rx480 best gpu under $200 ?

[othertomperson]

2 minutes ago, stconquest said:

...and a GTX 980 (brand new) for $200 is a great deal.

Does your mind get blown every time we see a generational change? "The same thing we had before... FOR CHEAPER! OMG!"

[stconquest]

1 minute ago, othertomperson said:

Does your mind get blown every time we see a generational change? "The same thing we had before... FOR CHEAPER! OMG!"

More performance per dollar is exactly what I expect to see from each new generation.  What the fuck are you on about?

[othertomperson]

1 minute ago, stconquest said:

More performance per dollar is exactly what I expect to see from each new generation.  What the fuck are you on about?

It clearly isn't because if it were you'd see this as wholly unremarkable. If this card were anything less than 390 level performance at a 380 price point something would be seriously wrong.

 

Why do you think Nvidia opened with GP100 even though that chip was likely never to be seen in a consumer graphics card at all?

[stconquest]

1 minute ago, othertomperson said:

It clearly isn't because if it were you'd see this as wholly unremarkable. If this card were anything less than 390 level performance at a 380 price point something would be seriously wrong.

 

Why do you think Nvidia opened with GP100 even though that chip was likely never to be seen in a consumer graphics card at all?

I fail to remember me saying that this launch was remarkable in any way...

 

Please, tell me all about why Nvidia opened up with Tesla.  I care so much.  You are on it today.

[othertomperson]

4 minutes ago, stconquest said:

I fail to remember me saying that this launch was remarkable in any way...

 

Please, tell me all about why Nvidia opened up with Tesla.  I care so much.  You are on it today.

You're going on about how a 980's performance, with a 980's TDP (even though this is on a much smaller process and should really be blowing the 980 out of the water here, but it isn't) for more than a 380 is a great deal. The only revelation here is that the x80 tier card is $20 more than it was last generation.

 

And sure thing! Nvidia opened with GP100 because they wanted to show what Pascal could do. They wanted to show HBM2.0, their fp64 performance. Then they launched their second tier GPU to show that this was even more powerful than what they had before. Whereas AMD went with something that performs like the second hand 290Xs you can get off of ebay for less money. These launches shouldn't be about just selling low end stuff, they should be about generating hype and excitement and this card just doesn't do that. Especially coming on the heel of Pascal.

[stconquest]

13 minutes ago, othertomperson said:

You're going on about how a 980's performance, with a 980's TDP (even though this is on a much smaller process and should really be blowing the 980 out of the water here, but it isn't) for more than a 380 is a great deal. The only revelation here is that the x80 tier card is $20 more than it was last generation.

 

And sure thing! Nvidia opened with GP100 because they wanted to show what Pascal could do. They wanted to show HBM2.0, their fp64 performance. Then they launched their second tier GPU to show that this was even more powerful than what they had before. Whereas AMD went with something that performs like the second hand 290Xs you can get off of ebay for less money. These launches shouldn't be about just selling low end stuff, they should be about generating hype and excitement and this card just doesn't do that. Especially coming on the heel of Pascal.

I see.  How about you form a real argument. 

 

So you don't agree the direction AMD has taken.  That has nothing to do with me.

 

Great information about GP100 there.  That really changes my whole outlook.

 

If you have not guessed, I am trying to figure out what point you are trying to make. 

 

I never stated this launch of Polaris was anything amazing.  This "seems" to be your whole shtick with me.  So you are wrong. 

[Dackzy]

37 minutes ago, othertomperson said:

You're going on about how a 980's performance, with a 980's TDP (even though this is on a much smaller process and should really be blowing the 980 out of the water here, but it isn't) for more than a 380 is a great deal. The only revelation here is that the x80 tier card is $20 more than it was last generation.

 

And sure thing! Nvidia opened with GP100 because they wanted to show what Pascal could do. They wanted to show HBM2.0, their fp64 performance. Then they launched their second tier GPU to show that this was even more powerful than what they had before. Whereas AMD went with something that performs like the second hand 290Xs you can get off of ebay for less money. These launches shouldn't be about just selling low end stuff, they should be about generating hype and excitement and this card just doesn't do that. Especially coming on the heel of Pascal.

Just gotta correct you the 980 TDP is 165.

[BrightCandle]

We used to get clock speed scaling, then we hit a thermal wall where voltage was increasing power consumption in a square relationship. Nowadays all chips are limited by thermal dissipation, you can't have a CPU more than about 140W without extreme cooling, you can't have a GPU more than 300W and 250W is definitely preferred to ensure its quieter.

