Intel vs. AMD - the topic ... the poll #2

[AndreiD]

Poor AMD ... as vrea sa scoata si ei un K8L reusit ca sa mai stranga si ei ceva profit...
Parca s-au schimbat rolurile... Intel vine cu solutii energy efficient si performante iar amd cu goblini power hungry ... 'om trai si 'om vedeam.

[moonwalker]

Eh, să nu generalizăm de la Radeon R2D2, consumul pe Athlon e în continuă scădere. Pe 65nm oferta single core Lima vine cu TDP de 35W, dual core Brisbane vine la 65W sau EE la 45W, iar pe segmentul mobil Turionul a primit un boost de frecvenţă fără modificarea TDP-ului. În plus, procesoarele AMD pt desktop dau dovadă în continuare de un potenţial destul de bun de undervolting.

Şi în grădina Intel se mai răceşte ceva, E4400 vine într-o nouă revizie - M0. Sunt curios ce consum efectiv are.

[savantu]

Hai sa vedem ce zice Intel despre toata afacerea :

May 16 (Bloomberg) -- Intel Corp. Chief Executive Officer Paul Otellini said his chips' speed will outpace Advanced Micro Devices Inc.'s even after the company introduces new models.

``I expect us to have microprocessor performance leadership,'' Otellini said at Intel's annual meeting today in Santa Clara, California. Intel's lead will be ``substantial'' in chips for desktop parts, he said.
....
Intel had 74 percent at the end of the fourth quarter, the lowest level in about 11 years. Otellini and Chairman Craig Barrett said Intel will prevent Advanced Micro from gaining back sales, and will keep adding designs and manufacturing techniques.

Intel is still following the dictum of co-founder Andy Grove that ``only the paranoid survive,'' Barrett said.

``There is still ample paranoia at Intel,'' he said. ``Dr. Grove lives on with his doctrine.''

http://www.bloomberg.com/apps/news?pid= ... nU9iQGCdns

[moonwalker]

Well, you know what they say... perfect paranoia is perfect awareness.

[SoboLAN]

Sa zicem ca poate noile procesoare de la AMD vor avea ceva spor de performanta fata de concurentii lor de la Intel. Problema e ca, daca aceste Phenom-uri nu vor avea un potential de overclocking de cel putin 40-50%, atunci nu se vor vinde foarte bine, in ciuda probabilei performante stock superioare. Vorbesc aici de procesoarele desktop. Nicaieri nu se vorbeste de potentialul de overclocking al Phenom-urilor, dar poate ca nimeni nu cunoaste nimic despre asta. Daca vor avea potential mic de overclocking, atunci s-a compromis activitatea AMD pe piata asta (poate chiar pe toate).

Asta ar fi foarte rau, pentru ca practic AMD ar iesi de pe piata procesoarelor. Si stim cu totii ca fara AMD suntem pierduti. Asta pentru ca Intel nu a facut niciodata nimic fara un foarte sanatos sut in c*r. Daca K8 nu ar fi avut succesul pe care l-a avut (si Intel nu ar fi simtit ca pierde incet dar sigur din cota de piata), noi nu am fi vazut arhitectura Core poate nici in 10 ani.

[moonwalker]

SoboLAN, Brisbane are deja potenţial de overclocking de 40-50%, aşadar noile procesoare K8L, K10 sau 10h cum au început unii să le spună, au toate şansele să ofere cel puţin un potenţial de overclocking similar, poate chiar mai mare judecând după faptul că procesul de producţie pe 65nm a avut timp să se maturizeze.

Acesta este şi motivul pentru care întrebam mai devreme de A64 single core Lima, este cea mai nouă revizie disponibilă de la AMD bazată pe procesul lor de 65nm, şi un indicator bun pentru progresul făcut până acum. Din păcate nu am găsit nimic relevant.

[myx4ever]

dupa cum stim pe AMD abia incepand cu a 2a revizie lucrurile incep sa fie cu adevarat interesante.

