Sandy Bridge news from IDF

[antiacid]

ok, so basically, after I bought an i7 920 D0, I won't need to upgrade until ivy bridge / late sandy bridge (for the 4c/8t or 6c/12t unlocked processors). Thanks intel, now I can forget about upgrades on the cpu side for another year at least.

[Sampsa]

Intel has told some official info about Sandy Bridge overclocking. They also list BCLK adjusting in < 1 MHz increments but rumors say you can raise it very little.

PDF: https://intel.wingateweb.com/us10/sc...og/catalog.jsp

Source: Muropaketti.com

[RPGWiZaRD]

Yea looks like they've already started to promote their "K" SKUs if you want to overclock. Memory in 266MHz increments hmm... sounds hard to match the optimal CPU clock with optimal RAM OC.

[Chri$ch]

link doesn´t work

[FlanK3r]

resave pdf at another free link pls

[kl0012]

Link to the full IDF catalog:
https://intel.wingateweb.com/us10/sc...og/catalog.jsp

[FlanK3r]

some donwload it and uplod it...here dont see too it

[BenchZowner]

http://rapidshare.com/files/41945416...2_103.pdf.html

[kl0012]

Intel's senior principal engineer explains SB ring architecture:
http://www.theregister.co.uk/2010/09..._interconnect/

[PatRaceTin]

hmmm wait...................

[-Sweeper_]

[-Sweeper_]

sweeet

PI 1M





PI 32M





CB 11.5

[RPGWiZaRD]

9.5 - 18.3% advantage in those tests. Well at least it's something. Still hope for better overclockability though.

The ring architecture looks like a smart move.

[Tenknics]

Intel's senior principal engineer explains SB ring architecture:
http://www.theregister.co.uk/2010/09..._interconnect/

really really awesome read. thanks man

ring bus looks pretty dope..

"Our bandwidth provided by this ring for each element connected to the ring connect gives 96 gigabytes per second ... if you're talking about running at 3GHz," Kahn said. "The multi-bank last-level cache for a four-core product provides upwards of 380 gigabytes per second. This is 4X of what existed in previous generations — even the two-core product is 190 gigbytes per second."

looks like the ring bus is going to be a mainstay for intel cpu uARCHs from here on out:

The new ring interconnect, according to Kahn, has a bright future: "This ring, in a very similar implementation, is going to go in the Sandy Bridge-EP and -EX with their large numbers of cores, and we see this ring scaling forward into the next generation as well."

And not just the next generation: "We actually believe this architecture is going to be scalable into any forseeable future that we have with Intel for the client space," Kahn said, "and, believe me, we are looking quite far ahead into the future, and we're going to be able to maintain this ring architecture for a long time."

so obviously ivyB, Haswell for sure and beyond? wow


edit: Also from here on out, Intel is no longer calling L3 Cache, L3 cache... but LLC - Last Level Cache...found that interesting

[Manicdan]

9.5 - 18.3% advantage in those tests. Well at least it's something.

they said that HT was suppose to get alot better, which means i was expecting the CB scores to be much, much higher. it looks like HT is the same, but some IPC is the only thing really that went up

also we cant see the memory settings and that plays a big part too.

honestly guys, why would you switch to SB if you already have an i7? is 15% better IPC and a supposed higher OC really worth it, or would you guys just wait for more cores before you do a full upgrade?

[informal]

I think many i7 users will just wait for six core unlocked SB ,since these quads are not that big of an upgrade compared to the Lynnfield line.

[R101]

To think that from 2006 and c2d we came from 3.6 GHz to 2.5 for ~ 14s SPI 1M,
not a bad improvement at all . Not to mention that SB clocks higher, too.

EDIT:
Also - this is a 2500 es? 2600 also has more cache, so couple of % there, too?

[JumpingJack]

I think many i7 users will just wait for six core unlocked SB ,since these quads are not that big of an upgrade compared to the Lynnfield line.

I think you are right, the bulk majority of these CPUs will go into OEM systems and sold as complete packages, with 20% more performance at the same price point SB should put significant downward price pressure on existing AMD systems and older generation Intel systems.

[-Sweeper_]

I think many i7 users will just wait for six core unlocked SB ,since these quads are not that big of an upgrade compared to the Lynnfield line.

probably... at least the performance per clock went up more than nearly zero or up to 10% as some people (without any knowledge) were speculating....

sandy 4c/8T is a nice upgrade for those on the LGA775 platform though...

[-Sweeper_]

To think that from 2006 and c2d we came from 3.6 GHz to 2.5 for ~ 14s SPI 1M,
not a bad improvement at all . Not to mention that SB clocks higher, too.

EDIT:
Also - this is a 2500 es? 2600 also has more cache, so couple of % there, too?

yeah.. and SB-EP will have more cache (per core) than bloomfield/westmere...

[orangekiwii]

My q9550 just took a crap on me and I've temporarily replaced it with a e3300 celeron at 4ghz (won't go past no matter what voltage even though it only takes about 1.3 to get to 3.9 )

These can't come out soon enough, going from a stock q9550 to this celeron is like night and day. Just sitting on desktop feels sluggish in comparison, not to mention anything with any form of multi threading runs pitifully slow in comparison.

Was planning on just keeping my q9550 out through this gen, but theres no point in buying a replacement one when I can upgrade sockets for only 200$ more.

[informal]

probably... at least the performance per clock went up more than nearly zero or up to 10% as some people (without any knowledge) were speculating....

sandy 4c/8T is a nice upgrade for those on the LGA775 platform though...

I said up to 10% and i was off by 5%(and the gains vary ,going to below 10%).Better than some others who expected miraculous Conroe-like jumps .

For those on LGA this is logical choice if they want to go intel again,especially having the upgradeability in mind(why go 1156?).

[-Sweeper_]

I said up to 10% and i was off by 5%(and the gains vary ,going to below 10%).Better than some others who expected miraculous Conroe-like jumps .

For those on LGA this is logical choice if they want to go intel again,especially having the upgradeability in mind(why go 1156?).

most CPU intensive tasks (SB vs lynnfield) I've seen show a 15-20% boost (same clocks), even more in some handpicked (specially FP) tests, similar or better than the core -> nehalem (without HT) transition, not bad at all..

core i5 2500k will probably be a popular choice for gamers (still on LGA775) with OC in mind...

[informal]

Not in AT preview and not in coolalers latest preview.But yeah,there are always corner cases where you can maybe find 20% increase,but that's corner case. Like someone else posted in this thread,according to coolalers latest results,gains are in 9.5-18.3% range.Pick an average and you get close to 15%. That's not bad at all,but not like some people were predicting.

edit: to add,SB does have better Turbo,so this might sway some i7 owners who don't want to mess with manual OCing.This is a good feature for average users.

[-Sweeper_]

Not in AT preview and not in coolalers latest preview.But yeah,there are always corner cases where you can maybe find 20% increase,but that's corner case. Like someone else posted in this thread,according to coolalers latest results,gains are in 9.5-18.3% range.Pick an average and you get close to 15%. That's not bad at all,but not like some people were predicting.

AT preview had turbo off on SB and turbo on on lynnfield, if you do the math you'll see more than ''up to 15%'' in many tests..

coolaler's latest results also show >15% perf. increase at equal clocks in everything but 3DMark, so....