AMD vs Intel

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.

slvr_phoenix

Splendid
Dec 31, 2007
6,223
1
25,780
But I'm not talking about 5-6 years old chip. The P3 is no more current as well as the thunderbird. We have NOW Prescott and A64.
Sorry, but as wusy said, you don't seem to understand architecture. I'm not talking about 5 year old chips. I'm talking about the fact that even the A64s and Opterons of today are all just reworkings and improvements to the same architecture that the original Athlon core was based on. You can walk right back through the changes very easily to their beginnings. Just as the PM is based on the P3's architecture. (And really, we could trace that one even further back.)

And without getting that basic fundament right, the rest of your post falls to pieces and isn't even worth commenting on.

<pre><font color=orange><i>Jesters do oft prove prophets.</i> -Regan in
King Lear (Act V, Scene iii) by William Shakespear</font color=orange></pre><p>@ 189K -> 200,000 miles or bust!
 

pat

Expert
So first of all, you claim that AMD forced Intel to use AMD's 64-bit set. Is that really a great accomplishment? What this means is that AMD had to spend the money, time, and work to get software companies to open up to the idea of using 64-bit software (which still isn't mainstream, and won't be for some time). Then Intel comes along and says "ok, *tweak*, now our hardware can be used by those software companies too...Thanks for blazing the trail, AMD".

Ok.. I will try to make my point clear this time. I've been using 8 bits cpu..then 16 bit..then 32 and now, we have 64 bits. What I mean is that it takes someone to start the trend. This time , it is AMD. I remember the trill it was from switching from 8 to 16 bits computer. They were fast.. And then , going to 32 bits. Wow... I like technological avancement.. always going further. And now, not yesterday or tomorrow, but now, AMD set the pace to newer trend.

We all know the Wright brothers, they set the path to modern aviation. But before them, there was other man who attempt to fly in funny machine.. Maybe they were the AMD of aviation, setting up a trend even if others were thinking they were idiots.

The Wright brothers got the first flying machine with engine .. They should thanks all those who got terrible idea and get killed in their machine.. they did teach them, not the thing to do, but the thing to avoid. And that matter too.

<font color=red>Sig space for rent. make your offer.</font color=red>
 

pat

Expert
I might not have used the right term in my posts, but I still believe that AMD did make thing happen, for the better or the worst.

My point is, if you want to do something, start something..


AMD did start the 64 bits train... Is that for the best, or the worst, I dont care, they just did it, and if trend in CPU keep going as it always had, then, 64 bits will be better than 32 that was better than 16 which was better than 8.

AMD did design their CPU with multi processor/multicore in mind. They knew it was going to happen, they were ready.

I know that I look like an AMD fanboy. Well, maybe I am.. because I'm excited to see new thing and ideas happening in the computer field, other than high MHz..

<font color=red>Sig space for rent. make your offer.</font color=red>
 

gomerpile

Distinguished
Feb 21, 2005
2,292
0
19,810
I remember the day of the 100 MHZ main boards hitting the market I believe that Intel was the first to produced the 100 mhz with the Intel 820 chipset designed for rdram ram but that was expensive and was not selling, out came the memory translator hub on the pc3-2000 main board but this was a call back and then the slot 1 processor was a failure and disappointed almost ever Intel user and that is when Intel sat still and watched AMD climb the ladder and today in my opinion AMD won the gamers hearts. That is what started the war between Intel and AMD. I too was an Intel fan and truly believed that they were the best, I was wrong. I switched to AMD design and was amazed by the performance of the processor and main boards.

I do believe that Intel has made systems what they are today, because of its failures systems got better for the user, I lost trust with Intel processors and chipsets on the expense of my dollars after the 820 chipset.
 

P4Man

Distinguished
Feb 6, 2004
2,305
0
19,780
>So first of all, you claim that AMD forced Intel to use AMD's
> 64-bit set. Is that really a great accomplishment?

Oh yes it is. For several reasons. First, and maybe most importantly, AMD64 has guaranteed the long term viability of x86. You might have forgotten, but not much more than 5 years ago, most people expected the world to abandon x86 in favor of IA64 (Itanium) as the need for 64 bit computing would make x86 obsolete. Intel can claim all they want, I'm quite certain their plan was to move IA64 down to the desktop where and when 64 bit became a requirement, and sooner or later, that will include pretty much all (non embedded) markets. If it wheren't for AMD64, and intel feeling the heat, we all would have had substantial less choice than we have now. You really can not overestimate the importance of this; its even felt in the high end RISC market where Power, Sparc, IPF and others are being more and more pushed into shrinking niches with x86 (64) growing into higher end every year.

If AMD had not successfully pushed AMD64, 5 or so years from here they wouldn't have a market to compete in anymore. x86 would have dried up, IA64 too patent ridden to make a clone. 32 bit cpu's just wouldn't sell anymore. Yes AMD64 *is* a significant achievement.

