMD cards are great choices under $275.
Maybe you're disappointed, though I don't know why. The 470 and 480 top the charts on performance/$.
Add in massive savings on Freesync over Gsync and the fact you can mine (should you choose) and have the card pay for itself and the AMD cards are great choices under $275.
Competition is always good for the consumer. What we need is competing products at all price points from Nvidia and AMD. Right now Nvidia is making a killing with USD 400 - USD 450 GTX 1070, USD 650 - USD 700 GTX 1080 and USD 1200+ GTX Titan. AMD has a difficult road ahead as Polaris has been disappointing.Whether the problems with Polaris are design or process issues related remains to be seen. It looks like AMD and Nvidia might both use Samsung for few of their GPU products. So we are likely to see a faster ramp to FINFET GPUs than 28nm Kepler (GTX 600 series) or 28nm GCN (HD 7000 series). This is good for the consumer. What we as consumers need is faster Polaris revisions which fix the clock/power issues and a strong Vega lineup so that the consumer gets the best bang for buck in 2017 irrespective of which GPU vendor he chooses.
I will take a guess. 640 sp, 128 bit GDDR5, 32 ROPs, 112 - 128 GB/s, 45-60w, 100 sq mm. Performance slightly better than GTX 960. This chip will cause AMD a lot more pain.
Vega isn't Polaris. We'll have to see how it's efficiency is.Do i have to state the obvious. perf/watt and the ramifications for the higher end Vega GPUs which are going to be power limited to 250w-275w(max) .
Actually its 970 performance for $280.This is why its disappointing.You continually post Polaris has been disappointing. Disappointing to whom? Except for power usage it's exactly as promised. Is that what you are talking about? Besides that it's selling well and performs exactly where AMD targeted it to (~970 performance for $200). Even better in DX12 and Vulkan.
Actually its 970 performance for $280.This is why its disappointing.
So basically same performance as a 2 yr old GPU but only $50 cheaper.That really paints the entire picture.
He's upset because he's looking at this from a technical achievement perspective, not from a product perspective. Price anything low enough and it'll be a good deal to some.
You continually post Polaris has been disappointing. Disappointing to whom? Except for power usage it's exactly as promised. Is that what you are talking about? Besides that it's selling well and performs exactly where AMD targeted it to (~970 performance for $200). Even better in DX12 and Vulkan.
Add to that the only time nVidia releases product at a good perf/$ ratio is when they have to target an AMD product. Looking at the 1060 I'd say it was successful at that as well.
Vega isn't Polaris. We'll have to see how it's efficiency is.
imo AMD way overhyped Polaris with 2.5x perf/watt improvement claims and we all now know it was plain lies. AMD now resembles a company which has only hype and marketing to sell what are basically unimpressive products.
And that can be said for NVidia as well. Where is the async compability for Maxwell, the slides that NVidia used to promote their 900 series products included async compute, oh and what about the 4GB 970 fiasco? Don't tell me NVidia are the good guys here.
You continually post Polaris has been disappointing. Disappointing to whom? Except for power usage it's exactly as promised. Is that what you are talking about? Besides that it's selling well and performs exactly where AMD targeted it to (~970 performance for $200). Even better in DX12 and Vulkan.
Add to that the only time nVidia releases product at a good perf/$ ratio is when they have to target an AMD product. Looking at the 1060 I'd say it was successful at that as well.
Vega isn't Polaris. We'll have to see how it's efficiency is.
If you use the worst performing game and highest priced card sure then it looks bad. Overall it is a very good buy, there is a reason its constantly sold out.
I don't think its about being the good guy or being the bad guy. The real issue here is that clearly AMDs smaller R&D funding is being exposed or in fact shows just how much money is required in this industry to be relevant.
And thats his point because the GCN architecture is getting very old in the tooth e.g. it still uses gen 1 rasterisers even in Polaris! Their shader core hasn't seen any dramatic changes for a lack of a better term. Most of it has seen small optimizations or tweaks hence why we don't really see better perf/watt as opposed to nVIDIA. In the long term this could be a problem because if Volta completely negates the so called "DX12/Vulkan" advantage, what can AMD do (while being more power efficient)? In the meantime, nVIDIA has had the luxury of releasing several GPU architectures that outperform their competition during its lifetime and has successfully done it each generation regardless of the longevity equation.
You continually post Polaris has been disappointing. Disappointing to whom?
Then by that analogy Pascal is a failure. The only thing it brings to the table is a shrunk Maxwell with higher clock speeds, a tad faster RAM, dynamic scheduling and preemption. But it lacks true hardware async compute and does not see as big of a performance boost in DX12/Vulkan as modern AMD cards do. So from a technical acheivement standpoint I can wholeheartedly say that am not impressed. Pascal also costs more than prior generation products in the same family, GTX1070 vs 970, 330$<380$(In reality 450$) and GTX 1080 vs 980, 550$<600$( In reality 700+$).
if Volta completely negates the so called "DX12/Vulkan" advantage, what can AMD do
Do you have inside info on that reason? You insinuate it's because of demand, do you have any proof?
Where's the $500 performance for $200? Isn't that what I heard AMD promise?
https://forums.overclockers.co.uk/showpost.php?p=29795238&postcount=516
https://forums.overclockers.co.uk/showpost.php?p=29840469&postcount=1436
https://forums.overclockers.co.uk/showpost.php?p=29714968&postcount=69
I'm sorry I forgot where you provided any sources or proof yourself.
Well considering it woops the 980 in DX12 or Vulkan, right there.
In the long term this could be a problem because if Volta completely negates the so called "DX12/Vulkan" advantage, what can AMD do (while being more power efficient)?