octoberasian
2[H]4U
- Joined
- Oct 13, 2007
- Messages
- 4,081
I keep reading a lot of criticism about BD lately being posted here and on other forums. It seems to revolve around two things mostly: Sandy Bridge and IPC.
It's as if everyone is expecting it to match or beat Sandy Bridge processors. And, if it doesn't do that then the processor itself fails and won't be successful against Intel. Others are expecting it to be just below SB. And the rest? Everyone else is expecting it to falter and just match or barely beat Intel's previous generation Socket 1366/1156 Core i7/i5 CPUs.
Very few from what I've read have high hopes for it that it will be both successful and affordable compared to Intel's offerings even if it just matches S1366 Core i7 950/975/990X performance.
So, the question I ask and answer as honestly and realistically about this is:
Well, I can understand that a lot of the criticism against Bulldozer are justified just by looking at past processors that AMD has released-- Phenom I and Phenom II. While Intel was ten steps ahead, AMD was ten steps behind, figuratively speaking.
However, I wonder if many of the critics realize several things regarding this. AMD does not have the financial support that Intel has nor the R&D size like Intel. GlobalFoundries might help in some way in the future, but I don't seem to think it will help in the present. For example, with 28 nm Ivy Bridge coming out in late 2011/early 2012, GlobalFoundries will only have started 28nm production by 2012 with Fab 8.
I'm assuming that the next CPU coming out of AMD (and possibly their next GPU after Southern Islands) will be entirely 28 nm post-2012. However, Intel will already be ahead of them by the time the first signs of such an AMD CPU is out on the market possibly around 2013 to 2014 at the earliest.
I think, and this is my personal opinion, if Intel hadn't [indirectly bribed] computer makers like Dell with incentives to only to use Intel CPUs, and I think AMD would have had more revenue to research and develop better CPUs that would have performed better than what Phenom I and Phenom II is today. And, I think Intel held them back in the market. Less revenue from sales, less money for R&D, and less R&D, lesser CPUs is my thinking.
I honestly would think that AMD would be in a better position with CPUs on par with Intel's latest and current offerings instead of beating a previous generation Intel CPU if they had more financial backing to commit to making better CPUs.
I think that is where the criticism becomes unjustified because many don't realize or understand that AMD is not Intel. AMD has, as of now, an operating income of $848 million dollars compared to Intel's at $16 billion dollars US.
It's a huge disparity.
It's the equivalent of David versus Goliath, with David having nothing more than stones and sticks to fight back. So what do you do? AMD is fighting where it can fight and that's at affordability. I've stuck by AMD since the Athlon XP days because it's more affordable than what I can get with an Intel CPU. For $350 after shipping and tax, I can get 8 GB of RAM, a board, and an AMD CPU. For $315, I can get an Intel Core i7 2600K with no motherboard and no RAM. I'm looking at $450 to $500 by adding RAM and a board.
It may not perform closely to a 2600K but at least it works and it means I spent much less, and that's where AMD Bulldozer will excel in.
For example, a top-end, 8-core FX-8130P at $320 is considerably cheaper than a 6-core/12-thread Core i7 990X. The only problem is that the pricing is on par with a Core i7 2600K which may lead to the thinking that the top-end AMD 8-core CPU may not perform as well as the 2600K and be considerably worthless from at that price point. Thus, that's where the fear and worry about AMD's next CPU begins given the performance of past CPUs from AMD.
Point of all this is: Let's at least give AMD the benefit of a doubt here and let the benchmarks of Bulldozer when it's released decide whether AMD has succeeded or not.
It's as if everyone is expecting it to match or beat Sandy Bridge processors. And, if it doesn't do that then the processor itself fails and won't be successful against Intel. Others are expecting it to be just below SB. And the rest? Everyone else is expecting it to falter and just match or barely beat Intel's previous generation Socket 1366/1156 Core i7/i5 CPUs.
Very few from what I've read have high hopes for it that it will be both successful and affordable compared to Intel's offerings even if it just matches S1366 Core i7 950/975/990X performance.
So, the question I ask and answer as honestly and realistically about this is:
How do you increase IPC in a processor without making it:
My thoughts on all this?a) complex
b) costly
c) have high TDP?
