Message boards :
Technical News :
new server specs (Oct 28, 2010)
Message board moderation
Previous · 1 · 2 · 3 · Next
Author | Message |
---|---|
Oleg A. Kobytev Send message Joined: 28 Mar 03 Posts: 8 Credit: 23,327,002 RAC: 1 |
HP ProLiant BL460c G7 with two blades + SAN (such as HP LeftHand or EVA) + VMware - it would be much better... It seems to me, ridiculously to use server's disk capacity for data storage. OLEG. |
Michael Reaves Send message Joined: 4 Jun 99 Posts: 23 Credit: 22,801,220 RAC: 0 |
They should be getting at least the government/education pricing |
Dimly Lit Lightbulb 😀 Send message Joined: 30 Aug 08 Posts: 15399 Credit: 7,423,413 RAC: 1 |
I'm not an IT person but 96GB RAM sounds like a dream. |
X-Files 27 Send message Joined: 17 May 99 Posts: 104 Credit: 111,191,433 RAC: 0 |
Wouldn't this be a step back (cpu wise) - from 4 x six-core 2.13GHz Xeon (I believe its Xeon E7530) to 2 x quad-core E5620? |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
Wouldn't this be a step back (cpu wise) - from 4 x six-core 2.13GHz Xeon (I believe its Xeon E7530) to 2 x quad-core E5620? Guessing as to the motivations: Depends. Databse activity is typically I/O bound to start with, but in this case the newer 4core chip & the 6core chip have the same total L3 cache, so that's more total cache area per core for the quad which would lead to less thrashing (of memory & disk) with larger tasks, that's a big deal, as well as the higher processor frequency (Base 2.4GHz instead of 1.866GHz). Rated Total Power Dissipation is lower (80 instead of 105W), so substantially cooler running though probably amounts to about the same processing power overall. The big one though is probably the price diffference. Judging by the current tray prices ( $1446 for the six core, versus $387 for the quad ) it's difficult to justify the hex core of similar overall processing capacity, when for less money you can get 3 quad processors instead of one six-core. If one of those quads 'blows up' then it's not going to break the bank either. "Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions. |
Helli_retiered Send message Joined: 15 Dec 99 Posts: 707 Credit: 108,785,585 RAC: 0 |
.... If one of those quads 'blows up' then it's not going to break the bank either. I pushed (overclocked) every Intel CPU since Pentium 60 to their limits - and beyond - but had never killed any Intel CPU... Helli A loooong time ago: First Credits after SETI@home Restart |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
I pushed (overclocked) every Intel CPU since Pentium 60 to their limits - and beyond - but had never killed any Intel CPU... Me neither, but then I don't run my machines in a hot server closet with ~250,000 or so active users, and stock cooling (or HP designed) . I think that ~20% lower disspiation for similar capacity might have been a factor in the choice, given that HP probably won't want to support some custom cooling mods under warranty. "Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions. |
Helli_retiered Send message Joined: 15 Dec 99 Posts: 707 Credit: 108,785,585 RAC: 0 |
Well, you're right. I forgot that such rackmount Server have a poor Cooling Solution for 24/7 under full Load. Helli A loooong time ago: First Credits after SETI@home Restart |
ivan Send message Joined: 5 Mar 01 Posts: 783 Credit: 348,560,338 RAC: 223 |
2 Xeon quad core E5620 processors. Nice. Be aware that if you are using the Intel chipset and running the memory at 1333 MHz, adding memory to the second bank reduces all memory access to 1000 MHz. According to Supermicro at least... That's why I specified my last Xeon server at 6x4 GiB. |
OzzFan Send message Joined: 9 Apr 02 Posts: 15691 Credit: 84,761,841 RAC: 28 |
Wouldn't this be a step back (cpu wise) - from 4 x six-core 2.13GHz Xeon (I believe its Xeon E7530) to 2 x quad-core E5620? The old Xeon 7xxx series uses the antiquated "Front Side Bus" technology, severely limiting the ability for the memory to feed all 6 cores (the Xeon 7xxx series doesn't have HT), even if you're using the multi-FSB, one for each socket. The QuickPath Interconnect of the Xeon 5xxx series should have no bottleneck in feeding the 4 cores + HT (8 possible threads per socket). Additionally, the Xeon 7xxx series has a shared L2 cache situation, in that each set of cores (3x dual cores = 6 cores) shares a 3MB L2 cache, increasing the need for cache-coherency (if one core changes the contents of the L2 cache, it must update it's "table of contents" so that the other core it is sharing the L2 cache with knows about the changes), which takes additional CPU cycles. The Xeon 5xxx series dedicates a 512KB cache per core, only needing to be split with a "virtual" core in HyperThreading. The higher 16MB L2 cache found in the Xeon 7xxx series is great for DB applications, but that can be offset by the additional threads and higher throughput of the QPI bus and increased memory capabilities. Jason touched upon another important point: heat dissipation. The Xeon series uses less wattage, thus needing less cooling than the 6-core beasts. This can translate into electrical savings per server, allowing them to fit more servers into their closet with less heat for future considerations. In all, the Xeon 5xxx series should be better for the project than the Xeon 7xxx series. |
Randy Send message Joined: 17 May 99 Posts: 6 Credit: 520,295 RAC: 0 |
Twins? New name Gemini? |
John A. Cummings Send message Joined: 28 Aug 99 Posts: 1 Credit: 2,889,667 RAC: 33 |
Two servers, two names from Gemini--Castor and Pollux. |
Patrick M. ORourke Send message Joined: 21 Aug 02 Posts: 28 Credit: 643,752 RAC: 0 |
I was going to post about the twins but then I read the first rule, pretty good for a first post. Patrick M. O'Rourke |
Dirk Villarreal Wittich Send message Joined: 25 Apr 00 Posts: 2098 Credit: 434,834 RAC: 0 |
Two servers, two names from Gemini--Castor and Pollux. Orion is a nice name..... |
kaydavis Send message Joined: 15 May 99 Posts: 10 Credit: 1,023,550 RAC: 0 |
Watch out for these CPU's and make sure that the "Green" technologies are not killing you.... http://www.brentozar.com/archive/2010/10/sql-server-on-powersaving-cpus-not-so-fast/ Yeah I know you are running a *.nix and mysql not the dreaded MickySoft stuff,, but when the system BIOS starts deciding you don't need full power... watch out Kay |
Matt Giwer Send message Joined: 21 May 00 Posts: 841 Credit: 990,879 RAC: 0 |
Two servers, two names from Gemini--Castor and Pollux. Yes, it is. Unvarnished Haaretz Jerusalem Post The origin of the Yahweh Cult |
ManwithnoPlan Send message Joined: 26 May 03 Posts: 1 Credit: 16,286,845 RAC: 0 |
@$o |>0/|/3 read again... http://ark.intel.com/MySearch.aspx?s=t&FamilyText=Intel%C2%AE%20Xeon%C2%AE%20Processor&CoreCountMin=6 |
MarkJ Send message Joined: 17 Feb 08 Posts: 1139 Credit: 80,854,192 RAC: 5 |
Two servers, two names from Gemini--Castor and Pollux. Also the name of the cat in Men In Black. BOINC blog |
OzzFan Send message Joined: 9 Apr 02 Posts: 15691 Credit: 84,761,841 RAC: 28 |
@$o |>0/|/3 read again... Thanks for pointing that out. I was only aware of the Xeon 74xx (Dunnington) series. I didn't know about the 75xx+ series (Nehelem EX). I guess it could be considered a step back in processing power, but it's still a step forward in reliability considering the existing Xeon 7530 motherboard was a test unit, and if I recall correctly the guys in the lab said one of the CPU sockets didn't work. |
APCyberax Send message Joined: 6 Jun 01 Posts: 29 Credit: 24,078,024 RAC: 48 |
Two servers, two names from Gemini--Castor and Pollux. that would also make a nice logo for the server :) |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.