AMD vs. Intel Strange Conspiracy theory?

Page 4 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.

smartkid95

Distinguished
Sep 2, 2009
153
0
18,710
This is a strange one that I have been wondering and looking into, and observing for a long time. I was told by a source that is obviously not dependible but humor me, that intel cpu's are no match to amd due to the theory (it's rediculous but humor it) intentional replacement,and the cpu is designed to get slower after time which is why they never release cpus on the same mobo platform like amd and theychange constantly to get more money. I know this a lame conspiracy but I wan to get some honest opinions about that. Is there any truth to that at all or is is trash. the biggest question is I want to know the history of amd vs. intel because it also was claimed that overtime amd cpu's will work better than intel and amd has always been better. Intel fanboys rejoice because i'm ready for anyting I have strong feeling so don't be afraid to attack me. amd fanboys are wecome to have their say as well. have at it.



As a side note, I'm not really swayed either way as a fanboy to either company, my decision is decided by the almighty dollar. so far amd has consistantly been the better value enjoy. feel free to flame coment the post on my blog about this too.
http://rokk-itscientistblog.blogspot.com/
 
Solution
Complete garbage. CPUs don't get slower over time. Operating systems however can get slower over time, but this is easily remedied with a reinstallation of the OS.
The x86-64 architecture is a superset of the x86 architecture. You can't "remove" the x86 part of it because all 64-bit programs still use 8- 16- and 32-bit instructions and registers in amongst all the 64-bit stuff.

The only way to do it is to go to a brand new architecture, which is what Intel tried with Itanium. But the marketplace has ruled that backward compatibility is more important than the potential performance benefits of a purely RISC architecture.
 

Upendra09

Distinguished
Sminlal

Yes new arch is what i mean and iknow about itanium, but what i am saying is it is time to move on.

Very seriously answer this question

What else can we do in x86 bit architecture?
by this i mean, can we make processors faster with higher clocks like 1990s where the 5 mhz advantage gave me 15 fps extra in doom

or will better arch help make procs faster, like PII getting better instruction set and new procs get better and get better instruction sets

more cores equal faster as soon as SW catches up......... but how many cores will one need? 12?20? why would u need so many cores? to run every bk app?



Now think about how x64 opens up the doors for innovation

AMD get smart and do something in the next few years
 

croc

Distinguished
BANNED
Sep 14, 2005
3,038
1
20,810


www.top500.org

Read that site's various lists and charts, THEN come back and tell us how much we will benefit by going the RISC path. You do know what RISC means, right?

DEC went with their RISC only technology, and went broke. Intel bought out all of the Alpha IP from Compaq after the DEC sell-off, but gave up on their version (IA-64) after a few years. Sun still uses RISC (Fujitsu produced SPARC processors) but got bought out by Oracle. NEC still makes RISC chips (sort of) but they are vector-based and no one (besides NEC) has a use for them. HP used to make RISC processors, but quit flogging the dead horse, mostly. Only IBM still makes RISC chips in any quantity, and IBM does pretty well with them in the supercomputer space.
 
If you want to move on, there are plenty of high performance 64-bit architectures out there waiting for you. You have your pick of Itanium, SPARC, PowerPC, and others. If you think x86 is at a dead end then why aren't you using one of them? Oh that's right - it's the same reason nobody else uses them either - they won't run all the programs you know and love.

What else can we do in x86 bit architecture?
The hardware designers have pretty much run into a dead end in terms of extracting parallelism from a single instruction stream. After tricks like branch prediction, register renaming, out-of-order issue and retirement, speculative execution, etc there really isn't very much more you can do to get multiple things going on at once out of a single stream of instructions. This isn't just an x86 or 32-bit issue, it applies equally well to 64-bit and RISC as well.

In fact, that's why Intel introduced "EPIC" (Explicitly Parallel Instruction Computing) with the Itanium. They basically said "the hardware's doing all it can, if you want to get any more parallelism out of an instruction stream then you're going to have to write smarter compilers". (Turns out that it's no easier for the compiler writers do to that than it was for the silicon designers).

So for the last few years the trend has been to add additional complete CPU cores, increase cache sizes, integrate more functions onto the CPU die (memory controller, GPU, etc.), add more capable vector processing (SSE instructions), and look for ways to reduce power consumption. Again, this isn't an x86 thing, it's true of all architectures.

