View Full Version : Why is AMD where it is?
stratcatprowlin
09-29-04, 04:55 PM
I think this was a really a great article.Also the prime reason I could not convince my friend the other day to save some money on an AMD system.
http://overclockers.com/articles1112/
AMD just has to face it...they will remain the underdog untill they can convince the larger audience of Joe Sixpacks that they are on-par with Intel - and that takes time, alot of time. Not to mention that Joe likes cheap, fast processors, which i believe at this time is a bit of a problem for AMD.
This is what I wrote to ed:
"I think that AMD is currently in the fiscal quandary of their current A64 lineup not being ahead enough in performance to justify the greatly decreased value per dollar. If AMD keeps pricing their products this way, they may end up like 15K hard drives versus 7.2K - one is obviously quicker, but not enough to justify the enormously inflated pricetag. However, if AMD begins pricing hammers down in the future, it may turn for the worse - since every time they have done that, the cost per unit goes through the floor and lose money. I suppose they feel they lose less money by selling few units at a high price than tons at too low a price. They can't slip to a happy medium because intel/market pressure will force their prices down to a below-optimal level, it seems.
I have built and used many Athlon XP systems, and personally still don't see any real need to shift to AMD64. It's the exact same reason I didn't go over to P4s when they were doing rather better than XPs - mobiles came out that could overclock to 2.5GHz+ and bring any perfomance advantage an overclocked p4 would have to nearly zero at nearly half the price. $100 Mobile XP 2600+ versus $170 P4 2.4C? The choice is simple.
This effect is even more exaggerated with the hammers, as the price disparity is even greater. I chime the tone of most bells when I say that until there is a SIGNIFICANT performance advantage, i.e. at least 50%, I won't upgrade. And I'll wait until that 50% faster part comes down in price before buying.
I recently invested a chunk in building an AMD dually with mobile bartons. Granted, it requires registered RAM, which isn't cheap, but still comes out cheaper than a 3500+ / 939 system. And two mobile bartons running at ~2.5GHz definitely have a lot more F@H computing power than a single 3500+ A64, even if it's overclocked on phase change anyway. :)
I'm not downplaying the A64; it's clearly superior to its predecessor for obvious reasons. It's simply that value per dollar is too much in the pits to make me (or many other people in my boat) to switch."
... if you believe that, you probably also believe the U.S. government has a fleet of alien spacecraft parked in a Roswell lot.
You mean they don't? OMG! My entire belief system is in shambles!
...Intel processors bursting into flames and killing users.
Oh my...I don't think I'll buy an intel any time soon... :)
Great article. Spot on.
CPUonNO2
09-30-04, 01:27 AM
Terrific article, hit the nail on the head, and fun to read!
c627627
09-30-04, 03:47 PM
Ed, you got PM re Inq author.
def a great article, and i 2nd the statement that he hit the nail on the head...
sup3rcarrx8
09-30-04, 09:02 PM
Whoa. That is like the greatest article ever written on AMD and why their speeds are where they are now. :thup:
[just a thought]
Don't you think AMD has guys that get paid figuring these things?
What will AMD do when it is confronted by this kind of conclusions again and again (more will follow, definatly). It may be great, it may be smart, it may be well-written. But the succes of it depends on what AMD does with it.
jiggamanjb
10-01-04, 03:30 AM
I hate reading, but that article was great. Just gave me more reasons to stick with AMD for life!
matttheniceguy
10-01-04, 05:15 PM
That precott get's any hotter and it just might burst out into flames and burn people's house down....
evowatercooling
10-01-04, 10:19 PM
This is what I wrote to ed:
This effect is even more exaggerated with the hammers, as the price disparity is even greater. I chime the tone of most bells when I say that until there is a SIGNIFICANT performance advantage, i.e. at least 50%, I won't upgrade. And I'll wait until that 50% faster part comes down in price before buying.
Alex Koik-Cestone (cV)"
I agree. A big problem that the A64 is having is the fact that they are not much faster then the XP. And since the XP systems have been very cheap for a very long time, anyone who buys AMD has no need to upgrade. You can run just about any program, any that the majority of buyers are using, with an XP 2600 and a good video card.
One thing that will give the chip some boost will be pci express. It would be a shame to buy the new video cards and have them be on the AGP bus. If you ever did upgrade in the future the $500-$600 video card that should be good for a long time is useless. I think a big part of computer sales in the future will come from the peripheral add ons required and not the computing power. This would mean AMD needs to force the user to upgrade if they want to use certain hardware because the processors now will be good enough for quite a while.
Snipester
10-02-04, 03:43 PM
AMD's marketing is crappy i think.
Opteron Sempron? WTFon?
First it was pentium then it was pentium II then it was Pentium III and then by a surprise attack it was Pentium 4....
Right now i have a system built from components my brother had thought were broken but the real question is...
Which socket 754/9027338924? on Athlon 64 or opteron or what?? Do i do to get a motherboard that works?
The transition phase is the worst part and AMD is in it with its CPU's AND as well as PCI-Express which makes it that much more confusing.
If AMD was more clear on it products then maybe people could understand more clearly yadayadayada blahblahblah.
Intel had to go through transformations but it was very easy to follow what was going on. (mostly from socket size changes) Obviously CPU's are going to rediculous speeds anymore but AMD is just letting intel back in the game.
I think AMD needs a tune and a new sleek name to be able to really compete for real marketshare against intel. Until then... its all intel. Remember gold star? (if you don't it was a reallllly ****tay korean manufacturer?) thats LG now... (lucky goldstar)
stratcatprowlin
10-02-04, 04:09 PM
I think AMD needs a tune and a new sleek name to be able to really compete for real marketshare against intel.
