The Official PC building thread -3rd Edition

Thanks guys. Very nice to see crucial coming back up in the world. I have some older Ram of theirs. They used to be very good with ram. As of late, I've heard the opposite with the newer DDR2/DDR3 modules. Perhaps they're making a comeback. Quite the leap from the intel drive. I almost wish I had waited No big deal though. I'm sure my drive will impress me.

 

i'll never buy crucial memory again as long as i live. ive gone threw 4 sets of mem before i got 1 set that worked and i only used them for about a week cause it was about the same time span the others went out. their ddr2 1066 mem was so crappy and they new it they offered me some 2 year old double sided ddr2 800 mem in replacement.yea gave me some $35 cheap mem and thought it was a fair deal.. yea they went bad also.. well 1 stick did


ok maybe im stupid here and no i havent done any research on the topic but how does intel make a AMD's complier codes? doesnt AMD make their own? so im supose to beleive that intel has screwed with all benchmarking test (the hundreds of different ones) to make their product look better? nope dont buy it. like i said i dont really know much about it, TBH sounds like one company/persons just upset with there lack of technology and putting the blame on someone else. typical human way.
another conspercy theory...that like toyota blaming ford for making their faulty gas pedal????????? i know i'll take the beat down now and wont post for awhile again.

 

Rob,
Come on man, wake up and smell the coffee! The software is used to compile the programs used in a computer. Most use Intel's compiler, so all those programs will force AMD's CPU to run at reduced performance by using the slowest codes that work.

The lawsuit and decision is a matter of public record, based on the facts presented in the suit. Why do you think that Intel paid off AMD to the tune of 1.25 Billion dollars so quickly? There is no conspiracy! Intel broke the law, and got caught! While it's perfectly legal to optimize codes for your own products, it's not legal to cripple another companies products performance to make your own product appear better than it really is!

Respectfully,
Russ

 

S-ATA3 for SSDs is eventually going to need to become compulsary, but for the time being it is of little benefit. As for the Crucial in particular, I've not heard anything that makes theirs particularly revolutionary. I'm still sticking with Intel as my recommendation for the moment as they are pushibg by far the best performance where it counts (random and small-sector write). The size of the drive hints it'll cost an absolute fortune though, so it doesn't excite me for the time being. X25-MG2s are expensive enough as it is.
I still don't trust Crucial for memory, it's too soon since their last absolute failings.

As far as the Intel antitrust goes, I'll say again, companies are always doing stuff like thus. Yes, it's wrong, and yes, Intel should be punished for it, but it's far from unprecedented, and sadly, it's just expected in today's business environment.

 

Rob, what has happened here is a competitive advantage intel has achieved by getting a program compiler popularised. Since most people tend to use Intel's compiler for their software, Intel have the ability to do whatever they like with the performance of all these programs on various CPUs. Intel therefore chose to reduce the performance of AMD CPUs at these particular tasks. Cripple is a strong word, but it does have an appreciable impact. Doing this is wrong, however, as I said before, it does not stop AMD users from being able to do anything, it just means that programs written with Intels in mind will run better on Intels. They have used an illegitimate means of doing such, but that's all that's happened. In many other cases, the Brand B product isn't even able to do what the Brand A product can because it's outright disabled. That hasn't happened here, and until it does, I'm not boycotting Intel. If it were enough to make me do so, I would have done it a long time ago, as it's stuff like this that brought about AMD's formation in the first place, they were ex-Intel employees who wanted to do things their own way.

 

Sam i understand what your say and what russ is saying, but i just dont believe AMD doesnt have any, hell even 1 program that shows anythign different than what has been shown with the programs that are being used... if i made a product and wanted more people to buy it and didnt have a way to show off what i can do, i myself would create or have someone for me do just that. just so i could prove what i can do. beside that, everyone here knows that the benchmarking stuff is just a numbers game, there are to many ways to make the NUMBER say what you want them too. like everyone else has said its what happens in real life/time. and anyone with any common sense knows how well AMD performs. ive never had a amd (well a old amd duron) but i know they are good CPU's if they wasnt why are so many millions of them being sold??? once again common sense. ok yea intel has all the NUMBERS to prove what they can do. so what that dont mean crap if i buy the product and it falls on its face, as im sure it has and does and will do again. its that way for both companies. the one thing that has kept me away from AMD is all the locked cores and little tricks you have to do to get them to work properly. the guy i work with that just did the I5 build his amd 9850 wasnt a quad like it said it was. it was only 3 cores once broke down. the so called 3rd core was bigger than the other 2 so AMD called it a QUAD.. now isnt that lieing to the public???

 

The AMD Quad cores were more quad cores than Intel's. Until the new i5 and i7 architectures, Intel's quads were almost 'copy and paste' of their dual cores, whereas AMD CPUs were designed to be quads from the outset. Your friend was probably commenting that the 9850 was so slow it performed like it only had three cores
As for numbers, they're nice, but real world numbers are where it matters. SuperPi is hardly a real-world instance, and nor is 3dmark.

 

FRIKKIN AYYYY!! That's what I was trying to say earlier I think. LOL.

