Article: Sandy Bridge for Servers
By: David Kanter (dkanter.delete@this.realworldtech.com), July 28, 2011 11:01 am
Room: Moderated Discussions
As one reader pointed out, we omitted mention of the speed of QPI 1.1...but this was for a good cause, saving some juicy tidbits for this week's article.
In this piece, we discuss the system architecture (aka uncore) of the upcoming Sandy Bridge-EP. This includes a discussion of changes to the ring interconnect compared to consumer Sandy Bridge variants. We also examine some of the physical design choices that were described at ISSCC earlier this year, particularly in the area of clock domains.
"Intel's Sandy Bridge-EP arrives late this year to take on AMD's Bulldozer in 2 and 4-socket servers. It offers up to 8 cores with a new system architecture including 20MB L3 cache, 4 DDR3 memory controllers and faster 8GT/s QPI 1.1 links. Sandy Bridge-EP is also the first server CPU to integrate PCI-E 3.0 on-die, with up to 40 lanes - a significant bandwidth and power efficiency advantage.
This article compares the system architecture and design to previous approaches and shows that Sandy Bridge-EP will be a compelling upgrade for 2-socket servers and attractive for certain 4-socket systems, particularly those with large I/O needs."
http://www.realworldtech.com/page.cfm?ArticleID=RWT072811020122
Comments, feedback and suggestions are welcome.
David
In this piece, we discuss the system architecture (aka uncore) of the upcoming Sandy Bridge-EP. This includes a discussion of changes to the ring interconnect compared to consumer Sandy Bridge variants. We also examine some of the physical design choices that were described at ISSCC earlier this year, particularly in the area of clock domains.
"Intel's Sandy Bridge-EP arrives late this year to take on AMD's Bulldozer in 2 and 4-socket servers. It offers up to 8 cores with a new system architecture including 20MB L3 cache, 4 DDR3 memory controllers and faster 8GT/s QPI 1.1 links. Sandy Bridge-EP is also the first server CPU to integrate PCI-E 3.0 on-die, with up to 40 lanes - a significant bandwidth and power efficiency advantage.
This article compares the system architecture and design to previous approaches and shows that Sandy Bridge-EP will be a compelling upgrade for 2-socket servers and attractive for certain 4-socket systems, particularly those with large I/O needs."
http://www.realworldtech.com/page.cfm?ArticleID=RWT072811020122
Comments, feedback and suggestions are welcome.
David
| Topic | Posted By | Date |
|---|---|---|
| Sandy Bridge-EP article online | David Kanter | 07/28/11 11:01 AM |
| Sandy Bridge-EP article online | David Ball | 07/28/11 08:44 PM |
| Sandy Bridge-EP article online | David Kanter | 07/28/11 09:50 PM |
| PCIe3 cache interaction | Rohit | 07/28/11 10:48 PM |
| PCIe3 cache interaction | anon | 07/29/11 06:48 AM |
| Sandy Bridge-EP article online | savantu | 07/28/11 10:46 PM |
| Sandy Bridge-EP article online | David Kanter | 07/29/11 01:56 AM |
| Sandy Bridge-EP article online | IntelUser2000 | 07/30/11 05:11 AM |
| Sandy Bridge-EP article online | EduardoS | 07/29/11 03:37 PM |
| Sandy Bridge-EP article online | Will Smith | 07/28/11 11:55 PM |
| Sandy Bridge-EP article online | David Kanter | 07/29/11 12:13 AM |
| Sandy Bridge-EP article online | Will Smith | 08/04/11 01:19 AM |
| High speed PHYs | David Kanter | 08/04/11 02:45 AM |
| GX++ | Michael S | 08/04/11 12:02 PM |
| GX++ | Brian O' Hanlon | 08/04/11 02:33 PM |
| I meant exactly what I wrote (NT) | Michael S | 08/04/11 03:09 PM |
| Thanks | Brian O' Hanlon | 08/04/11 03:28 PM |
| GX++ | Aaron Spink | 08/04/11 05:54 PM |
| GX++ | Michael S | 08/05/11 03:44 AM |
| GX++ | Michael S | 08/05/11 04:14 AM |
| GX++ | Gabriele Svelto | 08/05/11 08:44 AM |
| GX++ | Brian O' Hanlon | 08/05/11 09:34 AM |
| GX++ | David Kanter | 08/05/11 09:30 AM |
| GX++ | Michael S | 08/06/11 10:05 AM |
| GX++ | David Kanter | 08/04/11 08:29 PM |
| GX++ | Michael S | 08/05/11 03:02 AM |
| GX++ | David Kanter | 08/05/11 09:45 AM |
| GX++ | Jesper Frimann | 08/04/11 09:57 PM |
| GX++ | Michael S | 08/05/11 08:27 AM |
| GX++ | S. Rao | 08/05/11 10:13 AM |
| GX++ | Michael S | 08/06/11 10:27 AM |
| GX++ | S. Rao | 08/06/11 03:29 PM |
| GX++ | Michael S | 08/07/11 12:52 AM |
| GX++ | S. Rao | 08/08/11 07:39 PM |
| GX++ | Jesper Frimann | 08/07/11 01:08 AM |
| GX++ | Michael S | 08/07/11 02:52 AM |
| GX++ | Jesper Frimann | 08/07/11 11:39 PM |
| High speed PHYs | Will Smith | 08/05/11 01:38 AM |
| High speed PHYs | Michael S | 08/05/11 02:50 AM |
| High speed PHYs | Will Smith | 08/05/11 04:54 AM |
| High speed PHYs | Michael S | 08/05/11 07:00 AM |
| High speed PHYs | David Kanter | 08/05/11 09:41 AM |
| High speed PHYs | Will Smith | 08/05/11 12:18 PM |
| High speed PHYs | David Kanter | 08/05/11 01:04 PM |
| High speed PHYs | Will Smith | 08/06/11 01:05 AM |
| Market Segments | Brian O' Hanlon | 08/06/11 10:39 AM |
| Market Segments | Will Smith | 08/06/11 12:25 PM |
| Market Segments | Brian O' Hanlon | 08/06/11 03:11 PM |
| High speed PHYs | Michael S | 08/06/11 10:48 AM |
| High speed PHYs | Will Smith | 08/06/11 12:19 PM |
| High speed PHYs | Alberto | 08/07/11 04:30 AM |
| High speed PHYs | Brian O' Hanlon | 08/07/11 04:11 PM |
| High speed PHYs | Alberto | 08/08/11 01:53 AM |
| Posts removed, user banned | David Kanter | 08/08/11 09:01 AM |
| High speed PHYs | Michael S | 08/08/11 03:29 AM |
| High speed PHYs | Alberto | 08/08/11 05:23 AM |
| High speed PHYs | David Kanter | 08/08/11 09:04 AM |
| High speed PHYs | Alberto | 08/08/11 04:08 PM |
| High speed PHYs | anon | 08/09/11 01:42 AM |
| High speed PHYs | Michael S | 08/09/11 02:30 AM |
| High speed PHYs | Michael S | 08/06/11 11:07 AM |
| Sandy Bridge-EP article online | Michael S | 08/04/11 11:21 AM |
| Sandy Bridge-EP article online | Will Smith | 08/05/11 12:21 AM |
| [O.T.] Confused about 10GE-Cu phys | Michael S | 08/07/11 10:42 AM |
| [O.T.] Confused about 10GE-Cu phys | David Hess | 08/07/11 11:51 AM |
| [O.T.] Confused about 10GE-Cu phys | Michael S | 08/08/11 12:53 AM |
| [O.T.] Confused about 10GE-Cu phys | president_ltd | 08/08/11 02:06 AM |
| [O.T.] Confused about 10GE-Cu phys | president_ltd | 08/07/11 03:15 PM |
| [O.T.] Confused about 10GE-Cu phys | Michael S | 08/08/11 12:46 AM |
| [O.T.] Confused about 10GE-Cu phys | president_ltd | 08/08/11 02:04 AM |
| [O.T.] Confused about 10GE-Cu phys | ajensen | 08/08/11 03:45 AM |
| [O.T.] Confused about 10GE-Cu phys | president_ltd | 08/08/11 04:30 AM |
| [O.T.] Confused about 10GE-Cu phys | David Hess | 08/08/11 05:26 AM |
| [O.T.] Confused about 10GE-Cu phys | Michael S | 08/08/11 06:43 AM |
| [O.T.] Confused about 10GE-Cu phys | president_ltd | 08/08/11 12:22 PM |
| [O.T.] Confused about 10GE-Cu phys | Adrian | 08/11/11 05:41 AM |
| LLC latency | Michael S | 07/29/11 04:11 AM |
| LLC latency | Alberto | 07/29/11 09:44 AM |
| on-die IOH and hot-plug PCIe | Michael S | 07/29/11 04:44 AM |
| Sandy Bridge-EP article online | forestlaughing | 07/29/11 07:48 AM |
| Sandy Bridge-EP article online | David Kanter | 07/29/11 10:08 AM |
| Possible error | Kevin G | 07/30/11 01:04 AM |
| Possible error | David Kanter | 07/30/11 11:34 AM |
| Possible error | Michael S | 07/30/11 12:08 PM |
| Possible error | Kevin G | 07/31/11 01:50 AM |
| Typo or mistake found. | I.S.T. | 08/02/11 02:50 AM |
| Typo or mistake found. | David Kanter | 08/02/11 09:10 AM |
| Typo or mistake found. | I.S.T. | 08/02/11 11:00 AM |
| Typo or mistake found. | David Kanter | 08/02/11 01:14 PM |
| Possible error | Michael S | 07/30/11 12:03 PM |
| Sandy Bridge-EP article online | intel user | 07/31/11 11:21 AM |
| I/O Bandwidth | Brian O' Hanlon | 07/31/11 08:13 PM |
| Read the rest of the article... | David Kanter | 07/31/11 11:15 PM |
| Read the rest of the article... | Brian O' Hanlon | 08/01/11 12:40 AM |
| Read the rest of the article... | David Kanter | 08/01/11 12:55 AM |
| Read the rest of the article... | Brian O' Hanlon | 08/01/11 02:41 AM |
| OT: OP asked about rendering, so benchmark it! :-) | JMK | 08/09/11 12:00 PM |
| Sandy Bridge-EP article online | John Meyer | 09/06/11 06:04 AM |
| Sandy Bridge-EP article online | David Kanter | 09/06/11 02:11 PM |
| Memory fairness | Mark Roulo | 09/06/11 02:35 PM |
| Memory fairness | John Meyer | 09/07/11 03:29 PM |
| Memory fairness and rings | David Kanter | 09/07/11 08:46 PM |
| Sandy Bridge-EP article online | Tom Abramson | 09/07/11 08:49 AM |
| Sandy Bridge-EP article online | David Kanter | 09/07/11 06:08 PM |



