Vega 10 — 4096SP AMD Polaris/Vega Architecture Patent

[csbin]

http://appft.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d=PG01&p=1&u=%2Fnetahtml%2FPTO%2Fsrchnum.html&r=1&f=G&l=50&s1=%2220160085551%22.PGNR.&OS=DN/20160085551&RS=DN/20160085551


Patent PDF：http://pan.baidu.com/share/link?uk=2483816949&shareid=890948965&fid=295684580717454

A compute unit configured to execute multiple threads in parallel is presented. The compute unit includes one or more single instruction multiple data (SIMD) units and a fetch and decode logic. The SIMD units have differing numbers of arithmetic logic units (ALUs), such that each SIMD unit can execute a different number of threads. The fetch and decode logic is in communication with each of the SIMD units, and is configured to assign the threads to the SIMD units for execution based on such differing numbers of ALUs.

 

[guskline]

Interesting find. Notice file date of September, 2014 but just now, March 24, 2016 , published.

 

[Headfoot]

Another duplicate thread by csbin. This has been posted and discussed at length in this thread: http://forums.anandtech.com/showthread.php?t=2469674

 

[Bacon1]

Another duplicate thread by csbin. This has been posted and discussed at length in this thread: http://forums.anandtech.com/showthread.php?t=2469674

You mean when it was posted ~30 minutes after this thread was made in that thread?

 

[Head1985]

Why the hell everyone saying 4096Sp is VEGA10?Do you know that on 14nm That sku is around 300-350mm2 big so its GP 104 Size.
Its GP104 big pascal?Nope..

 

[master_shake_]

Why the hell everyone saying 4096Sp is VEGA10?Do you know that on 14nm That sku is around 300-350mm2 big so its GP 104 Size.
Its GP104 big pascal?Nope..

huh?

 

[Saylick]

Head1985 is just saying that Vega 10 should be a chip bigger than 400 mm^2, or at least Hawaii-size. If so, then Vega 10 should have more than 4096 SPs. If anything, Vega 11 should be around 350 mm^2 and should have ~4096 SPs.

 

[Osjur]

You mean when it was posted ~30 minutes after this thread was made in that thread?

Sure, if it means that 04-09-2016, 12:07 AM is 30min earlier than this thread. That's when it first came to light on the other thread.

 

[Stuka87]

You mean when it was posted ~30 minutes after this thread was made in that thread?

You need to read your time stamps better

http://forums.anandtech.com/showpost.php?p=38158216&postcount=107

 

[Bacon1]

You need to read your time stamps better

http://forums.anandtech.com/showpost.php?p=38158216&postcount=107

The only pictures from it I saw was http://forums.anandtech.com/showpost.php?p=38158614&postcount=115

which was posted after this thread .

Anyway so much misinformation in the speculation threads I don't view them often.

 

[Stuka87]

The only pictures from it I saw was http://forums.anandtech.com/showpost.php?p=38158614&postcount=115

which was posted after this thread .

Anyway so much misinformation in the speculation threads I don't view them often.

The link I posted has the link to the patent. A photo doesn't need to be posted in order to make it a valid post.

 

[Bacon1]

The link I posted has the link to the patent. A photo doesn't need to be posted in order to make it a valid post.

I never said it was an invalid post. And if anything, the first post to that patient was on page 1:

http://forums.anandtech.com/showpost.php?p=38154078&postcount=13

But again, I don't go reading patients so missed the post, like many others and apparently including yourself

 

[Mahigan]

I look forward to what RTG have in store for us because it would seem that Polaris isn't merely midrange if these patents hold up for the Polaris Gen.

 

[swilli89]

I look forward to what RTG have in store for us because it would seem that Polaris isn't merely midrange if these patents hold up for the Polaris Gen.

Well regardless if people end up buying Pascal or Polaris, we need Polaris 10 to be at least competitive with Pascal GP104. AMD has strongly indicated that this card is meant to change the perf/$ landscape and I'd wager we will see this card launch at $299 or $349 which will limit how much NV can charge for Pascal.

 

[Mopetar]

NV probably doesn't care too much about the price for GP104 as unless AMD really goes all out with Vega, NV will still have the performance crown and can charge the usual high-end price premium.

I expect AMD to charge at least $350, with some version possibly going for $400, if they can get a good bin. AMD needs to start making money, so as much as they like to talk about performance per dollar, I don't think they can afford to drastically undercut on price.

 

[airfathaaaaa]

NV probably doesn't care too much about the price for GP104 as unless AMD really goes all out with Vega, NV will still have the performance crown and can charge the usual high-end price premium.

I expect AMD to charge at least $350, with some version possibly going for $400, if they can get a good bin. AMD needs to start making money, so as much as they like to talk about performance per dollar, I don't think they can afford to drastically undercut on price.

can you give me the lotto numbers too? your tarrot readings seems accurate