![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Considering Going back to INTEL just 1 question?
- Thread starter newls1
- Start date
Chickan said:Leaked? I thought we had all decided that was BS hogwash marketingAnd 10 monthes away to boot.
Turn off alternate reality mode please.This isn't "Conspiracies on trial" or Evil Inside (TM). See things how they are , stop using coloured glasses.
The figure comes from a document which was a technical study on multicore processors and the reasoning behind going multicore , meant to be used as material for universities and technical seminars.
It was not a press release or anything like that .
Btw, for your info , Merom taped out in may-june timeframe and was showed at Intel's autumn IDF running Windows&Linux.Samples are sent to developers right now. "Oh my God" ! they said.
Looking at benchmarks here it looks like Dothan can hold its own even at stock, granted the Fx is beating it but not by alot, thats on a Aopen board with PCI-E. Dothan isnt that hard to game on, only down side is mobos for it cost a little more then other mobos. For me in canada a Fx-55 would be over 1000Cdn while a P-m 770 can be had for under 600CDN.
- Joined
- Feb 1, 2005
- Messages
- 31,350
Flak Pyro said:
Those benchmarks show the P-M 770 ~ A64 4000+ (old FX-53). 770 is priced at $440 according to pricewatch. 4000+ at $330. A64 FTW price/performance at STOCK speed
.kirbyrj said:Those benchmarks show the P-M 770 ~ A64 4000+ (old FX-53). 770 is priced at $440 according to pricewatch. 4000+ at $330. A64 FTW price/performance at STOCK speed
hey your right, i never really paied much attention to the other AMD cpu's i was mainly looking at the Fx-55. I dont know I think there is just a coolness factor of using a Mobile chip, it sets your rig apart from everyone elses, if Yonah is as good as Dothan for gaming i may give it a shot.
robberbaron
Supreme [H]ardness
- Joined
- Sep 6, 2004
- Messages
- 6,101
savantu said:
Xbit had no trouble reaching it!
http://www.xbitlabs.com/articles/cpu/display/pentiumd-820_3.html
Look at the 640. 134W. It's TDP is 84W. 84/134 = 62%. So I was wrong, some chips are incredibly underrated.
The 820, however, was right on the money with 130W.
That 3500+ has a 67W TDP, by the way. It "burns" at 67% of the TDP. I trust AMD's thermal envelopes way more than Intel's.
sabrewolf732
Supreme [H]ardness
- Joined
- Dec 6, 2004
- Messages
- 4,776
Nasgul said:
Every time you post god kills a kitten.
AMD's platform is just as stable as Intels. In fact, dothan is the only thing that can compete with amd's right now, and that's simply because it uses a short stage arc. very similar to the pentium 3, which is similar to the original athlon which the a64 is derived off of. See a trend? Netburst, booooo. Also that ;link doesn't use nvidia's drivers which multithread the driver or something and give a good boost in performance which is what he is talking about. So that link is invalid./I was playing on a dothan based intel and then played on a fx57 and they where very comparable. But i think the AMD was just a tad faster.sabrewolf732 said:Every time you post god kills a kitten.
sabrewolf732
Supreme [H]ardness
- Joined
- Dec 6, 2004
- Messages
- 4,776
Big Worm said:
agreed.
robberbaron said:
No it doesn't because a it's not just the temps but the power it's drawing. Also undervolting slows the processor down and a Pentium-M at 1.6GHz, like you're talking about would kick the shit out of it. Dual core wouldn't make much sense because more threads mean more power drain and heat, yes for AMD as well.
TDP for X2 is 110W and one site measured the 4400@93W, now if I could just fine the link.
Old news.
With Dual Cores, it will not need Hyperthreading.
Donnie27
Flak Monkey
Gawd
- Joined
- Feb 15, 2005
- Messages
- 651
If your selling the FX-57 then that should give you enough money for a mid-range P4 and some phase change cooling. That should just about even things out.
- Joined
- Jun 11, 2003
- Messages
- 6,578
Flak Monkey said:
Why even go through the trouble? If he was willing to OC I would say get the cheap dualcore Opteron and the phase change.
sabrewolf732
Supreme [H]ardness
- Joined
- Dec 6, 2004
- Messages
- 4,776
Flak Monkey said:
............. barely. Even then, I would rather stick with the fx57.
robberbaron said:
The way Prescott handles voltages is completly different from A64.Prescott uses a microcontroller and has a variable voltage while A64 uses a fixed one.
Because of this , crude measurements are not indicative at all of actual power consumption because they are greatly influenced by capacitor and power regulator performance or better said inefficiency. ( To better understand , the more power you need the more innacurate the measurement is )
Btw , it is pretty easy to see how much power a P4 can burn.Unlike AMD , Intel's tech docs present a clear picture on how voltages and amps scale.And I trust that more than XbitLab's test.
As a sidenote I was told that the funny part was on how much work the processors did under burn tests.Instead of using a useless power virus , better use something which can actually measure how much work the cpu does.
Because you see, if you cpu uses 45w and mine 135w , but I do 3x more work , then it's pretty level...
brucedeluxe169
2[H]4U
- Joined
- Aug 24, 2004
- Messages
- 3,720
why do people always jump on others that talk up the performance of Conroe and Merom? Isn't it undeniable that they will own when they come out? They will get better performance per clock than even the Dothan, and Conroe will be specifically enhanced for the desktop, with double the L2 cache and a way faster FSB. Those are just 2 of the improvements among many more....
Isn't it obvious that Conroe will be very very fast? Even if it isn't faster clock for clock than the Dothan (highly unlikely....), if it comes in with the near 3ghz speeds that Intel is claiming, it will kill AMD's procs unless they can get past 3ghz themselves (doubtful on anything other than the highest end FX chips).
Isn't it obvious that Conroe will be very very fast? Even if it isn't faster clock for clock than the Dothan (highly unlikely....), if it comes in with the near 3ghz speeds that Intel is claiming, it will kill AMD's procs unless they can get past 3ghz themselves (doubtful on anything other than the highest end FX chips).
That's funny. The same line was used when the Prescott Pentium 4 was months away. On paper it looked like it would "undeniably own" with double the L2 cache, additional L1 cache, SSE3, and lots of internal improvements. Only a few weeks from release, the problems started surfacing, like its long pipeline and poor thermal and power figures. So don't be too quick to declare a winner, especially since the Conroe/Merom pair is largely a new architecture. We have no pertinent history to draw from.
xonik said:
Right. When the Pentium 4 was released it seemed like a step in the wrong direction. It wasn't until the Northwood was released that the architecture started showing what it could do. Then Prescott was certainly lackluster.
So I'd agree, lets wait and see.
newls1
Supreme [H]ardness
- Joined
- Sep 8, 2003
- Messages
- 4,628
OH, I forgot to say this FX57 has been on PHASE cooling (-60c) and have seen 3.52GHz, but sold the cooler, and am now buying another oneFlak Monkey said:
brucedeluxe169
2[H]4U
- Joined
- Aug 24, 2004
- Messages
- 3,720
BAH guys, that was a low blow! using the presHOTT angle on me....
ehh, I'm happy with my preshott (though it was donated to me....)
sure, intel misstepped with netburst, but with conroe being based on what is one of the best CPU designs of all time, how can they misstep? It will basically be a Pentium M with a better manufacturing process and made for the desktop, so it will have way higher clock speeds.
I know the preshott fiasco leaves a bad taste in ppl's mouths, but I think its going to be pretty damn hard for Intel to mess up the Pentium-M architecture.
ehh, I'm happy with my preshott (though it was donated to me....)
sure, intel misstepped with netburst, but with conroe being based on what is one of the best CPU designs of all time, how can they misstep? It will basically be a Pentium M with a better manufacturing process and made for the desktop, so it will have way higher clock speeds.
I know the preshott fiasco leaves a bad taste in ppl's mouths, but I think its going to be pretty damn hard for Intel to mess up the Pentium-M architecture.
brucedeluxe169 said:
Preshott was a mis-step, Netburst wasn't IMHO. Northwoods of all kinds were good and it was still more of them than Prescotts. I'd have loved it if Intel would have just built a 90nm Northwood C, call it a Northwood-D and add 1MB of L2 to it before moving to Preshott. Just think, if these were the 512K L2 versions, Intel could have gotten aways with two of these and had a Smithfield with less than 110 million transistors.
Donnie27
I disagree. The reason why Prescott was Preshott was because of Netburst. Netburst implementation meant unusually long pipeline stages. Even though this technically allowed higher Hz scalability, it meant that per cycle, the chip itself was doing less work. Also, the scalability was proven untrue with the Prescott series. The chip was inefficient. Even while beating AMD's clockspeed by a good amount, the work per cycle was what caused discrepancy, and the work per cycle was tied to the Netburst implementation.Donnie27 said:
Notice that NetBurst is pretty much abandoned now. In retrospect, if Intel had improved on the P6 architecture (The pentium III line), then they would have had a much superior chip. The current Pentium M is the incarnation of the improved Pentium III. It does more work per cycle, while pushing NetBurst to the wayside.
NetBurst is why 6 years of research at Intel was wasted. Had they improved with the P6 architecture, they would probably be better than AMD right now.