[bolovan]

SoboLAN stai linistit...1-2% din procesoarele vindute sint cumparate de pasionatii de overclocking? Asta in cazul in care sint foarte optimist...
In fapt, din 300 de sisteme de care stiu eu la ora acutala...nici un sistem nu are overclock-ing facut...

[CANARIS]

Sincer, eu mă mir că Intel şi AMD nu s-au înţeles cumva să se pună contra OC-ului… sau să vină cu o soluţie de gen “Vreţi OC? Ok, puteti cumpara orice procesor, indiferent de frecvenţă, cu multiplicatorul deblocat. Chiar orice model. Garanţie? 0.”

[moonwalker]

Pentru că este mai profitabil.

Raţionamentul este simplu. Overclocking face doar segmentul de utilizatori entuziaşti, iar dacă le câştigi inimile acestor oameni, ai parte de o publicitate gratuită extraordinară, şi mai important decât atât, restul utilizatorilor tind să le urmeze exemplul şi sfatul şi să cumpere chipurile producătorului recomandat de entuziaşti.

Să luăm chiar exemplul actual. Majoritatea celor de pe forum recomandă aproape oricui sisteme Intel, iar la întrebarea "de ce să cumpăr un C2D când un X2 cu performanţe egale costă mai puţin?" li se răspunde "da, dar C2D se overclockează mai bine!".

Iată deci cum reuşeşte Intel să vândă procesoare la fel de performante dpdv IPC*Frecvenţă cu ale competiţiei la preţuri mai scumpe - procesoarele lor se overclockează mai bine. Acesta este motivul suprem, produsul Intel capătă aura irezistibilă a învingătorului şi este alegerea implicită chiar dacă utilizatorii de rând fac rareori OC.

Aşadar puteţi sta liniştiţi, producătorii de procesoare au la dispoziţie mecanisme foarte interesante şi totodată foarte eficiente prin care recuperează financiar vânzările scăzute de procesoare cu frecvenţe crescute.

[CANARIS]

Mda... ai dreptate. Fair enough.

[savantu]

Un articol must read despre Intel si modus operandi al echipei israeliene.

http://www.forbes.com/2007/05/17/perlmu ... r=yahootix

[CANARIS]

A relentless arguer, Perlmutter has long questioned the Intel orthodoxy that speed is the best benchmark for measuring a chip's value. This has worked out well for his employer. Four years ago Intel's speedy Pentium processors were running out of gas. Intel couldn't push their design much further without sucking up too much power and generating too much heat. Computers and laptops seemed to be getting hot enough to double as hibachis--and Intel's business was getting cooked by rival Advanced Micro Devices

Mai bine mai târziu decât niciodată…

[savantu]

Un articol de Ed , prietenii stiu de ce...

http://www.overclockers.com/tips01155/

[AndreiD]

Un articol de Ed , prietenii stiu de ce...

http://www.overclockers.com/tips01155/

AMD sucks big time dupa acel articol...

"All powerpoint but no product" .. titlul zice totul...

[van helsing]

Să luăm chiar exemplul actual. Majoritatea celor de pe forum recomandă aproape oricui sisteme Intel, iar la întrebarea "de ce să cumpăr un C2D când un X2 cu performanţe egale costă mai puţin?" li se răspunde "da, dar C2D se overclockează mai bine!".

Mda! Un E 6400 are performante apropiate cu un AMD x2 6000(poate chiar mai bune),doar ca X2 este mai scump cu 1 milion.Daca introduci in ecuatia aceasta si capacitatea de OC a celor doua procesoare,X2 ramane mult in urma.Nu inteleg de unde concluzia aceasta ca un X2 are performante egale la aceleasi costuri.
Sa fim seriosi in acest moment AMD miroase urat.Tastarile acestea subiective imi amintesc de vremurile cand Intel mirosea la fel de urat si persoane cunoscute acestui forum tastau opinii similare.

[moonwalker]

van helsing, ai scos afirmaţia mea din context şi ai tras şi o concluzie greşită.

Nu inteleg de unde concluzia aceasta ca un X2 are performante egale la aceleasi costuri.

Postul tău este însă exact genul de replică la care mă refeream. Entuziast fiind, tu iei în considerare şi acorzi mare greutate capacităţii de overclocking. Pentru majoritatea utilizatorilor de rând această capacitate nu contează la propriu, dar ei vor alege totuşi chipul recomandat de entuziast, măcar pentru senzaţia de "am cumpărat cel mai bun produs de pe piaţă".

Cât despre problema preţului, amândoi ştim că ai ales exemplul care te avantaja, anume cel de la sfârşitul spectrului de frecvenţe posibil pentru Athlon X2. Acolo e normal să se vada cu uşurinţă diferenţa între oferta AMD vs. Intel, pentru că X2 începe să respire greu pe când C2D mai au încă o margine bună de frecvenţă, mai ales dacă luăm în considerare produsul IPC C2D/IPC X2*Frecvenţă.

Aşadar, hai să lăsăm referirile la tastări subiective şi să citim mai atent posturile celorlalţi, şi mai ales subiectul pe care au postat.

[x_bush]

E6850 si Q6600 vor avea acelasi pret;voi pe care l-ati alege? E6850 pt frecventa si fsb mai mare sau Q6600 pt cele 4 nuclee?

[AndreiD]

Q6600 categoric.

[savantu]

Cloverton vs. Barcelona

http://www.uberpulse.com/us/2007/05/int ... u_ship.php

[Neon Biker]

Intrebarea mea este: De unde a facut rost Intel de 4 proace Barcelona?

[moonwalker]

Nu a făcut rost, a folosit rezultatele publicate de AMD.

Singura mică-mare problemă este că acele rezultate aveau ca scop unic o demonstraţie de creştere a performanţei faţă de actualele Opteron HE fără modificarea consumului de putere. Sistemul cu 16 nuclee Barcelona consumă cât sistemul cu 8 nuclee Opteron HE dar are un scor aproape dublu în benchmark-ul POV Ray. În ambele cazuri vorbim de procesoare optimizate pentru un cosum minim de putere şi tactate la sub 2Ghz.

Chiar şi aşa răspunsul AMD este foarte slab, ideea lor despre a fi "more talkative" rezumată la nişte slide-uri de PowerPoint cu o lipsă crasă de detalii tehnice despre sistemele respective le va face mai mult rău decât bine.

[Neon Biker]

Mda, intr-adevar e foarte suspecta tacerea asta a AMD-ului. Cu A64 au batut toba un an intreg inainte sa-l lanseze, acum insa tac malc. Nu stiu nici ce au de clocit atata vreme. Intel deja a lansat 2 nuclee noi (Prescott si Conroe) in timp ce AMD nimic. Bine... Prescott a fost un esec, dar tot nucleu nou se cheama.

[IL Dotore]

Iata ceva mai multe detalii: The clock speed of AMD Barcelona was......

[CANARIS]

Pentium 4 631 @ 8,180 GHz vs QX6700 @ 5,006 GHz

[CANARIS]

As if the latest Athlon 64 X2 CPUs weren't energy-efficient enough, AMD plans to release 45-watt versions for these CPUs. One of the reasons behind this move is that AMD intends to reposition the entire Athlon 64 X2 CPU family towards entry level users, immediately after the upcoming Phenom processors are released. In addition, the new energy-efficient lineup of Athlon 64 X2 CPUs won't be figuring any performance rating in the official denomination. Thus, AMD is readying the Athlon 64 X2 BE-2000-series.
AMD claims that it has already begun sampling the upcoming Athlon 64 X2 BE-2000-series and these should remain based on AMD’s 65nm Brisbane-core. As far as energy is concerned, the 45-watt TDP ratings will make the Athlon BE-2000-series as efficient as the previous single-core processors. (...)

(...) we further find out that the company schedules new price cuts for its server chips. AMD makes "a strong value proposition against the competition and these new prices will enable (users) to offer (their) clients AMD based server solutions at competitive prices." This being said, Opteron 100/1000 CPUs are expected to be 10 to 50 per cent cheaper by June 4th.

Sursa

[savantu]

Salariu CEO la AMD pentru 2006 : 12.8m$ + 9.28m $ = 22.08m$

Retributie invers paralela cu performanta , dar ce mai conteaza asta...

http://theinquirer.net/default.aspx?article=40009

Intel CEO Paul Otelini a primit 6.18m $ in 2006...

[CANARIS]

Not all CEO's are Steve Jobs, to take 1$ per year...

[IL Dotore]

AMD Unveils New CPU Model Naming & AM2+ Phenom FX

Leslie Sobon - Director, Desktop Business, AMD provided us some updates on Phenom today. According to her, there is no special meaning for G, S and L. G stands for Premium where Phenom X4 will compete against Kentsfield and Phenom X2 will compete against Conroe. S stands for Intermediate while Athlon 64 X2 will compete against the Intel Pentium 2000 series. As such, there will be AMD Phenom X4 GP-7xxx, Phenom X2 GS-6xxx, Athlon X2 LS-2xxx and Sempron LS-1XXX. She mentioned that there will be both Socket 1207+ as well as Socket AM2+ Phenom FX for more platform versatility and the FX series will retain back their model naming eg. AMD Phenom FX-xx.

[AndreiD]

AMD Athlon X2 BE-2350: Mainstream X2s with 45W TDP

AMD introduces mainstream dual core X2s with a 45W TDP and that cost less than $100. How does that sound?

Full Review :
http://www.anandtech.com/cpuchipsets/sh ... spx?i=3003

[savantu]

Hai sa mai postam unele stiri si sa mai trezim forumul :

Increasing Talk Barcelona Server Chip Will Be Delayed
Last update: 6/5/2007 12:37:38 PM

By Donna Fuscaldo
Of DOW JONES NEWSWIRES

Speculation is mounting that Advanced Micro Devices Inc.'s (AMD) much awaited Barcelona server chip will be delayed until the fall, which could pressure margins and weigh on its waning competitive position.
On Monday supercomputer maker Cray Inc. (CRAY) issued a revenue warning for the year, citing a delay in the availability of a component, which analysts said is likely the Barcelona server chip. Meanwhile, on Tuesday Citigroup analyst Glen Yeung said in a research report out of Taiwan's Computex Trade show, that "several sources," including server vendors, distributors and motherboard makers, confirmed Barcelona will be delayed to September/October from June/July.
Officials at AMD weren't immediately available to comment.
Barcelona is viewed by many on Wall Street as a saving grace for AMD, which has been pummeled by a fierce price war that's weighed severely on its gross margins. Barcelona is seen as a way for AMD, Sunnyvale, Calif., to boost its gross margins and make further inroads into the corporate server market, where it doesn't have that strong of a position.
A delay, even for a few months, could hurt AMD's bruised reputation with corporate customers and open the door for Intel Corp. (INTC) to steal more market share.
"If you look at all the important product introductions AMD has done in the past 12 months, it missed every single one," said Hans Mosesmann, an analyst at Nollenberger Capital Partners, who has a sell rating on AMD. A delay in Barcelona would be a "significant" miss and could push out the financial benefit until 2008, the analyst said.
Mosesmann suspects the bigger problem for AMD isn't a delay in the timing of the chip's roll out, but that the chip may not be up to snuff.
"AMD's graphic processor was delayed a couple of times...and it was significantly below Nvidia's class of product in terms of performance power," said Mosesmann. "Being late is bad. Being late with potentially a lousy product could be a disaster."
In its research report Tuesday, Citigroup's Yeung said AMD is showing a sample at Computex, but that yield problems is the likely reason for the delay in the chip going into volume production. According to Yeung, the delay could hurt gross margins and gives Intel more opportunity to gain share.
"Barcelona is generally viewed to be AMD's best opportunity to stem server share loss to Intel," said Yeung in the research report. "While AMD's recent notebook win at Toshiba may balance any negative revenue implications to our current model. Barcelona's delay has negative implications for AMD's gross margin."
...

