is AMD a better bet than Intel

[X3M]

I was looking for the same answer a wile ago. What I got is; if you are doing mainly 3d than AMD is faster because of its fast floating point calculations. For compositing and video Xeons are faster. I cannot prove it, but this is the best answer that I had got... :)

[GregHess]

1) Depends on Individual App in question.

2) Depends on the software your using.

3) Depends on the program your using.

If you list the apps your using, your budget, and whether you routinely upgrade your machine (or buy new ones), we can better answer this type of question.

[mrcz]

Depends on the application. Speaking about CPU only, Lightwave has optimization for iP4 for example, so you get short render for same money (look in tomshardware (http://www.tomshardware.com))
I have both Athlon and P4 and it is absolutely true.
P4 is not so hot as Athlon so you can use quieter cooler.
But on the other hand the chipsets for both platform are different. You will be probably limited to 1GB RAM (2x512 DDRAM )on standard intel chipset (because who have enough money for 2x1024 DDRAM ) etc.

IMHO you have to set up your priorities.

Easy question not so easy answer.

[GregHess]

P4 is not so hot as Athlon so you can use quieter cooler.


This statement was true until recently. Now Pentium IV processors (starting with the 2.8) are as Hot if not HOTTER then Athlon CPU's.

Here's some quick data from www.intel.com and www.amd.com...

The Pentium IV 3.06 HT produces 81.8 Watts.
The Pentium IV 2.8 produces 68.4 Watts.

The 3000+ Barton produces 74.3 Watts of heat.
The 2800+ Barton produces 68.3 Watts of heat.


Just thought you should know.

[singularity2006]

why exactly doesn't Intel and AMD just stick with a larger contact die? Something perhaps the size of the CPU itself? Why did AMD start using something so small in the first place?

[GregHess]

Both AMD and Intel have similar die sizes. Just intel uses a headspreader on their die to provide for a better mounting surface.

Here's a shot of what it actually looks like.

http://www.overclockers.com/articles486/

I'm aware this is a williamante class CPU (Socket 423). The same basic heatspreader design is used on the northwoods

Found a Northwood photo.

http://www.hardocp.com/images/articles/104317361585CkZAmyBz_1_12_l.jpg

Amd choose not to go this route (why I don't really know) possibly because the heatspreader reduces cooling efficency about 4-6C.

I believe Opteron and Athlon 64 have heatspreaders.

[mrcz]

>>GregHess

Thanks for correction. You are absolutely right if you compare technical data only.
BUT having an optimized aplication like Lightwave, you must compare same performance - iP4 2.0 GHz (54.3W) versus Athlon XP 3000+(74.3W)

So from that point of view is Pentium less hot then Athlon.

[Dave Black]

I switched from AMD to intel after I saw a demonstration on tomshardware.com between a p3, a p4 and (at that time) a thunderbird.

The test was:

3 identical pcs, running quake 3.

Each PC would have it's CPU heatsink and fan removed...This is what happened:

P3: Crashed, on reboot everything was fine.

P4: Clocked it's self back and just slowed down. Once heatsink was replaced, CPU when back to normal running speed.

AMD: Chip caught on fire in about 3 secs. Toasted the MoBo, and the memory.

So this is how I see it. Even if the 2.8+ P4 is hotter, it still won't burn down your house.

Do you want to trust the fan on your CPU?

Ok, so it's fear-mongering. hehe..

And, please, if this has changed, I stand corrected.

-3DZ

:D

[GregHess]

And, please, if this has changed, I stand corrected.

I'm sorry you based your purchasing decision on a website reknown for falsifying data and information.

Starting with the KT333 chipset, the Athlon's have had overheating and shutdown protection systems on a variety of motherboards. The original article on craphardware.com has been proven incorrect by a wide array of hardware websites.

iP4 2.0 GHz (54.3W) versus Athlon XP 3000+(74.3W)

The performance gap is not that large. More likely a 2400+ vs a 2.0 P4.

[Dave Black]

Thanks, Greg, that's really good to know.

I was of the understanding that tomshardware was a pretty good site. I'll have to look into it further. I really value your input, and I'd love to know what sites you think are honest and more reputable...

So AMD is looking up then...

-3DZ

:D

[GregHess]

Do a search on this forum for...

tomshardware.

There is a thread called

"Where do you get your info?"

That has a variety of site recommendations from a host of cgtalk members.