The only way that will happen is if they change it their name to Intel.:)
Good points all, except for one glaring error. AMD has allot more than 17% marketshare in the desktop market. It is these days somewhere in the high 40's. We are now seeing AMD topping 50% a few times this year. However, this is all a moot point. Success in this business will NOT be detimined in the desktop market at all, but in the server market. Currently, AMD has about 7% in this market. What really catches ones attention is that they have gotten this market share in just one year after the Opteron's release. AMD is making big inroads in the mid-range server market almost completely at the cost of Intel. Just my 2 cents worth...
nECrO
9mmCensor
10-02-04, 07:08 PM
Good points all, except for one glaring error. AMD has allot more than 17% marketshare in the desktop market. It is these days somewhere in the high 40's. We are now seeing AMD topping 50% a few times this year. However, this is all a moot point. Success in this business will NOT be detimined in the desktop market at all, but in the server market. Currently, AMD has about 7% in this market. What really catches ones attention is that they have gotten this market share in just one year after the Opteron's release. AMD is making big inroads in the mid-range server market almost completely at the cost of Intel. Just my 2 cents worth... nECrO
Welcome to the forums!
Um, my understanding was that there are more desktops than servers by a huge margin. So there is way more money in desktops than servers, for CPUs.
9mmCensor
10-02-04, 07:09 PM
The only way that will happen is if they change it their name to Intel.:)
1ntel should work...
I also think the article expresses what's going on perfectly. It's like some guy wrote down what we were all thinking, amazing how that happens...
One thing I don't like about AMD recently is the price of the A64s like was mentioned before. Their main strongpoint was the value they provided in their athlons, and that seems to have gone down the drain too.
speed bump
10-04-04, 07:54 PM
My favoriute part of the article was, when Ed makes the comment that basically goes with Intel on the ropes becuase of Prescott what does AMD do? They start beating themselves in the face.
Really right now is a good time to buy a great system for cheap. They aren't the greatest setups but a P4 3.2ghz on a socket 478 is pretty cheap, as well as a AMD a64 3200 on socket 754. They just areb't future proof. I bought my Mobile cuase it was a good value, but now I almost want to sell my current rig and build a cheap evil socket 478 P4 becuase it seems to be the best value right now.
What I don't like about AMD right now is they are going to get around to giving us the stuff that Intel already offers. I must say that before I drop alot of money into an AMD system it will be Dual core, PCI-E, and DDR2 on a futureproof socket.
AMD pushing themselves into a niche market by focusing on servers and high-end enthusiasts is a dream at best - they simply don't have the money to maintain their fabs and push massively upmarket at the same time.
If they try to force it, they may well end up bankrupt. Fortunately, some sense is coming in and the policy is cracking.
As for complexity, most average consumers don't have to worry about sockets and such because HP, Dell, etc will have figured it out for them. It's easier for the consumer to pay more for less, however, with that amount of optionery and confuzzlement available to the OEMs.
Many not-so-enthused whitebox folk and computer builders may just go the LGA775/s478 route because it's less convoluted. That is where the real loss is in that regard.
I agree. A big problem that the A64 is having is the fact that they are not much faster then the XP. And since the XP systems have been very cheap for a very long time, anyone who buys AMD has no need to upgrade. You can run just about any program, any that the majority of buyers are using, with an XP 2600 and a good video card.
One thing that will give the chip some boost will be pci express. It would be a shame to buy the new video cards and have them be on the AGP bus. If you ever did upgrade in the future the $500-$600 video card that should be good for a long time is useless. I think a big part of computer sales in the future will come from the peripheral add ons required and not the computing power. This would mean AMD needs to force the user to upgrade if they want to use certain hardware because the processors now will be good enough for quite a while.
The real value for amd64 will come in when it supports features that intel already does (i.e. PCI express) and then some (dual core). Features, not necessarily performance. We all know that the A64 is the technically far better chip, but it's the technical baseline that most people buy computers for. It's the reason why server boards don't typically have a boatload of USB ports and accommodations for many inexpensive expansion cards, as opposed to consumer desktop boards.
PCIe will give amd that me-too reason as well as features such as sse3 and such. I've noticed in the past that intel always tends to advertise features less prominently when amd has adopted them as well. If AMD had a wifi platform similar to centrino, for instance, intel wouldn't/wouldn't be able to display the stupid pink butterfly at every major worldwide air/sea/train terminal.
Of course that'll won't happen in the near future because AMD simply doesn't have enough money to compete heavily in the chipset business. But analagous things could happen and the results would be...funny.
When 64-bit comes into use, people will eventually figure out that intel's support is quirky at best and AMD's kicks more than I can describe here. Athlon will pull ahead of Pentium, Opteron ahead of Xeon and possibly Itanic. Again.
And AMD will royally screw up pricing and marketing strategy at another golder opportunity.
Again.
And this cycle will continue until something really bad happens to AMD and certain people in upper management get fired.
Susquehannock
10-06-04, 09:46 PM
Yeah, Intel sure has the general population brainwashed that
high MHZ is the only measure of performance. Never mind that AMD
CPU are capable of 50% more calculations per cycle than Intel
(6 vs. 9) ... everyone is hooked on high MHZ numbers.
It's kind of like saying "My engine can run at 8,000rpm".
So what? ... The REAL question is, how much Torque/Horse Power does it have? ;)
A friend asked me to build his first new system in over four years. And try as I
may, he just wouldn't go for an nForce2 rig. So I ended up building him a 3.0ghz
Intel instead.
Short story ... he about sh*t when we compared my benchmark scores to his.
Instead of buying a $280 CPU he could have bought a $42 chip like mine (1700 T-bred B dLt3c)
and OC it to run damn near as fast.
Yeah, Intel sure has the general population brainwashed that
high MHZ is the only measure of performance. Never mind that AMD
CPU are capable of 50% more calculations per cycle than Intel
(6 vs. 9) ... everyone is hooked on high MHZ numbers.
It's kind of like saying "My engine can run at 8,000rpm".