And continued

Meanwhile, on Tuesday Citigroup analyst Glen Yeung said in a research report out of Taiwan's Computex Trade show, that "several sources," including server vendors, distributors and motherboard makers, confirmed Barcelona will be delayed to September/October from June/July.

And the piece de resistance ( Citigroup report ) :

Computex 2007: Day 1 Implications for Semiconductors; Lowering Estimates for AMD

Barcelona Delayed — While much speculated in the press, several sources (server vendors, distributors, and board makers) have confirmed that Barcelona, AMD’s much-awaited new server processor, has been delayed. From an original launch in June/July, Barcelona is now not expected until September/October (although we did see a sample device at the show). Yield problems for the estimated 286mm2 die was most frequently cited as the source of the delay.

Negative AMD Margin Implications— Barcelona is generally viewed to be AMD’s best opportunity to stem server share loss to Intel. While AMD’s recent notebook win at Toshiba may balance any negative revenue implications to our current model, Barcelona’s delay has negative implications for AMD’s gross margin. We note that server processors carry as much as 1500bps-2000bps greater gross margin than do notebook or desktop processors. We currently model a gross margin recovery of 1250bps in 4Q07 from 1Q07 levels. We tweak our AMD estimates lower to reflect risk to this assumption. Our 2007 EPS is revised to $(2.59) from $(2.48), no change to 2008.

Door Opens Wider for Intel Server MPU’s — Meanwhile, Barcelona’s delay opens the door wider for Intel to consolidate its share gain, particularly as it prepares to launch 45nm server processors in 4Q07. Intel currently expects its 45nm Harpertown quad-core processor to comprise ~8% of its sales in 4Q07. ....

8% ii excelent...

...
Door Opens Wider for Intel Server MPU’s (cont). Interestingly, one boardmaker we met with suggested that their tests indicated that AMD’s 65nm Barcelona only bridges the gap between Intel’s 65nm Clovertown. The release of 45nm Harpertown is expected to re-establish the lead for Intel, emphasizing the importance of any delay by AMD. Moreover, we remind investors that Intel will lower prices of its 65nm Clovertown server processors on July 7th to drive further penetration of quad-core processors ahead of AMD’s Barcelona launch.

Si distractia continua :

AMD took the opportunity at Computex to showcase server platforms running the Quad-Core AMD Opteron processors, codenamed Barcelona, with the help of MSI, Supermicro, TYAN and Uniwide.

The AMD quad-core processors are designed to drop-in (following a BIOS upgrade) with all existing AMD Opteron processor-based systems using DDR2 memory, while also enabling new platform capabilities such as those being demonstrated today.

One vendor demonstrated Barcelona to DailyTech running at 1.6GHz. Current AMD Barcelona samples are not scaling too well. AMD partners confirmed the highest running, POST and OS capable, Barcelona processor is 2.0 GHz.

2GHz ??!

Intel a demonstrat in martie Harpertown ruland la 3.33GHz , FSB 1600 si 12MB L2...Avand in vedere cele de mai sus , AMD abia va reusi sa egaleze Clovertownul actual ( presupun ca cel de 2.66 ) iar modelul de 3GHz este inaccesibil.
Si ca fapt divers , spre sfarsitul verii va fi lansat Seaburg , un nou chipset pt Xeon care va imbunatatii semnificativ performanta in lucrul cu memoria , calcaiul lui Ahile pentru Xeon la momentul de fata.

[IL Dotore]

AMD's Budapest is late but it denies Barcelona is balked :

A REPORT ON the Dow Jones wire confirmed that while the Budapest quad core chip will be delayed, Barcelona is still on time and will be rolled out this summer.

That follows earlier reports of Cray delaying a supercomputer because AMD was unable to ship a processor it was using on time.

