You think I'm really a 100% Durango fan? lol
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
Rumor: PS4 GPU based on AMD's GCN 2.0 architecture?
- Thread starter artist
- Start date
You think I'm really a 100% Durango fan? lol
if PS4 has those additional ACEs, is it possible it may be more efficient than durango?
Not criticising MS' slides, they were talking about comparisons with 360 and general increased efficiency (just that us as an audience took it out of context and then started making jizz jokes)
Not criticising MS' slides, they were talking about comparisons with 360 and general increased efficiency (just that us as an audience took it out of context and then started making jizz jokes)
Yeah.
Yeah.
Well actually I don't know what a "100%" fan is, but you're definitely a pretty big one, and as far as I can tell the most active in the speculation threads. Theres nothing wrong with that, I'm pretty surprised you would consider yourself otherwise though.
I want both consoles to do well. The better the competition, the better it is for all gamers.
But if MS wants to give the core to Sony, then they deserve what's coming to them if all the negative rumors outside specs are true.
If you're talking about stuff like locking out used games and all that jazz, it's not going to be an uncommon thing in anyone's future. Microsoft is just the first one to have it omnipresent in the console space, in terms of what's leaked.
ALAN ULTIMATE
Neo Member
If used games get blocked.....
This technically means everytime your machine breaksdown and is replaced under warranty, all of your exsisting games immediately become very expensive coasters.
But don't worry... there's plenty more at your local game store.
Just saying.
This technically means everytime your machine breaksdown and is replaced under warranty, all of your exsisting games immediately become very expensive coasters.
But don't worry... there's plenty more at your local game store.
Just saying.
You think MS would be this dumb? All games will be tied to your online account.
ALAN ULTIMATE
Neo Member
Good thinking Sherlock, I forgot about that. LOL
ALAN ULTIMATE
Neo Member
Less load on the CPU I believe.
oldergamer
Member
Inefficient? No. it depends how its utilized.
Lord Error
Insane For Sony
Last VGLeak from three days ago oddly pointed to a 10+8 CU split (render/compute + compute). Was that discussed, or am I just interpreting this wrong:
http://www.vgleaks.com/orbis-gpu-compute-queues-and-pipelines/
http://www.vgleaks.com/orbis-gpu-compute-queues-and-pipelines/
Galacticos
Banned
bingo.
weekend_warrior
Banned
Holy shit!!
AgentP
Thinks mods influence posters politics. Promoted to QAnon Editor.
Yes, you are.
Lord Error
Insane For Sony
What does it mean? There wasn't any discussion about it in the thread about that leak here.
AgentP
Thinks mods influence posters politics. Promoted to QAnon Editor.
Lord Error
Insane For Sony
I saw it just now, and I even searched for '10' through the whole thread and couldn't find any discussion about this. This is what led me to think that I just misinterpreted it, and that the 10/8 thing was really something else, but I still don't understand what it refers to.
AgentP
Thinks mods influence posters politics. Promoted to QAnon Editor.
The 10 pipelines are not the same as the compute units (CUs). Look at the flow chart. All pipelines flow down to the CUs, this just means they can kick off more compute only commands in parallel. More efficient use of the CUs.
See;
There is no split because the CUs are still a unified pool. Both the graphics pipeline as well 8 ACE or compute pipelines can use the 18 CUs.
Lord Error
Insane For Sony
I see, I get it now.
McHuj
Member
Because the latency from the ESRAM to the GPU's L2 cache should be much, much lower than from any ram (GDDR5 or DDR3) to the L2 cache. The L2 is small ( 768 KB in Tahiti for example), so in theory the GPU could blow through that quickly and if you do, the GPU just sits idle and waits until new data arrives.
Obviously, if you prefetch all your data you can hide the latency, but that isn't always possible. Depends on what you're computing and how you're computing it. My guess is that the ESRAM really helps performance when you can't prefetch into the L2 cache. But that's just my speculation.