brucedeluxe169
2[H]4U
- Joined
- Aug 24, 2004
- Messages
- 3,720
Follow along with the video below to see how to install our site as a web app on your home screen.
Note: This feature may not be available in some browsers.
Drexion said:Intel just didnt surprise AMD, they surprised nearly every one. No one was really expecting them to come back with such a bang, conroe procs just totally annihilates everything else out there today. Its not just next gen, its like next-next gen yet available next 2 weeks. .
reddhaus said:I don't think AMD was blindsided.
AMD knows what Intel is doing and vice versa. It's just how the industry works.
Intel will be King for a while and then it will be AMD's turn again.
"blindsighted" isn't very accurate, unless you count over 2 years of seeing where Intel was heading as a surprise.Drexion said:Intel just didnt surprise AMD, they surprised nearly every one.
pxc said:"blindsighted" isn't very accurate, unless you count over 2 years of seeing where Intel was heading as a surprise.
The announcement of Tejas cancellation (death of NetBurst) was the first notice, the rise and performance demonstrations of Dothan then Yonah were the next huge clues and then there was the miscellaneous chatter by Intel over the years of CPU power (65W... targetted 2 years ago). It's been 11 months since Intel introduced NGMA and 5 months since AMD said "Intel has no idea what we have". Well, we know what both of them have now.
BTW, you can look up that Conroe was the replacement for Tejas (and several correct rumors like significantly shorter pipeline) as far back as 2004. AMD should know more than the average Joe who reads up on this kind of stuff. IMO, the problems are problems AMD doesn't want to discuss. Ed brings up at least one with the transistion to 65nm. By the time AMD has a next-gen core, so will Intel (Nehalem).
AMD has an Analyst Day every year where they are very unsecretive. AMD presents roadmaps and it's not "anyone's guess" on the 65nm, K8L or next gen core are due. All are scheduled to a month (December for 65nm) a quarter (Q2'07 for K8L) or half year (2H'07 for next gen... highly doubtful).Obi_Kwiet said:AMD seems to like being secretive. Their "new" thing isn't out yet, and it's anyone's guess as to when it will or what it will be. In either case AMD will probably wait until the Conroe launch to give Intel.
Uninformed comments like these remind me of Germany in 1945. Rumors about secret weapons so powerful that they would turn the tide. When would the Wehrmacht unleash it? When Allied forces reached Berlin, of course.Obi_Kwiet said:AMD seems to like being secretive. Their "new" thing isn't out yet, and it's anyone's guess as to when it will or what it will be. In either case AMD will probably wait until the Conroe launch to give Intel. the least room possible to maneuver. We do not know what AMD has.
duby229 said:Ed is an Intel !!!!!!. He always has been. They pay him well.
.
Killa_2327 said:Not that AMD is bad, I absolutely love AMD, but when people are saying "whats AMD? Are they new? These are faster then Intel?" theres an issue. .
brucedeluxe169 said:so at this point, is the enthusiast, and general consumer sentiment that .... AMD really does NOT have an answer for Conroe?
That is really the question at hand.... I dont want to hear about 2008 and beyond, nor do I want to hear about pre July 2006.... basically, as it stands, there is *no* answer to Conroe....
Obi_Kwiet said:That statement would imply something new. AMD has not released any new processors. Same as 5 months ago, but with a few 200MHz bumped speed bumps. AMD seems to like being secretive. Their "new" thing isn't out yet, and it's anyone's guess as to when it will or what it will be. In either case AMD will probably wait until the Conroe launch to give Intel. the least room possible to maneuver. We do not know what AMD has.
We knew what Intel was going to do sence the Centrino came out. It was actually kind of supprising that it didin't happen sooner.
Killa_2327 said:$80-$130 Athlon 64's
$150-$250ish? Athlon X2's
To beat conroe? Nothing that we know of, but when I was thinking about getting a Socket A mobo and cpu for a cheap fileserver, price drops occured. You can now build a great pc for around $600(i'm talking gaming rig, x1800XT, gig 'o ram, decent hd), definitely a great answer to Conroe imo. Wait til the drops on X2's occur, tons of people will be buying them.
So even if my pc that costs $600ish aint as good as your conroe, it probably costed half the price and can play many games on high settings, talk about a deal.
brucedeluxe169 said:point taken, smart man
hopefully AMD doesn't bleed too much into the red, without them, we'd be using the functional equivalents of Williamettes
Yashu said:AMD is not stupid... they know where their bread and butter is...
Lets educate:
AMD developed a SERVER architecture and adapted it to the desktop for you guys. Alot of people forget this...
Intel has developed another desktop architecture. As long as intel keeps pumping R&D into itanium, you will not see an intel desktop chip that competes with opteron in large servers... Big companies don't like overlap. Opteron is not a competitor to the conroe arch. it is a competitor to Itanium. Opteron scales very well, it is a true 64bit platform... it is everything in the server world that conroe is not and wasn't intended to be.
Conroe will go anywhere intel desktop chips are already going. And with vista release on the horizon, intel better prove that 64bit as a memory address extention only, is good enough to maintain that status quo.
conroe is wicked fast 32bit, but scale beyond 2 cores and then add in true 64bit support requirement and AMD still looks good on the back ends.
And before you give me talking points about how it's all RISC internally and EMT64 is just as good, ect... lets see some real benchmarks... as of yet the NDA isn't even up yet, and intel has done their very best, so far, to draw attention away from 64bit. It is certainly widely known that EMT64, currently, does not offer many % performance benefit (if at all), where AMD64 shows improvement all around... those DEC Alpha engineers were not put to waste.
I am not trying to tell you to pick sides... but that, I beleive, intel has done a good job *hyping* this release, where it looks to be the same old intel achitecture with just a better CPU, and in the server world, intel needs more then a faster 32bit CPU to make people want to chuck their opteron systems...
Now, if news was broken that intel, suddently stopped production and R&D on all itanium line... then AMD might be queued to worry that intel had a killer new serverside arch on the way.
so your arguments are ridiculous.
Yashu said:dude, man...
EMT64 on intel is not the same as AMD64... this is well documeted, and the benchmarks back it up... on linux and windows x64, there shows no improvement on intel platform currently and there is marked (25%+) improvement on AMD64...
hell, look at the recent H benchmark test of vista ready PCs... intel EMT64 only shows an improvement on one single benchmark... and on all others the results are the EXACT same as 32bit. AMD64 shows a performance increase all around. If the implimentation is the same then explain these results... and it's not just the H benches... like I said, on linux AND windows... benchmarks show EMT64 is mostly emulation + memory address extention.
You can go on all day about internal RISC and micro ops... and registers... all that stuff is fine, except the benchmarks show that EMT64 is not a full implimentation... the documentation shows that EMT64 is not a complete implimentation.
and if intel did the same thing with conroe then it is a huge mistake. x64 vista will eventually allow the masses to gain a performance increase on their athlon64s... for free, basically...
Intel, once again, gives us the same thing... a faster 32bit CPU on the same platform.
Also, are you trying to say that the arch. will scale evenly to 4 cores? I am not sure... currently with opteron, I can have a dual socket system... with 4 cores... and run 4 processes at full speed... I am not sure you can even do this with current SMP xeon even with 2 processes... I think you lose a measurable % with every added processor over 1 on the xeon platform.
Buisness isn't all games... like I said... it's great intel has a fast new 32bit CPU... but as long as they are develping itanium, they are not going to overlap and compete with themselves in that market!
Yashu said:...
in other words...
AMD64 in 64bit long mode, can handle a 64bit word length in one work cycle.
EMT64 breaks this 64 bit word length into more then one work cycle...
....
Hacked or not, the implementation is not important what matters is the resultant performance in the end utilizing 64Bit Code.Yashu said:fair enough. However, I am making an educated assumption because:
The benchmarks have already shown what I said to be 100% fact with current EMT64 on netburst.
Core2 is based mostly on core duo/solo that is already on the market. core is a 32bit only architecture. Unless core 2 is a completely new design from the bottom up (and it is not), EMT64 will still be just a cheap method of memory address extention for intel.
something you have to understand is this x64 stuff hit intel way by suprise... Intel put their stock in SSE... and they do SSE pretty well... that was their solution before AMD64 came about.
I know SSE is not the same... but intel created SSE to allow 128bit optimised calculations in a 32bit world..
and AMD is not innocent either... AMD has tacked on SSE support in the same way that intel has tacked on emt64...
there is a big difference between supporting the registers and the processor being designed from the beginning and built around it.
Well my benches above already disprove that theory, there is performance gained with Woodcrest moving to 64Bit Extensions, regardless if they are tacked on or not.Yashu said:yes that makes sense...
but you have the core architecture already being used in notebooks... and it is entirely 32bit.
now I thought that core2 was an enhanced version of this, not a redesign... suggesting that intel has tacked on EMT64 in the same way as they did with netburst, primarily as an extended memory adress purpose.
I understand that the final performace is the key... however, if customers are gaining performance on their existing AMD64 platforms by moving to 64bit, and the same migration on intel platform shows no performance increase... many people are going to wonder what the big deal about conroe really is... but at the same time... many people will not notice or care as long as it is faster.
I know I would prefer my processor to be able to take full advantage of 64bit word length.
I think intel has a good product on the desktop... but I am not sure it is quite as attractive in the server world *yet*. Opteron is a good product... and there aren't yet compelling reasons to convert platforms, say, if you use AMD already.
also... these coming AMD price cuts will help them some. it's not as if athlon64 is too slow to be useful yet.
Yonah is not Conroe. Conroe doesn't appear to be the EM64T hack that Prescott is.Yashu said:but you have the core architecture already being used in notebooks... and it is entirely 32bit.
coldpower27 said:Well my benches above already disprove that theory, there is performance gained with Woodcrest moving to 64Bit Extensions, regardless if they are tacked on or not.
Theoretically better is worthless, AMD has access to a 12.8GB/s of bandwidth with Socket AM2, but AMD's still loses to Conroe with it's "outdated" FSB architecture that can't even take advantage of all the banwidth Dual Channel DDR2-800 provides. Same issue applies with "true" 64Bit Support the AMD propaganda machine is churning out for AMD K8's. What matters in the end is better real world performance, not theoretically better.
Woodcrest is looking like it can take the 2P segments fine, it's the 4P -16P? or so systems that AMD should maintain a competitive edge against Intel considering Tigerton is not arriving till 2007, with Intel's MP offering still being based on NetBurst.
And it is incorrect to say there are no gains with EM64T, it just usally somewhat less I would say on NetBurst vs K8, K8 benefits a tad more, but Pentium 4's also do benefit form X86-64 too.
The price cut argument can be easily applied to basically the rest of the modern mainstream Pentium 4 90nm/65nm NetBurst based products, they aren't too slow at the majority of tasks either.
They wil help some, but for the high end desktop, above 200US AMD's price/performance is basically finished. Intel will have both the performance, power consumption, as well as performance/dollar crowns.
As well, Yonah is still based largely on Pentium M technology, with a few minor improvements, and is definitely not based on Core Architecture, it carries the Core brand name though, so that is why you mgiht be getting confused.