>What this means is that AMD had to spend the money, time,
>and work to get software companies to open up to the idea of
> using 64-bit software (which still isn't mainstream, and
>won't be for some time

AMD invested in it, off course, but it really wasn't that expensive. From a hardware POV, adding 64 bit extentions is almost trivial. Of course, it does require a very significant change of the core, but if you are designing a new one anyway, extending it is easy. From the software support side, AMD only did a minimal effort, simply because they can't afford much more, but Jerry going on trial in defense of Bill Gates was a very cheap way to get MS to support their ISA :D

Then there is another relatively big feat about AMD64: server space. Its not only its performance than made Opteron crack into the tier one server market, it was also its 64 bit ability. Opteron now holds 10-15% of the server market, a highly lucrative niche AMD previously held about no market share in at all. It now has its chips sold and marketed by Sun, HP, IBM, Cray and other high end server OEMs. It gets motherboard support from companies like Tyan, iwill, supermicro, it gets chipset support by nVidia, ATI, Serverworks (!),... Contrast that with a couple of years ago, where taiwanese motherboard vendors would not even dare publically sell a desktop board for AMD chips, and we only had white box boards with either AMD chipsets or VIA crap. This change is fairly phenomenal, and AMD64 played a significant role in this change.


Nah, IMO, extending x86 to 64 bits really isn't much of a technical achievement; "anyone" can do that, but making it a success, and getting intel to adopt it achieved quite a lot for AMD:
-it ensured its core market (x86 cpus) will continue to exist for the foreseeable future.
-it killed off any hopes for intel to make Itanium, and their associated ~$10B investment mainstream and/or profitable
-it gave them a significant foothold in the lucrative high end server market
-it gave them the image of being an industry leader, not just a follower or cloner which is extremely important to get adopted by big corporations.

All that for no more than a maybe a (few) hundred million $ (my WAG) is a fantastic deal.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
 

mpasternak

Distinguished
Apr 27, 2005
533
0
18,980
I think the important thing to realize between AND and IMTEL is that without one or the other we're all royally SCREWED

they're direct competitors. Copetition breeds innovation. both have come up with wonders and both have made flops. (anyone else remember their first generations of AHTLONS exploding and smoke comming out of thir computers or 2nd degree burns on their fingers from it?.. just me? sheesh)

they both have influenced the PC market today. who's ahead now doesnt mean they'll be ahead 2 years from now. and vice versa.
 

wolverinero79

Distinguished
Jul 11, 2001
1,127
0
19,280
10-15% of the server market? What?

Who's your source?

Here's a clickly that shows in May 2005 that Operton had 5.7% of the x86 server space (so that doesn't include the non-x86 server space)...

<A HREF="http://computerworld.com.sg/ShowPage.aspx?pagetype=2&articleid=1299&pubid=3&issueid=49" target="_new">http://computerworld.com.sg/ShowPage.aspx?pagetype=2&articleid=1299&pubid=3&issueid=49</A>

An interesting quote
""Opteron is still a viable product," said Joseph Gonzalez, a Gartner analyst. "You can run a lot of your older applications on it. It's also a little bit lower-priced than some of the Xeon offerings," he said. "It's not going to make a huge inroad into Intel's sales, but it's going to bring in a lot of revenue for AMD.""

I believe only AMD's most extreme optimistic projections have them at 12% by the end of the year. I'm curious where you got your numbers from.

I'm just your average habitual smiler =D<P ID="edit"><FONT SIZE=-1><EM>Edited by Wolverinero79 on 06/03/05 03:34 PM.</EM></FONT></P>
 

Atolsammeek

Distinguished
Dec 31, 2007
1,112
0
19,280
Here one thing Wolverinero79. If you read anything on Tomshardwaer site about Dual core cpu and Hypertreading you understand there differnt.
 

wolverinero79

Distinguished
Jul 11, 2001
1,127
0
19,280
Well obviously, but the concept and the pavement needed by software companies in order to support a multi-threading application environment are similar. You could theoretically argue that other things really started the multi-thread revolution, but it seems like (and at least what the industry recognizes) is that HT was the start of the push of multi-threading to average joe consumers (instead of giant servers or workstations). Dual-core is another step forward, but is building upon the work done by HT.

I'm just your average habitual smiler =D
 

HansGruber

Distinguished
Apr 28, 2005
238
0
18,680
An interesting quote
""Opteron is still a viable product," said Joseph Gonzalez, a Gartner analyst. "You can run a lot of your older applications on it. It's also a little bit lower-priced than some of the Xeon offerings," he said. "It's not going to make a huge inroad into Intel's sales, but it's going to bring in a lot of revenue for AMD.""
Those are just trying to lick Intel, or are thingking K6 era, or Slot A Athlons.. :)

"Opteron still viable"

In case some don't know, the Opteron is best x86 server CPU by a <A HREF="http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2397&p=10" target="_new">HUGE</A> margin.
I guess that it probably makes it "still viable", and that it "can run a lot of your older applications"..