So what? ... The REAL question is, how much Torque/Horse Power does it have? ;)
A friend asked me to build his first new system in over four years. And try as I
may, he just wouldn't go for an nForce2 rig. So I ended up building him a 3.0ghz
Intel instead.
Short story ... he about sh*t when we compared my benchmark scores to his.
Instead of buying a $280 CPU he could have bought a $42 chip like mine (1700 T-bred B dLt3c)
and OC it to run damn near as fast.
I really agree with the first part. As much as i disagree with the latter of your post.
Comparing an extremely overclockeed 1700+ to a stock 3.0Ghz is just no deal. Either compare them both at stock, or both overclocked. The 3.0C needs alot more FSB to start shining. Also, what benches did you use? Dragonprince made a slip with that here (http://www.ocforums.com/showthread.php?t=331860) because he was comparing an overclocked amd vs. a stock intel. The 3.0C is likely to do 3.5Ghz or even 3.75 with a slight voltage increase - try that, then rebench vs. your 1700+ ;)
DrSlinky
10-07-04, 08:40 AM
I really agree with the first part. As much as i disagree with the latter of your post.
Comparing an overclockeed 1700+ to a stock 3.0Ghz is just no deal. Either compare them both at stock, or both overclocked. The 3.0C needs alot more FSB to start shining. Also, what benches did you use? Dragonprince made a slip with that here (http://www.ocforums.com/showthread.php?t=331860) because he was comparing an overclocked amd vs. a stock intel. The 3.0C is likely to do 3.5Ghz or even 3.75 with a slight voltage increase - try that, then rebench vs. your 1700+ ;)
My machine doesn't crap out until just over 3.7. This is using air cooling and PC3500. Also a 3.0C
Tell me what benchmarks you want me to run, and i'll post results, stock vs. OC, when i get home from school.
I have had a 3.0C myself, and noticed a great improvement running it on 3.75Ghz. 3dmark2001 and sandra profit the most from it. I was just trying to tell that to Susquehannock.
Susquehannock
10-12-04, 08:15 PM
I really agree with the first part. As much as i disagree with the latter of your post.
Comparing an extremely overclockeed 1700+ to a stock 3.0Ghz is just no deal. Either compare them both at stock, or both overclocked. The 3.0C needs alot more FSB to start shining. Also, what benches did you use? Dragonprince made a slip with that here (http://www.ocforums.com/showthread.php?t=331860) because he was comparing an overclocked amd vs. a stock intel. The 3.0C is likely to do 3.5Ghz or even 3.75 with a slight voltage increase - try that, then rebench vs. your 1700+ ;)
Cool ... you're right. Not really a good comparrison. Still, on a mhz to price
basis the OCed 1700 was probably a better deal. That was to be my main
point up^ there.
:)
Cool ... you're right. Not really a good comparrison. Still, on a mhz to price
basis the OCed 1700 was probably a better deal. That was to be my main
point up^ there.
:)
True, but then even a duron 1200 @ 1600 would outmatch the t bred, seeing is as its even cheaper...Thereby, the t bred has a dramatically low MFLOP / FSB bandwidth compared to almost any P4.
Do they still sell the xp1700+ where you live? The slowest XP i can get here is 2200+, they just ''dont sell'' the rest anymore...which is really too bad :eh?:
Susquehannock
10-13-04, 12:09 AM
Hmmm, didn't know the Duron would OC by over 1000 mhz. ... Nice! :)
I've read the 1600 Duron would OC well, but you had to do the bridge mod
to double the cache to 128k, right?
Even so they weren't any cheaper than the 1700 Athlon. At least not
around here. The 1600 Duron still goes for around $47.
There are a few places that sell the 1700 & 1800 t-bred. E-mailed several places
recently and none would say if you get an A or B core.
eXcaliberPC.com had the 1800 1.5v t-bred B (locked) for around $55 until last week. (gone now)
All my 1700 & 1800 were bought well before they started locking the multis. Purchased a
couple $42 1700 during a "Mother's Day special". One dUt3c ... one dLt3c. :D
Dylruss
10-13-04, 11:32 AM
Wow, it isn't a war on whats better, (even though when I had my bone stock 2400+ was better then my bone stock 2.4C, in almost all ways).
The article was well written though. It is near impossible to change the psyche of what people think about Intel or AMD, barring a catastophic event. I say well done Ed.
DocGiggs
10-13-04, 12:03 PM
As many people in the know realize.. computers themselves are not going to get much faster in the years to come. Moore's law should never have been called a law but theory because it won't hold true. Everything has physical limtations. Both Intel and AMD are reaching that wall. It's not about performance, compatibility or reliability anymore. It's as Ed says, it's about brainwashing.
McDonalds isn't the top fast food chain in the world because it makes the best food or even the cheapest. It's top because it's advertising brainwashed a few generations of people into believing it WAS the best and the cheapest. Regardless if that was true or not. Intel has done the same thing. I saw it day in and day out when I was working at either Circuit City, Bestbuy, or CompUSA. I've owned only TWO intel chips and that was a p133 and later a 333 celeron I overclocked to 500. Everything else after has been an AMD. Either a K6 or K7, to my current A64. I wouldn't have even upgraded to the A64 if my motherboard hadn't blown out for it.
But at those jobs, people wouldn't care if I recommended AMD. If it wasn't Intel, Sony, or Dell they wouldn't buy it. Period. No matter the arguement or the facts presented to them. People are just that dumb. It's hard to beleive sometimes but it's the fact. People don't want to think about it. Unless they get SERIOUSLY burned by a product they buy they will continue to buy that product. Sometimes even if they do get seriously burned.. it will only cause a moments hesitation before buying it again.
Don't get me wrong, I LOVE AMD and hope they stick around for as long as Intel does. Otherwise, we'd all be paying for $4000 computers again that would be worth $85 two years later (in the case of my p133 I purchased). There would be no "budget" computers anymore. And any company trying to start this late in the game is asking for an arse kicking. No one is dumb enough now to be another competitor to either Intel or AMD. Especially not to Intel. Smart shoppers are the minority and not the rule. Suckers are born every minute and that is the rule.
