Intel vs AMD

Inane baanter or words of wisdom - you decide...
User avatar
mattley
Baaaaaaa!
Baaaaaaa!
Posts: 6253
Joined: Thu Aug 01, 2002 3:50 pm 0
Location: Watford
Contact:

Post by mattley »

hairy_pete wrote:If your system is unstable, you built it wrong, or you got a faulty or under-spec component. Your fault, or your supplier's. Not AMD's.
I've never thought it was AMD's fault, but it's undoubtably AMD's problem.

...VIA I'm especially looking at you......

Now I'll concede that because AMD are percived are better value for the gamer, there may be a tendancy among their customer base to scrimp on other parts, especially motherboards, but it doesn't change the fact that AMD based systems are generally fussier than Intel.
vi vi vi the text editor of the beast.
User avatar
Stevo
Baaaaaa!
Baaaaaa!
Posts: 3545
Joined: Mon Jan 12, 2004 11:00 am 0
Location: Perth, WA

Post by Stevo »

mattley wrote:
hairy_pete wrote:If your system is unstable, you built it wrong, or you got a faulty or under-spec component. Your fault, or your supplier's. Not AMD's.
I've never thought it was AMD's fault, but it's undoubtably AMD's problem.

...VIA I'm especially looking at you......

Now I'll concede that because AMD are percived are better value for the gamer, there may be a tendancy among their customer base to scrimp on other parts, especially motherboards, but it doesn't change the fact that AMD based systems are generally fussier than Intel.
Via chipset sux...........n-force is the way to go
The most important people in your life are taken away too soon and the arseholes never go a away.
User avatar
hairy_pete
Baaaaaaa!
Baaaaaaa!
Posts: 7217
Joined: Mon May 27, 2002 8:10 pm 0
Location: A Tiny Village in Bavaria

Post by hairy_pete »

Having access to the (internal) errata sheets for everything we're talking about here, I can only laugh at statements like these :?

Let me say it like this: There are no components in the market right now where something would be horribly wrong or utterly incompatible with something else. Past that, every single component from every single vendor DOES have its share of minor hardware bugs and driver issues. You need to know your stuff in order to compose a system that actually works - and there will always be the luck-of-draw risk of combining two things that just won't work together. No matter who made them.

The desktop PC market suffers two major problems:

(1) Cheap boxes are often built way too cheap. Yes this affects AMD systems more than Intel's, simply because Intel has nothing to offer in the "rock bottom" price range. From underperforming power supplies, RAM with broken bit cells, cheap cooling fans all the way to downright poor assembly, there's lots of corners to be cut.

(2) Stuff gets shipped as soon as final silicon is available. At that point, firmware and drivers are far from being finished, and that shows.

So, if you build your own system from reasonably specced parts, and if you stayy away from bleeding edge presented-last-week stuff, you're safe with anyone's - Intel, AMD, VIA, SiS, ULi, Transmeta, whatever.
User avatar
Dio
Sheepie Whore
Sheepie Whore
Posts: 18887
Joined: Mon May 31, 2004 1:08 pm 0
Location: Newcastle UK

Post by Dio »

I'd add an additional problem with stuff that's far away from the bleeding edge: as components move into more price-sensitive market segments, there is added incentive to cut corners to reduce cost.

e.g. it's much harder to buy good CD-R blanks now than it was five years ago, because even the quality vendors have to some extent cut corners to reduce costs.

This certainly seems (to me) to be affecting Socket 478 parts a bit.
User avatar
manicx
Baaaaa!
Baaaaa!
Posts: 1209
Joined: Tue Jan 14, 2003 10:50 am 0
Location: Greece
Contact:

Post by manicx »

hairy_pete wrote:Yet another "AMD is unstable" spinner.

If your system is unstable, you built it wrong, or you got a faulty or under-spec component. Your fault, or your supplier's. Not AMD's.
Certain software/driver can interact badly with the CPU. As far as I am concerned, this is nothing wrong with building the system. Latest drivers, enermax PSUs, kingston SIMMs, thermatake fans, Asus mobos, Asus gfx cards .... Hardly underspec components. We both paid a fortune, space in the box is the best you can find. Still he has problems. SP2 for example had several problems with AMDs...
Mainboard's temperature readings are hardly ever accurate, and never comparable from one kind of board to the next. That's not a valid point either.
We used a special temp reader not the built in readers that you find on the coolers. Not sure if there is a more accurate component than a specialised temperature reader (used excessively in my work to monitor temps into our servers).
User avatar
hairy_pete
Baaaaaaa!
Baaaaaaa!
Posts: 7217
Joined: Mon May 27, 2002 8:10 pm 0
Location: A Tiny Village in Bavaria

Post by hairy_pete »

Exactly what problems do you think did SP2 have with AMD processors? Show me Microsoft Knowledge Base articles, no FUD please.