[Dave Black]

Thanks, Greg.

If it helps, none of that stuff about tomsharware was happening when I put my machine together. That sucks.

Just goes to show me that I'm not done stepping on landmines...

Sorry if I pissed you off, man. I really thought that the test I saw made alot of sense.

I really hope I don't go looking and decide to switch back to an AMD...that would suck. ;)

-3DZ

:D

[FreeQ]

Kill me if you want. But I can't resist to be quiet.

Alpha Processors have the most human ability of designing the core. I*tel kills it for pleasure. Not for money, just they can do it. :shame:

Don't ask anyone who is the best. Ask who is evil?

AMD just another crap X86yng. :shame:

I'm sorry for my ...yng.

PS: AMD cores has less transistor and less cycles than intel cores. But price/performance ratio is amazing. Why? AMD says we used EV6 bus system... maybe there is Alpha core inside. ;)

[elvis]

freeq: were you drunk when you typed that? what the hell are you talking about???

"Alpha Processors have the most human ability of designing the core"?????

"AMD just another crap X86yng"????

get off the weed, man!

[FreeQ]

I'm not drunked yet ;)
I mean RISC is better than CISC always ;)

http://www.amd.com/us-en/Corporate/VirtualPressRoom/0,,51_104_543~69124,00.html

This news cool.:applause:

[elvis]

Originally posted by FreeQ
I mean RISC is better than CISC always

"always"? sorry, but that's the biggest load of tosh i've heard today.

this is the second warning i dish out for wild statements in the last 10 minutes. clarify your conditions, express the situation clearly and reaffirm your statement. your just tossing throw-away lines out at the moment.

[liumm]

I'm kinda noob,sorry if this is stupid.

So...what are all the processors we are using now---CISC or RISC?

[FreeQ]

I love you Elvis :) even I don't get you.


lium;
AMD, Intel and VIA are CISC.


----
Please look this pathetic situation!
http://www.theinquirer.net/?article=8691
I know inquirer is a exagger but this is true!

Do you know 1 GHz Alpha paper-out date? Shame on you i*tel. :shame:

[elvis]

CISC (http://www.acronymfinder.com/af-query.asp?String=exact&Acronym=CISC&Find=Find) "Complex Instruction Set Computer", and RISC (http://www.acronymfinder.com/af-query.asp?String=exact&Acronym=RISC&Find=Find) is "Reduced Instruction Set Computer".

the analogy is this: there are to types of ways to tackle a problem. you can either do the whole problem in one step, or break it down into smaller steps and tackle each step independantly.

RISC chips, as their name implies, contain a greatly reduced number of "instructions", or features. these are usually broken down into simpler mathematical instruction sets (multiply, devide, add and subtract, shift left, shift right, etc). a complex task is broken down into it's more fundamental components, and each component tackled at a simpler level. RISC chips include some older architecture like the old MIPS chips, as well as modern day ALPHA chips and the IBM/Motorolla PowerPC chips found in Apple systems.

CISC takes the other approach, making the insturction sets on the chip more complex, enabling them to do more in a single cycle. Often CISC manufacturers will include things like MMX, SSE and the like into the instruction sets to again increase the complexity of a task that can be completed by a circuit in a single pass. CISC chips of the modern era are your typical x86 insturction sets such as the modern dat intel and AMD chips.

often you'll find people like our friend FreeQ above will throw such statements as "RISC always beats CISC" into the wind, and be pretty happy with themselves. often RISC manufacturers can lower production costs of their chips and increase the bandwidth of their systems, as in the case of ALPHA and PowerPC chips. this in turn results in more data being able to be processed simultaneously, but at the expense of the reduced instruction set.

CISC on the other hand is much lower bandwidth (nt by design, but generally by cost ratio), but compensates by computing more stuff in one pass.

the analogy of these methods is often described as "shallow and wide" versus "deep and narrow" in terms of both computing power and clockspeed.

certainly, in days of old anyone worth their salary given an indefinite budget would choose an alpha processor running some sort of proprietry unix as a workgroup server. these days the lines have blurred slightly and the MUCH cheaper RISC alternative starts to become very tasty to the large corporates who are constantly looking at reducing their IT budgets.

today there's no certain answer. one only has to look at the adobe website to see the "intel prefferred" pages to see that CISC has overtaken RISC once more to claim the performance crown in graphical processing. it's a heavily debated argument on the mac vs PC camps every day!

there is no simple answer. to say one is better than the other is comparing apples and oranges, ATi and NVIDIA, Mac and PC. depending on everything from the applications you are using on them to the operating system and kernel running below them.

horses for course folks.

oh, and FreeQ what I said above was basically a request for you to say what i just said above. state the case and situation to back yoru argument, and stop just throwing statements into the air.

