EU files anti trust against Google to reduce budget deficit

From the Old Lady-

http://www.nytimes.com/2010/12/01/technology/01google.html?_r=1&hpw

Google’s dominance on the Internet has been a sore point in Europe, where it controls more than 80 percent of the online search market, compared with about 66 percent in the United States, according to comScore, a research firm.

and

If Google is found in violation of European competition law, the commission has the power to fine it up to 10 percent of its annual revenue, which totaled more than $23 billion last year.

Before settling last year, Microsoft had paid fines of about $2.4 billion over the past decade in a long-running antitrust case in Brussels that focused on the Windows operating system.

In another case, the commission fined Intel about $1.45 billion for abusing its dominance in the computer chip market.


——————————————————————————————

Maybe Google should ask the European Union to buy a groupon for anti trust cases.

Related

11 Ways to Beat Google

 

 

Amazon goes HPC and GPU: Dirk E to revise his R HPC book

Looking south above Interstate 80, the Eastsho...
Image via Wikipedia

Amazon just did a cluster Christmas present for us tech geek lizards- before Google could out doogle them with end of the Betas (cough- its on NDA)

Clusters used by Academic Departments now have a great chance to reduce cost without downsizing- but only if the CIO gets the email.

While Professor Goodnight of SAS / North Carolina University is still playing time sharing versus mind sharing games with analytical birdies – his 70 mill server farm set in Feb last is about to get ready

( I heard they got public subsidies for environment- but thats historic for SAS– taking public things private -right Prof as SAS itself began as a publicly funded project. and that was in the 1960s and they didnt even have no lobbyists as well. )

In realted R news, Dirk E has been thinking of a R HPC book without paying attention to Amazon but would now have to include Amazon

(he has been thinking of writing that book for 5 years, but hey he’s got a day job, consulting gigs with revo, photo ops at Google, a blog, packages to maintain without binaries, Dirk E we await thy book with bated holes.

Whos Dirk E – well http://dirk.eddelbuettel.com/ is like the Terminator of R project (in terms of unpronounceable surnames)

Back to the cause du jeure-

 

From http://aws.amazon.com/ec2/hpc-applications/ but minus corporate buzz words.

 

Unique to Cluster Compute and Cluster GPU instances is the ability to group them into clusters of instances for use with HPC

applications. This is particularly valuable for those applications that rely on protocols like Message Passing Interface (MPI) for tightly coupled inter-node communication.

Cluster Compute and Cluster GPU instances function just like other Amazon EC2 instances but also offer the following features for optimal performance with HPC applications:

  • When run as a cluster of instances, they provide low latency, full bisection 10 Gbps bandwidth between instances. Cluster sizes up through and above 128 instances are supported.
  • Cluster Compute and Cluster GPU instances include the specific processor architecture in their definition to allow developers to tune their applications by compiling applications for that specific processor architecture in order to achieve optimal performance.

The Cluster Compute instance family currently contains a single instance type, the Cluster Compute Quadruple Extra Large with the following specifications:

23 GB of memory
33.5 EC2 Compute Units (2 x Intel Xeon X5570, quad-core “Nehalem” architecture)
1690 GB of instance storage
64-bit platform
I/O Performance: Very High (10 Gigabit Ethernet)
API name: cc1.4xlarge

The Cluster GPU instance family currently contains a single instance type, the Cluster GPU Quadruple Extra Large with the following specifications:

22 GB of memory
33.5 EC2 Compute Units (2 x Intel Xeon X5570, quad-core “Nehalem” architecture)
2 x NVIDIA Tesla “Fermi” M2050 GPUs
1690 GB of instance storage
64-bit platform
I/O Performance: Very High (10 Gigabit Ethernet)
API name: cg1.4xlarge

.

Sign Up for Amazon EC2

Enterprise Linux rises rapidly:New Report

Tux, as originally drawn by Larry Ewing
Image via Wikipedia

A new report from Linux Foundation found significant growth trends for enterprise usage of Linux- which should be welcome to software companies that have enabled Linux versions of software, service providers that provide Linux based consulting (note -lesser competition, lower overheads) and to application creators.

From –

http://www.linuxfoundation.org/news-media/announcements/2010/10/new-linux-foundation-user-survey-shows-enterprise-linux-achieve-sig

Key Findings from the Report
• 79.4 percent of companies are adding more Linux relative to other operating systems in the next five years.

• More people are reporting that their Linux deployments are migrations from Windows than any other platform, including Unix migrations. 66 percent of users surveyed say that their Linux deployments are brand new (“Greenfield”) deployments.