But an AMD rep told the wire that Barcelona was still on target, despite claims from a financial analyst at the Computex show that this chip was likely to be delayed too.

AMD Desktop CPU Price Cut July 9th :

One such change will be announced on July 9th, when the AMD desktop processor family will have a reduction in suggested prices. Savings for the AMD Athlon™ 64 X2 (dual-core) processor family will be about 20% and the AMD Athlon 64 (single-core) processor family will be adjusted by approximately 15%. If you are looking for a great value, look to the AMD Sempron™ processor, which will also offer savings of about 15%.

[CANARIS]

Deja încep să cred că e ceva putred în Danemarca… mai ales că AMD tace mâlc şi nu suflă o vorbuliţă care să ne convingă cât de super vor fi noile lor procesoare.

E ceva în neregulă care nu-i în regulă.

[savantu]

Si prajitura de pe tort : in Cinebench K10 este cu 4,3% mai slab clock/clock decat C2D.

Cinebench completed the default benchmark in 27 seconds for the 1.6 GHz K10; 17 seconds for the Intel Xeon X3220. The Kentsfield Xeon was 58% faster with a 50% higher clock frequency for Cinebench.

http://www.dailytech.com/Quick+and+Dirt ... le7574.htm

Unde sunt aia care sustineau cum k10 va rupe Core ?

[CANARIS]

Măi… nu mai loviţi în cel căzut la pământ… nu-i fair play. Măcar în vremea NetBurst-ului, Intel avea cota de piaţă şi conturile din bancă drept consolare; AMD nu prea mai are nimic.

Nici măcar gură, din câte văd: când procesoarele lor le râdeau în nas celor de la Intel, AMD nu pierdea nici o ocazie să bată toba. Ori acu tăcerea lor spune multe.

[Tzeb]

Am mai postat linkul asta acum un an, dar merita revizitat pentru a primi o doza de realitate.


Pentru ce faceau ei "arogante" dinastea la vremea aia? Oare nu realizau ca sunt ca o musculita care bazaie prin fata monstrului ?
Chestii de genul ala pot irita la maxim si intr-o zi monstrul a zis - "Gata, hai sa ii ingropam !"
Aproape 2 ani mai tarziu uite ce au patit... deabia se mai tarasc de ti se face si mila de ei. Uneori nu e cruda viata?

[savantu]

Viata-i cruda ?

Vi se pare...



http://www.legitreviews.com/article/521/1/

2 Barcelona scot in jur de 16000...

[Arclite]

http://www.dailytech.com/Quick+and+Dirt ... le7574.htm. Nu prea spune mare lucru pt ca shi K8 in Cinebench era de la egal la cu Intel iar in restul apl nici nu se compara cu core 2-ul http://www.hwupgrade.com/articles/cpu/2 ... 0+_10.html

[savantu]

Intel primeste licenta SLI : http://theinquirer.net/default.aspx?article=40123

[CANARIS]

(1) AMD has just announced a 9-watt processor, a Sempron 2100+ to be more precise. This particular chip is obviously not performance-oriented, but is intended for fanless embedded designs.


(2) AMD to announce new naming system next week


(3) Test AMD Athlon X2 BE-2350
On that note, the AMD Athlon X2 BE-2350 performed exactly as we expected today. Unlike other low power consuming chips where clock speed is severely compromised, the BE-2350 performs as a similarly clocked previous generation X2. Performance-wise, this is exactly what AMD intended, but using a lot less power than before. Comparing it to the AMD Athlon 64 X2 Dual-Core Processor 5000+ Socket AM2 (65W), it did trail in all of the benchmarks but there is a 500MHz frequency difference between the two. Coupled with our MSI K9AGM2-FIH, we did not experience any dropped frames in FFDShow while viewing a DVD we were up-converting.


(4) Intel shows off 8-core 45nm Penryn benchmarks


(5) Intel Penryn quad-core spy pics


(6) All C2D models