Good article, very truthful, but in the end though it was a geek article written by a geek for geeks. That's a shame too.
deception``
10-13-04, 01:07 PM
There are many points in this article that I do agree with. It is very true in assuming that AMD is essentially in competition with itself as it releases core after core and socket after socket. Essentially they are competing with themselves for A64 sales in the form of Newcastles, Clawhammers, COs, CGs, Semprons, s754, s939, and so on. This is probably AMD's biggest problem in terms of increasing their A64 sales.
However, I do not think that AMD is wrong for increasing the price of their cpus. Despite the fact that they are supported by our community and most overclockers, the firm has lost a lot of money on their AXP sales. Therefore, they need to look to make some profit somewhere. I understand people's desire for a substantial performance difference when upgrading from AXP to A64, but there definitely is a margin, especially compared to Prescott and Northwood P4's. Hell, there is practically no performance increase to be had when changing from Intel's P4C to P4E, and their 915/925 chipsets still trail behind most comparable Athlon 64 lineups. With that being said, there is a performance difference to be had that is certainly larger than that of a similar P4 upgrade, but it is not at the 50% like some are waiting for. Realisitically, however, I do not see such a performance leap being achieved on either side for at least another six months or so. It is true that AMD is having trouble with producing higher-clocked Athlon 64's, but Intel is currently having their fair share of issues with the Mhz game. This is why Intel has decided to back off from Mhz ratings in favor of a more AMD-esque PR system.
Essentially, AMD needs to do two things. First of all, it needs to stop competing with itself. Instead of all these different sockets and cpu types, AMD simply needs to stick with one or two sockets and market those in an effort to push a more uniform transition in the market. Secondly, AMD needs to make people realize that their Athlon 64 lineup is worth the price difference over their 32-bit offerings. Essentially, they need to wipe away this notion we have of them being a "budget company." This won't be a problem to most of the uninformed buyers, but on the OEM/enthusiast community we must realize that over time we simply won't be able to get away with paying $90 for blazing fast performance anymore. I symphathize with many people who want to pay the least for the most performance (as I am currently on AXP-M myself) but I do not mind paying 150-200 for an AMD64 cpu that will outperform a similarly priced P4C/E. My plans to upgrade are currently contingent on the release of DFI's s939 Lanparty, which will come with all the bells and whistles as my current mobo.
In the end, I do think AMD has a lot of work cut out for themselves. This time around, however, the company has a very mature lineup and is ready to step out of the value moniker for a more mainstream/enthusiast persona. The firm itself needs to make it's 64-bit offerings a bit more uniform, and we as a community (those cheap geeks out there) could stand to cut them some slack for pricing their A64's competitive to P4's, especially given their outstanding performance. Let's hope that AMD keeps up the good work...
deception``
AMD's marketing is crappy i think.
Opteron Sempron? WTFon?
First it was pentium then it was pentium II then it was Pentium III and then by a surprise attack it was Pentium 4....
Right now i have a system built from components my brother had thought were broken but the real question is...
Which socket 754/9027338924? on Athlon 64 or opteron or what?? Do i do to get a motherboard that works?
The transition phase is the worst part and AMD is in it with its CPU's AND as well as PCI-Express which makes it that much more confusing.
If AMD was more clear on it products then maybe people could understand more clearly yadayadayada blahblahblah.
Intel had to go through transformations but it was very easy to follow what was going on. (mostly from socket size changes) Obviously CPU's are going to rediculous speeds anymore but AMD is just letting intel back in the game.
I think AMD needs a tune and a new sleek name to be able to really compete for real marketshare against intel. Until then... its all intel. Remember gold star? (if you don't it was a reallllly ****tay korean manufacturer?) thats LG now... (lucky goldstar)
Excuse me? Did you say pentium4 2.4ghzA, B, C and pentium4 2.8ghzB, C, E were easy to follow?
deception``
10-13-04, 02:17 PM
Excuse me? Did you say pentium4 2.4ghzA, B, C and pentium4 2.8ghzB, C, E were easy to follow?
I agree completely. I am not trying to flame intel whatsoever, but their cpu revisions are becoming less and less unclear. Furthermore, the bitter fanboyism found within your post is definitely something that this thread can do without. Let's try to keep these forums as objective as possible.
Back to the topic at hand: yes, intel has had a number of different cpu types which are just as confusing as AMD's own. Let's break them down shall we? Beginning with deftor's list:
2.4a = 400 mhz fsb northwood
2.4b = 533 mhz fsb northwood
3.06 = 533 mhz fsb HT northwood
3.0 = 800 mhz fsb HT northwood
2.4c = 800 mhz fsb northwood
2.4A (Prescott) = 533 mhz prescott
2.8,3.0,3.2 E = socket 478 prescott
LET'S NOT FORGET THE NEW SOCKET!?!?
P4 520 = 2.8 Ghz LPA 775 prescott
P4 530 = 3.0 Ghz LPA 775 prescott
P4 540 = 3.2 Ghz LPA 775 prescott
P4 550 = 3.4 Ghz LPA 775 prescott
P4 560 = 3.6 Ghz LPA 775 prescott
Add to that the number of cpu and motherboard combinations one can have, including 865/875 LPA 775 hybrid motherboards, and you have one confusing bunch. I am not saying AMD is perfect here, but misleading information is a strategy found common amongst both companies.
deception``
The main reason why Pentium 4's benefit so much more from a dual channel memory config than the Athlon 64 or XP is exactly the same reason why they perform better with a larger level 2 cache than Athlon XP's and Athlon 64's.
Due to the extremely long pipeline of the P4, there is a much greater chance for branch misprediction, and hence, often data stored in the L2 cache is not required and the necessary data must be pulled from the main system RAM.