• Among the early adopters who are operating in cloud environments, 70.3 percent use Linux as their primary platform, while only 18.3 percent use Windows.

• 60.2 percent of respondents say they will use Linux for more mission-critical workloads over the next 12 months.

• 86.5 percent of respondents report that Linux is improving and 58.4 percent say their CIOs see Linux as more strategic to the organization as compared to three years ago.

• Drivers for Linux adoption extend beyond cost: technical superiority is the primary driver, followed by cost and then security.

• The growth in Linux, as demonstrated by this report, is leading companies to increasingly seek Linux IT professionals, with 38.3 percent of respondents citing a lack of Linux talent as one of their main concerns related to the platform.

• Users participate in Linux development in three primary ways: testing and submitting bugs (37.5 percent), working with vendors (30.7 percent) and participating in The Linux Foundation activities (26.0 percent).

and from the report itself-

download here-

http://www.linuxfoundation.org/lp/page/download-the-free-linux-adoption-trends-report

Revolution R for Linux

Screenshot of the Redhat Enterprise Linux Desktop
Image via Wikipedia

New software just released from the guys in California (@RevolutionR) so if you are a Linux user and have academic credentials you can download it for free  (@Cmastication doesnt), you can test it to see what the big fuss is all about (also see http://www.revolutionanalytics.com/why-revolution-r/benchmarks.php) –

Revolution Analytics has just released Revolution R Enterprise 4.0.1 for Red Hat Enterprise Linux, a significant step forward in enterprise data analytics. Revolution R Enterprise 4.0.1 is built on R 2.11.1, the latest release of the open-source environment for data analysis and graphics. Also available is the initial release of our deployment server solution, RevoDeployR 1.0, designed to help you deliver R analytics via the Web. And coming soon to Linux: RevoScaleR, a new package for fast and efficient multi-core processing of large data sets.

As a registered user of the Academic version of Revolution R Enterprise for Linux, you can take advantage of these improvements by downloading and installing Revolution R Enterprise 4.0.1 today. You can install Revolution R Enterprise 4.0.1 side-by-side with your existing Revolution R Enterprise installations; there is no need to uninstall previous versions.

Download Information

The following information is all you will need to download and install the Academic Edition.

Supported Platforms:

Revolution R Enterprise Academic edition and RevoDeployR are supported on Red Hat® Enterprise Linux® 5.4 or greater (64-bit processors).

Approximately 300MB free disk space is required for a full install of Revolution R Enterprise. We recommend at least 1GB of RAM to use Revolution R Enterprise.

For the full list of system requirements for RevoDeployR, refer to the RevoDeployR™ Installation Guide for Red Hat® Enterprise Linux®.

Download Links:

You will first need to download the Revolution R Enterprise installer.

Installation Instructions for Revolution R Enterprise Academic Edition

After downloading the installer, do the following to install the software:

  • Log in as root if you have not already.
  • Change directory to the directory containing the downloaded installer.
  • Unpack the installer using the following command:
    tar -xzf Revo-Ent-4.0.1-RHEL5-desktop.tar.gz
  • Change directory to the RevolutionR_4.0.1 directory created.
  • Run the installer by typing ./install.py and following the on-screen prompts.

Getting Started with the Revolution R Enterprise

After you have installed the software, launch Revolution R Enterprise by typing Revo64 at the shell prompt.

Documentation is available in the form of PDF documents installed as part of the Revolution R Enterprise distribution. Type Revo.home(“doc”) at the R prompt to locate the directory containing the manuals Getting Started with Revolution R (RevoMan.pdf) and the ParallelR User’s Guide(parRman.pdf).

Installation Instructions for RevoDeployR (and RServe)

After downloading the RevoDeployR distribution, use the following steps to install the software:

Note: These instructions are for an automatic install.  For more details or for manual install instructions, refer to RevoDeployR_Installation_Instructions_for_RedHat.pdf.

  1. Log into the operating system as root.
    su –
  2. Change directory to the directory containing the downloaded distribution for RevoDeployR and RServe.
  3. Unzip the contents of the RevoDeployR tar file. At prompt, type:
    tar -xzf deployrRedHat.tar.gz
  4. Change directories. At the prompt, type:
    cd installFiles
  5. Launch the automated installation script and follow the on-screen prompts. At the prompt, type:
    ./installRedHat.sh
    Note: Red Hat installs MySQL without a password.

Getting Started with RevoDeployR

After installing RevoDeployR, you will be directed to the RevoDeployR landing page. The landing page has links to documentation, the RevoDeployR management console, the API Explorer development tool, and sample code.

Support

For help installing this Academic Edition, please email support@revolutionanalytics.com

Also interestingly some benchmarks on Revolution R vs R.

http://www.revolutionanalytics.com/why-revolution-r/benchmarks.php

R-25 Benchmarks

The simple R-benchmark-25.R test script is a quick-running survey of general R performance. The Community-developed test consists of three sets of small benchmarks, referred to in the script as Matrix Calculation, Matrix Functions, and Program Control.

R-25 Matrix Calculation R-25 Matrix Functions R-Matrix Program Control
R-25 Benchmarks Base R 2.9.2 Revolution R (1-core) Revolution R (4-core) Speedup (4 core)
Matrix Calculation 34 sec 6.6 sec 4.4 sec 7.7x
Matrix Functions 20 sec 4.4 sec 2.1 sec 9.5x
Program Control 4.7 sec 4 sec 4.2 sec Not Appreciable

Speedup = Slower time / Faster Time – 1   Test descriptions available at http://r.research.att.com/benchmarks

Additional Benchmarks

Revolution Analytics has created its own tests to simulate common real-world computations.  Their descriptions are explained below.

Matrix Multiply Cholesky Factorization
Singular Value Decomposition Principal Component Analysis Linear Discriminant Analysis
Linear Algebra Computation Base R 2.9.2 Revolution R (1-core) Revolution R (4-core) Speedup (4 core)
Matrix Multiply 243 sec 22 sec 5.9 sec 41x
Cholesky Factorization 23 sec 3.8 sec 1.1 sec 21x
Singular Value Decomposition 62 sec 13 sec 4.9 sec 12.6x
Principal Components Analysis 237 sec 41 sec 15.6 sec 15.2x
Linear Discriminant Analysis 142 sec 49 sec 32.0 sec 4.4x

Speedup = Slower time / Faster Time – 1

Matrix Multiply

This routine creates a random uniform 10,000 x 5,000 matrix A, and then times the computation of the matrix product transpose(A) * A.

set.seed (1)
m <- 10000
n <-  5000
A <- matrix (runif (m*n),m,n)
system.time (B <- crossprod(A))

The system will respond with a message in this format:

User   system elapsed
37.22    0.40   9.68

The “elapsed” times indicate total wall-clock time to run the timed code.

The table above reflects the elapsed time for this and the other benchmark tests. The test system was an INTEL® Xeon® 8-core CPU (model X55600) at 2.5 GHz with 18 GB system RAM running Windows Server 2008 operating system. For the Revolution R benchmarks, the computations were limited to 1 core and 4 cores by calling setMKLthreads(1) and setMKLthreads(4) respectively. Note that Revolution R performs very well even in single-threaded tests: this is a result of the optimized algorithms in the Intel MKL library linked to Revolution R. The slight greater than linear speedup may be due to the greater total cache available to all CPU cores, or simply better OS CPU scheduling–no attempt was made to pin execution threads to physical cores. Consult Revolution R’s documentation to learn how to run benchmarks that use less cores than your hardware offers.

Cholesky Factorization

The Cholesky matrix factorization may be used to compute the solution of linear systems of equations with a symmetric positive definite coefficient matrix, to compute correlated sets of pseudo-random numbers, and other tasks. We re-use the matrix B computed in the example above:

system.time (C <- chol(B))

Singular Value Decomposition with Applications

The Singular Value Decomposition (SVD) is a numerically-stable and very useful matrix decompisition. The SVD is often used to compute Principal Components and Linear Discriminant Analysis.

# Singular Value Deomposition
m <- 10000
n <- 2000
A <- matrix (runif (m*n),m,n)
system.time (S <- svd (A,nu=0,nv=0))

# Principal Components Analysis
m <- 10000
n <- 2000
A <- matrix (runif (m*n),m,n)
system.time (P <- prcomp(A))

# Linear Discriminant Analysis
require (‘MASS’)
g <- 5
k <- round (m/2)
A <- data.frame (A, fac=sample (LETTERS[1:g],m,replace=TRUE))
train <- sample(1:m, k)
system.time (L <- lda(fac ~., data=A, prior=rep(1,g)/g, subset=train))

Sector/ Sphere – Faster than Hadoop/Mapreduce at Terasort

Here is a preview of a relatively young software Sector and Sphere- which are claimed to be better than Hadoop /MapReduce at TeraSort Benchmark among others.

http://sector.sourceforge.net/tech.html

System Overview

The Sector/Sphere stack consists of the Sector distributed file system and the Sphere parallel data processing framework. The objective is to support highly effective and efficient large data storage and processing over commodity computer clusters.

Sector/Sphere Architecture

Sector consists of 4 parts, as shown in the above diagram. The Security server maintains the system security configurations such as user accounts, data IO permissions, and IP access control lists. The master servers maintain file system metadata, schedule jobs, and respond users’ requests. Sector supports multiple active masters that can join and leave at run time and they all actively respond users’ requests. The slave nodes are racks of computers that store and process data. The slaves nodes can be located within a single data center to across multiple data centers with high speed network connections. Finally, the client includes tools and programming APIs to access and process Sector data.

Sphere: Parallel Data Processing Framework

Sphere allows developers to write parallel data processing applications with a very simple set of API. It applies user-defined functions (UDF) on all input data segments in parallel. In a Sphere application, both inputs and outputs are Sector files. Multiple Sphere processing can be combined to support more complicated applications, with inputs/outputs exchanged/shared via the Sector file system.

Data segments are processed at their storage locations whenever possible (data locality). Failed data segments may be restarted on other nodes to achieve fault tolerance.

The Sphere framework can be compared to MapReduce as they both enforce data locality and provide simplified programming interfaces. In fact, Sphere can simulate any MapReduce operations, but Sphere is more efficient and flexible. Sphere can provide better data locality for applications that process files or multiple files as minimum input units and for applications that involve with iterative/combinative processing, which requires coordination of multiple UDFs to obtain the final result.

A Sphere application includes two parts: the client program that organizes inputs (including certain parameters), outputs, and UDFs; and the UDFs that process data segments. Data segmentation, load balancing, and fault tolerance are transparent to developers.

Space: Column-based Distbuted Data Table

Space stores data tables in Sector and uses Sphere for parallel query processing. Space is similar to BigTable. Table is stored by columns and is segmented on to multiple slave nodes. Tables are independent and no relationship between tables are supported. A reduced set of SQL operations is supported, including but not limited to table creation and modification, key-value update and lookup, and select operations based on UDF.

Supported by the Sector data placement mechanism and the Sphere parallel processing framework, Space can support efficient key-value lookup and certain SQL queries on very large data tables.

Space is currently still in development.

and just when you thought Hadoop was the only way to be on the cloud.

http://sector.sourceforge.net/benchmark.html

The Terasort Benchmark

The table below lists the performance (total processing time in seconds) of the Terasort benchmark of both Sphere and Hadoop. (Terasort benchmark: suppose there are N nodes in the system, the benchmark generates a 10GB file on each node and sorts the total N*10GB data. Data generation time is excluded.) Note that it is normal to see a longer processing time for more nodes because the total amount of data also increases proportionally.

The performance value listed in this page was achieved using the Open Cloud Testbed. Currently the testbed consists of 4 racks. Each rack has 32 nodes, including 1 NFS server, 1 head node, and 30 compute/slave nodes. The head node is a Dell 1950, dual dual-core Xeon 3.0GHz, 16GB RAM. The compute nodes are Dell 1435s, single dual core AMD Opteron 2.0GHz, 4GB RAM, and 1TB single disk. The 4 racks are located in JHU (Baltimore), StarLight (Chicago), UIC (Chicago), and Calit2(San Diego). The inter-rack bandwidth is 10GE, supported by CiscoWave deployed over National Lambda Rail.

Sphere
Hadoop (3 replicas)
Hadoop (1 replica)
UIC
1265 2889 2252
UIC + StarLight
1361 2896 2617
UIC + StarLight + Calit2
1430 4341 3069
UIC + StarLight + Calit2 + JHU
1526 6675 3702

The benchmark uses the testfs/testdc examples of Sphere and randomwriter/sort examples of Hadoop. Hadoop parameters were tuned to reach good results.

Updated on Sep. 22, 2009: We have benchmarked the most recent versions of Sector/Sphere (1.24a) and Hadoop (0.20.1) on a new set of servers. Each server node costs $2,200 and consits of a single Intel Xeon E5410 2.4GHz CPU, 16GB RAM, 4*1TB RAID0 disk, and 1Gb/s NIC. The 120 nodes are hosted on 4 racks within the same data center and the inter-rack bandwidth is 20Gb/s.

The table below lists the performance of sorting 1TB data using Sector/Sphere version 1.24a and Hadoop 0.20.1. Related Hadoop parameters have been tuned for better performance (e.g., big block size), while Sector/Sphere does not require tuning. In addition, to achieve the highest performance, replication is disabled in both systems (note that replication does not afftect the performance of Sphere but will significantly decrease the performance of Hadoop).

Number of Racks
Sphere
Hadoop
1
28m 25s 85m 49s
2
15m 20s 37m 0s
3
10m 19s 25m 14s
4
7m 56s 17m 45s

Windows Azure vs Amazon EC2 (and Google Storage)

Here is a comparison of Windows Azure instances vs Amazon compute instances

Compute Instance Sizes:

Developers have the ability to choose the size of VMs to run their application based on the applications resource requirements. Windows Azure compute instances come in four unique sizes to enable complex applications and workloads.

Compute Instance Size CPU Memory Instance Storage I/O Performance
Small 1.6 GHz 1.75 GB 225 GB Moderate
Medium 2 x 1.6 GHz 3.5 GB 490 GB High
Large 4 x 1.6 GHz 7 GB 1,000 GB High
Extra large 8 x 1.6 GHz 14 GB 2,040 GB High

Standard Rates:

Windows Azure

  • Compute
    • Small instance (default): $0.12 per hour
    • Medium instance: $0.24 per hour
    • Large instance: $0.48 per hour
    • Extra large instance: $0.96 per hour
  • Storage
    • $0.15 per GB stored per month
    • $0.01 per 10,000 storage transactions
  • Content Delivery Network (CDN)
    • $0.15 per GB for data transfers from European and North American locations*
    • $0.20 per GB for data transfers from other locations*
    • $0.01 per 10,000 transactions*

Source –

http://www.microsoft.com/windowsazure/offers/popup/popup.aspx?lang=en&locale=en-US&offer=MS-AZR-0001P

and

http://www.microsoft.com/windowsazure/windowsazure/

Amazon EC2 has more options though——————————-

http://aws.amazon.com/ec2/pricing/

Standard On-Demand Instances Linux/UNIX Usage Windows Usage
Small (Default) $0.085 per hour $0.12 per hour
Large $0.34 per hour $0.48 per hour
Extra Large $0.68 per hour $0.96 per hour
Micro On-Demand Instances Linux/UNIX Usage Windows Usage
Micro $0.02 per hour $0.03 per hour
High-Memory On-Demand Instances
Extra Large $0.50 per hour $0.62 per hour
Double Extra Large $1.00 per hour $1.24 per hour
Quadruple Extra Large $2.00 per hour $2.48 per hour
High-CPU On-Demand Instances
Medium $0.17 per hour $0.29 per hour
Extra Large $0.68 per hour $1.16 per hour
Cluster Compute Instances
Quadruple Extra Large $1.60 per hour N/A*
* Windows is not currently available for Cluster Compute Instances.

http://aws.amazon.com/ec2/instance-types/

Standard Instances

Instances of this family are well suited for most applications.

Small Instance – default*

1.7 GB memory
1 EC2 Compute Unit (1 virtual core with 1 EC2 Compute Unit)
160 GB instance storage (150 GB plus 10 GB root partition)
32-bit platform
I/O Performance: Moderate
API name: m1.small

Large Instance

7.5 GB memory
4 EC2 Compute Units (2 virtual cores with 2 EC2 Compute Units each)
850 GB instance storage (2×420 GB plus 10 GB root partition)
64-bit platform
I/O Performance: High
API name: m1.large

Extra Large Instance

15 GB memory
8 EC2 Compute Units (4 virtual cores with 2 EC2 Compute Units each)
1,690 GB instance storage (4×420 GB plus 10 GB root partition)
64-bit platform
I/O Performance: High
API name: m1.xlarge

Micro Instances

Instances of this family provide a small amount of consistent CPU resources and allow you to burst CPUcapacity when additional cycles are available. They are well suited for lower throughput applications and web sites that consume significant compute cycles periodically.

Micro Instance

613 MB memory
Up to 2 EC2 Compute Units (for short periodic bursts)
EBS storage only
32-bit or 64-bit platform
I/O Performance: Low
API name: t1.micro

High-Memory Instances

Instances of this family offer large memory sizes for high throughput applications, including database and memory caching applications.

High-Memory Extra Large Instance

17.1 GB of memory
6.5 EC2 Compute Units (2 virtual cores with 3.25 EC2 Compute Units each)
420 GB of instance storage
64-bit platform
I/O Performance: Moderate
API name: m2.xlarge

High-Memory Double Extra Large Instance

34.2 GB of memory
13 EC2 Compute Units (4 virtual cores with 3.25 EC2 Compute Units each)
850 GB of instance storage
64-bit platform
I/O Performance: High
API name: m2.2xlarge

High-Memory Quadruple Extra Large Instance

68.4 GB of memory
26 EC2 Compute Units (8 virtual cores with 3.25 EC2 Compute Units each)
1690 GB of instance storage
64-bit platform
I/O Performance: High
API name: m2.4xlarge

High-CPU Instances

Instances of this family have proportionally more CPU resources than memory (RAM) and are well suited for compute-intensive applications.

High-CPU Medium Instance

1.7 GB of memory
5 EC2 Compute Units (2 virtual cores with 2.5 EC2 Compute Units each)
350 GB of instance storage
32-bit platform
I/O Performance: Moderate
API name: c1.medium

High-CPU Extra Large Instance

7 GB of memory
20 EC2 Compute Units (8 virtual cores with 2.5 EC2 Compute Units each)
1690 GB of instance storage
64-bit platform
I/O Performance: High
API name: c1.xlarge

Cluster Compute Instances

Instances of this family provide proportionally high CPU resources with increased network performance and are well suited for High Performance Compute (HPC) applications and other demanding network-bound applications. Learn more about use of this instance type for HPC applications.

Cluster Compute Quadruple Extra Large Instance

23 GB of memory
33.5 EC2 Compute Units (2 x Intel Xeon X5570, quad-core “Nehalem” architecture)
1690 GB of instance storage
64-bit platform
I/O Performance: Very High (10 Gigabit Ethernet)
API name: cc1.4xlarge

Also http://www.microsoft.com/en-us/sqlazure/default.aspx

offers SQL Databases as a service with a free trial offer

If you are into .Net /SQL big time or too dependent on MS, Azure is a nice option to EC2 http://www.microsoft.com/windowsazure/offers/popup/popup.aspx?lang=en&locale=en-US&offer=COMPARE_PUBLIC

Updated- I just got approved for Google Storage so am adding their info- though they are in Preview (and its free right now) 🙂

https://code.google.com/apis/storage/docs/overview.html

Functionality

Google Storage for Developers offers a rich set of features and capabilities:

Basic Operations

  • Store and access data from anywhere on the Internet.
  • Range-gets for large objects.
  • Manage metadata.

Security and Sharing

  • User authentication using secret keys or Google account.
  • Authenticated downloads from a web browser for Google account holders.
  • Secure access using SSL.
  • Easy, powerful sharing and collaboration via ACLs for individuals and groups.

Performance and scalability

  • Up to 100 gigabytes per object and 1,000 buckets per account during the preview.
  • Strong data consistency—read-after-write consistency for all upload and delete operations.
  • Namespace for your domain—only you can create bucket URIs containing your domain name.
  • Data replicated in multiple data centers across the U.S. and within the same data center.

Tools

  • Web-based storage manager.
  • GSUtil, an open source command line tool.
  • Compatible with many existing cloud storage tools and libraries.

Read the Getting Started Guide to learn more about the service.

Note: Google Storage for Developers does not support Google Apps accounts that use your company domain name at this time.

Back to top

Pricing

Google Storage for Developers pricing is based on usage.

  • Storage—$0.17/gigabyte/month
  • Network
    • Upload data to Google
      • $0.10/gigabyte
    • Download data from Google
      • $0.15/gigabyte for Americas and EMEA
      • $0.30/gigabyte for Asia-Pacific
  • Requests
    • PUT, POST, LIST—$0.01 per 1,000 requests
    • GET, HEAD—$0.01 per 10,000 requests

SAS/Blades/Servers/ GPU Benchmarks

Just checked out cool new series from NVidia servers.

Now though SAS Inc/ Jim Goodnight thinks HP Blade Servers are the cool thing- the GPU takes hardware high performance computing to another level. It would be interesting to see GPU based cloud computers as well – say for the on Demand SAS (free for academics and students) but which has had some complaints of being slow.

See this for SAS and Blade Servers-

http://www.sas.com/success/ncsu_analytics.html

To give users hands-on experience, the program is underpinned by a virtual computing lab (VCL), a remote access service that allows users to reserve a computer configured with a desired set of applications and operating system and then access that computer over the Internet. The lab is powered by an IBM BladeCenter infrastructure, which includes more than 500 blade servers, distributed between two locations. The assignment of the blade servers can be changed to meet shifts in the balance of demand among the various groups of users. Laura Ladrie, MSA Classroom Coordinator and Technical Support Specialist, says, “The virtual computing lab chose IBM hardware because of its quality, reliability and performance. IBM hardware is also energy efficient and lends itself well to high performance/low overhead computing.

Thats interesting since IBM now competes (as owner of SPSS) and also cooperates with SAS Institute

And

http://www.theaustralian.com.au/australian-it/the-world-according-to-jim-goodnight-blade-switch-slashes-job-times/story-e6frgakx-1225888236107

You’re effectively turbo-charging through deployment of many processors within the blade servers?

Yes. We’ve got machines with 192 blades on them. One of them has 202 or 203 blades. We’re using Hewlett-Packard blades with 12 CP cores on each, so it’s a total 2300 CPU cores doing the computation.

Our idea was to give every one of those cores a little piece of work to do, and we came up with a solution. It involved a very small change to the algorithm we were using, and it’s just incredible how fast we can do things now.

I don’t think of it as a grid, I think of it as essentially one computer. Most people will take a blade and make a grid out of it, where everything’s a separate computer running separate jobs.

We just look at it as one big machine that has memory and processors all over the place, so it’s a totally different concept.

GPU servers can be faster than CPU servers, though , Professor G.




Source-

http://www.nvidia.com/object/preconfigured_clusters.html

TESLA GPU COMPUTING SOLUTIONS FOR DATA CENTERS
Supercharge your cluster with the Tesla family of GPU computing solutions. Deploy 1U systems from NVIDIA or hybrid CPU-GPU servers from OEMs that integrate NVIDIA® Tesla™ GPU computing processors.

When compared to the latest quad-core CPU, Tesla 20-series GPU computing processors deliver equivalent performance at 1/20th the power consumption and 1/10th the cost. Each Tesla GPU features hundreds of parallel CUDA cores and is based on the revolutionary NVIDIA® CUDA™ parallel computing architecture with a rich set of developer tools (compilers, profilers, debuggers) for popular programming languages APIs like C, C++, Fortran, and driver APIs like OpenCL and DirectCompute.

NVIDIA’s partners provide turnkey easy-to-deploy Preconfigured Tesla GPU clusters that are customizable to your needs. For 3D cloud computing applications, our partners offer the Tesla RS clusters that are optimized for running RealityServer with iray.

Available Tesla Products for Data Centers:
– Tesla S2050
– Tesla M2050/M2070
– Tesla S1070
– Tesla M1060

Also I liked the hybrid GPU and CPU

And from a paper on comparing GPU and CPU using Benchmark tests on BLAS from a Debian- Dirk E’s excellent blog

http://dirk.eddelbuettel.com/blog/

Usage of accelerated BLAS libraries seems to shrouded in some mystery, judging from somewhat regularly recurring requests for help on lists such as r-sig-hpc(gmane version), the R list dedicated to High-Performance Computing. Yet it doesn’t have to be; installation can be really simple (on appropriate systems).

Another issue that I felt needed addressing was a comparison between the different alternatives available, quite possibly including GPU computing. So a few weeks ago I sat down and wrote a small package to run, collect, analyse and visualize some benchmarks. That package, called gcbd (more about the name below) is now onCRAN as of this morning. The package both facilitates the data collection for the paper it also contains (in the vignette form common among R packages) and provides code to analyse the data—which is also included as a SQLite database. All this is done in the Debian and Ubuntu context by transparently installing and removing suitable packages providing BLAS implementations: that we can fully automate data collection over several competing implementations via a single script (which is also included). Contributions of benchmark results is encouraged—that is the idea of the package.

And from his paper on the same-

Analysts are often eager to reap the maximum performance from their computing platforms.

A popular suggestion in recent years has been to consider optimised basic linear algebra subprograms (BLAS). Optimised BLAS libraries have been included with some (commercial) analysis platforms for a decade (Moler 2000), and have also been available for (at least some) Linux distributions for an equally long time (Maguire 1999). Setting BLAS up can be daunting: the R language and environment devotes a detailed discussion to the topic in its Installation and Administration manual (R Development Core Team 2010b, appendix A.3.1). Among the available BLAS implementations, several popular choices have emerged. Atlas (an acronym for Automatically Tuned Linear Algebra System) is popular as it has shown very good performance due to its automated and CPU-speci c tuning (Whaley and Dongarra 1999; Whaley and Petitet 2005). It is also licensed in such a way that it permits redistribution leading to fairly wide availability of Atlas.1 We deploy Atlas in both a single-threaded and a multi-threaded con guration. Another popular BLAS implementation is Goto BLAS which is named after its main developer, Kazushige Goto (Goto and Van De Geijn 2008). While `free to use’, its license does not permit redistribution putting the onus of con guration, compilation and installation on the end-user. Lastly, the Intel Math Kernel Library (MKL), a commercial product, also includes an optimised BLAS library. A recent addition to the tool chain of high-performance computing are graphical processing units (GPUs). Originally designed for optimised single-precision arithmetic to accelerate computing as performed by graphics cards, these devices are increasingly used in numerical analysis. Earlier criticism of insucient floating-point precision or severe performance penalties for double-precision calculation are being addressed by the newest models. Dependence on particular vendors remains a concern with NVidia’s CUDA toolkit (NVidia 2010) currently still the preferred development choice whereas the newer OpenCL standard (Khronos Group 2008) may become a more generic alternative that is independent of hardware vendors. Brodtkorb et al. (2010) provide an excellent recent survey. But what has been lacking is a comparison of the e ective performance of these alternatives. This paper works towards answering this question. By analysing performance across ve di erent BLAS implementations|as well as a GPU-based solution|we are able to provide a reasonably broad comparison.

Performance is measured as an end-user would experience it: we record computing times from launching commands in the interactive R environment (R Development Core Team 2010a) to their completion.

And

Basic Linear Algebra Subprograms (BLAS) provide an Application Programming Interface
(API) for linear algebra. For a given task such as, say, a multiplication of two conformant
matrices, an interface is described via a function declaration, in this case sgemm for single
precision and dgemm for double precision. The actual implementation becomes interchangeable
thanks to the API de nition and can be supplied by di erent approaches or algorithms. This
is one of the fundamental code design features we are using here to benchmark the di erence
in performance from di erent implementations.
A second key aspect is the di erence between static and shared linking. In static linking,
object code is taken from the underlying library and copied into the resulting executable.
This has several key implications. First, the executable becomes larger due to the copy of
the binary code. Second, it makes it marginally faster as the library code is present and
no additional look-up and subsequent redirection has to be performed. The actual amount
of this performance penalty is the subject of near-endless debate. We should also note that
this usually amounts to only a small load-time penalty combined with a function pointer
redirection|the actual computation e ort is unchanged as the actual object code is identi-
cal. Third, it makes the program more robust as fewer external dependencies are required.
However, this last point also has a downside: no changes in the underlying library will be
reected in the binary unless a new build is executed. Shared library builds, on the other
hand, result in smaller binaries that may run marginally slower|but which can make use of
di erent libraries without a rebuild.

Basic Linear Algebra Subprograms (BLAS) provide an Application Programming Interface(API) for linear algebra. For a given task such as, say, a multiplication of two conformantmatrices, an interface is described via a function declaration, in this case sgemm for singleprecision and dgemm for double precision. The actual implementation becomes interchangeablethanks to the API de nition and can be supplied by di erent approaches or algorithms. Thisis one of the fundamental code design features we are using here to benchmark the di erencein performance from di erent implementations.A second key aspect is the di erence between static and shared linking. In static linking,object code is taken from the underlying library and copied into the resulting executable.This has several key implications. First, the executable becomes larger due to the copy ofthe binary code. Second, it makes it marginally faster as the library code is present andno additional look-up and subsequent redirection has to be performed. The actual amountof this performance penalty is the subject of near-endless debate. We should also note thatthis usually amounts to only a small load-time penalty combined with a function pointerredirection|the actual computation e ort is unchanged as the actual object code is identi-cal. Third, it makes the program more robust as fewer external dependencies are required.However, this last point also has a downside: no changes in the underlying library will bereected in the binary unless a new build is executed. Shared library builds, on the otherhand, result in smaller binaries that may run marginally slower|but which can make use ofdi erent libraries without a rebuild.

And summing up,

reference BLAS to be dominated in all cases. Single-threaded Atlas BLAS improves on the reference BLAS but loses to multi-threaded BLAS. For multi-threaded BLAS we nd the Goto BLAS dominate the Intel MKL, with a single exception of the QR decomposition on the xeon-based system which may reveal an error. The development version of Atlas, when compiled in multi-threaded mode is competitive with both Goto BLAS and the MKL. GPU computing is found to be compelling only for very large matrix sizes. Our benchmarking framework in the gcbd package can be employed by others through the R packaging system which could lead to a wider set of benchmark results. These results could be helpful for next-generation systems which may need to make heuristic choices about when to compute on the CPU and when to compute on the GPU.

Source – DirkE’paper and blog http://dirk.eddelbuettel.com/papers/gcbd.pdf

Quite appropriately-,

Hardware solutions or atleast need to be a part of Revolution Analytic’s thinking as well. SPSS does not have any choice anymore though 😉

It would be interesting to see how the new SAS Cloud Computing/ Server Farm/ Time Sharing facility is benchmarking CPU and GPU for SAS analytics performance – if being done already it would be nice to see a SUGI paper on the same at http://sascommunity.org.

Multi threading needs to be taken care automatically by statistical software to optimize current local computing (including for New R)

Acceptable benchmarks for testing hardware as well as software need to be reinforced and published across vendors, academics  and companies.

What do you think?


%d bloggers like this: