Shanghai 45nm K10.5 launches on the 13th

[tmjuju]

http://www.fudzilla.com/index.php?option=com_content&task=view&id=10482&Itemid=1

 

It looks like that Shanghai for servers actually has much to offer and that AMD’s memory interface really plays an important role against Harprtown based Quad core server than we expected. The 45nm Shanghai with 4x512KB L2 cache and 6MB of shared L3 cache shows hard time even to mighty Dunnington, the six core server CPU.

 

We looked at the spec.org scores of the Spec 2006 test and it looks that Shanghai at the same frequency is about fifteen percent better than Barcelona, which is certainly better than most expected.

 

Using Specint_rate2006 an integer test, at the same frequency, Shanghai is about the same performance as Harpertown. This would likely be true for database applications but you cannot really be sure based only on this numbers. For HPC applications (Using SPECfp_rate2006), Shanghai at 2.7Ghz is faster than all current Intel processors (2-socket and 4-socket). This is something that we haven’t seen coming.

 

Dunnington should be significantly faster than Shanghai on database applications, but Shanghai will have an advantage in virtualization performance. AMD is also fortunate that 2S Nehalem doesn't come out until Q1 2009 and it might capture a massive market share based on its early arrival. The scores are below and SPECint_rate2006 is an integer test while SPECfp_rate2006 is a floating point test.

 

The summary of the scores is below and the rest is here. The bigger is better.

 

SPECint_rate2006 (base/peak) SPECfp_rate2006 (base/peak)

 

Shanghai 2.7GHz (2384, 2S) 113/136 105/118

Shanghai 2.7Ghz (8384, 4S) 202/249 188/210

 

Barcelona 2.5Ghz (2360, 4S) 92/106 83/90

Barcelona 2.5Ghz (8360, 4S) 169/197 153/170

 

Harpertown 3.2Ghz (X5482, 2S) 145/156 86/93

Harpertown 2.66Ghz (E5430, 2S) 119/129 69/74

 

Dunnington 2.66Ghz (X7460, 2S) 144/158 74/78

Dunnington 2.66GHz (X7460, 4S) 267/290 141/154

[nikosl]

Κοινώς μέχρι να βγάλει τη νέα σειρά η ΙΝΤΕl η AMD θα έχει το πάνω χέρι στους σέρβερς.

[FarCry]

Κοινώς μέχρι να βγάλει τη νέα σειρά η ΙΝΤΕl η AMD θα έχει το πάνω χέρι στους σέρβερς.

 

οι 6 πύρινοι έχουν βγει!!!!

[tmjuju]

οι 6 πύρινοι έχουν βγει!!!!

 

Από όσο θυμάμαι σε 6 μήνες θα βγουν…

Και ανάλογα τα sockets…

[FarCry]

Από όσο θυμάμαι σε 6 μήνες θα βγουν…

Και ανάλογα τα sockets…

 

κάτσε πως το εννοείς αυτό? γιατί ο shangai ποτε θα μπει? ανακοινώθηκε μετά τις intel

[tmjuju]

Προσθήκη στο καλάθι αγορών?

 

Κλικ…

2S

http://www.pcconnection.com/IPA/Shop/Product/Detail.htm?sku=9240271&br=14

http://www.buy.com/prod/amd-opteron-quad-core-model-2382-without-fan/q/loc/101/209765523.html

 

4S+

http://www.buy.com/prod/amd-opteron-quad-core-model-8382/q/loc/101/209765531.html

http://www.pcconnection.com/IPA/Shop/Product/Search.htm?term=8382%20&DefSort=Y

 

I rest my case!

Και εάν όλα πάνε όπως συνήθως περίμενε 2S σε ~έξι μήνες και 4S+ σε κάνα χρόνο. Μακάρι να έρθουν και γρηγορότερα δηλαδή.

 

Mε αυτά και με αυτά - ‘ανακοίνωση’ ‘επίδειξη’ ‘marketing’ - μας έχουν κατά-μπερδέψει…

[FarCry]

αμάν κιόλας? καλά συγγνώμη δηλαδή η intel ανακοινώνει και βγάζει στην αγορά μετά από 6 μήνες?