ASUS boards? Meh. Here's a company whose quality has gone down the drain. I mean it.
User avatar
D.K. Tronics
Baaaa!
Baaaa!
Posts: 630
Joined: Fri Jan 03, 2003 2:06 am 0
Location: UK

Post by D.K. Tronics »

Don't apply old concepts to new architecture. There is no FSB in A64. The RAM frequency is divided down from CPU core clock
Umm, not as far as I'm aware.
If this were the case, I wouln't have a bootable system due to my cheap PC3200 RAM and it's 210 mhz ceiling.
Also, again as far as I'm aware, HTT is often called FSB in some boards, and visa versa.
On Nforce 3 250 boards, they use both HT AND FSB, with RAM dividers.
So I can set a 235 FSB, 4x HT, and 166 mhz RAM divider.
The available RAM options are usually 200, 166, 133, 100
What this means is 1:1, .83:1, .665:1, and .5:1 respectivly with 1 being your HTT/FSB.
My RAM is currently at 166, so thats a .83:1 divider.
So, 235 FSB x .83 gives me 195 mhz that my RAM is now running at. Which also allows me to run with a 1T command, and 2x2x8 timing.
Perhaps, this is just an Nforce 3 option, but it's definately there.
Programs like cpu-z, are reporting a 235 FSB, as well.
User avatar
Dio
Sheepie Whore
Sheepie Whore
Posts: 18887
Joined: Mon May 31, 2004 1:08 pm 0
Location: Newcastle UK

Post by Dio »

hairy_pete wrote:Exactly what problems do you think did SP2 have with AMD processors? Show me Microsoft Knowledge Base articles, no FUD please.
I had to disable execute protect in the boot string or my machine was borked due to a series of complicated interactions between A64, SP2 and my firewall software.
User avatar
Firefox
Sheepie Whore
Sheepie Whore
Posts: 11412
Joined: Fri Apr 26, 2002 6:15 pm 0
Location: Sheffield

Post by Firefox »

hairy_pete wrote:ASUS boards? Meh. Here's a company whose quality has gone down the drain. I mean it.
/me looks at my Asus motherboard with VIA chipset... :oops:
User avatar
manicx
Baaaaa!
Baaaaa!
Posts: 1209
Joined: Tue Jan 14, 2003 10:50 am 0
Location: Greece
Contact:

Post by manicx »

hairy_pete wrote:Exactly what problems do you think did SP2 have with AMD processors? Show me Microsoft Knowledge Base articles, no FUD please.
There you are:

http://support.microsoft.com/?kbid=873158
This problem occurs when an incompatible version of the common language runtime is installed on a computer that uses an AMD64 processor.
As far as I am concerned, this is a feature for AMDs and the Itanium CPUs. Since I doubt there are several people with Itaniums, the bug reports came mostly from AMD users.
Here's a company whose quality has gone down the drain. I mean it.
Just like most companies and most products out there. Economies of scales ruined quality. Still, much better quality than several hardware manufacturers tho.
User avatar
hairy_pete
Baaaaaaa!
Baaaaaaa!
Posts: 7217
Joined: Mon May 27, 2002 8:10 pm 0
Location: A Tiny Village in Bavaria

Post by hairy_pete »

D.K. Tronics wrote:
Don't apply old concepts to new architecture. There is no FSB in A64. The RAM frequency is divided down from CPU core clock
Umm, not as far as I'm aware.
If this were the case, I wouln't have a bootable system due to my cheap PC3200 RAM and it's 210 mhz ceiling.
Also, again as far as I'm aware, HTT is often called FSB in some boards, and visa versa.
On Nforce 3 250 boards, they use both HT AND FSB, with RAM dividers.
So I can set a 235 FSB, 4x HT, and 166 mhz RAM divider.
The available RAM options are usually 200, 166, 133, 100
What this means is 1:1, .83:1, .665:1, and .5:1 respectivly with 1 being your HTT/FSB.
My RAM is currently at 166, so thats a .83:1 divider.
So, 235 FSB x .83 gives me 195 mhz that my RAM is now running at. Which also allows me to run with a 1T command, and 2x2x8 timing.
Perhaps, this is just an Nforce 3 option, but it's definately there.
Programs like cpu-z, are reporting a 235 FSB, as well.
Again, your picture of this new world is wrong.

