Why Barts is really VLIW4, not VLIW5 (and more on HD 5830/6790 being mainly 128-bit)

Ah, but you forget, supposedly Beyond3D is totally an AMD fan-site as evidenced by Bo's rantings on whatever that other forum was.
Bo_Fox is continuing to rant on ABT forums after he got permanently banned from here. Although I feel like I need to reply to one thing…

Bo_Fox at ABT said:
One guy was actually so afraid to get in trouble for sharing the links to this thread here at ABT - that's how bad it is.
I wasn't afraid—I was cautious. If I was truly afraid I would not have posted the links.
 
Thank you AlexV for doing what everyone quietly was waiting to happen.

I have limited time to read my favourite forum and going through nonsense and noise he was generating was getting tiring.
When presenting him with facts proving to the contrary of his belief he was reiterating to personal insults or twisting these facts to suit his own agenda.

So I only passively watched this ongoing situation and refrained myself from posing not to feed the obvious ...

Once again thank you and everybody who contributes to this forum with their knowlege and passion!
 
He is just another conspiracy theorist. These crazies come out of the woodwork regularly.
 
He's had plenty of chances so far (just looking at the size of the thread and how long it's been active), but it doesn't seem to be sinking in that this isn't the place for irrational outbursts and arguments. Anyways, we should all probably move on from this. There did seem to be a neat discussion at some point.
 
I liked hearing details on why the only VLIW4 chip ended up being Cayman. Related to the loss of 32nm and that Cayman was the design with the most effort put into it at the time so was most worth back porting to 40nm. Cayman is special and unique!
 
I liked hearing details on why the only VLIW4 chip ended up being Cayman. Related to the loss of 32nm and that Cayman was the design with the most effort put into it at the time so was most worth back porting to 40nm. Cayman is special and unique!

It would be interesting to gauge how Trinity factored into all of this too. After all, it will carry Cayman's legacy a bit more into the future, and they probably wanted the hardware locked in at a reasonable point in time. It also probably helps their case that Cayman based hardware has seen some driver development and has some presence in end-users hands, thus making it a relevant optimization target. Otherwise Trinity would've been caught in a rather unpleasant situation.
 
It would be interesting to gauge how Trinity factored into all of this too. After all, it will carry Cayman's legacy a bit more into the future, and they probably wanted the hardware locked in at a reasonable point in time. It also probably helps their case that Cayman based hardware has seen some driver development and has some presence in end-users hands, thus making it a relevant optimization target. Otherwise Trinity would've been caught in a rather unpleasant situation.

If they really had to they would have reworked Cayman and, I'm sure, Trinity as well.
They decided enough of the work was already done, the benefits outweighed the cons, and moved forward.
 
Tee hee. I've just had a quick look over at that ABT thread mentioned on the last page.

Do you have to be deluded to post over there, or does it just recommended? Very odd.
 
It also probably helps their case that Cayman based hardware has seen some driver development and has some presence in end-users hands, thus making it a relevant optimization target. Otherwise Trinity would've been caught in a rather unpleasant situation.

Well, AMD has the unique advantage to pick one of three (distinctiuous) mature architectures in their portfolio, when a customer wants a custom design. Each architecture occupies it's own optimal space, and all are DX11/GL4. Not that I think it was intended to have them 3, or that someone will ask ...
 
Well, AMD has the unique advantage to pick one of three (distinctiuous) mature architectures in their portfolio, when a customer wants a custom design. Each architecture occupies it's own optimal space, and all are DX11/GL4. Not that I think it was intended to have them 3, or that someone will ask ...

I'm not convinced that VLIW(4/5) has any significant advantage over GCN.
 
Back
Top