 

ox τι βάψαμε....

[tmjuju]

Και η AMD το ίδιο κάνει.

Απλά τα τελευταία χρόνια το marketing μας έχει πεθάνει… πρώτα ανακοίνωση του κωδικού ονόματος, μετά της τεχνολογίας, μετά φωτο του πύρινα, μετά closed benchmarks, μετα NDAa , ανακοίνωση partners, ανακοίνωση τιμών και ονομάτων, engineering sample to trusted reviewers, FUD, κ.ο.κ.

(όχι απαραίτητα με αυτή τη σειρά)

 

Για αυτό λέω ότι όλοι μας έχουμε μπερδευτεί με αυτά και με αυτά.

[FarCry]

LINPACK: Intel's Nehalem versus AMD Shanghai

 

We don't pretend that the comparison is completely fair: the Nehalem platform uses unbuffered RAM which has slightly lower latency and higher bandwidth than the Xeon "Nehalem" will get. But we had to satisfy our curiousity: how does the new "Shanghai" core compare to "Nehalem"?

 

 

 

http://www.anandtech.com/weblog/showpost.aspx?i=528

http://www.anandtech.com/weblog/showpost.aspx?i=529

 

 

-----Προστέθηκε 13/12/2008 στις 05 : 04 : 20-----

 

Nehalem

 

The CRN Test Center has reviewed evaluation units of the chip giant's latest processors and motherboards over the past several weeks. Early results show nearly historic levels of improvement over previous generations of processors. The Test Center took a first look at the Core i7-965 Extreme Edition, installed in Intel's "SmackOver" motherboard, the DX58SO Extreme Series.

 

From the moment the power was turned on, it was obvious that this was a fast machine. A complete boot of Vista took only 43 seconds, with most of the time (27 seconds) in the POST phase of the boot.

 

Using Primate Labs' GeekBench2, our system scored an incredibly high 7,998. This CPU scored the highest the Test Center has ever recorded on Geekbench.

Shanghai

 

In our tests, we used a Super Micro server loaded with two Opteron 2384 processors, which run at 2.71 GHz each. The system had 16 GB of RAM, and Microsoft (NSDQ:MSFT) Windows Server 2008 64-bit with SP1.

 

Using Primate Labs' Geekbench2 for benchmarking, the system reached a score of 7,912—one of the highest scores achieved by a server in our lab. Considering the unit had more memory than some of the others we compared it to, performance was still outstanding, especially when price and power consumption are factored in.

http://www.crn.com/hardware/212200536?pgno=4

 

 

-----Προστέθηκε 13/12/2008 στις 05 : 05 : 59-----

 

 

Για αυτό λέω ότι όλοι μας έχουμε μπερδευτεί με αυτά και με αυτά.

 

 

 

-----Προστέθηκε 13/12/2008 στις 01 : 36 : 01-----

 

 

 

http://www.amdzone.com/phpbb3/viewtopic.php?f=52&t=135802&st=0&sk=t&sd=a

 

 

-----Προστέθηκε 13/12/2008 στις 04 : 54 : 57-----

 

 

EPIC SCALING NEHALEM EP

 

All of that is gone with Nehalem EP, replaced with the sleek new Quick Path Interconnect and on-die memory controllers. The result is truly epic bandwidth and performance scaling. The big number many will be waiting for is the SPECfp_rate_base2006 base rate. So here it is: 160. (Performance estimated using preliminary hardware provided by Intel; for details of testing see this link.)

 

To put that into context, the Intel's current Penryn-based Xeon dual-socket platform fails to hit 90, even running at 3.4GHz. Perhaps even more significantly, AMD's shiny new 45nm Shanghai (Opteron 8384) chips in dual-socket 2.7GHz trim score just 105 points. Even a four-socket Shanghai rig is only good for just under 190. (Competitive data retrieved from www.spec.org on 24 November 2008.)