 

TDP = noise and it is also limited by the power delivery mechanisms setup by the PSU, its not OK to just consume whatever you want and be inefficient as all modern design depend on efficiency as its their limiting factor more so than die size.

 

A 1070 is > 980 ti performance at 150W

A RX 480 is 390 performance at 150W

 

Making the 1070 much more efficient, 30-50% more fps per watt.

 

If you play this out to the high end cards then if you go to 600mm^2 die (like a 980 ti) then Nvidia gets to 315 watts all else being equal. Drop the clocks a bit and a 600 mm^2 die is possible with a similar but much larger design. If you do the same calculation for the much smaller RX 480 however to a 600mm2 die (a Fury) then its 420 watts, that isn't going to be possible, that is dramatic clock speed reductions and an uncompetitive card as a result expensive and slow. So it looks like unless AMD solves its power efficiency problem they are going to be limited to about 400mm^2 on 14nm with this design and that will pull 300 Watts, ie be a hot and loud card like the Fury was. Nvidia on the other hand will go all the way to 600mm^2. Not only that but Nvidia will be at 1.6Ghz and AMD will be around 1.1Ghz and clock speed is a big heat producer.

 

These cards don't play in the same space of performance, they aren't even remotely the same die size and AMD isn't going to compete on performance. That isn't the point, its just this card looks far too hungry for what it is, it should be 100W for that performance level but something hasn't scaled well for them and this is actually a bad omen because it indicates that we might see problems with thermals for the more powerful cards.

 

[stconquest]

14 minutes ago, BrightCandle said:

 ...it should be 100W for that performance level but something hasn't scaled well for them and this is actually a bad omen because it indicates that we might see problems with thermals for the more powerful cards.

I get this... we will have to wait to see more of these GPUs to have a better understanding.

 

But where do you get this:

 

14 minutes ago, BrightCandle said:

A RX 480 is 390 performance at 150W

 

[othertomperson]

49 minutes ago, stconquest said:

I see.  How about you form a real argument. 

 

So you don't agree the direction AMD has taken.  That has nothing to do with me.

 

Great information about GP100 there.  That really changes my whole outlook.

 

If you have not guessed, I am trying to figure out what point you are trying to make. 

 

I never stated this launch of Polaris was anything amazing.  This "seems" to be your whole shtick with me.  So you are wrong. 

If you can't even understand the very simple English I am using, why do you keep responding and arguing? If you can't be mature enough to have a conversation about a fucking graphics card without assuming the other person has a "shtick with" you, then you probably shouldn't be on the Internet. 

[BrightCandle]

6 minutes ago, stconquest said:

I get this... we will have to wait to see more of these GPUs to have a better understanding.

 

But where do you get this:

 

 

Its a 5 tflop card with 256GB/s bandwidth, that puts it in and around the 390 which is a 5.1 tflop part with 364GB/s bandwidth. We don't have a lot more to go on currently but AMD is talking in all the same terms in CUs and such with this card and IPC has already been counted into this calculation so I am kind of basing it on that. They could be better or worse than that but it seems the most likely competition to this card and it also fits very well with the sort of scaling they ought to get in terms of transistors to produce the SP count they have

[stconquest]

 

38 minutes ago, othertomperson said:

If you can't even understand the very simple English I am using, why do you keep responding and arguing? If you can't be mature enough to have a conversation about a fucking graphics card without assuming the other person has a "shtick with" you, then you probably shouldn't be on the Internet. 

 

1 hour ago, othertomperson said:

Does your mind get blown every time we see a generational change? "The same thing we had before... FOR CHEAPER! OMG!"

 

See, what happens when you directly insult someone... they get pissed at you.  You assumed something about me and phrased a derogatory statement, directed at me.  This how you have an argument, not a conversation. 

 

I know I am knit-picking, but you offered nothing in regard to my question to BrightCandle.  That is where you felt like jumping in, but did not know how to deal with that particular question...  maybe because you are not BrightCandle, maybe because you don't understand simple english.

 

 

43 minutes ago, BrightCandle said:

Its a 5 tflop card with 256GB/s bandwidth, that puts it in and around the 390 which is a 5.1 tflop part with 364GB/s bandwidth. We don't have a lot more to go on currently but AMD is talking in all the same terms in CUs and such with this card and IPC has already been counted into this calculation so I am kind of basing it on that. They could be better or worse than that but it seems the most likely competition to this card and it also fits very well with the sort of scaling they ought to get in terms of transistors to produce the SP count they have

Thanks for clarifying.

