Discussion in 'Frontpage news' started by Guru3D News, Oct 17, 2011.
Yeah it's shame BD didn't recapture the market
Glass Half Full
So they should have lots of room to easily tweak / optimize the architecture! ...or is that how AMD gets the 20-30% IPC increase chart for the next couple years? 10% Piledriver (hand design tweak one area of the die) etc.
This doesn't apply to engineering. What he's saying is that they use computers to actually engineer for them, instead of doing that by hand. Obviously they don't make each CPU by hand..
Please take it easy on us AMD users... We are going through some hardship.. :bawl:
When AMD goes down I won't feel quite as bad if I know that the monumentally stupid management get their lives ruined as well.
The FX 8150 would actually be worth buying at a really low price tag if it used 20% less base power and was 20% faster clock for clock. Who knows what other tremendous screwups were done to cut corners. As it stands no matter how they price it won't be worth buying.
ATI is what's keeping AMD "afloat" these days....
If HD7K turns out like Bulldozer....nVidia might be the only option without heavy funding from Intel and the US Gov't.....
HD7900 series is getting a brand new GPU architecture....at least according to all the marketing crap AMD has sent out in regards to it.
Sadly, management will probably walk away in better financial shape than any of us.... It appears as though management is trying to setup AMD to be bought out....
Here's my biggest issue with Bulldozer, Bulldozer is in no way a bad platform.(Not a very good one either) Sure, it doesn't give the performance we expected out of it but it does have some uses. My biggest issue with Bulldozer though is that AMD is touting that it is the upgrade path to Phenom and k10 users, that it's the light at the end of the tunnel for k10 when in all actuality, it's not. I understand it's a new product, I understand AMD is just promoting their product but the product they're promoting is no better than even just about what I currently have!
Then they're response to the bad performance of Bulldozer, is the fault of the software. Windows 7 may not be able to fully make use of Bulldozer correctly, and some games may not be completely multithreaded but maybe you should create hardware that would have no complications with current software. What's the point in having this technology when it could take years to fully use it, and by the time there's something to use it fully there's something new and better right around the corner?
It's an upgrade in how new the core is, but it's a downgrade in performance, and a downgrade in power usage.
Software, at least in the consumer market, doesn't even make full use of the Core 2 Quad series yet.... Software just isn't advancing as fast as hardware has. If Intel were to only develope processors that the software could make full use of, we'd still be using the Core 2 Duo line....
Sometimes, for the sake of advancement, you have to break compatibility somewhere. This is part of the reason AMD can't out-perform Intel....developing a processor with the intention of backwards compatibility hinders advancement.
Looks like they cut too many corners to save money, and now it'll cost them.
As far as software level I was speaking of Windows 7 using and optimizing core usage properly.
So, instead of trying to be innovative, they should have done what they've been doing and worry more about backwards compatibility? Backwards compatibility is BAD for innovation.
Tell me, what good is the technology if there's nothing to use it, and by the time there is something the next best thing will be out giving out eh, let's just throw out a random figure, double the performance?
I'm not saying don't innovate, I never said or even implied that. What I'm saying is don't try to over complicate things when it's not needed. Windows Developer preview has shown that it can fully make use of Bulldozer and core parking works for it, but at best in some applications it only gives it 10% increase in performance, but then when the competition gets the same increase or more it doesn't really work out too well.
Let's also not forget about the power usage for Bulldozer despite giving sub-par performance. The competition from Intel and even Thuban and Deneb are using less power than Bulldozer for the same, little bit less, or more performance.
It's all so insanely preposterous that it seriously does seem like a conspiracy.
Automation is not bad in itself. It could fasten development of modules to test new architectures and decide what to do, or set up a working 'prototype product' then customize it so the next version will be faster, yet won't cost too much on development.
The real problem i would say is that AMD took too much time to do the switch and advertised it as the next big thing. None should call the next big thing something that didn't get all corners rounded for performance.
20% bigger and 20% slower doesn't means it's going to be 20% faster, it could be 40%, 50%, or just 5%. it all depends of exactly where they put those, what are the implication they bring (temperature, leakage, etc), what they bottleneck.
And well, i wouldn't blame backward compatibility, imo everything to be blamed here is a mediocre engineering decision to relay too much on software.
So if I take this former employee's word for it, AMD got lazy? How the heck did that happen? They already know they are behind in performance when compared to Intel, why take that risk? Seriously something is wrong in the workforce in AMD.
cut costs to maximize profits. look where that got them. also, amd has stated many times they have no intentions on competing performance wise with intel.
What would happen if IBM got into the x86 market?
then we would have some real competitionl. but they are making sh!t tons of money on their server side. so i think they are happy with what they got.
They got in bed with Cyrix back in the day. I owned an IBM 6x86 PR120+.
This I would like to see. Some of IBM's server stuff is running 16 threads at 4GHz IIRC. That would be epic pwnage performance.