In a run-up for the HotChips22 conference, AMD’s Jonathan Fruehe started a series of blog articles on AMD Blog, answering 20 questions on Bulldozer. Jonathan posted fourth and final post, containing the last five questions from selected readers.
This is just another example from this long question and answers session, giving insight into why AMD canned the original 45nm design:
"How much resemblance does your today?s Bulldozer architecture have to the original design?? – Tye
As you are aware, when we initially designed "Bulldozer," we were working with a more modular processor design. The original "Bulldozer" design was 45nm. But as development progressed, it became clear that the 45nm design that we had been working on was not going to be as competitive as we would have liked.
With the world watching your every move, all product decisions become very public [how many articles have you seen about ?Bulldozer? in the press?] It is never an easy decision to make substantial changes to your product roadmap, but sometimes you have to make the tough call because it is the right thing to do. It helps when your partners are behind you and agree with the changes – sometimes a little short-term pain is necessary for a long-term gain.
We obviously aren?t going to get into specific design changes, but we think that the 32nm "Bulldozer" can bring a lot of benefit to our product due to smaller transistor size [which can help drive down the power envelope]. By going for lower power, we hope to give you more room for compute cores, FP capabilities and more."
In case you haven’t read the series, we would wholeheartedly recommend that you go through them, as they gave invaluable insight into how AMD is positioning Bulldozer and which markets company ceded to Intel. For instance, this was an answer from the question what AMD plans to do with a small niche of enthusiast workstations:
"Client systems will be available in single socket version. The two socket client market is very small and getting smaller every quarter. It hit its peak a few years ago with dual core [giving you 4 total cores], but quad core processors essentially crushed that market. Now the fact that you will get 8 cores in a single socket will probably wipe out what little is left in that market. It is hard to justify putting the resources into a market that was, at its peak, .8% of the total client market and exited Q4 of last year at .4% of the client market.
AMD OpteronTM processors will be available in dual socket systems, but those will not have the overclocking capability that you are looking for. Server customers do not manually overclock their systems; they need to ensure reliability. While you can run a processor outside of its specified operating range (overclocking it to a higher frequency], when you do that you take on some risks.
First, you diminish the useful life of the processor. By how much is hard to say because all processors are different. But overclocking increases the potential for failure, not something you want on a server. Secondly, when you overclock, sometimes your results are not what you expected. 2+2=5. Not a problem if you are playing a game, but a real problem if you are using a server to run automated systems, make business decisions or keep financial records straight.
Marketing AMD OpteronTM Processors as "overclockable" would not help us in growing server share in the commercial market (and most likely hurt us], so it is not something that we will be pursuing."
You can read the whole series by clicking on following links:
AMD Answers 20 Questions on Bulldozer, Part I
AMD Answers 20 Questions on Bulldozer, Part II
AMD Answers 20 Questions on Bulldozer, Part III
AMD Answers 20 Questions on Bulldozer, Part IV