 

i will try and find the artical he showed me where it broke down the 9850 and showed the inside of it. he isnt here tonight to ask him about it or i would.

as for the numbers game goes. they are nice to look at thats all. just like over in the scan thread. those scans mean crap its just something to go by as to see what you have. its all about what the product does when im useing it. thats what sells me. yea the numbers get your attention buy performance is the seller. thats why i find it so hard to beleive that AMD doesnt or cant create some software showing their abillities, 1. they dont feel the need to prove themselves like that,2. they dont want the truth to be shown that they are in fact not as powerful as intel. i hope it is the first of the 2, but who knows?

 

Intel didn't create software to 'big themselves up' though, they merely created the compiler that everyone else uses to write their programs with. Had AMD had the same success they would be in a similar situation, but probably not getting fined since they don't abuse privileges like Intel do. Or so we think.

 

thats my whole point right there if AMD was so worried that intel was doing this then why didnt AMD make their own complier to show off what they can do???

 

My thinking too, I'd offer my insight but in reality, I don't tend to look at this side of the industry much, so I don't have anything to say. For all I know they might have in the past...

 

Rob,
You would have to show me that Quad core! I've never even heard anything about it, and that's pretty remarkable in itself. I've never once heard a word about any triples being represented as Quad Cores, and the triples were first reported here on AD by me, long before they were produced.

As far as the one core being bigger, that's nonsense. The cores are all the same size. He was probably looking at the on chip cache in a picture. There was a 9850 Quad core, but no triple core 9850 at all! Your friend doesn't know what he's talking about!

As far as unlocking cores, you are being a bit silly! Unlocking cores is only a sometimes Bonus when buying AMD Quad core based Dual or Triple core CPUs. AMD makes no claims for them, and they are not sold with the promise of being a Quad core. Given that the wafer yields have gotten so good that many triples are Quads with a core turned off, with some chips it's almost a certainty that you can unlock the shut down cores. It's easy enough to do if you buy one!

The Intel compiler in question takes the information embedded in the software from the CPUID, and in the case of higher end AMD CPUs, uses the slowest code that will run, effectively slowing down the performance of the CPU. The people making the software had no idea that this was going an. It just became a known and accepted fact that AMD was slower than Intel, all because of this illegal code, and the can of worms it opened. It's as simple as that! My Athlon x2 7750 was faster than my E6750 in everything but benchmarks, and I dearly love my Quad core, as it's easily twice as fast as the dual Core 7750! It will be nice to see what it can really do, when they get this mess all sorted out!

Russ

 

Rob,
The AMD chips do pretty well against the more expensive Intels, even when crippled. When all this is straightened out, I'm sure there will be a number of surprises when both Intel and AMD are compared honestly, without the slowdowns that some software that used the Intel compiler impose on the AMDs. Had someone not had the idea to change AMD's CPUID to a Genuine Intel and then check the performance, the world may never have known there was a problem deliberately created by Intel!

Russ

 

Here's some stuff I was reading.. seems pretty good and maybe a little different than what RUss posted.

http://www.agner.org/optimize/blog/read.php?i=49

http://www.amd.com/epd/desiging/fusionpartners/prodbytype/3.developme/11.compiler/

http://developer.amd.com/CPU/OPEN64/Pages/default.aspx

http://www.thefreecountry.com/compilers/cpp.shtml

From what I read, briefly that is, it seems that LINUX has something that will "measure" AMD speed.... is that true Linux users?? Anyway just trying to find something that makes sense in all of this.

 

Sam,
In effect, it has disabled some on chip instructions or at least crippled them so badly that they run slower than mud, effectively negating the function.

Russ

 

once again all im saying is IF amd knew/knows INTEL is/was doing this why didnt/DONT they come out with their own complier, or their own software to measure what they can do???? its that simple. if i owned a company and my arch rival was slinging mud about me and my product i would do everything i could to prove them wrong... unless what the rival was saying is true???? it dont get no easier than that... i know its a proven fact that for day to day stuff there isnt much difference in the 2 cpu's except for the numbers game that has caught so many people in there loop. dont get excited over a man made number. ive been to missouri, you have TO SHOW ME... let me see the performance thats all anyone has to see and then you know the numbers DO lie. i cant see it any other way

 

Rob,
While it's been suspected for years, no one had any proof that Intel was manipulating the conditions to exclude AMD from better performance. It was only discovered by accident, when someone wrote some code to change the Genuine AMD CPUID to an Intel CPUID, and ran the same benchmarks, only to discover that they got much better scores with the AMD when it was identified as an Intel.

As far as the performance goes, any AMD owner will tell you that the numbers lie. I figured that out by the end of my first full day of using my 7750BE, as it was noticeably faster than the E6750, doing the same work. The 7750BE was at least as fast and in many instances faster than the E6750, running at a slower clock speed. My 630Quad is more than twice as fast, and at the moment falls behind Will's 9550, 60,000 MIPS to my 54,000 MIPS. I'll see how long that advantage lasts after I mod the NB and VRM heatsinks with Ceramique. By doing that, I'm pretty sure I can lower the motherboard temps enough to reach 3.8GHz or higher, and that's with the "Intel compiler problem" still there! Time will tell!

Russ

 

The one thing that's being overhyped here is the suggestion that AMD CPUs are slower than Intels entirely due to this compiler incident. That isn't true. Not every program is written using Intel's compiler. The programs that have a massive advantage for Intels are the examples here. Even when this matter is resolved, AMD CPUs aren't suddenly going to be as good as or better than Intels, they are still behind. Eliminating this scenario merely eliminates the excessively biased programs.