Even if everyone decided to switch tomorrow to a pure 64-bit architecture, these issues would not really change.
 

verndewd

Distinguished
Mar 27, 2009
640
0
18,990


good to know
 

Upendra09

Distinguished
i see so x64 isn't as promising as it seems?

Then where aer we going to go? won't making CPUs be dead end within the near future because how many cores will we need? how fast can clocks go before everything runs at the same speed?
 
I have no doubt that there are some very smart people trying to figure out where to go next. But whatever they come with, there's a good chance it can be applied to the x86-64 architecture just as easily as any of the other 64-bit architectures.
 

cheesesubs

Distinguished
Oct 8, 2009
459
0
18,790


clockrate is sh*t. the future processor will meassure from benchmarking and ipc instead of clockrate or multicore. forced menufacturer inverting better arch in the market than just raising up clock/sqeeze more core in one die.

it will be a architecture war between intel and amd as they will inventing new arch every year. which preventing intel from be god damn lazy again and amd from been cheap by loading outdated arch into multicore solution. the architecture will be much advance than we think today. it is a good thing push two company much foward
 

croc

Distinguished
BANNED
Sep 14, 2005
3,038
1
20,810


There are some applications that DO take advantage of RISC processors better than x86-64 processors, such as Oracle (for one). In a large, high speed transaction based setting Oracle will have a lower TCO on a RISC system than it will have on an x86-64 system.
 

randomizer

Champion
Moderator

Chances are they are already working on it :)



But it will still be x86.
 
Totally agree - although I'm tempted to paraphrase the old line about statistics with: "Lies, damn lies, and TCO"... It's my experience that TCO figures usually favour the company who's quoting them. ;)

In most cases RISC processors definitely CAN be more efficient and faster than a CISC architecture like x86 (be it 32- or 64-bit) - no question about it. And they're used in the high-end server marketplace because the costs and workloads involved make it feasible for companies to produce purpose-designed software such as Oracle, SAP, etc. in versions which run natively on RISC CPUs.

But there's a huge chicken-and-egg problem in the desktop marketplace because most software companies don't want to spend money producing RISC versions when they're such a small fraction of the consumer market, and consumers don't want to buy RISC because there's hardly any software available. Even the enthusiast market, which could probably track down the scarce hardware, isn't really interested.

And so we're stuck with x86. AMD pretty much killed the one chance we had to jump to a new 64-bit RISC architecture by grafting 64-bit registers and instructions into the x86 architecture. I don't say this as a criticism of AMD - it was a brilliant move on their part. But it was probably the last nail in the coffin for those wanting to move mainstream desktop computing onto a more modern CPU architecture.
 

keithlm

Distinguished
Dec 26, 2007
735
0
18,990


By "Market" I take it you really mean "Microsoft". Because the average home or office user doesn't care. The only time they care is if the software they have won't work on the different platform. And in truth just about everytime Microsoft has had a windows update people have often ended up buying new software anyway; so in reality that reason is not a good excuse. (Other than being "kind of" compatible helped in transitions.)

Optimally it would have been best to ditch the x86 compatibility about 12 years ago. (I had an SGI O2 workstation with a MIPS back then.)
 

randomizer

Champion
Moderator

Many businesses use pathetically old software which they don't feel they should have to upgrade (hence the reason they keep the same OS for years as well). The problem is that a sudden switch to a radically different architecture would mean that there is no transition period where older software can be slowly phased out. It will either work or it won't work, and only enthusiasts will bother to buy the new architecture for quite some time because (generally) we aren't as interested in backwards compatibility as others, we just want performance.

Microsoft wouldn't want to pull a Vista^3 compatibility failure on 100% of their customers and force them to upgrade, that would be suicidal.
 
Until a few years ago that would have been pretty much true. But look what's happened - even Apple has now switched from PowerPC to Intel x86. It's not just Microsoft any more - it's now very, very difficult to find a non x86 desktop from ANY hardware or software vendor.

Apple has actually switched architectures twice now - once from Motorola 68000 to PowerPC and then again to Intel. They have a tremendous advantage because of their mostly closed platform design and the much more limited amount of software that has to be dragged along to a new processor and OS. They could probably do it again, but I think it's pretty much impossible to accomplish that kind of switch in the Wintel space.

The switch from 32-bit to 64-bit provided a terrific opportunity for a new architecture, and that was Intel's intention when they introduced Itanium. But AMD did an end run around them with x86-64, and in the process guaranteed that they'd maintain a significant share of the desktop marketplace.
 
Status
Not open for further replies.