Do not get the 9600xt, cancel your preorder. Revision 2 is comming!

You misunderstand me. I knew exactly what ATi was doing with the 9600 replacement of the 9500's. While everyone else was going "What the heck are they doing?" I knew exactly why they had. My confusion is, it seems like they could have made a .13 card with the same amount of pipelines and transistors as, say a 9800 pro, and made it their ultimate high end card. So the only reason I can see, is that they needed a mid range card so they didn't bother with the idea. And the next chip will be along those lines.

I'm thinking that ATI didn't want to switch to .13 earlier because they could have had problems, like NVIDIA. When you can produce a card on a proven technology, why take the risk to go to a new process which might cause problems and delays? The R420 will be made on .13 :)

Talldude is right on money. The more transistors, the more complex the process of manufacturing. Testing out a new process on a less complex design is a far better idea than possibly backing yourself into a corner by moving everything to an unproven process and risk have chip shortages. The 150nm process is very mature and good yeilds could be produced on this process.

Yeah, I guess you're right. Given the results of the 9600xt I think they just might have done incredibly well if they had gone .13 on the 9800's;)

I can see the 9600xt using 650mhz DDR since the all in wonder 9600P will have 650mhz. :)

But can we believe that website???

Quote:

---

Originally posted by ATilaptops
You misunderstand me. I knew exactly what ATi was doing with the 9600 replacement of the 9500's. While everyone else was going "What the heck are they doing?" I knew exactly why they had. My confusion is, it seems like they could have made a .13 card with the same amount of pipelines and transistors as, say a 9800 pro, and made it their ultimate high end card. So the only reason I can see, is that they needed a mid range card so they didn't bother with the idea. And the next chip will be along those lines.

---

Shrinking to a new process it a DARN tough thing to accomplish. That's why in the CPU world, it's most often done midway through a processor's run, when they know that they can still push out more chips on the old process if they need to, but have time to get the new process right on something that they know(an old chip).

In the graphics world, there are no old chips to learn on. This means that a process shrink almost always occurs concurrently with a new core design. This is doubly tough. The 9600 was the result of this. They had a new laptop core that was an ideal candidate for a process shrink, it's less complex, and it also benefits from the lower heat and power more than a desktop chip.

After they started to get the chip into production, they decided to use it not only in laptops but also as a mainstream desktop chip (because it had better separation from the high-end chips and wouldn't eat all the market for the high-end chips like the NV Ti-4200 did to the 4600, and the R9500P did to the R9700)

You have to admire the engineering and politics behind the new chip.

Quote:

---

Originally posted by sapasion
Believe it when you see it

---

w3rd.