:D

Am i starting to sound like a fan boy ? :|
 

P4Man

Distinguished
Feb 6, 2004
2,305
0
19,780
>10-15% of the server market? What?
>Who's your source?

Ahmm... can't find any sources to back up that claim :)
I seem to recall I read a report stating just that, but it most likely was just of some subniche where opteron has a 10+% marketshare. Probably 2 and/or 4 way servers. Remember, in volume (not revenue), the bulk of the server sales are low end, glorified 1 socket destops that companies like Dell sell as entry level file/print/whatever "server". Often even with Celerons. If you look at 2 and 4 socket systems, I expect Opteron to do a whole lot better, though I can't find any numbers.

One more thing: keep in mind Opteron has only been enjoying significant tier 1 support for a relatevily short time. Sales cycles in these markets are also long, so you're not seeing the whole picture yet. check back in another quarter or two.

Anyway, none of this hardly changes my argument, unless you want to argue Opteron/AMD64 has not enabled AMD significant (for them) leverage in the high margin server market.

As for the Gonzales/gartner quote.. doh!. Wheren't Gartner the guys that predicted Itanium crossing over with x86 by 2004 ? They are good at getting industry numbers, but their "<b>anal</b>ysis".. well, is just that :)

= The views stated herein are my personal views, and not necessarily the views of my wife. =
 

Cybercraig

Distinguished
Dec 31, 2007
2,058
0
19,780
the fact that even the A64s and Opterons of today are all just reworkings and improvements to the same architecture that the original Athlon core was based on.
No doubt about it. The first look at the A64 design as posted on THG made it clear.

Abit IS7 - 3.0C @ 3.6ghz - Mushkin PC4000 (2 X 512) - Sapphire 9800Pro - TT 420 watt Pure Power
Samsung 120gb ATA-100 - Maxtor 40gb ATA - 100
Sony DRU-510A - Yellowtail Merlot
 

endyen

Splendid
The first look at the A64 design as posted on THG made it clear.
That is an old and rather feable argument, and I have grown too sick of it.
What that shows is that Amd is using the same basic layout as they did with the Tbred b. Layout is such a small part of architecture as to make thier deduction laughable.
If components were interchangable at that level, exactly 0 parts from an xp chip could be used on an A64. Everything about the components has changed, except where they are positioned on the chip.
Why have they kept the positioning? Because it helps minimize crossover current. That's a real problem, that Intel just cant seem to get thier teeth around.
Would Intel like to see Amd change thier layout? You bet, then Amd's leakage current problems would be almost as bad as Intel's
Does keeping something that works mean Amd's architecture is unchanged? Only to those who haven't looked a little deeper.
 

pat

Expert
No no, it's not about layout. It's about how the logics and the flow diagram works.
Much of how the processor interprets the given data from the days of the first Athlon is somewhat the same to Athlon64.

Ask yourself why they kept the name 'Athlon'.

Being all designed to interprete x86 code, you cannot expect big change in logic...

They keep the Athlon name because it is a well recognized name now..




<font color=red>Sig space for rent. make your offer.</font color=red>
 

fishmahn

Distinguished
Jul 6, 2004
3,197
0
20,780
A couple years ago, I read an article on the internal architecture changes between the Athlon XP and Athlon 64 CPUs. If I remember right (I'll go find the article or a similar one soon), aside from the add'l instructions (SSE II/III, 64-bit registers, etc.) and the ODMC, they took the basic XP logic units unchanged, and enhanced the decoder, scheduler and branch predictors to increase the utilization of the LUs.

Mike.

Edit to add link to an article re. A64: <A HREF="http://www.cpuid.org/K8/index.php" target="_new">http://www.cpuid.org/K8/index.php</A> Turns out it wasn't completely unchanged from the K7, but largely unchanged...
<P ID="edit"><FONT SIZE=-1><EM>Edited by fishmahn on 06/06/05 11:40 AM.</EM></FONT></P>
 

Cybercraig

Distinguished
Dec 31, 2007
2,058
0
19,780
That's what it looked like to me too! A souped up XP! There's no question it works great, but I don't think it was all that radical. I just don't feel much like hunting up the documentation. I must be too "feeble". [action]smacksEndyenlikebiatch[/reaction/shotby9mm]

Abit IS7 - 3.0C @ 3.6ghz - Mushkin PC4000 (2 X 512) - Sapphire 9800Pro - TT 420 watt Pure Power
Samsung 120gb ATA-100 - Maxtor 40gb ATA - 100
Sony DRU-510A - Yellowtail Merlot
 

fishmahn

Distinguished
Jul 6, 2004
3,197
0
20,780
The core wasn't radical, it was all the other integration that was... ODMC, HyperTransport, Exclusive L1/L2 cache (I think that's what it's called - where it doesn't duplicate the data in a lower-level cache), effectively enlarging the cache without extra transistors. Can't do much to a core that is already a nicely balanced (balance between clockability and IPC) RISC core with 6 (I think its 6 - 3 ALU, 3 FPU) execution engines. Just tweaks...

Mike.