As much as you dislike Netburst from a technical standpoint, Intel gained (and then lost) much marketshare from it. And of course, it didn't stop Intel from enjoying spectacular profits as usual. I'm sure we will see many elements of the Netburst architecture for years to come, like the FSB and cache designs.Genem said:
Genem said:
I couldn't disagree with you more than anything I've read the last 8 months. First you confuse Netburst and Prescott. That's just wrong headed. Netburst started slow(Willy), gained ground (Northwood A&B), turned out to be pretty damned good (NorthwwodC) and then crashed because Intel Got greedy (Preshott). Now trying say say Netburst is only Prescott is, well, BULL! Northwood still makes up most of the Netburst processors sold with the longest run being the Northwood C that launched in the Second quarter of 03. Sure Prescott was a Screwup/Misstep/slipup and etc... But to try and say that's all Netburst was, you're only fooling yourself.
Intel would bet an average of 2.6 billion made during each of the last 21 quarters that you're wrong.
Donnie27
Genem said:
That's BS.
First of all , triyng to do things differently isn't a stupid thing.It's a calculated risk and it paid off for intel.Valuable lessons were learned and unlike AMD , they have the courage to pull out a new architecture.I wonder how will AMD fare when K7/8 is abandoned.
Secondly , the concepts behind Netburst aren't dead at all , they are alive and fine.For your info , Xenon , the Xbox CPU has a 40 stage pipeline , similar to Cell while Prescott had 31.In theory the ideal pipeline length is 55 according to Intel studies.
Doing less work / cycle isn't stupid , it's simple another way of doing things.
Prescott was designed to ramp fast to 5Ghz.In order to do that , it's internal critical paths were using relaxed timings and so were the caches.The 90nm altough remarkably good couldn't keep up and Prescott stopped at 3.8 , far short of its goals.All the measures taken to reach high frequencies backfired.
At 5Ghz it would have been top notch and would have smoked a 2.6 A64 no doubt.
Thirdly , you can't beat a dead horse (P6 ) until it's up again.P6 was nice but it is dead.Pentium M changed it considerably , Merom and Conroe are complety different and new ( I might add ) and the 2008 Nehalem is also a brand new microarch.
Genem said:
As others pointed out, all Pentium 4's and Pentium D's as well as their Celery counterparts are all Netburst core processors. The chip doing less work per clock cycle isn't as accurate as saying it's performance per clock cycle isn't as good. By no means is it "doing less" The processor isn't so much lazy as it's ineffecient.
Netburst was a calculated risk by Intel. R&D comes up with all kinds of things, they present those technologies and the marketting people and executives decide what to use. By the time the P6 microarchitecture was replaced, it was done so because it couldn't scale any more with clock speed. Another die shrink was too expensive to impliment and wouldn't have produced the needed results anyway.
The P6 is still alive and well in Banias, Dothan and Yonah. However, Yonah on a clock for clock comparison basis still gets beaten by a 2.0GHz Athlon 64.
When Intel came out with Netburst, all Intel was going up against were early Athlon XP's. They were slightly slower, but the chipsets sucked for AMD at the time. Plus, they still had more pull with OEMs. Intel was doing fine. Especially in a market space where MHz mattered most. It didn't actually matter which was really faster. When Intel came out with Northwood they were able to provide better performance alot of the time than the Athlon XP could. What Intel wasn't prepared for was going up against an on-die memory controller. They never thought of such a thing, and still have no plans to do it due to the inflexibility that such a design suffers from. Now they are pushing the quality and features of their platform, vs raw performance itself like they used to be able to do.
Intel has some interesting things on their roadmap, which of course changes all the time. Only time will tell wether or not they can 1up AMD again in performance.
I was one of the original dissenters from this school of thought, and in my P4 vs. the G4e series I tried to make a plausible technical case for why the P4's designers had made some of the design decisions that they did. I ultimately managed to convince myself and not a few others that the P4's deeply pipelined design was, in fact, performance-driven and not marketing-driven.
That was then, and this is now. As it turns out, the P4 bashers were right. Revelations from former members of the P4's design team, as well as my own off-the-record conversations with Intel folks, all indicate that the P4's design was the result of a marketing-driven focus on clock speeds at the expense of actual performance and scalability.
from http://arstechnica.com/articles/paedia/cpu/pentium-2.ars
I kinda doubt that the P4 was purely an R&D department working entirely on it's own. Cause when it came out the general public basically equated clockspeed with performance. Most people didn't know how a chips particular architecture could effect the relative performance of each clock cycle and probably still don't today.
That being said I've seent some comparisons of the Pentium M vs. the A64. The P-M is impressive in a desktop setting and I would consider the Pentium M based desktop especially if I wanted a quieter machine. I think that the FX chips and the P-M are close enough in performance that you can make your decision basically on personal preference. Although today I doubt the Intel chips are really more stable than the AMD chips. I think that AMD has come along way in the stability department.
edited for spelling.