There is no Front Side Bus - CPU and Northbridge are one chip.

Clock generation starts with a base frequency (what that silly BIOS sells you as "FSB"). HTT frequency is multiplied UP from there, so is CPU frequency - independent from each other. RAM bus frequency is divided DOWN from CPU frequency.

So for example, with an Athlon 3000+ and PC2700 RAM on socket-754, you got 200 MHz base frequeny, HTT on x4, CPU on x9, RAM on /11. When Cool&Quiet steps the CPU down to 800 MHz, it does so by changing the CPU multiplier to x4 and the RAM divider to /5.

If this had PC3200 RAM, then base clock input and RAM clock output would happen to be the same frequency - but the latter is not derived from the further, as was the case in northbridge-centric architectures.

Similarly to the temperature issues I mentioned above, all that BIOS mess is to present the users with terms they've heard of and "protect" them from having to comprehend new stuff. Your post is a fine demonstration of how this actually has the opposite effect.
User avatar
hairy_pete
Baaaaaaa!
Baaaaaaa!
Posts: 7217
Joined: Mon May 27, 2002 8:10 pm 0
Location: A Tiny Village in Bavaria

Post by hairy_pete »

manicx wrote:
hairy_pete wrote:Exactly what problems do you think did SP2 have with AMD processors? Show me Microsoft Knowledge Base articles, no FUD please.
There you are:

http://support.microsoft.com/?kbid=873158
This problem occurs when an incompatible version of the common language runtime is installed on a computer that uses an AMD64 processor.
As far as I am concerned, this is a feature for AMDs and the Itanium CPUs. Since I doubt there are several people with Itaniums, the bug reports came mostly from AMD users.
Ah, that. As you can see for yourself, and as the alert message itself quite clearly says, that's not a "problem of SP2 with AMD processors". It's a software bug that AMD64 "Execute Disable" feature helped detect - in fact, the very kind of (potentially exploitable) software bug the XD feature was designed to prevent. Thank you for demonstrating AMD's superiority.
It crashes with XD-enabled E step Intel Prescotts just as well, btw.
User avatar
Dio
Sheepie Whore
Sheepie Whore
Posts: 18887
Joined: Mon May 31, 2004 1:08 pm 0
Location: Newcastle UK

Post by Dio »

I'm not attacking AMD, I have many of their systems and am on good terms with several people there.

But the fact is while I've no arguments that my problems were down to application bugs, they only occurred with AMD systems with SP2. It is not AMD's fault, and I don't expect it to persist, but it's still a problem that I've experienced.
Last edited by Dio on Wed Jan 12, 2005 3:09 pm 0, edited 2 times in total.
User avatar
manicx
Baaaaa!
Baaaaa!
Posts: 1209
Joined: Tue Jan 14, 2003 10:50 am 0
Location: Greece
Contact:

Post by manicx »

:lol: Sorry, a CPU never highlights a bug. Debugging does. However, the bug exists when used in a particular CPU so people who have AMDs have to do the workaround.

As for the Prescott, I hardly heard any people mentioning a problem. Me having a prescott, didn't have a single problem. My friend did tho...
User avatar
hairy_pete
Baaaaaaa!
Baaaaaaa!
Posts: 7217
Joined: Mon May 27, 2002 8:10 pm 0
Location: A Tiny Village in Bavaria

Post by hairy_pete »

That's because you don't have an E step Prescott, neither does the general public. Intel retrofitted the XD feature only recently, after AMD introduced it and MS found it useful ...

Of course do bugs exist that are only revealed by certain hardware. There is plenty of code out there that "works fine" for everyone but the next guy - the most popular reason being corner-cutting where proper software engineering would have been needed. This is no news, at least not if your business is creating firmware and drivers *hint hint*.

The "execute disable" is a CPU hardware feature, preventing code execution in areas marked as "data". This was introduced to make buffer overflow exploits impossible, and has the side effect of kicking out any software that purposefully attempts to execute code buried inside their data. This is a major coding style mistake, going unnoticed until confronted with a CPU that detects it.

Windows XP does have a "whitelist" for the XD feature, btw. Just add your faulty software there after it's been kicked out once.
Post Reply