A Dual channel memory config allows the P4 to continue to operate more smoothly as it doesn't have to wait as long or often for data to be fed to it. This is also the reason why larger level 2 caches benefit the P4 so much and not Athlon XP/64's. A detailed explanation of the Pentium 4 architecture can be found over at Ace's Hardware. I'd recommend you check it out here: http://www.aceshardware.com/list.jsp?id=4
If Intel went with a integrated memory controller it would really scr*w AMD. Intel's architecture would benefit from it more.
Amusing really. AMD was first out with dual channel: it benefitted Intels P4 architecture more than the Athlon XP.
AMD is once more first out with an integrated memory controller: it will once more benefit Intel's P4 architecture more.
I suspect Intel is waiting for their FB-DIMM memory module technology to be implemented before going with an integrated memory controller. Likewise, I suspect AMD is also waiting for the FB-DIMM instead of going straight to DDR2. This will mean the end of all socket changes instead of changing whenever memory technology changes.
It would mean that regardless of the RAM type being used, the memory controller would not need to be modified and neither would the memory slots on the mainboard. It would not matter whether the RAM type is DDR, DDR2, Rambus XDR or QBM RAM, the controller only has to talk to the FB-DIMM.
The FB-DIMM itself talks to whatever type of chips are used on the module.
Check out the following two links for details:
http://www.theinquirer.net/?article=15214
http://tech-report.com/onearticle.x/6563
This will definitely help out the better performing but currently unsupported memory types such as QBM (6.4Gbit/module) or XDR RAM(12Gbit/module).
Funnily enough, this technology combined with faster memory types might actually benefit AMD more than Intel as the Athlon 64 currently uses a 2GHz HyperTransport Point to Point interconnect offering far more bandwidth, and fewer bottlenecks than Intel's 1066MHz shared system bus.
Likewise, the reason AMD & Intel cannot conform with the BTX form factor when using an integrated memory controller is no longer an issue when using FB-DIMM technology. The FB-DIMM specification allows for much longer signal traces on the mainboard, so the RAM will not have to be placed in close proximity to the memory controller.
Because FB-DIMM technology requires less mainboard traces we will probably start to see boards with support for quad channel RAM, though I think this would be overkill when fast memory technologies such as Kentron QBM and Rambus XDR is available.
Beyond this, I really do see a brighter future for Intel at the moment than AMD. Both companies have run out of steam when it comes to ramping clock speeds. The current solution that both companies seem to be taking is to add more Level 2 cache. There’s only one problem here: More L2 cache gives the Pentium 4 a bigger performance increase than it does the Athlon 64.
In the meantime, AMD has released an Athlon 64 4000+ processor that runs at the same 2.4GHz as the 3800+. The difference: 1MB level 2 cache instead of 512K. The performance difference: practically nil. It goes without saying that the 4000+ rating is unjustified as the size in cache increase doesn't benefit the Athlon 64's architecture all that much.
Intel will be increasing the size of the cache on their mainstream Pentium 4's to 2MB next year. Due to the Pentium 4's architecture, this will bring fairly significant performance increases.
Everything is not yet lost for AMD however. Their K8 architecture was designed from the ground up to be dual core capable. They are expected to be significantly faster than Intel’s initial dual core solution.
In the end, only time will tell who wins out…… I'll leave the rest up to you. I prefer to argue with the technical points....
Was my post too big? If so, I'll try to keep them smaller in the future. Feedback plz.....
deception``
10-20-04, 11:34 PM
Well said. A bit technical for some, but I found it thorough and entertaining.
deception``
Well said. A bit technical for some, but I found it thorough and entertaining.
deception``
I'm glad someone enjoyed it..... No one ever seems to reply to my posts..... :eh?:
I've always hoped that these detailed posts would be a cause for more debate, but they only ever seem to stop them..... :(
evowatercooling
10-21-04, 01:09 AM
That is a very interesting read. I was wondering, will having a stop on the line to the dimm eventually be a bottleneck for faster memory. If it has to make a stop and decide where it needs to go, the information will obviously take longer then if it didnt have any stops.
That is a very interesting read. I was wondering, will having a stop on the line to the dimm eventually be a bottleneck for faster memory. If it has to make a stop and decide where it needs to go, the information will obviously take longer then if it didnt have any stops.
There is expected to be a bit of an increase in memory latencies with FB-DIMMs. I don't really see how it can be a problem as you just have to modify the FB-DIMM controller chip.
Even if the standard changes, so long as it still talks the same language to the memory controller, there shouldn't be a problem.
Likewise, each memory controller will likely be tweaked to perform best with whatever memory is mainstream in these modules, but it will still talk the same language......
Also, with the error correction capabillities built into this technology, well, all I can say is I'm all for it...... :p
Does what I've just said make sense???? :beer:
In any case, as my favorite saying goes: Only time will tell..... :D
evowatercooling
10-21-04, 06:54 AM
Yeah that makes perfect sense. I believe that an increase in latency would have to happen. But if they possible increase in frequency is worth it then yeah I am all for it too.
What would be much better is if they could make memory chips the same way they make cache chips. They are inherently much faster. The only problem is they take up an enormous amount of space compared to dram, and they cost a lot to produce. If they ever managed to make that a viable solution though, that would be a revolution in memory, not just an evolution.
deception``
10-21-04, 08:50 AM
I'm glad someone enjoyed it..... No one ever seems to reply to my posts..... :eh?:
I've always hoped that these detailed posts would be a cause for more debate, but they only ever seem to stop them..... :(
It was late....look for me to make a contribution later today.
deception``
Yeah that makes perfect sense. I believe that an increase in latency would have to happen. But if they possible increase in frequency is worth it then yeah I am all for it too.