[KayosIII]

Heres the deal....

AMD kicks Intels but when it comes to processing double presision floating point numbers.... Which makes it good for CAD and many 3D packages. (FreeQ - this is largely due to technology taken from the Alpha)

Intel is ahead when It comes to single presision floating point numbers. (Mostly used for games - but possibly some 3d applications)...

AMD is I think slightly ahead with Integers.

Intel is better for apps that use a lot of streaming instructions - Video, Audio that sort of thing... Especially if they are SSE2 optimised...

Ultimately you should go for price/performance rather than raw performance.

[TheClick]

Yo FreeQ,
Quit Babysittin the blunt and pass that weed, dawg!

[elvis]

Originally posted by KayosIII
Ultimately you should go for price/performance rather than raw performance.
my philosophy exactly. for my own personal use i've never yet spent more than AU$2500 on a system (roughly US$1000 to $1500). that always seems to be the price to performance sweet spot at any point in time for me.

[FreeQ]

I'm really sad now. Cuz I can't explain myself clearly. It's y fault and I'm sorry.

Since I took my forst SlotA Athlon 850 MHz price/performance is my philosophy too. But I talk about conduct/bussines thing.

I believe [IMHO] Athlon is a RISC. Builded by cooperate with Alpha engineers. Not only I/O Bus system. That desig used in main core also.

I'm not againts to AMD or Intel. I'm against to theft.

Elvis, you're right about CPU marketing history. Do you think if CISC do not push the market as much like as before then what can do RISC design nowadays?

BTW, I can assure you I never talk about my thinks on this board again. I really sad about my fault, may be I will go to buy itanic to punishment myself!

[FreeQ]

Originally posted by TheClick
Yo FreeQ,
Quit Babysittin the blunt and pass that weed, dawg!


I 'm sure this is fake nick. I never 'clicking' you!

[elvis]

Originally posted by FreeQ
I believe [IMHO] Athlon is a RISC. Builded by cooperate with Alpha engineers. Not only I/O Bus system. That desig used in main core also.
athlons are certainly not RISC. they are an x86 cpu, and by design and definition RISC. just because they had a few alpha engineers working on their bus design doesn't mean they are RISC! the instruction set of a computer and it's BUS system are two completely different topics.

it's pretty black and white: Reduced or Complex. i think you're getting your brand names confused with a simple definition here. :)

I'm not againts to AMD or Intel. I'm against to theft.
careful what you insinuate. anyone can make a CPU. IBM invented x86, yet intel and AMD are the major chip manufacturers today. there are hundreds of smaller companies making x86 compatible chips out there.

there's a big difference between intellectual property theft and making a alternative product that adheres to a standard. and without the latter, there's no such thing as an open market, and you end up with Apple Macs being sold at stupidly high prices.

Elvis, you're right about CPU marketing history. Do you think if CISC do not push the market as much like as before then what can do RISC design nowadays? see my comment above, as wella s my original statement. chip design is a moot argument if you aren't going to talk about the chip's role in a system.

BTW, I can assure you I never talk about my thinks on this board again. I really sad about my fault, may be I will go to buy itanic to punishment myself!
lord no! don't do that! i wouldn't wish itanic onmy worst enemy! :)

seriously tho: your thoughts and ideas are more than welcome. progress can't be made if people don't discuss (read: argue) about things. without alternative ideas, no alternatives are ever reached! </soapbox>

[KayosIII]

AMD is RISC... or rather it has a risc core.... It has a outerlayer to decode the extra X86 instructions...

I don't know about stealing -- The Alpha's technology was split 3 ways when Dec sold (I think it was to Compaq)... Intel, AMD and Samsung all have rights to the technology.

Of course there had been claim and counter claim between Intel and DEC for quite some time prior to this... I am not savy with the details of this -- somebody fill us in....

I believe that the Alpha is still avaible - AFAIK NT4 is the last Windows to run on it.... Of course you can use DEC Unix or Linux on it.... I think that it still has kick arse performance - especially with double precision floats.

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.