Final Words

Unlike a product review, there’s very little I can do to conclude here. There’s no buying recommendation, no performance to summarize. Even as an analytical piece there’s not much for me to conclude based on what I’ve learned at this point. When I wrote The RV770 Story I was convinced that ATI had embraced a new, refocused approach to GPU design, only to learn that they nearly threw out all of the learnings with the RV870.

The Northern Islands GPUs, due out later this year, were surely designed before anyone knew how RV870 would play out. Much less whether or not Fermi/GF100 would be this late.

I’m not sure any of what we’ve seen thus far in the history leading up to the RV770 or RV870 can tell us what we should expect from Northern Islands. While we can’t conclude about ATI’s future products, I do believe I have learned a considerable amount about how AMD’s graphics division works.

Carrell told me that the process of doing a product is not a logical process. There's logic in it, but it's not a logical process. It's an argumentative process. Not in the sense of having conflicts, but rather developing new data when the data isn't all there. When companies like AMD and NVIDIA do a product the engineers don't know all of the answers, and the knowledge they do have isn't binary - it's probability, it's weight, it's guesses. Sometimes they guess right, and sometimes they guess very wrong. The best they can do is to all weigh in with their individual experiences and together come up with the best group of guesses to implement. Over the years it seems that ATI has learned to, as much as possible, have all members of its team bought in to the product they're building.

The graphics team’s dedication and experience in jumping to new process technologies seems to have paid off with this generation. The move from TSMC to Global Foundries will surely challenge them once more. It’s not all about process technology though. The team’s focus on schedule and execution was a much needed addition to the company’s repertoire.

Carrell Killebrew helped turn ATI from a traditional GPU company with a poor track record, to one that could be known for its execution. The past three product generations have been executed extremely well. Regardless of whether you're an AMD, Intel or NVIDIA fan, you must give credit where it's due. The past couple of years have shown us a dramatic turn around from the graphics group at AMD. To go from the shakiness of the R500 and R600 GPUs to solidly executing on the RV670, 770 and 870 year after year is praiseworthy. I almost wonder if AMD’s CPU team could learn from the graphics group's execution. I do hope that along with the ATI acquisition came the open mindedness to learn from one another.

Preventing Espionage at AMD: How The Eyefinity Project Came to Be


View All Comments

  • ThomasS31 - Monday, February 15, 2010 - link would be very interesting if nVidia tell us the story openly, what happened with this long delayed newly designed chip.

    Anand staff, please try and persuade them to tell all details! :)
  • XiZeL - Monday, February 15, 2010 - link

    dont want to look like a fanboy here but these stories just make me like ATI even more, make me feel like part of the familiy...

    why dont you guys ever write these about nVidia?
  • Anand Lal Shimpi - Monday, February 15, 2010 - link

    The only reason I'm able to write something like this is because of how open/honest/unmarkety Carrell Killebrew is. And how trusting AMD PR is that they allow me to talk to him without any sort of limitations in place.

    Until recently, there hasn't been a similar contact for me at NVIDIA. That has changed in the past few months and I've already reached out to him to see if he is willing to allow me the same opportunity to talk about Fermi.

    If I can make it happen, I will :)

    Take care,
  • XiZeL - Tuesday, February 16, 2010 - link

    thanks for the reply :)

    just for the record, as a costumer and someone who really likes technology, these kind of articles will help me take my final descision when bying a product...
    yeally looking forward to seeing the same from Nvidia and try to understand their aproach on building a chip :)

    thanks again
  • Mat3 - Monday, February 15, 2010 - link

    I really enjoy reading stuff like this. One request for the next time you're talking to ATI guys: can you ask them about Fast14 and why it didn't work for them? Reply
  • Iketh - Monday, February 15, 2010 - link

    6900 series = Pagan
    6800 series = Sarigan
    6700 series = Saipan
    6600/5500 series = Tinian
    6400 series = Rota

    Im probably way off lol
  • hyvonen - Monday, February 15, 2010 - link

    Leaker! You'll be reported. Reply
  • XtremeOne - Monday, February 15, 2010 - link

    Thank You for this beautiful and insightful article. Like many before me, i registered just to say that. Anandtech is one of the sites I fell very lucky to know about. Some of your articles are a bit "techie" for me, but this one is practically impossible to stop reading. :D Thank you Anand. Reply
  • Iketh - Monday, February 15, 2010 - link

    it's a shame there cant be more articles like this... having a close relationship with one company allows it every year or two, but there are many companies in this field which could produce many more stories... i feel Anand is slowly tapping into a gold mine Reply
  • jstall - Monday, February 15, 2010 - link

    This is a fantastic article, nice to get some insight into the the RD and thought process just as much as it is to see performance charts. Be nice to see a little more of this.

Log in

Don't have an account? Sign up now