Feeds

back to article HP forges 32-socket Itanium iron

What's going on with Hewlett-Packard's Itanium-based Integrity server lineup? It has been over a year since HP got its first machines using Intel's "Tukwila" quad-core Itanium 9300 processors into Integrity machines, and the company has still not delivered machines spanning 32 and 64 sockets. This is the kind of big iron that …

COMMENTS

This topic is closed for new posts.
Silver badge
Unhappy

Lashing together?

Quote

Bresniker tells El Reg that OpenVMS 8.4 will span across all of the four blades in the most recent Tukwila blade servers, delivering 32-core scalability, and that customers who need more scalability than that can do VAXclusters lashing multiple blade boxes together

Sigh.

Come on El Reg. Just ask anyone who has used VMSClusters in production, they are not lashed together.

VAXClustering has been around since 1983. I witnessed some of the FCC RFI Testing of the kit in a place called 'The Bubble' near Marlboro, MA. I was testing a VAX 11/730 combo.

Since the release of the Alpha CPU the clustering was renamed VMSClusters.

If this is HP still saying VAX Cluster then I'd be rather worried if I were an HP customer.

Even nearly 30 years after its introduction VMSClustering gives a whole lot of features that those in the Unix/Linux world can only drool at in envy. A fully implemented OOTB cluster wide file system will be IMHO its lasting legacy.

2
0
Heart

HP and Intel continue to divest

This is why we moved our standard from HP to IBM Power. HP still only has a 16 socket box which has half the cores of Power7 and each core is less than half the performance of Power7 cores.

TPC-H is a direct comparison which shows Power is 2.35X better performance per core.

So did we want to but an extremely expensive systems from HP with 16 sockets, or would we rather buy a great price performance Power7 four socket box and get more performance and a fraction of the 3rd party software costs? We did end up putting the p770 as the preferred database consolidation server to increase utilization of our expensive middleware.

bootnote.....HP does not use the 5 QPI links for the SX3000 chipset. The architecture drawings show that one QPI goes to I/O, and one QPI each go to the dual SX3000 chips. Two QPI links are dead. So you have a disjointed architecture. The BL890 needs either 8 QPI links per chip or the SX3000 and the SD2 only uses 3 out of 5.

0
0
Silver badge
FAIL

Meanwhile, over at IBM FUD Central.....

".....IBM got its big, bad Power 795 AIX box out the door last fall....." So, you admit that IBM didn't release their top-end offering at the same time as their initial Pee7 servers, that IBM staggered their launches. So, when IBM do it it's no cause for comment, but when hp stagger their launches it's headline news? Please try for impartiality, if just for appearence's sake.

"....was due in the first half of 2011...." What, it's July already? Oh, no it's not. Try again, TPM.

"....You cannot slap OpenVMS 8.4, the most current release, on any Integrity machine, and in fact, it is only supported on the BL8X0 i2 blades....." Now that bit did confuse me. I personally don't work with OpenDinosaur, but other parts of our group do, and I know one team that currently have some proof-of-concept rx2800s running OVMS 8.4. A quick check of the hp Quickspecs website seems to say rx2800i2 is hp-ux only, but I have access to a copy of the order placed by my colleagues with the OVMS 8.4 High Availability bundle factory-installed and fully-supported. So I'm guessing the Quickspecs site is out-of-date, just like your information. I'm guessing the IBM FUD material is too?

0
3
Go

Let's compare

February 2010 IBM announces Power7 at events worldwide the largest in NYC while Intel announces with HP in the audience at a trade show in San Fran.

IBM started selling Power7 systems in March 2010. The p770 and p750 I believe 8 socket and 4 socket.

IBM started selling the Power blades in April. HP started selling Tukwila systems at the end of April. The 2-8 socket blades shipping immediately and superdome2 was announced.

Sept. IBM starts selling the 32 socket p795 which is a board upgrade from the p6 595. HP starts selling the 16 socket Superdome2 which is a forklift upgrade.

Keep in mind that the 16 socket SD2 only has half the cores/socket and less performance per core. So in reality HP started shipping the equivalent of an IBM 3.4 socket box.

16 sockets/only half the cores/2.35 the ibm advantage by core as shown by tpc-h.

So HP might bring out the 32 socket box soon which according to HP documents is just a couple of cables but its is not comparison to what Power7 is doing.

It's no wonder why Oracle is dumping Itanium as they performance test on both platforms. As far as increasing the per core license factor well that was just cruel. Glad we started moving off HP so it wont be cruel to us.

cheers

0
0
Silver badge
FAIL

RE: Let's compare

Nope, sorry, I don't want to compare looniness!

"February 2010 IBM announces Power7..." You forgot the bit where IBM were forced to announce Pee7 early, when they didn't have any servers ready, because they knew the new Xeons and Tukzilla were coming. They pulled the announcement forward over a month, whereas hp had servers ready to go the same April as they announced them.

"....Sept. IBM starts selling the 32 socket p795 which is a board upgrade from the p6 595..." You forgot the bit where you have to recompile your p595 app on new AIX v7.1 if you wanted to get the extra oomph from those Pee7 chips, otherwise you actually got better performance per core staying on Pee6. That doesn't compare to Integrity, where I can take an app off any Integrity box running hp-ux 11i v3 and run it already optimised on the new blades or the new Superdome2, no need for any recompilation.

".....a forklift upgrade....." LOL! An IBM troll trying to whine about a fork-lift upgrae, when just about every Power update since day one has meant not only a fork-lift upgrade but also a new version fo AIX! Please do think and compare that to hp's record of ten years' development in the same Superdome chassis. It was inevitable that hp would have to change the frame some time, just as it is inevitable that the Pee8 will mean a forklift upgrade for p795 owners. Loonies that live in glass houses......

"....It's no wonder why Oracle is dumping Itanium...." The wonder is why they did not dump SPARC, seeing as 3 times as much Oracle licences go out on hp Integrity than SPARC, and more than go out on Power. In fact, more IBM software gets sold with Integrity than with IBM's own Power, a fact that IBM Software salesgrunts are happy to admit. What you don't want to wonder about is what will happen if Oracle decide to stop developpment on Power - DB2 on Power is too weak a product to replace Oracle.

"....Glad we started moving off HP...." I'm quite comfortable in betting you never touched Integrity, hp-ux or anything without an IBM logo on it.

0
1
KBP
Happy

OpenVMS 8.4 supports a wide variety of Integrity hardware, not just BL8x0 i2 blades

Re: "You cannot slap OpenVMS 8.4, the most current release, on any Integrity machine, and in fact, it is only supported on the BL8X0 i2 blades" <

Actually, OpenVMS 8.4 supports a wide variety of Integrity machines -- see the complete list in the OpenVMS 8.4 Software Product Description at http://h71000.www7.hp.com/doc/spdalphaintegrity84.pdf

2
0
Silver badge
Happy

RE: OpenVMS 8.4 supports a wide variety of Integrity hardware, not just BL8x0 i2 blades

Confirmed, I sent the list to one of our resellers and they checked with hp. Looks like hp will have to update their Quickspecs and IBM will have to send TPM some new FUD.

0
0
WTF?

neat.. so what are you going to run on it?

Oracle 11g.. yes indeed, for a while unless Larry decides to double the license cost again. (Which conveniently doubles the already excessive maintenance, too.) Nevermind that you won't get a version beyond 11g.

MS SQL IA64? Sure, if 2005 suits your fancy. Neat features you want in 2008 R2? Oh, drat.

Sybase... bwahaha. Sorry, I couldn't actually write a proper sentence about that.

DB2 it is, then. And why are you running DB2 on anything besides AIX??

1
0
Alert

db2?

Doesn't run particularly well on AIX either!

0
0
Bronze badge
Alert

OracleRdb ?

I suspect quite a few of them will run OracleRdb at this stage, just as they always have done - it is by far the best database available on OpenVMS.

Sadly, OracleRdb has been shot by Oracle as well - at this stage - so those folks at some REALLY big clients have several years work cut out for them converting their software to some "other database" on some "other OS". Not a pleasant thought I am sure, having tried it personally.

HP has wanted OpenVMS (and Tandem) to go away as quickly as possible ever since they acquired them, now their strategy is enabled by Oracle's poke in the eye. What luck for the HP execs, soon they can move on to more stock options and more interesting things that don't work. They haven't been able to kill OpenVMS themselves, so Larry lends a helping hand by decommissioning the 2 databases that a very, very large percentage of OpenVMS customers rely upon. Good to go Larry! No long term database upgrade path is about as clear a message as is possible for OpenVMS customers.

The "Alphacide" and the Itanium or nothing plan was always the death of OracleRdb (and NonStopSQL) - now I think the clock is counting down to the last decommissioning of what truly is the best piece of database engineering out there. Sadly OracleRdb requires OpenVMS, which despite its superiority in so many respects, has zero visibility in the IT journalists world, and even less in the potential customer world - thanks HP, we appreciate it sooooo much, not.

I earn my living in the SQLServer world now. The worst database I have used since Oracle V3.x (IIRC), and that, if you are old enough to remember, was particularly nasty

Sailing off into early retirement to let he unwilling, led by incompetent solve the problems created by the "vision" of Carley, and ego of Larry.

2
0

IBM P795, 256 cores is chicken sh*t.

Linux has those 1024 core SGI Altix servers. IBM P795, 256 cores pales in comparison. Thus, IBM P795 scales utterly bad. True/false?

.

Also, my suspicion is that Intel worked with HP to learn how to do high RAS cpus as in Itanium. But now Intel has learned RAS and incorporated RAS into Intel Xeon, so Intel is not interested in Itanium anymore. Intel used HP. But the sad thing is that Itanium is a superior chipset, much cleaner and not bloated. Intel is wrong when they believe that RAS gives Enterprise reliability. Itanium will always be more reliable than Intel Xeon, no matter how much RAS the Xeon cpu gets.

0
1
Headmaster

Well...

It's not like the POWER 795 a big SMP machine made for commercial business type of workloads competes with the SGI altix which is made for HPC. They are targeted at two very different markeds.

But when that is said then sure a 128 socket, 1280 core Altix will deliver a lot of compute power, but

if you look at the specint and specfp rates for the POWER 795 versus the Altix

http://www.ideasinternational.com/Benchmark-Top-Ten/SPECint2006rate

and

http://www.ideasinternational.com/Benchmark-Top-Ten/SPECfp2006rate

Then the POWER 795 is not that much behind, for example on specfprate then five times the cores only/chips only gives you twice the throughput.

But why don't you compare the POWER 795 to the M9000 ? The throughput difference is what.. a factor of 7 per chip on specintrate ? and what a factor of 8 on specfprate ? *cough*

// Jesper

0
0

@Jesper Frimann

I am just trying to say to all the Linux fanboys that the SGI Altix machine is basically a big cluster and it is not comparable to a big SMP machine such as P795 or the M9000. But the Linux fanboys dont believe me - they say that Linux scales better than AIX and Solaris together. Which is wrong. Scaling a cluster is easy.

Regarding the M9000 vs P795 - everyone agrees that the POWER7 is superior to the CPU in M9000. That is not the question. The question is if the Linux fanboys can claim that Linux SGI Altix server with 256(?) cpus is a SMP machine. It is not.

.

Regarding the Itanium. Here are some interesting links. Intel is relocating the Itanium engineers to Xeon.

http://www.xbitlabs.com/news/cpu/display/20110503125538_Intel_HP_Unsure_About_the_Future_of_Itanium_Kittson_May_Be_Last_Survivor.html

http://www.xbitlabs.com/news/cpu/display/20110417112218_Intel_Relocates_Itanium_Engineers_to_Xeon_Projects_Sources.html

Intel claims that Xeon has better RAS than Itanium:

http://www.xbitlabs.com/news/cpu/display/20110413162720_Intel_Xeon_Now_Has_Equal_or_Better_Reliability_and_Performance_than_Itanium.html

I really doubt that, as x86 is too buggy. I really believe Intel wanted to learn RAS from HP, and then kill off Itanium. Which is quite dumb, because x86 is too buggy and will never has as good RAS as Itanium.

0
2
Anonymous Coward

"Itanium will always be more reliable than Intel Xeon"

Justify, with real examples with real detail (please don't mention cosmic rays), or expect to be rightly ignored. Preferably don't compare with Xeon, preferably compare with AMD64, but Xeon's largely caught up now.

I've heard this RAS claim since before IA64 failed to conquer the "industry standard 64bit computing" sector. I've never seen any meaningful facts to back it up. I'd welcome some evidence, if there is any. HP/Intel and their customers might too.

Also bear in mind that much of the RAS stuff in today's systems comes from *software* as well as hardware. Put something better than Windows on an AMD64 and you could do very nicely.

0
0
This topic is closed for new posts.