What would be much better is if they could make memory chips the same way they make cache chips. They are inherently much faster. The only problem is they take up an enormous amount of space compared to dram, and they cost a lot to produce. If they ever managed to make that a viable solution though, that would be a revolution in memory, not just an evolution.
Cache, RAM and hard drives will become obsolete once they figure out molecular memory. It's only a few years of.
They expect it to be in production by 2007. The problem will then be whether they can create an interface for it that is fast enough..... ;o)
As far as I can tell, the main reason for the FB-DIMM is to allow the implementation of new memory technologies for rapidly without the need for changes to chipset/memorycontroller/DIMM slots.
I like it as it means we're likely to see better memory technologies such as QBM and XDR RAM which would ordinarily never be supported in computers as their proprietry.
I don't care if they're proprietry, I just want something thats faster!
MonroeM
10-22-04, 12:26 PM
They expect it to be in production by 2007.
And by that time, we'll all be using Mac's anyhow. :D
Just kidding. Anyway, I liked that article, it's only one view... but very accurate.
I've got a niave buddy who puts all his time and effort into his AMD system. He's got an AXP 2200+ and as soon as he got it, he wanted to test it against mine. We have always had this rivalry some way or another, so we went ahead and fresh-installed everything. Our systems were fairly comparible, he had (past-tense since our systems have changed) a AXP 2200+, me w/ my old P4 2.6B, both w/ dual-channel 512mb CL2.5 RAM, him w/ a severly O/C FX5700u, me w/ my stock 5900xt O/C (BFG), all latest drivers, no tweaking besides the O/C in his card to get as closest to mine as he could. He was quite adamant that he was going to whoop my system's ass.
We ran the benchmarks and I get a 14,000 in 3DMark2001SE and he gets a 10,000 in the same app. Needless to say, he was quite disappointed w/ his mark.
I'm not sure my point of telling that story but my overall opinion of AMD geeks is that they're usually the people who root for the underdogs in general. They buy an AMD CPU + motherboard... and then go buy an ATI video card. I usually bump into a new geek at school at least one time a week and I notice after I ask, "What CPU do you have?" and they answer, "An AMD-something-or-other." I usually wait for their "oh-yeah-and-I've-got-an-ATI-video-card" addition to my question, which happens at least 7/10 times.
Anyway, but my point is... from my personal experiences w/ other geeks, I've noticed that the ones who go w/ AMD are usually misinformed to some degree; that their 64bit processor is going to run UT2004 quicker or get them faster rendering times in 3D Studio Max. The majority of them seem like... if another geek told them that AMD is better than Intel, they'll buy an AMD, not because it's better, but because geeks can be quite gullible between themselves (IMO).
Anyway, I could go on about how that buddy of mine I mentioned above is the exact same way but I'll stop here.
EDIT: and just for clarification, I'm not saying Intel is better by any means. IMO, the majority of people who have Intel-based systems probably bought it from Dell, Gateway, or HP; they probably don't know they have an Intel system and most likely don't care; just as long as it works (I do freelance computer consulting and this is the gengeral consensus I've noticed) and runs their programs. I personally agree w/ the brainwashing opinion by large corporations like MS, Intel, and McDonalds. In the end, what matters is that you do your homework w/ an unbiased opinion (this goes for any comparison between products) and pick whichever will work best for you at the lowest cost.
Cheers.
DrSlinky
10-22-04, 12:39 PM
And by that time, we'll all be using Mac's anyhow. :D
I agree!
http://a1112.g.akamai.net/7/1112/492/2002091464/www.wired.com/news/images/full/strawberry_f.jpg
...I've noticed that the ones who go w/ AMD are usually misinformed to some degree;
Or they just have an eye for value. But let's not be too inflamatory, shall we? Each platform has its technical strengths and weaknesses. One of AMD's enduring marketing strengths has been its ability to produce good, reliable processors at a very low price. The departure from this formula is the basis for the article and subsequent rebuttal (http://overclockers.com/articles1112/) to which this thread originally referred.
MonroeM
10-23-04, 03:35 AM
Here is what I normally see in my everyday experiences w/ customers.
1) Those who have a Dell, HP, Gateway, Compaq, or <insert brand name here> usually have a problem w/ their computer and want it fixed, plain and simple. They call me because of many possible reasons, <company>'s support sucks, their warranty has expired, they can’t find their manual, or they just don't care who they get to fix it as long as it's fixed and they can get back to using Quicken.
2) Sometimes people ask me about their computer, usually informing me that they have no clue when it comes to computers, and I like to explain in layman's terms what exactly they have and the capabilities of it. When I mention "Intel", the customer usually nods his/her head in communication that he/she has at least heard of the company and has been "trained/brainwashed" into thinking it's good (as I agree w/ the post previously in this thread).
3) When I mention an AMD, they usually raise an eyebrow.
4) Then I get the occasional gamer in need of help w/ his (haven't seen many 'hers') PC. At least half of the time w/ AMD users, I notice that they are the certain kind of person who only purchased a system w/ an AMD-brand CPU because someone, somewhere told them it was better than the other guy. Sometimes they will out-right say someone told it was better and sometimes I get, and I quote, "Well in theory, AMD is better at computing the same amount of data as Intel... just at a lower speed."
5) I rarely need help from an Intel gamer, however I have in the past but it was simple things like, "WTF d00d? Why does my hard drive keep clicking while I'm playing games?"
6) And like I posted above, the majority of the Intel users I get have no idea what a CPU, GPU, or RAM is and usually don't care just as long as they can get their "Microsoft Windows" to run and edit their Word documents. (I exaggerate of course; I only had one customer so far that has referred to Word as "Microsoft Windows").
Anyway, but I digress. In my personal experiences, half of my AMD customers (usually gamers oddly enough) are misinformed.