That was then, and this is now. As it turns out, the P4 bashers were right. Revelations from former members of the P4's design team, as well as my own off-the-record conversations with Intel folks, all indicate that the P4's design was the result of a marketing-driven focus on clock speeds at the expense of actual performance and scalability.
from http://arstechnica.com/articles/paedia/cpu/pentium-2.ars
I kinda doubt that the P4 was purely an R&D department working entirely on it's own. Cause when it came out the general public basically equated clockspeed with performance. Most people didn't know how a chips particular architecture could effect the relative performance of each clock cycle and probably still don't today.
That being said I've seent some comparisons of the Pentium M vs. the A64. The P-M is impressive in a desktop setting and I would consider the Pentium M based desktop especially if I wanted a quieter machine. I think that the FX chips and the P-M are close enough in performance that you can make your decision basically on personal preference. Although today I doubt the Intel chips are really more stable than the AMD chips. I think that AMD has come along way in the stability department.
edited for spelling.
Araanor said:
Not at all LOL! You're giving AMD way the hell too much credit, but what's new? Intel killed Netburst as we know it because they couldn't fully correct Heat, Power Draw/Dissipation and not get them to 5GHz as was planned. Desktop Processors being killed didn't have a damned thing to do with AMD. AMD doesn't build enough Desktop Hammers to cause Intel to change anything=P The Server market is where AMD can do more damage and infuence things. Lower volumes in the server markets mean the very excellent Opterons don't have to be out numbered like their Desktop brothers.
AMD needs a new Architecture because they need something that has better yields and can RAMP up faster than their current Product. Opterons and Semp(R)ons saved them from the Red this last quarter, sheesh!
Donnie27
Sir-Fragalot said:
1. Yonah gets beaten some of the time, not all of it and that's by X2 on the review.
2. Intel did risk it on Netburst, just as AMD did with Hammer. Intel knew all about IMC, since they Fabbed/Built Apha's and tested it with Timna before Hammer ever hit the market. Maybe Intel knew of the troubles AMD would have and dumpped it.
3. All companies raodmaps change=P
With that said, my Athlon64 3500+ (mainly for games) and Asus A8N SLI is Ordered and in route to me. So before the AMD faightful jump me and this will be updated to A64 Dual Core next year. I had to get a Guitar so that cut into my Budget or I would have gotten maybe a 4400+/Opteron 175 and or for sure at least a 3800+.
Donnie27
xonik said:
They tested it against Dual Core, NOT A64 Single core. I can link to GamPC running plain old A64 Single if you like? Yes, this is stated in the review as well.
It's way too early to give Yonah a Thumbs UP or DOWN=P The Processor and Board are SAMPLES not shipping products. As much as I think Anandtech is biased, even they tried to point that out "in the review". There are many other "GAFFS" to talk about. Let's wait for the Reviews instead of the Previews OK? I absolutely don't trust ANYTHING Anadtech posts by itself. I read reviews that are Pro, Nutr and Con, Anandtech is 100% con when it comes to Intel.
I just don't like the pin change and I'm not touching it for that reason.
Donnie27
RaphaelVinceti
Gawd
- Joined
- Apr 7, 2005
- Messages
- 600
Donnie27 said:
Intel dumped Netburst because it was a mistake, and AMD was killing them because of it. (When i mean killing, i mean, AMD was starting to gain larger market share, and Intel is like microsoft, in that Intel doesnt like someone encroaching on their turf) If AMD hadn't been killing them, Intel would have probably kept netburst and tried to make it work, just because they could, they had the time, and they didnt have to worry about another company taking their profits.
and AMD is probably designing a new architecture, but, right now, they are focusing on their Dual core options, which are, btw, FAR better than intels options. Two P4s slapped together is not the way a dual core should be, AMD's offering is much better.
How can you say that Intel canned Netburst because it was a mistake and because AMD was killing them because of it, and then go on to say that Intel would have stuck with NetBurst if it weren't for AMD? How does it benefit Intel's bottom line to pour money into a currently unfeasible architecture while they have a perfectly viable architecture (P6/PM) that has proven performance and dramatically lower power and thermal requirements?RaphaelVinceti said:
Intel realized that the NetBurst architecture in practice would alienate a great deal of their customers--embedded computers, industrial computers, notebook computers, home theater computers, and blades. At best, the increasing power and thermal demands were costing R&D money as well as higher repair and customer service costs with Intel's OEMs.
Let me ask you--would you take the high road when the low road is easier and safer?
skully said:
The stability of Intel processors is only achieved when combined with Intel chipset based motherboards. AMD's former stability issues were the fault of 3rd party chipset firms and motherboard builders. Nvidia deserves most of the credit for making the AMD Athlon platform more competetive in the areas of features and stability.