The New AMD Building Thread

Maw: I think perhaps you underestimate Intel's wealth. It would take an astonishing amount of money to 'cripple' Intel right now, even in the current financial scenario. Intel have done al they should be required to do, remove the code from the latest version of the compiler - every program from now on should be on even ground. Having to rewrite all the existing ones should really be an instructional - Intel should produce a bulletin for the software developers on how to recompile the software or whatever, but this is up to the software companies to implement, not Intel, it just isn't a doable task otherwise, and not every company is either going to still exist, or want to bother with all this.

 

Sammy hit the nail on the head with that one, the only thing I would add is I would demand they do it right away not years from now.

 

It could possibly be something done within the OS. Ala AMD's dual core optimiser. It can't be quite as profoundly screwed up as it seems.

Maybe I'm wrong though

 

If it's as simple as that, I agree. I do not know much about compilers or cpu micro architecture so I don't know if it's just a case of enabling/disabling SSE2/3 or something more and as such I tried to not comment on that part. If it's as simple as it sounds then intel may have violated the agreement they signed which is quite disappointing (in both cases) and I would be one very upset AMD owner right now.

Unfortunately it will take years for all the software compiled on the "old intel" compiler to be replaced and we won't see AMD on a real level playing field for quite some time.

I probably have, although I would imagine replacing all the programs compiled with their "special" compiler would take an astonishing amount of money.

 

Sam,
I think some sort of a patch that either turns off the offending bits, or changes the CPUID so programs think it's an Intel is what the Government is after. That's the very reason Intel's solution is not acceptable to our government! Intel would enjoy an unfair advantage for several years to come, until all the old software is upgraded or replaced. That's why the courts called Intel back to Court in January, because Intel's solution won't change anything for several years to come! The agreement that both companies signed means now, not 3 or 4 years from now! Intel can't defend it's position, because they have agreed to stop doing it, and the Courts insist that it be done now! It's not a difficult thing to create a patch to eliminate the problem, and the Courts know that! After all, that's how the problem was discovered in the first place!

Russ

 

Recompiling every bit of software ever written for several years is bound to throw up a fair few problems. I'd be surprised to see it implemented.

 

Red_Maw,
I think that's what the Courts have threatened Intel with. Either get it done now the easy way, or risk being ordered to recompile any and all software that's affected by Intel's actions. I'm almost positive that Intel will comply! They would be very foolish not to. I'm sure that's what this Federal Trade Commission suit is all about, to put more pressure on Intel to comply!

Russ

 

Wow so "ffurther down the rabbit hole" so to speak. Intel have really dug themselves a hole here. I can't say they've let me down too badly as they make a good product besides, but this business with crippling AMD chips is unacceptable.

Does this mean that instead of being a clock-for-clock equal, my 940 is actually potentially MUCH FASTER than my Q6600? My mind reels.

 

depends on the app, depends how criples it was on that perticular app,a nd depends if this is even true at all.

 

Well I don't doubt it's true. This is a major issue that the US government has forced itself into. I think we'll be seeing a global OS patch sometime soon.

 

no major site has said anything about it, NOTHING! there were all over the 1.25 billion payout.

where does it show the government intervenign abotu this?

 

I'm interested to see what comes of this, if anything, in terms of performance, but I don't expect much to change at all with regard to how powerful CPUs are. Even if Intel do release a patch for this (It's possible the story is being kept quiet due to an ongoing legal process) I'm doubtful it will change the state of play for AMD vs Intel - I just think that in X program where the Intels absolutely smash the AMDs for no reason, they'll catch up to where they are in all the others. I do not expect the X4 630 to suddenly outperform the Q8300, the X4 940 to suddenly outperform the Q9400, the X4 965 to suddenly outperform the Q9650, the X2 7750 to suddenly outperform the E5200, and so on.

 

No I don't think so either. It won't be like a major performance upgrade. But I might see some nice improvements in things where AMD was initially lax.

 

Sam,
I just got to thinking about what changes might occur, and I think you might be right! However, you know me when I start thinking! LOL!!

Just suppose that AMD has had to do certain tweaks or make certain exceptions in order for their CPUs to run acceptably under the current conditions. Suppose AMD has had to work around these bottlenecks and roadblocks the compiler library has placed on them by using slower code? After all, I'm sure that AMD has had to make some compromises because of that slower code, so what happens if AMD can undo some of those compromises and regain some of their lost power they've had to give up in certain areas to run acceptably in areas more crucial to overall performance? I don't doubt that AMD has had to rob from Peter, to pay Paul in order to perform better, so once the roadblocks are removed, they will no longer have to do so. At that point AMD can optimize their CPUs in the same manner as Intel does.

I'm not expecting miracles here, but rather a refinement that a level playing field will allow. I know Shaff would like to believe that all this is BS, because he doesn't respect the sites reporting what's going on here, but there is far too much information out there to be ignored. Look at it from a business standpoint. It is known fact that Intel paid AMD 1.25 Billion Dollars. No huge corporation forks out that kind of money without a fight, yet Intel just handed over the money with very little comment, preferring to beat the drums for the Agreement both companies signed. Things like that don't happen without reason. The agreement between the two companies itself, is a matter of public record! It happened, so it can't be BS.
http://download.intel.com/pressroom/legal/AMD_settlement_agreement.pdf

I would almost bet the Farm that Intel's lawyers told them that all they have to do is remove the offending code Library from their latest release of their Compiler, and Intel could wash their hands of the whole situation. Apparently Our Courts don't see it that way, as they rightfully shouldn't, and called Intel back to Court early last week to explain "The facts of Life" to them, and told them what they want done, along with a few veiled threats of what the court will order if they don't. The Government and the courts have their own computer experts, so they already know that this can be an easy fix. Intel can huff and puff all they want, but in the end they will comply with the court's order! We'll have our patch, and AMD will be able to optimize and refine the performance of their CPUs, in much the same manner as Intel now optimizes theirs.

Russ

 

This could breathe some extra life into my 940. I'm considering a graphics card upgrade in the future, and I'd like to stretch my CPU a bit further. I have in no way been disappointed by the performance yet. But it would be nice to know it's faster than I initially though.

 

You may struggle to keep a perfect 60fps, but a 3.7Ghz X4 940 is unlikely to dip into the realms of unplayability any time soon in anything except for the ridiculousness that is Arma 2.

 

Welcome back KOTTER

 

Heres my take after 21/2 weeks of building.The 945 or 965 are the processors to run whether on am+2 or Am3 boards. The 955 has to many quirks for my taste.Chris

 

That makes no sense. The 955 is the exact same chip with a different stock multiplier. It's no better or worse than a 940/945 or 965.

 

Chris,
First, welcome back!

I haven't heard about any quirks with the new 955 with the C3 stepping. In fact, quite the opposite. There are 2 versions, both Black Editions. There is the new one with the C3 Stepping.

http://www.newegg.com/Product/Product.aspx?Item=N82E16819103808
And the older one with C2 Stepping.

The ones with the GMBOX at the end of the model number are C3 stepping for all the Phenom II x4s. Here's the older 955BE. The model number ends in GIBOX.

http://www.newegg.com/Product/Product.aspx?Item=N82E16819103674

I've been keeping my eyes open because I'm planning on one of the new 955BEs for my next CPU, and I'm eager to hear about any quirks!

Best Regards,
Russ