When a customer wants me to build them a new system, it always comes down to money, and the less money they have, the more likely I will purchase them an AMD CPU. The price, you cannot beat, however I am personally still on the fence about the more expensive AMD CPUs. 64bit applications are nil at this point in time so a 64bit processor is the same, IMO. The architecture is unique and I like that, however when it comes to an overall do-everything CPU, I turn to Intel.
The focus of this thread is the article on the front page and I completely agree w/ Ed about AMD's financial status. Intel is a giant; everyone knows that and being a giant probably means it has enough assets to invest in it's product's future in the competing market. AMD, being the underdog (as I stated in my previous post) simply does not have the funds to put out a product worthy of turning the head of the mass market... and not just the geeks.
Sorry for the long-winded post. Just my opinion on the matter... as you can tell by the lack of my name at the end of the post, so like many other users do. :p
...AMD, being the underdog (as I stated in my previous post) simply does not have the funds to put out a product worthy of turning the head of the mass market... and not just the geeks.
When you consider the momentum behind the Intel marketing machine, it would definitely require an earth-moving advance in technology to allow AMD to overcome it. Being "just about as good" or even "as good" just won't cut it...
DrSlinky
10-23-04, 10:10 PM
When you consider the momentum behind the Intel marketing machine, it would definitely require an earth-moving advance in technology to allow AMD to overcome it. Being "just about as good" or even "as good" just won't cut it...
In the desktop market, apparently, it is. With the exception of Dell, all the major manufacturers have made large moves towards AMD. If you want a cheap PC, you're looking at a Celeron. If you want a high end PC/multimedia machine, you'll probably look at a P4, although the A64s are creepy into that market. For most buyers, who want a good all around PC at a resonable price, it's going to be an Athlon XP. And because of this, the Athlon name is catching on with the average consumer. The "best bang for the buck" concept is seriously affecting manufacturer choices, which in turn almost force people to buy an AMD. If AMD could get Dell to use their chips instead of 533fsb P4s, they could easily dominate the desktop market every week.
If AMD has any real problem in the consumer level playing field, it's lack of chips to compete with centrino based machines. Mobile Athlon XP and A64 machines are as common, if not more common, than mobile P4s these days.
In the desktop market, apparently, it is...
This (http://www.theregister.co.uk/2004/09/09/amd_vs_intel_retail/) definitely bears that out. It's interesting, however, that the dominance of the desktop market still apparently only provides AMD with 15.5% (http://news.com.com/Intel+loses+market+share,+AMD+gains/2100-1006_3-5303705.html) of the overall processor market. Does this mean that nearly 70% of the market is comprised of laptops? Or is the server market that robust?
DrSlinky
10-24-04, 09:38 AM
From my own experience, laptop sales have begun to outpace desktop sales, at least where I've worked. It's become an attractive alternative to a desktop and a LCD screen, for about the same price.
As far as servers go, I haven't the foggiest idea. Not something I've had the opportunity to sell.
In the desktop market, apparently, it is. With the exception of Dell, all the major manufacturers have made large moves towards AMD. If you want a cheap PC, you're looking at a Celeron. If you want a high end PC/multimedia machine, you'll probably look at a P4, although the A64s are creepy into that market. For most buyers, who want a good all around PC at a resonable price, it's going to be an Athlon XP. And because of this, the Athlon name is catching on with the average consumer. The "best bang for the buck" concept is seriously affecting manufacturer choices, which in turn almost force people to buy an AMD. If AMD could get Dell to use their chips instead of 533fsb P4s, they could easily dominate the desktop market every week.
If AMD has any real problem in the consumer level playing field, it's lack of chips to compete with centrino based machines. Mobile Athlon XP and A64 machines are as common, if not more common, than mobile P4s these days.
I don't agree with this view at all. If you look at desktop home/home office PC's, yes AMD are outselling Intel. But not a single one of the 3 big tier 1 manufacturers (HP, IBM & Dell) offer a AMD based corporate desktop solution.
This is probably because most businesses don't really care about performance with their desktops. I mean, how fast does a computer need to be to run Word or Excel? So for the most part they go with Celerons.
Business for the most part tend to be a conservative lot. You mention the name Intel and they're quite happy to go this way. You mention AMD and they give you a blank look and ask what that is. It generally takes quite a bit of persuasion to get a client to go that route. This is especially true if they've heard of AMD's reputation from the K6 days.
AMD was the cheaper than Celeron budget solution. As such, components used in building systems based on the K6 platform were substandard resulting in unreliable computer systems. The public in turn associated this unreliabillity with AMD based systems.
There are still people who ask about this reliabillity when considering an AMD system. Also there's a general belief that AMD's processors run hotter than Intels. While thats no longer correct, the public perception remains the same.
It will take quite some time before AMD can break into the corporate market. The people making decisions here have normally been doing so for some time, and have long memories. They will generally stick with a proven company/technology than take a risk on something new.....
I personally believe that AMD currently has a technically superior product. But with corporate sales, perception, not the facts, is what counts
MonroeM
10-25-04, 10:42 AM
I agree w/ part of what you said, mjw21a. AMD definitely needs to re-establish themselves in the market somehow. Someone suggested a name-change, which I personally think might help a little. AMD needs to get a good foothold in the market in order to really start competing w/ Intel on a more level playing field, whether that be a far-superior product or a company make-over. I agree, consumers cling to bad experiences w/ products. If AMD wants to change consumers' opnions, AMD needs to do something big, whatever that may be.
I also agree w/ AMD "currently having a technically superior product." A 64bit processor is definitely a move towards the future... however the question is: does anyone require it at this point in time? That subject is definitely up for debate. I personally think (as I know many others do) a 64bit processor doesn't quite belong in an 32bit dominated market, but someone had to put out the first mainstream desktop 64bit processor, which is probably why AMD did it, to wipe a little dirt from it's name.
Just my two cents.
Cheers.
If AMD can establish itself in the server market then I imagine corporate attitudes towards their products will change, and with this will come brand recognition.