My best assumption here has to be something what Intel is doing with Ivy Bridge S2011 which is coming out on 28nm process. Unfortunately, I don't see AMD coming anywhere near that anytime soon with 32nm being their next process while Intel is ahead. And, sadly, this is going to give Intel enough reason to price the CPU higher than AMD due to fearing little to no competition from AMD. But, at the same time decreasing TDP yet increasing architecture complexity in the newer Intel CPU.
Also, is native 4-core/8-core design better than Intel's native 4-core/6-core with 4 or 6 virtual threads? Native cores versus virtual threads/cores in other words.b) costly
c) have high TDP?
My best assumption here has to be something what Intel is doing with Ivy Bridge S2011 which is coming out on 28nm process. Unfortunately, I don't see AMD coming anywhere near that anytime soon with 32nm being their next process while Intel is ahead. And, sadly, this is going to give Intel enough reason to price the CPU higher than AMD due to fearing little to no competition from AMD. But, at the same time decreasing TDP yet increasing architecture complexity in the newer Intel CPU.
It seems the general consensus I'm getting is that a lot of people believe AMD's approach to a native 4-core/8-core design might not be enough to beat Core i7 SB or S1366 Core i7 due to what might be a possible lower IPC per core in Bulldozer. Thus, it comes to the next question.
If AMD managed to increase the IPC per core but at the same time made BD more complex, would anyone be willing to buy it if it were to cost more?
Well, I can understand that a lot of the criticism against Bulldozer are justified just by looking at past processors that AMD has released-- Phenom I and Phenom II. While Intel was ten steps ahead, AMD was ten steps behind, figuratively speaking.
However, I wonder if many of the critics realize several things regarding this. AMD does not have the financial support that Intel has nor the R&D size like Intel. GlobalFoundries might help in some way in the future, but I don't seem to think it will help in the present. For example, with 28 nm Ivy Bridge coming out in late 2011/early 2012, GlobalFoundries will only have started 28nm production by 2012 with Fab 8.
I'm assuming that the next CPU coming out of AMD (and possibly their next GPU after Southern Islands) will be entirely 28 nm post-2012. However, Intel will already be ahead of them by the time the first signs of such an AMD CPU is out on the market possibly around 2013 to 2014 at the earliest.
I think, and this is my personal opinion, if Intel hadn't [indirectly bribed] computer makers like Dell with incentives to only to use Intel CPUs, and I think AMD would have had more revenue to research and develop better CPUs that would have performed better than what Phenom I and Phenom II is today. And, I think Intel held them back in the market. Less revenue from sales, less money for R&D, and less R&D, lesser CPUs is my thinking.
I honestly would think that AMD would be in a better position with CPUs on par with Intel's latest and current offerings instead of beating a previous generation Intel CPU if they had more financial backing to commit to making better CPUs.
I think that is where the criticism becomes unjustified because many don't realize or understand that AMD is not Intel. AMD has, as of now, an operating income of $848 million dollars compared to Intel's at $16 billion dollars US.
Code:
[URL="http://en.wikipedia.org/wiki/Advanced_Micro_Devices"]Advanced Micro Devices - Wikipedia, the free encyclopedia[/URL]
[url]http://en.wikipedia.org/wiki/Intel[/url]
It's the equivalent of David versus Goliath, with David having nothing more than stones and sticks to fight back. So what do you do? AMD is fighting where it can fight and that's at affordability. I've stuck by AMD since the Athlon XP days because it's more affordable than what I can get with an Intel CPU. For $350 after shipping and tax, I can get 8 GB of RAM, a board, and an AMD CPU. For $315, I can get an Intel Core i7 2600K with no motherboard and no RAM. I'm looking at $450 to $500 by adding RAM and a board.
It may not perform closely to a 2600K but at least it works and it means I spent much less, and that's where AMD Bulldozer will excel in.
For example, a top-end, 8-core FX-8130P at $320 is considerably cheaper than a 6-core/12-thread Core i7 990X. The only problem is that the pricing is on par with a Core i7 2600K which may lead to the thinking that the top-end AMD 8-core CPU may not perform as well as the 2600K and be considerably worthless from at that price point. Thus, that's where the fear and worry about AMD's next CPU begins given the performance of past CPUs from AMD.
Point of all this is: Let's at least give AMD the benefit of a doubt here and let the benchmarks of Bulldozer when it's released decide whether AMD has succeeded or not.
Last edited: