back to article All the sauce on Big Blue's hot chip: More on Power7+

The Hot Chips 24 conference hosted by Stanford University is next week, and IBM, Oracle, Advanced Micro Devices, Fujitsu, and Intel are expected to talk tech relating to just-announced or impending processors. But Big Blue seems unable to contain its enthusiasm for the Power7+ chip that it will talk about alongside its next- …

COMMENTS

This topic is closed for new posts.

AIX for databases...

I want to see the benchmarks from SAP & other vendors. With our main application, much (not all) of the IBM market-texture resulted in benchmark improvements over Itanium, SPARC and even x64.

The key here is whether IBM can continue to justify (for key, not all) applications, it's premium price.

3
0
Anonymous Coward

Re: AIX for databases...

Rough rule of thumb. Power 7 will outperform x86 7500 by around 50% on standard OLTP. X86 costs less than Power and Oracle charges a higher core factor for Power, but, if you are using VMware, Oracle will not allow a hard partition for x86. You can hard partition Power. Itanium and T Sparc are not close to either. It is funny that Oracle is claiming that Sparc is the highest performing chip for everything but "integer math", yet the Sparc's core multiplier is 1/4th that of Power and 1/2 that of x86. Oracle would be losing billions in licensing on their DB if that were true and people moved to Sparc. I guess luckily for them, it isn't true.

0
0

Re: AIX for databases...

"...It is funny that Oracle is claiming that Sparc is the highest performing chip for everything but "integer math"..."

Do you have a link where Oracle is claiming that SPARC is fastest on everything, except integer math? I have never seen such a claim, and I follow SPARC and Oracle closely.

On the other hand, IBM claimed that having many slower cores as Niagara was a bad thing and the future lied in having 1-2 very strong cores, because "databases are best run on a strong cores rather than many weak cores". So where are those POWER cpus with 1-2 cores running at 7-8GHz or even beyond that? Why does the newer POWER cpus have many lower clocked cores? POWER6 had 2 cores running at 5GHz. POWER7 should have 2 cores running at 6-7 GHz, and POWER7+ should have 2 cores running at 7-8GHz. Have IBM changed their mind? Are many lower clocked cores not a dumb thing to do, anymore?

0
3
Anonymous Coward

Re: AIX for databases...

Larry Ellison said it in his keynote speech at OpenWorld this year.

"We're faster for Java, Mister Blue," Ellison said at Oracle World, touting Sparc chips. "You're faster for integer arithmetic. If you think companies do a lot of arithmetic, cool. We think they access a lot of data and run a lot of Java." "We're better than IBM in Java, and we're going to beat them in integer arithmetic, and then there will be nothing left,"

Note that "faster for Java" is based on Oracle's totally fraudulent T4 benchmarks that used many more cores, many times more disk, many times more memory, etc than the Power benchmark they were comparing it to. Oracle did everything possible to make that benchmark about anything besides processors. Oracle has since been fined by the national advertising council and forced to take their Sparc ads down because they were lies.

Sparc is toast. Oracle is well aware that it is, to borrow a phrase, "dead, dead, dead" and "a pig with lipstick... at best." Oracle is moving on to x86.

0
0
Anonymous Coward

Re: AIX for databases...

Link: http://wraltechwire.com/business/tech_wire/news/blogpost/10232172/ (you can watch the video if you find Larry's T4 speech on Oracle's website)

0
0
Headmaster

Re: AIX for databases...

The difference between being that POWER managed to keep increasing single core throughput.

POWER5+ at 2.2GHz 16 cores 8 chips 217 in specint_rate2006.

POWER6 at 4.7 GHz 16 cores 8 chips 484 in specint_rate2006

POWER6+ at 5.0 GHz 16 cores 8 chips 542 in specint_rate2006

POWER7 at 3.86 GHz 16 cores 2 chips 652 in specint_rate2006

So keb, lower Ghz doesn't always means slower cores.

// Jesper

2
0
Silver badge

RNG use

RNGs are also an important part of many applications, particularly in financial services or physics simulations that require randomness.

On-chip RNGs - particularly if they're true RNGs, using some electrical noise source - are much more likely to find application in crypto, particularly in generating randomness for SSL/TLS (most of that HTTPS) work. For financial and physical simulations, you generally want a PRNG that exposes its entire state, so that you have reproducible results (and can restart from any point). Unless, of course, you're planning to publish in the JIR.

1
0
This topic is closed for new posts.

Forums