[othertomperson]

2 minutes ago, stconquest said:

 

 

 

See, what happens when you directly insult someone... they get pissed at you.  You assumed something about me and phrased a derogatory statement, directed at me.  This how you have an argument, not a conversation. 

 

I know I am knit-picking, but you offered nothing in regard to my question to BrightCandle.  That is where you felt like jumping in, but did not know how to deal with that particular question...  maybe because you are not BrightCandle, maybe because you don't understand simple english.

 

 

Thanks for clarifying.

How the ever loving fuck did you manage to take an approximation of the 980's TDP as an insult?

 

Once again, if you are so thin skinned that you are crying about the TDP of a 28nm card, go home and get your porcelain arse off of the Internet.

[stconquest]

1 minute ago, othertomperson said:

How the ever loving fuck did you manage to take an approximation of the 980's TDP as an insult?

 

Once again, if you are so thin skinned that you are crying about the TDP of a 28nm card, go home and get your porcelain arse off of the Internet.

Wow, again.  You just can't stop. 

 

I am not thin skinned.  I just can't say what I want to say to you you *************.

[othertomperson]

2 minutes ago, stconquest said:

Wow, again.  You just can't stop. 

 

I am not thin skinned.  I just can't say what I want to say to you you *************.

You're now swearing at me because I said something you didn't like about a graphics card. Sounds pretty thin skinned to me.

[stconquest]

1 minute ago, othertomperson said:

You're now swearing at me because I said something you didn't like about a graphics card. Sounds pretty thin skinned to me.

gud

[TheRandomness]

2 hours ago, stconquest said:

...and a GTX 980 (brand new) for $200 is a great deal.

*cough* nvidigimp *cough*

If that does happen..

[stconquest]

1 minute ago, TheRandomness said:

*cough* nvidigimp *cough*

If that does happen..

I am hopeful for that little P10.

[BrightCandle]

I see this particular behaviour pattern a lot on forums so I am going to call it out for what it is. Whenever someone points out a potential weakness in a product, and its almost exclusive when attacking an AMD product but it happens for Nvidia but less often, the "fan" tries to derail the thread by inciting as many people as possible to try and get the thread locked. Its a common tactic, I have seen it soooo often on Anandtech and they take the bait from the trolls so often there its literally impossible to have a conversation about both vendors cards. Heck even just talking about one card and noting it does something badly is enough to just kill the entire thread and get a load of people ban holidays.

 

You can't get away from and I just report the ad hominen's and general attacks, these sorts of individuals ruin forums if you let them run rampant for too long, you just can't have a conversation around them.

 

Anyhow one other thing I found quite interesting is that for the size and performance estimate it looks like AMD is getting almost precisely the same scaling factor as Nvidia. Its within 1%. That is quite interesting considering the presumably large differences in the 14nm process they are using at Samsung verses the 16nm at TSMC. We wont know for certain until AMD releases actual transistor counts but I found that deriving both from the CU and transistor count and using the Nvidia transistor scaling from the 980 to produce the same result and that was interesting. So AMD got the same sort of transistor density scaling but a lot less power reduction, this could be quite an important difference in the processes themselves rather than anything AMD did wrong in the design.

[stconquest]

8 minutes ago, BrightCandle said:

I see this particular behaviour pattern a lot on forums so I am going to call it out for what it is. Whenever someone points out a potential weakness in a product, and its almost exclusive when attacking an AMD product but it happens for Nvidia but less often, the "fan" tries to derail the thread by inciting as many people as possible to try and get the thread locked. Its a common tactic, I have seen it soooo often on Anandtech and they take the bait from the trolls so often there its literally impossible to have a conversation about both vendors cards. Heck even just talking about one card and noting it does something badly is enough to just kill the entire thread and get a load of people ban holidays.

 

You can't get away from and I just report the ad hominen's and general attacks, these sorts of individuals ruin forums if you let them run rampant for too long, you just can't have a conversation around them.

 

Anyhow one other thing I found quite interesting is that for the size and performance estimate it looks like AMD is getting almost precisely the same scaling factor as Nvidia. Its within 1%. That is quite interesting considering the presumably large differences in the 14nm process they are using at Samsung verses the 16nm at TSMC. We wont know for certain until AMD releases actual transistor counts but I found that deriving both from the CU and transistor count and using the Nvidia transistor scaling from the 980 to produce the same result and that was interesting. So AMD got the same sort of transistor density scaling but a lot less power reduction, this could be quite an important difference in the processes themselves rather than anything AMD did wrong in the design.

If you are talking about me, all I did was ask you a question.