Also the "NDA timing" thing was never more than a conspiracy theory based on a coincidence.
The info albert was talking about from his engineers was the article on digital foundry yesterday.
In that article the ms tech guys said the gpu is from sea islands just like sony.
They did claim they altered the compute units in the gpu but gave no specifics.
They also said they had overhead in the cus for gpgpu.
With it coming straight from ms that the gpu is sea islands that rules out it being volcanic islands.
Now the alterations they claimed to make to the compite units could be from volcanic islands but they didnt specify
Albert hasnt said anything as far as i know about the article but he did say the info would be coming very soon on Twitter friday and then the article came out sunday. In the article they clear up alot of misconceptions about the xbox one and specifically talked about the points albert posted on neogaf.
So im pretty sure that article is what he was refering to.
The article said no 2nd gpu and the gpu was based off of sea islands with modifications to the compute units which they didnt go into detail about
Albert hasnt said anything as far as i know about the article but he did say the info would be coming very soon on Twitter friday and then the article came out sunday. In the article they clear up alot of misconceptions about the xbox one and specifically talked about the points albert posted on neogaf.
So im pretty sure that article is what he was refering to.
The article said no 2nd gpu and the gpu was based off of sea islands with modifications to the compute units which they didnt go into detail about
Albert hasnt said anything as far as i know about the article but he did say the info would be coming very soon on Twitter friday and then the article came out sunday. In the article they clear up alot of misconceptions about the xbox one and specifically talked about the points albert posted on neogaf.
So im pretty sure that article is what he was refering to.
The article said no 2nd gpu and the gpu was based off of sea islands with modifications to the compute units which they didnt go into detail about
Yes, whether it is surprising or not, the MS engineer let out the Xbox One GPU is based upon Sea Islands, which belongs to the AMD 8000 series and they haven't been launched yet. I think we should know more in a couple of days.
http://en.wikipedia.org/wiki/Sea_Islands_(GPU_family)
you know, you are too smart to be posting over at that moronic blog you know.....
Yea that was my point, rebranding doesn't give or take more information to us apart from that it's not VI based.
Adding cores never scales linearly. For their statement to be true on a GFLOPS basis (which it probably isn't), a scaling efficiency of 95.6% per core added would provide 12CUs at 853MHz with the same horsepower as 14CUs at 800. At that (bogus) scaling factor, 18 CUs at 800MHz would only be 7.4% faster than 12CUs at 853MHz.
The real scaling factor is guaranteed to be higher than that, but it will still be <1, and the scaling factor probably goes down with increasing cores, since the bottlenecks become even more important. You're going to have a lot more contention for a 256 bit memory bus at 12 cores than at 2. What this means is that in real terms, PS4 does not have, and never had, a 50% GPU advantage. What the actual advantage is, we may never know.
Adding cores never scales linearly. For their statement to be true on a GFLOPS basis (which it probably isn't), a scaling efficiency of 95.6% per core added would provide 12CUs at 853MHz with the same horsepower as 14CUs at 800. At that (bogus) scaling factor, 18 CUs at 800MHz would only be 7.4% faster than 12CUs at 853MHz.
The real scaling factor is guaranteed to be higher than that, but it will still be <1, and the scaling factor probably goes down with increasing cores, since the bottlenecks become even more important. You're going to have a lot more contention for a 256 bit memory bus at 12 cores than at 2. What this means is that in real terms, PS4 does not have, and never had, a 50% GPU advantage. What the actual advantage is, we may never know.
You misreadIsn't the 7850 at 860mhz with 16 CU like 10% faster than a full 7790 with 14 CU and 1ghz.?
How can the difference be 7% in 14 CU vs 18 CU at 800mhz.?
I think your math is off.
bkilian is saying MS's maths is offAdding cores never scales linearly. For their statement to be true on a GFLOPS basis (which it probably isn't), a scaling efficiency of 95.6% per core added would provide 12CUs at 853MHz with the same horsepower as 14CUs at 800. At that (bogus) scaling factor, 18 CUs at 800MHz would only be 7.4% faster than 12CUs at 853MHz.
Yes it is a sea islands product as opposed to a southern islands. Supposedly volcanic islands (Hawaii etc) will début this week. MS has a strange NDA timing. Penello said his engineers will reveal certain things this week and the timing of it is after the volcanic islands reveal. However, Bonaire was sold into the channel starting last March.
Why have an NDA on a part based on a part that's already been released 6 months earlier?
You misread
bkilian is saying MS's maths is off
Isn't the 7850 at 860mhz with 16 CU like 10% faster than a full 7790 with 14 CU and 1ghz.?
How can the difference be 7% in 14 CU vs 18 CU at 800mhz.?
I think your math is off.
Look ive got my day one xbox one almost paid off. Im happy with everything the ms techs explained in the df article. The launch games look great. I think people are putting to much into tflops.
There is no proof they put volcanic islands tech in the xbox one gpu but it has been modified to take advantage of dx 11.2 and it could be possible they put some things from volcanic islands in there.
Its become clear there is not gonna be some last minute revelation in hidden power.
7870 XT@600MHz with same Memory (2GB GDDR5@1375MHz), bandwidth, same CPU (Core i7 3770K@ 4.3GHz and 16GB of DDR3 memory@1600MHz) and 32 ROPs is only 24% faster (on average) than 7850@600MHz, Which leads to 97.5% efficiency per CU.
7870 = 20 CUs
7850 = 16 CUs
20/16 = 1.25, that means +25% flops/ALUs/TMUs at the same clock, but the same setup/raster/ROP/memory bandwidth. Getting a 24% speedup under these conditions look suspiciously like either a theoretical or a botched benchmark, in any case one basically completely determined by ALU/TMU throughput. You can't expect this across the board. Even if the 7850 had slower memory at stock speeds and assuming a wide range of benchmarks, it would be an extremely good showing for the downclocked 7870 giving no hint of any scalability issues.