Menu Content/Inhalt
Home

From Blade to CUDA (2012) Print
July 2012

This article is in the context of HPC- High Performance Computing.  HPC refers to a high level of computing power that works on a single application.  Many scientific research, engineering analysis, digital content creation, and finite analysis applications are such types of single applications.  Take bio-molecular behaviour simulation as an example.  The fastest HPC system took 1 day to simulate 69ns of behaviour in 2011.  Do scientists want to achieve a significantly longer period of simulation?  Definitely.  Can they use 2 or more computers to run the same application?  Yes, this has been the approach used for decades.  A HPC system is the equivalence of a million departmental servers - just for an idea.

Blade Server has been the de facto format for HPC up to 2010.  How much space will be needed to accommodate a million departmental servers?  How many cables will be running across such a football field of computers?  Blade is an approach to integrate a large number of servers into the space of a cabinet.  A small blade server takes only 7 height units of a 19” wide cabinet to provide the computing power of 40 Xeon processors.  Instead of cables running around to connect 20 or 40 Xeon servers, blade servers use printed circuits inside the cabinet to save space and cables.  These nodes of servers use 10G Ethernet or Infiniband for interconnection among themselves, and an integrated Gigabit Ethernet Switch for communicating with clients outside of the blade. 

The above approach seemed the best way to go forward till 2011 when a company not traditionally supplying central processing units (CPU) turned up with a newish scheme called CUDA- compute unite device architecture. Nvidia has been a market leader in graphics processors and its GeForce series of graphics cards are well established in the market.  Nvidia turned graphics processing units (GPU) into application computing use and created CUDA compilers to instruct the CPU to send parcels of computing work to the GPU. CPU and GPU are connected with PCI Express which is a high speed serial bus on the motherboard running at 20Gbps per lane in each direction with 16 lanes on version 3.  It is not just the speed of interconnection of processors that beats the blade approach. The high count of cores in the GPU is the most important.  As of July 2012, a NZ$900 GTX680 card provides 1536 cores for an idea.  HPC does not use GTX which is for consumers.  HPC uses uses Tesla which is another Nvidia brand.  

CUDA based HPC rose to fame when the Tianhe (Sky River) computer based in China was rated the fastest computer in the world.  This computer had less than 1/15th of the count of Xeon CPU as the runners-up computer but had 1/30th of the count in the form of Tesla GPU.  This implies that each Tesla provided 30 times the performance of a server CPU for typical HPC applications.

Nvidia alerted Compucon of the new supercomputing ranking event in November 2010.  The alert set the beginning of a journey for Compucon to visit CUDA. Compucon has spent 12 months to understand CUDA and Quadro which is another Nvidia brand for professional graphics processing.  When Nvidia released Tesla 2075 cards, Compucon spent another 3 months to find out how Tesla could help professional graphics processing and has obtained comfort from benchmarking Adobe Premiere Pro CS6 with PPBM6 in June 2012.  We feel that it is time that we promote CUDA HPC systems instead of blades.

The latest as at July 2012 CUDA HPC system platform is designed and developed by Supermicro in US based on Intel C602 chipset (so that Compucon did not steal the credit).  This system is 1U rack mounted and reasonably tiny in space requirements.  It runs a Xeon processor and up to two Tesla cards and provides for 6 hard disks that can be set up in arrays of 0 for speed, 1 for redundancy, and 10 for both.  Compucon will configure and build this model with a 6-core 2GHz 15MBL3 Xeon based on Sandy Bridge micro-architecture and one Tesla card with 448 Fermi CUDA cores and 6GB of GPU memory as the base.  Options include the second Tesla card and up to 256GB ECC of main memory.

The appeals of this model are many: space, cost, performance, and electricity consumption.  It is about 1/10th of the price of a blade server. All these appeals do come with one condition.  This HPC system is only good for applications that have been compiled with CUDA.  This is not really a problem because CUDA is being taught in over 500 universities around the world and CUDA is seen as a new standard for HPC today.

 
Thunderbird FX6100 Introduction Notes Print
July 2012

We have had Thunderbird as a performance system for knowledge workers in the AMD series and Jasper as a standard system for process workers for 20 years.  At the same time, we have Superhawk and Diamond correspondingly based on Intel technologies.  As of July 2012, Thunderbird will be Jasper based with high performance hardware.  This is a weird but sensible idea for a while.   We will explain the last statement.

Jasper is the only desktop system not having a CPU that integrates video graphics display processing.   Jasper has its graphics processing done by a chip integrated in the North Bridge on the motherboard.  This Jasper arrangement has been around for at least 10 years. 

Thunderbird A75, which is outgoing, is different to Jasper.  Thunderbird A75 does not have a CPU and a VGA chip separately but an APU which consists of both functions.  This is the same as what Intel had done to its desktop platforms ahead of AMD by about 12 months.  The APU for Thunderbird has 4 CPU cores and 2 dozens of GPU cores in one physical die.   Obviously this APU socket called FM1 is not compatible with AM3+ socket which is used for the CPU on a Jasper motherboard.

According to the latest market information, AM3+ is way more popular than FM1.  The reason seems rather obvious.  AM3+ covers the high and low end and FM1 is for the middle.  Intel does it differently.  Intel applies the combined processors to the middle and low end whereas discrete processors to the high end.  On comparison, the AMD scheme could be seen as confusing to customers.  In business management terms the middle is called the no-man’s land and the middle has a lower growth potential than either poles.

AMD has not withdrawn FM1 but is pushing FM1 down the ladder and working to bring out new FM2 for the middle.  FM1 is still available but only with lower rated APU which does not suit Thunderbird.  FM2 has no production schedule yet as of July 2012.  This situation has prompted Compucon to introduce Thunderbird FX6100 to succeed Thunderbird A75. 

FX6100 is an AM3+ CPU with 6 cores and 8MB L3 cache.  It is faster than A75 CPU.  Thunderbird FX6100 has a discrete EN201S card with 16 CUDA cores and 1GB dedicated memory on the card, and this card is faster than A75 GPU.  This discrete combination will produce higher performance than A75.  Nevertheless other minor features are not as generous as A75 as this motherboard has a smaller real estate than used for Thunderbird A75. 

We will monitor the AMD FM2 release schedule closely and make the best move.

 
CPD Seminar 2012-0724 IPVS Print
July 2012

Compucon CPD Seminar
24 July 2012 Tuesday Compucon House Albany

This is a Continuing Professional Development effort for our business partners and customers.  Attendance is free of charge but pre-registration is strictly necessary because our seating capacity is limited.  Please register by replying to this email.

4:00 – 4:15pm Outside the Square Session
4:15 – 4:40pm Computing for Fitness for Purpose
4:40 – 5:20pm IP Video Surveillance (IPVS) System
5:20 – 6:00pm Video Surveillance Virtual System
6:00 - 7:30pm Wine and Cheese

The theme for Outside the Square is air travel.  Air refers to the Earth and not Space.  When we travel by air, we want the trip to be safe, comfortable and as fast as possible.  Efforts have certainly been made by the 2 major civilian airplane makers on these 3 fronts.  This session will have a look of these issues, and poke into the 4th factor which may come into being in 10 years.

As there are many types of computer models as there are vehicles addressing different applications, this short session is an effort to elaborate fitness for purpose of computer models.  We will single out 2 computer types for discussion.  The first is desktop computing which emphasizes speed of performance.  We will point out that Thunderbird FX6100 has its special appeals and weaknesses among the myriads of desktop models.  The second is on the top end of computing which could be in a datacentre or research organisation.  We will explain the difference between homogenous computing (as in blade server) and heterogeneous computing (as in CUDA based systems).  The terms may sound very technical but they will be extremely easy to understand once we show you an analogy.  CUDA is winning the hearts and minds of top organisations and professionals across the world and we shall have a fair go at this word.

IP based technology has surpassed analogue based for video surveillance across the world this year and reached the mature phase of its adoption lifecycle.  IPVS is not a disruptive technology and it is in fact a foundation technology as big IT players such as Cisco have clinched onto the IPVS bandwagon to develop new products and services.  This session will explain the 3 subsystems of IPVS as in video capturing, connecting fabrics, and recording systems; and some impressive artificial intelligence of IP based systems.    We will explain where IT people have a major advantage over traditional system installers and inversely how traditional system installers can upgrade to master this new technology foundation stone.   We will provide insights from our direct involvements.  This session will not go into technical details and will stay on a conceptual level.  It is suitable for people interested in video surveillance and a general (not specialist) technical background is necessary. 

The last session is about a project blueprint that Compucon is about to launch.  The project is the design and installation of a regional video surveillance system.  Regional is a keyword here.  It means all videos captured within a region will be processed centrally although each camera site sees itself as a discrete and independent system.  Apart from installations at each customer site, other expensive hardware will be virtualized for sharing and this achieves cost saving and operational efficiency.  This is not the hype of clouds.  Clouds carry many uncertainties whereas the planned system offers certainties to customers.  North Harbour Industrial Estate is a well defined region and we plan to start the project in this region.  The same scheme can be applied across New Zealand as long as the conditions are satisfied. This session will introduce the scheme and how customers are better off than other schemes.  We will not go into technical details and will stay on a conceptual level.   

 
Microsoft Office 2010 Compare Editions Print
June 2012
Image
 
NVIDIA Quadro Recommendations for Professionals Print
June 2012

Question
What graphics card should I recommend to my client?

Answer
Does your customer work in a technical profession? Design, engineering, architecture, science and research fields can all make use of a professional graphic card. Please see the following guide: Quadro Recommendations


Preview:

Image
 
<< Start < Prev 121 122 123 124 125 126 127 128 129 130 Next > End >>

Results 1108 - 1116 of 2512