Server administrators have to be some of the most conservative bunch of people around. However many of them want more power in as little increased space as possible.
AMD's strongpoint here is with servers with mroe than 2 processors. Intel can overcome their shared system bus in a 2 processor environment simply by adding more level 2 & level 3 cache. With more than 2 processors on the other hand it doesn't work so well.
The AMD Opterons on the other hand scale exceptionally well due to their use of the HyperTransport interconnect. Also, since they don't rely on massively increases cache sizes, the Opteron runs much cooler than Intel's Xeon processors....
AMD's server marksetshare has grown to 7% in just one year. A sign of things to come? I don't know, only time will tell....
evowatercooling
10-26-04, 09:53 AM
I know that if I ever have controle over a server it will be all AMD.
But I am a computer engineer, not a mere Network Admin.
(j/k all. I am not bashing network admin just trying to get a rise from anyone that is:D)
I know that if I ever have controle over a server it will be all AMD.
But I am a computer engineer, not a mere Network Admin.
(j/k all. I am not bashing network admin just trying to get a rise from anyone that is:D)
Yeah, I'm putting together a nice dual Opteron 246 server for my workplace, soon as they pay for the parts that is..... Then I'll load F@H and fold for team 32 on it..... ;o)
And no, you won't cop any flak from me evowatercooling. Thankfully I have a sense of humour...... :beer:
evowatercooling
10-26-04, 05:23 PM
LOL. That would be one sweet gaming computer!!! How much Memory???
2048MB DDR400 Regestered ECC RAM with 4x 512MB modules (much cheaper than 2x 1024MB modules).
The MSI K8D Master-FA4RM has 12 DIMM slots so expansion isn't a problem. The integrated ATI Rage graphics would kind of make it suck though.... But then again, it's going to be our MS SBS server + Greentree Acounting server. How much graphics performance do you nee in a server? Not much I'd say.
evowatercooling
10-26-04, 06:54 PM
Yeah but could you imagine dual opteron with 12gb of ram and sli. It would be a long time before a game could bring that down. (and by long time i mean a couple of months :))
Yeah, but I won't be allowing any games near this server. The only weakness I'll give in to is folding@home......
Stabillity is my first priority on this. The system is going to sound like a jet taking off with my choice of heat sink and multiple system cooling fans.
I'm hoping the board has some fan control based on temps, but if not, it's the jet sound effect. Should be the same as my home desktop, until I go to the 120CFM Sunon fan that is..... ;o)
evowatercooling
10-26-04, 07:25 PM
Yeah I was just saying any computer with those specs.
I know what you mean about the noise. I just modified my power supply. The fans were running off of the 5 volt line, so I put the fans on the 12 volt. So loud now. Oh well. I want to get some delta 92mm fans. They give the best performace for the wattage. 70 cfm at only 4 watts. The 120 is 100 at 12 watts. They will not be instock until after christmas though. Also I have two 50 db fans on my cpu.
I reckon one of the 200CFM Sunon fans would be nice. I like Sunon as I can actually buy them here in Australia.
Don't know where I can het the Delta's though....
I'm going to the 120CFM Sunon to put on my TT PIPE101 this weekend. I'm hoping that at lower RPMs it won't be any louder than my 80CFM TT UFO fan, but will allow for better cooling when the CPU gets too hot. I love my Asus Q-FAN!
evowatercooling
10-26-04, 09:42 PM
Where can you get those 200 cfm fans. I want to take a look at them.
You want the specs or wheer to buy them? I can email the specs through to you (all Sunon fan specs are in 1 PDF).
Do you have an email address for this (I imagine so, I have 3 for this sort of thing)....
evowatercooling
10-26-04, 11:13 PM
Yeah you can send them to cmderaps@mtu.edu
Thanks a lot.
I've emailed the file through. It's called power_motor.pdf as apparently thats the name of their fan series.
Woohoo, I just got to 4 stars!!!!! :D
Apokalipse
11-02-04, 02:11 AM
the 3000+ in socket 939 is not very expensive, and it is a fast CPU too!
there are a lot of socket 939 motherboards more expensive than the 3000+
AMD don't really have all that much control on the motherboards made for their CPU's, that's why things like DDR2 and PCI-E aren't out yet for them
although PCI-E and DDR2 are not that much better than AGP and DDR1 at the moment, and that might be why AMD haven't got into them yet
evowatercooling
11-02-04, 09:02 AM
Actually, since the memory controller is integrated into the cpu, amd has total control over DDR2. With pci-e, they do not have quite as much control over when it gets implemented, but I am sure that they are pushing on the chipset makers very hard, and even helping them on getting out a pci-e chipset.
AMD relies on nVidia, VIA and SIS to implement new features to the AMD platform. Possibly ATI sometime down the track also if they start making AMD chipsets.....
...but I am sure that they are pushing on the chipset makers very hard, and even helping them on getting out a pci-e chipset.
Nforce4 (http://www.nvidia.com/page/nforce4_family.html) w/pci-express was released last week and should be on the shelves any day now.
Wow, looks like ATI will be the first to the party with an AMD PCI-E chipset! :attn:
Check out: http://www.theinquirer.net/?article=19424
As far as I know this is the first AMD chipset ATI has made! Looks like I've got one more option with chipsets now. Previously I'd only look at nVidia chipsets, but I imagine that ATI chipsets would be good also. I'll wait for the reviews, eh? :p
Hmmm, by the looks of Intel's roadmaps they're accelerating the launch of FB-DIMMs to sometime in 2006, but only with servers.
Proably meant to give AMD a hard time so they'll have to make multiple cores (integrated memory controller). One core for the server and a totally different core for the desktop.
Who's willing to bet that AMD will support the FB-DIMM for the desktop, while Intel just wants to keep the tech in servers????
Hey, did anyone see the other day that Rambus has made a FB-DIMM chip for their hell-fast XDR RAM????
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.