 

And remember, Nehalem EP's 160 point score is for a pre-production system running at 2.8GHz. Models humming a 3.2GHz tune will be available when the chip launches early next year. What's more, our testing was carried out on Windows Server 2008.

 

Run the same benchmark courtesy of the leaner, meaner Linux OS and you'd be looking at an even higher score.

 

In other words, as good as AMD's new Shanghai chip is, it appears that Nehalem EP will slap it around with a wet fish.

 

Just to confirm that our eyes did not deceive us, we also gave Nehalem EP a quick going over with the Stars Euler3D benchmark. It's a computational fluid dynamics simulation that majors on floating point performance.

 

Sure enough, Nehalem EP roasts all comers in this benchmark, too – it's twice as quick as a pair of 2.7GHz Shanghai processors (14.34 seconds to complete five instances versus 30.32 seconds).

 

Oh, and the platform's memory bandwidth is borderline biblical, clocking up 35GB/s in SiSoft Sandra's quick and dirty bandwidth benchmark.

http://www.techradar.com/news/computing-components/processors/first-look-intel-dual-socket-nehalem-ep-benchmarked-487131

[tmjuju]

Τα νούμερα είναι εντυπωσιακά, και μάλιστα σε κάποια workloads ο Nehalem είναι ΕΚΠΛΗΚΤΙΚΟΣ. Εκεί συμφωνούμε ως συνήθως.

 

Τα όποια fail-images τα θεωρώ γραφικά – ειδικά στον τομέα των servers. Επαναλαμβάνω ότι όταν συζητάμε για servers πρέπει να αντιμετωπίζουμε το θέμα σφαιρικά και να μελετάμε σωστά τις λύσεις που μας δίνει η αγορά.

 

Το μόνο το οποίο προσπαθώ να διευκρινίσω εδώ και αρκετά post είναι:

Σήμερα αγοράζεις Sanghai. Τόσο για μονή Mobo , όσο και για διπλή Mobo όσο και για τετραπλή. Δηλαδή εάν θες για τα Χριστούγεννα ένα 4socket server πάς και αγοράζεις sanghai. Το γιατί να θες ένα 4S για τα Χριστούγεννα είναι άλλο θέμα. LoL

 

Nehalem δεν μπορείς ακόμα να αγοράσεις για server. Θα μου πεις πάρε το desktop… Sorry! Άλλο ο server με τις buffered ecc ram κ.ο.κ. και άλλο το desktop.

To Q1 2009 θα μπορείς να αγορασεις Nehalem based Xeon EP . Οι EP είναι για 2S.

Για 4S θα πρέπει να περιμένεις μέχρι το Q3 2009, τότε θα βγουν οι Nehalem based Xeon EX.

 

 

Σωστά δεν τις λέω τις ημερομηνίες?

Εάν είναι διορθώστε με.

 

 

Εν τω μεταξύ για όποιον θέλει ένα 4S για τα Χριστούγεννα ας ρίξει μια ματιά και στους power…

[FarCry]

The Xeon 5570 is based on the same architecture as the Core i7. It is a 2.93 GHz quadcore CPU with 4 times a 256 KB L2-cache and one huge shared 8 MB L3.

 

The SAP numbers are absolutely astonishing, as Intel's dual socket is able to outperform quad socket opteron machines. Based on the scaling of Barcelona, we speculate that a quad Shanghai at 2.7 GHz would obtain the performance of the Dual Xeon 5570 w/o HT.The new Xeon 5570 outperforms the "old" 5450 by 119%!!!

 

http://www.anandtech.com/weblog/showpost.aspx?i=532

*Update (a few hours later): It seems that the SAP page was wrong about HT. It reported 8 threads on 8 cores on the Fujitsu Siemens Primergy Server. The certification page says otherwise: 16 threads on 8 cores

έτσι όπως πάει το πράγμα θα τελειώσει το θέμα τον servers στους Nehalem based Xeon EP. δε χρειάζεται να πάνε σε quad