VMware ESXi 5.1 New Features and vSphere 5.1 New Features

This blog is a followup to the VMware ESXi 5.0 New Features posting from just over a year ago. VMware has released to the public the details of new features in VMware ESXi 5.1 and vSphere 5.1 and I will cover those new 5.1 features here, though if you are new to the 5.x series, the prior blog may still be quite interesting also. Spoiler: one huge new "feature" in 5.1 is removal of the vRAM limits! Let's look at all this more...

New Features in VMware ESXi 5.1

vRAM Memory Limits Removed: the biggest non-feature "feature"!

What does it say about a product when the biggest "feature" is simply un-doing / correcting a blunder made by upper-management at a company? If you remember the fiasco surrounding the new vRAM memory limits imposed by ESXi/vSphere 5.0, you know to what I refer. VMware's attempts to squeeze more cash out of customers by imposing what amounted to a RAM-tax upon their robust server boxes backfired (i.e., irked customers, like me). And, they have now un-done that mistake. ESXi/vSphere 5.1 is supposed to now be priced (solely) on a per-CPU-socket basis rather than on a strange and ridiculous combo of sockets/virtual memory used/VMs-being-managed. That is a good thing: I actually stuck with ESXi 4.1 due to the 5.0 vRAM bull@#! So, version 5.1 is on my radar.

Support for Newer Hardware

Not surprisingly, this latest 5.1 release includes support for bigger and more recent computing hardware (both Intel and AMD). In addition, the virtualization hardware-abstraction layer has been upgraded to a new "Version 9 virtual hardware" that includes support for Intel's VT-x with Extended Page Tables virtualization assistance features and the AMD-V with Rapid Virtualization Indexing (RVI) (nee, "nested page tables"). This VT-x/EPT and AMD-V/RVI support is to partially reduce hypervisor and virtual machine (VM) guest operating system overhead imposed on the physical processors (your server's CPUs).

One nice feature that comes along with this latest 5.1 version is, unlike with the 5.0 release, it is possible to allow any VM generated on VMware ESX Server 3.5 or later to continue to run on ESXi 5.1 unchanged (i.e., without being forced to shut down, update to the version 9 virtual hardware, and restart). Of course, if you want the latest features of the VM and hypervisor that come with "version 9 virtual hardware", you will have to update your VMs to get it, but at least you have the option to postpone the virtual-hardware upgrade task until it is convenient.

New Adobe-Flash Web-Based Management Client for vSphere 5.1

Yes, you read right: a new Flash-based management client! (actually, it was written in Apache Flex, which uses Flash to run applications built with Flex). I personally am OK with this as I have worked with some very capable Flash-based applications. The old management client is still able to interact with vSphere 5.1 applications, but features that are new to vSphere 5.1 will only be available in the Flash-based web interface client. Sure, it means that you need Flash installed on whatever machine you plan to manage your virtualization setup from, but such is. I already need Flash for so many other things that this is a given.

The new UI is peppy, stable, and secure from what reviewers are saying so far. And, it offers an advantage of performing some potentially-long-running-tasks asynchronously (threaded) so as to prevent UI lockup that could occur in the previous management UIs. And, the fact is, Flash-based UIs should look and behave identically on any device that can run Flash — which surely cannot be said of HTML-based UIs!

Virtual Machine Hardware-Accelerated 3D Graphics Support

Maybe VMware read my past blog where I stated (how in ESXi 5.0) that I felt "something is amiss: where is the Nvidia CUDA / vGPU support in ESXi 5.0? Well, it turns out VMware is noticing the importance of offloading processing to GPUs after all:

With vSphere 5.1, VMware has partnered with NVIDIA to provide hardware-based vGPU support inside the virtual machine. vGPUs improve the graphics capabilities of a virtual machine by off-loading graphic-intensive workloads to a physical GPU installed on the vSphere host. In vSphere 5.1, the new vGPU support targets View environments that run graphic-intensive workloads such as graphic design and medical imaging.

Hardware-based vGPU support in vSphere 5.1 is limited to View environments running on vSphere hosts with supported NVIDIA GPU cards [well, duh] (refer to the VMware Compatibility Guide for details on supported GPU adapters). In addition, the initial release of vGPU is supported only with desktop virtual machines running Microsoft Windows 7 or 8. Refer to the View documentation for more information on the vGPU capabilities of vSphere 5.1.

NOTE: vGPU support is enabled in vSphere 5.1, but the ability to leverage this feature is dependent on a future release of View. Refer to the View documentation for information on when this feature will be available.

Hmmmm... I am not too keen on that final caveat / disclaimer (about "future release" and timeline), but it sure sounds better than the lack of information about NVidia GPU support in previous releases! I am definitely intrigued by this since I play around a bit with CUDA code, but I am not specifically seeing "CUDA" mentioned here. I wonder how far this "off-loading" goes?

Other New and Enhanced ESXi / vSphere 5.1 Features

In no particular order...

• Windows 8 desktop and Windows Server 2012 support. Nothing I personally plan to use in production anytime soon, but support is there for the latest Microsoft operating systems. I do have intentions of trying these latest OS offerings out, and VMs are the only way I would even consider it; so, good thing they are supported.
  
• ESXi 5.1 has improved CPU virtualization methods ("virtualized hardware virtualization", or VHV) that is supposed to allow near-native-access to the physical CPU(s) by your virtualized guest OS's. We all like more speed in our VMs, so this sounds like a plus.
  
• ESXi now has the ability to perform a VM-live-migration between two separate physical servers (running ESXi) without the need for both machines to be attached to the SAN. I need to read up on this and fully understand what that means... like, do I need a SAN at all anymore for this?
• CPU counter and hardware-assisted-virtualization information can now be exposed to guest operating systems (useful to developers that need to debug / tune applications meant to run in a VM).
  
• New Storage Features including: read-only file sharing on a VMFS volumes have been increased to 32 (from 8); Space Efficient Sparse Virtual Disks with automated mechanisms for reclaiming stranded space plus a dynamic block allocation unit size (tune-able to storage/apps needs); 5 Node MSCS Cluster (vs. 2 node); jumbo frame support for all iSCSI adapters (with UI support too); and, Boot from Software FCoE.
  
• The reliance on a shared "root" user account (for administrators) was eliminated and support was added for SNMPv3. Local users assigned administrative privileges automatically get full shell access and no longer must "su" (sudo) to root to run privileged commands. This makes for finer-grained auditing and monitoring, which is a plus in shared environments.
  
• With vSphere 5.1 Guest OS Storage Reclamation feature: when files are removed from inside the guest OS, the size of the VMDK file can be reduced and the deallocated storage space returned to the storage array’s free pool (utilizes new SE sparse VMDK format available with View); but note, this feature carries with it the same disclaimer that the NVIDIA stuff did — i.e., "dependent on a future release of View". Argghh. Wonder how far in the future that may be?
  

Conclusion

There are a fair number of new features in this latest release of ESXi 5.1 and vSphere 5.1 that are worth checking out, even though some significant ones are "dependent on future releases of View". The timing of this ESX / vSphere release goes along with the latest VMware Workstation, which I discuss here too: VMware Workstation 9.0 New Features of Interest — if you are interested in the desktop-product side of things.

Continue to read this Software Development and Technology Blog for computer programming articles (including useful free / OSS source-code and algorithms), software development insights, and technology Techniques, How-To's, Fixes, Reviews, and News — focused on Dart Language, SQL Server, Delphi, Nvidia CUDA, VMware, TypeScript, SVG, other technology tips and how-to's, plus my varied political and economic opinions.

Nvidia CUDA Toolkit 4.1 and Parallel Nsight 2.1

Nvidia's CUDA technology has been around for 5 years now, and only 6 months ago blogged about Nvidia's CUDA technology when the CUDA Toolkit 4.0 was released.  Nvidia is keeping up the pace of innovation with a substantive upgrade to both Nvidia CUDA Toolkit (with version 4.1) and Parallel Nsight (now at version 2.1).  

CUDA: What is it?

CUDA is NVIDIA’s parallel computing architecture that enables dramatic increases in computing performance by harnessing the power of the GPU (graphics processing unit) for applications including image and video processing, computational biology and chemistry, fluid dynamics simulation, CT image reconstruction, seismic analysis, ray tracing, and much more.  The current Nvidia (NASDAQ:NVDA) "Fermi" line of GPUs (Graphical Processing Units) provides incredibly powerful parallel computing within reach of most individual users and businesses through rather affordable Nvidia Graphics Cards (and, the upcoming Nvidia "Kepler" GPUs for early 2012 will only be better, faster, and more efficient).

Note: many of these latest CUDA features require a "Fermi"-based GPU (and, using the LLVM-based compiler does).  These cards are worth investing in if you plan to do any CUDA development.  You can get a Fermi-based CUDA-Capable Graphics Card that is quite affordable and power-efficient: I rather like my Quadro 600 card (~$160.00) which uses only 40W for 96 CUDA-processing cores; this card has been very capable for running all my development work on.

New in Nvidia CUDA Toolkit 4.1

LLVM Compiler / Toolchain Support

Nvidia CUDA Toolkit 4.1 now includes a new LLVM-based CUDA compiler along with over 1000 new image processing functions, plus a redesigned Visual Profiler.  Integrating the open source Low Level Virtual Machine (LLVM) toolchain support definitely has my attention (LLVM is a collection of modular and reusable compiler and toolchain technologies).

The first notable benefit of the LLVM compiler is that Nvidia claims this compiler delivers up to 10% faster performance for many applications (compared to their prior in-house developed C/C++ compiler).

But, what strikes me as the most (potentially) important aspect of this move to LLVM is that we could potentially soon see more (programming) language support for using CUDA outside of just C/C++ and/or additional CPU support.  Nvidia has apparently used the Clang C and C++ compilers within the LLVM framework and has hooked in support for the CUDA parallel development environment.

Although Nvidia's (CUDA C and CUDA C++) compiler modifications are not open-sourced, LLVM will provide a foundation for more easily adding language/processor support.  Given Apple's use of LLVM on ARM (platform), I have to wonder if this is going to be a build-target in the not too distant future.  There are also open-source projects for other programming languages to make use of the LLVM toolchain, so the potential does exist for accessing CUDA / GPU-support from other domain-specific languages eventually (perhaps Java, Python, etc) directly.

Other Major New Features in CUDA Toolkit 4.1
(from Nvidia website, with some added comments and details)

New & Improved “Drop-In” Acceleration With GPU-Accelerated Libraries

• Over 1000 new image processing functions in the NPP (Nvidia Performance Primitives) library — this brings to total number of NPP functions to 2200+. These GPU-accelerated functions (building blocks) for image and signal processing include capabilities geared toward arithmetic, logic, conversion, statistics, filters, and more; also, these can execute on the GPU at up to 40x (yes, 40 times!) the speed of Intel IPP (Integrated Performance Primitives).  This is great for media, entertainment, and visual processing applications.
• New Boost style placeholders in Thrust CUDA C++ template library which allow inline functors now.  Thrust includes optimized functions for sort, reduce, scan operations and so on.
• New cuSPARSE tri-diagonal solver up to 10x faster than MKL on a 6 core CPU; this also includes up to 2x faster sparse matrix vector multiplication using ELL hybrid format 
• New support in cuRAND for MRG32k3a and Mersenne Twister (MTGP11213) RNG algorithms 
• Bessel functions now supported in the CUDA standard Math library 
• CuFFT (Fast Fourier Transforms) library has a thread-safe API now (callable from multiple host-threads); also, substantial improvements in speed!
• CuBLAS level 3 performance improvements up to 6X over Intel MKL (Math Kernel Library)
• Batched-GEMM API for more efficient processing of many small matrices (i.e., 4x4 through 128x128 matrices; up to 4X speedup over MKL); up to 1 TFLOPS sustained performance (yes, a teraflop!  Wow)
• Average and rounded-average functions (e.g., hadd / rhadd - signed and unsigned)

Enhanced & Redesigned Developer Tools (On Windows, Mac, & Linux)

• Redesigned Visual Profiler with automated performance analysis and expert guidance (guided workflow and drill-down expert guidance); during an online presentation, this was described as "almost like having an Nvidia engineer in a box", which sure sounds handy!  You should benefit from the experience of those engineers, and be helped along through attaining best-practice outcomes with these built-in automated analyses/experts.
• Assert() in device code - helpful for debugging!
• CUDA_GDB support for multi-context debugging and assert() in device code
• CUDA-MEMCHECK now detects out of bounds access for memory allocated in device code
• Parallel Nsight 2.1 CUDA warp watch visualizes variables and expressions across an entire CUDA warp
• Parallel Nsight 2.1 CUDA profiler now analyzes kernel memory activities, execution stalls and instruction throughput
• Learn more about debugging and performance analysis tools for GPU developers on our CUDA Tools and Ecosystem Summary Page

Advanced Programming Features

• Access to 3D surfaces and cube maps from device code
• Enhanced no-copy pinning of system memory, cudaHostRegister() alignment and size restrictions removed
• Peer-to-peer communication between processes
• Support for resetting a GPU without rebooting the system in nvidia-smi

New & Improved SDK Code Samples

• simpleP2P sample now supports peer-to-peer communication with any Fermi GPU
• New grabcutNPP sample demonstrates interactive foreground extraction using iterated graph cuts (this is really neat!)
• New samples showing how to implement the Horn-Schunck Method for optical flow, perform volume filtering, and read cube map texture

New in Nvidia Parallel Nsight 2.1 for Visual Studio
Parallel Nsight is a powerful IDE-integration and development tool that allows you to perform the following types of procedures from within Microsoft Visual Studio:

• Debug CUDA Kernels directly on the GPU hardware
• Examine (potentially thousands of) threads that are executing in parallel
• Use on-target conditional breakpoints to locate errors
• Use the CUDA memory-checker
• Perform System-Trace activities to review CUDA activities that span your CPU(s) and GPU(s)
• Perform deep kernel analysis to find performance bottlenecks so you can optimize the code speedup that is possible with CUDA and massively parallel-processed code.
• Profiling capabilities including advanced experiments to measure memory utilization, instruction throughput, and stall conditions

Some of the new capabilities include:

• a "warp watch" ability to watch variables and expressions across an entire CUDA warp (a particular level of granularity that is very useful to watch)
• analyzing kernel memory (alloc/dealloc events, execution stalls, etc)

Summary: CUDA 4.1 Continues Nvidia's Great GPU-Accelerated Application Development Tools Improvements

This latest release of the CUDA Toolkit from Nvidia continues to make life easier for any of us that are into parallel-programming with modern GPUs.  Although GPU-computing can be a bit overwhelming and a requires a different mindset than programming desktop applications or designing a website, if you have an application that can benefit from the power of simultaneous operations, this is a technology worth diving into: it is nothing short of a transformational technology.

VMware ESXi 5.0 Release Features of Interest

I am a rather ardent fan of VMware (Public, NYSE:VMW) computing hardware virtualization products, especially VMware Workstation and VMware vSphere ESXi (a "bare iron" virtualization product). As a software developer, these tools give me the ability to maintain many simultaneous testing environments on any given desktop or server I employ for writing, testing, or deploying software — a huge time and cost saver for me.

-----------------------------------------------------------------------------------------------------------
UPDATE (2011-07-20): VMware has released their official guides to What's New in ESXi 5.0 and vSphere 5.0 since this blog was first written, and I have now written a followup blog (to this one) about the New Features in VMware ESXi 5.0 and vSphere 5.0 that goes into much more detail about certain aspects of the products and the new features.  But, I did not necessarily repeat every bullet-point from this initial post, so you may find some nice info on this page also.

-----------------------------------------------------------------------------------------------------------

The most recent official release of VMware's vSphere ESXi server virtualization / hypervisor product — a free bare-metal hypervisor — was version 4.1 (with update 1), though VMware ESXi 5.0 is nearing on the horizon now, and promises to introduce some features I have been longing for.  In case you need help migrating from 4.0 to 4.1 (in preparation for the upcoming 5.0 release), I wrote a blog about How to Upgrade VMware ESXi 4.0 to 4.1 recently.

I find version 4.x of ESXi to be quite capable for most of my needs, but there were a couple features in the Workstation virtualization product that I wanted to see on my server (via ESXi).  Some early feature-set "leaks" and "rumors" are hitting the technology blogs and other technology news sites now, and if what I am reading comes to fruition in the vSphere 5.0 version of the ESXi hypervisor, I will be quite pleased.

Note: I may not be the "typical" ESXi user, since I do not necessarily use the platform for all its enterprise-grade features, but I doubt I am alone in my using ESXi as a sort of "big brother" to my VMware Workstation desktops.  I am a software developer and SQL-Server database aficionado, and this leads me to using my Server platforms for what is a mix of enterprise-level database-design/testing in addition to actually writing software inside VM's hosted on vSphere ESXi.  I need some of the large-CPU-count limits and memory-size limits the product offers as well as its support for some more enterprise-grade hardware, but I also like features (introduced in v4.x) that included high-resolution graphics for my "server" VMs so that I could enjoy screen-real-estate when editing large blocks of code and testing software sporting robust GUIs.

New Features to Expect
So, with the above in mind, client-connected USB devices are something I would really like to see in the VMware ESXi 5.0 new features, as there are times I am using my server with ESXi to test some software that interacts with a USB-based product — be it a USB printer, USB scanner, USB camera, USB memory-stick, or USB external hard-drive.  From what I am reading, this feature is coming in vSphere 5.0!  In addition, some other noteworthy vSphere 5 / ESXi 5 features that are being talked about include the following (with the ones I listed first being of most interest to me):

• USB 3.0 device support — to me, this is a "must have" while on the topic of USB-connectivity these days.  I have seen first hand how much faster (than USB 2.0) my USB 3.0 is for large file-transfers to my Seagate FreeAgent GoFlex external USB 3.0 drive (using a Buffalo USB 3.O add-in card).  There is no comparison... the USB 3.0 is many multiples faster, and large-file transfers are done in a fraction of the time; as such, having this available within my ESXi 5.x virtual environment would be wonderful!
• Smart card reader support for VMs — again, this is complements my desire for more "workstation-like" features when I need them. 
• Swap to SSD (Solid State Disk)  — this should be really nice for speed-improvements.  I do not know the details, but I am a fan of SSD devices and their speed.  I blogged about my Intel X-25M benchmark results in a desktop I use for software development, and the SSD products have only gotten better, faster, and more affordable since then.  Being able to perhaps cache portions of frequently-accessed data using an SSD (much cheaper than RAM), will be nice indeed.
• Apple Mac OS X Server 10.6 "Snow Leopard" guest OS support — this is a nice-to-have for me, as I have been wanting to play with more Apple code without necessarily purchasing an Apple server.
• Non hardware-accelerated 3D graphics for Windows Aero support — again, a nice-to-have from where I stand, if I want to test Windows 7 guests on ESXi under certain circumstances
• a new browser-based vSphere Client — what I find most interesting with this is that it may be *extensible*, which would perhaps make it easier for open-source (OSS) add-ons of things like free backup software for ESXi 5.x and similar much-desired functionality.  I look forward to seeing how truly extensible this product is, especially as a software-developer that would love to write some handy plug-ins for things I commonly desire.
• iSCSI user interface support  — this could be interesting too.  I have been just starting to play with FreeNAS and NexentaStor and their iSCSI support and capabilities.  Anything that makes it easier to integrate this type of functionality with ESXi is welcome by me.
• Support for 2TB+ LUN  — in this day and age of ever-increasing data and storage requirements, coupled with ever-cheaper storage, the existing 2TB LUN limit seems a bit dated.  So, being able to provision for very, very large storage requirements is sure to help with things like massive databases and the like.  I personally do not keep any "test databases" of that size laying around, but I can see the day coming where I may want to do so for performance testing, benchmarking, and similar optimization work.  I have some multi-hundred-gigabyte SQL-Server databases for testing some of my algorithms on, so perhaps the multi-TB threshold is coming (though, I am not sure if the native OS will support that anyhow? hmmm...)
• 512 VMs  — I am not near that limit by any means, but I could see this making some sense if and when my vision of "a VM per application" ever comes to fruition; that is a vision I have written about elsewhere... perhaps I will publish it here soon.
• 160 CPUs and 2TB of RAM!  uhh... wow!  Sounds nice, and wish I had such a system just for "kicks".
• 32 vCPUs per VM

There were plenty of additional enterprise-focused features talked about on the web (for vSphere 5), but most of them will not be very applicable to me.  I will keep my eye out for more "official" release information from VMware, and once some of these rumors/leaks of new features can be confirmed, I will surely write about them (and/or test them out and write about them).  Stay tuned.

Something Amiss: Nvidia CUDA support in ESXi / vSphere?

I did not read anything about VMware implementing support for Nvidia CUDA technology in vSphere5 / ESXi 5.  I sure hope they are considering this, as I see parallel-computing with CUDA as a major and serious technology trend that can not be ignored, and that should be available to server-applications.  I have written a few blogs about CUDA-enabled graphics cards and the potential they offer us software-developers that can leverage the power of the many-GPU-cores for parallel operations; this is an emerging trend in high-powered software that needs to be more accessible via virtual machines.

The last time I researched whether there was any way to make the NVidia graphics cards and their CUDA processors available to ESXi virtual machines, I found that you were out of luck.  Even attempts to use an Nvidia CUDA card as a "pass through" device (to ESXi) would not work.  Basically, unless VMware implements (in their hypervisor) something to allow the CUDA processing architecture capabilities to be accessed through server-based virtual machines, any applications that implement such features will have to run outside of a VM.  

With server vendors offering dedicated parallel-compute modules (implemented using NVidia cards), there is bound to be more and more demand for VM support of CUDA.  Financial firms, medical technology, engineering firms, and other large corporations are embracing CUDA-based applications development, and this surely must translate into demand for CUDA-enabled virtual-machine infrastructures.  So, I sure hope this is addressed in v5 if not in a subsequent "point release" of ESXi 5.1 or 5.2 (though, it is a rather substantial feature for a dot-release perhaps).

I am waiting anxiously to see how EXSi CUDA support is addressed in upcoming releases.  Aside from that, ESXi 5.0 looks rather promising so far.

Nvidia CUDA Toolkit 3.2 - more Fermi optimizations

The latest Nvidia "Fermi" GPUs (Graphical Processing Units) are available in a wide range of powerful new Nvidia Graphics Cards. If you have been waiting for a good reason to upgrade your graphics card, Nvidia has certainly provided some great reasons to upgrade with their latest CUDA-capable Nvidia cards that are just amazing!

For over a year, I was patiently waiting for the new Fermi architecture to hit the shelves so that I could get the latest CUDA parallel processing power at an incredible performance-per-watt and price-per-performance level. I have been looking forward to writing some parallel-processing-optimized code, and these cards were exactly what I needed. I ended up buying a Quadro 600 to start with, since it has 96 processing cores and only uses 40 watts of power! So far, I have been quite impressed by this card's abilities, quiet operation, and price (well under $200). I can now perform "real time" graphical operations that were impossible before. It is like having a 5 or 10 year old multi-million-dollar supercomputer on my desktop for under $200.


Personal supercomputing for the masses!
Nvidia (NASDAQ:NVDA) has moved to a modern 40nm architecture for these new GPUs, which has allowed them to be much more power-efficient while cranking out tons of graphics horsepower for gaming and/or professional applications that make use of their stream-processors (aka, "CUDA cores") on the graphics card for high-performance computing (HPC) via massively-parallel-processed algorithms.

CUDA is NVIDIA’s parallel computing architecture that enables dramatic increases in computing performance by harnessing the power of the GPU (graphics processing unit) for applications including image and video processing, computational biology and chemistry, fluid dynamics simulation, CT image reconstruction, seismic analysis, ray tracing, and much more.

Get an NVidia Fermi-based Graphics Card
First, get hold of a new Fermi-based Nvidia CUDA Graphics card  to develop and run your new CUDA applications on.

Now you can start putting some new CUDA abilities to work using the latest Nvidia CUDA Toolkit 3.2 release that has some features specific to the new Fermi cards and architecture that you may want to check into...


Nvidia CUDA Toolkit 3.2 Release Highlights

New and Improved CUDA Libraries

• CUBLAS performance improved 50% to 300% on Fermi architecture GPUs, for matrix multiplication of all datatypes and transpose variations
• CUFFT performance tuned for radix-3, -5, and -7 transform sizes on Fermi architecture GPUs, now 2x to 10x faster than MKL
• New CUSPARSE library of GPU-accelerated sparse matrix routines for sparse/sparse and dense/sparse operations delivers 5x to 30x faster performance than MKL
• New CURAND library of GPU-accelerated random number generation (RNG) routines, supporting Sobol quasi-random and XORWOW pseudo-random routines at 10x to 20x faster than similar routines in MKL
• H.264 encode/decode libraries now included in the CUDA Toolkit

CUDA Driver & CUDA C Runtime

• Support for new 6GB Quadro and Tesla products 
• New support for enabling high performance Tesla Compute Cluster (TCC) mode on Tesla GPUs in Windows desktop workstations

Development Tools

• Multi-GPU debugging support for both cuda-gdb and Parallel Nsight 
• Expanded cuda-memcheck support for all Fermi architecture GPUs 
• NVCC support for Intel C Compiler (ICC) v11.1 on 64-bit Linux distros 
• Support for debugging GPUs with more than 4GB device memory 

Miscellaneous

• Support for memory management using malloc() and free() in CUDA C compute kernels 
• New NVIDIA System Management Interface (nvidia-smi) support for reporting % GPU busy, and several GPU performance counters

New GPU Computing SDK Code Samples

• Several code samples demonstrating how to use the new CURAND library, including MonteCarloCURAND, EstimatePiInlineP, EstimatePiInlineQ, EstimatePiP, EstimatePiQ, SingleAsianOptionP, and randomFog 
• Conjugate Gradient Solver, demonstrating the use of CUBLAS and CUSPARSE in the same application 
• Function Pointers, a sample that shows how to use function pointers to implement the Sobel Edge Detection filter for 8-bit monochrome images 
• Interval Computing, demonstrating the use of interval arithmetic operators using C++ templates and recursion 
• Simple Printf, demonstrating best practices for using both printf and cuprintf in compute kernels 
• Bilateral Filter, an edge-preserving non-linear smoothing filter for image recovery and denoising implemented in CUDA C with OpenGL rendering 
• SLI with Direct3D Texture, a simple example demonstrating the use of SLI and Direct3D interoperability with CUDA C 
• cudaEncode, showing how to use the NVIDIA H.264 Encoding Library using YUV frames as input 
• Vflocking Direct3D/CUDA, which simulates and visualizes the flocking behavior of birds in flight 
• simpleSurfaceWrite, demonstrating how CUDA kernels can write to 2D surfaces on Fermi GPUs

Financial Opportunities :
Nvidia (NASDAQ:NVDA) stock?

Since this blog also focuses on stock-market and investing opportunities, I have to point out that in my August 13th, 2010 blog entry about Nvidia Toolkit 3.1 news, I contemplated whether the new Nvidia Fermi cards were going to drive substantial sales/revenue-gains and associated profit-gains for Nvidia corporation.

When I wrote that blog entry in August, Nvidia stock was $9.39, and today it is $14.77 as I write this article. If you jumped in on this one, you have already made 57% in a mere 4 months! (update: May-2011; NVDA at $19.00+) The current trend-lines on the stock look good, as it is staying ahead of its moving-average trend-lines on a technical basis, so it may well have some decent upside remaining. Plus, we are rather early in the Fermi-chip-based GPU series from Nvidia. 50+% return on your NVDA stock may have you wanting to take some profits. The choice is yours... this stock has a long-history of being rather volatile, and it will likely have some up/down cycles during its future. I plan to maintain at least some of my position in NVDA as I still think they have the technology to beat when it comes to GPU-computing and supercomputing. Intel's "Sandy Bridge" products coming out in Q1-2011 may have a slight impact on Nvidia (since the new Intel CPUs will include an integrated and allegedly rather capable GPU onboard, which will perhaps suffice for mainstream users).


WHO is going to use these cards?
The thing that I see happening with these discrete graphics cards like the Fermi-based CUDA-capable Nvidia cards is simple: if you are a business and you want to compete, you best learn how to leverage the power of these cards. Period.

Wall Street already knows this (and, I do not mean in the price of NVDA stock),... they are using this technology to perform lightning-fast calculations for algorithmic trading, options pricing, and much more. ANY application that can be significantly enhanced (i.e., made faster and more robust) through parallel processing will be made so by the companies that are leading in any field. They WILL use these GPUs from Nvidia to accomplish that feat.

I am not investing in Nvidia for the fact that home "gamers" and the like enjoy their super-potent "GeForce" cards... I am in this because businesses are going to use TONS of these cards/GPUs in their "desktop supercomputers" for analyzing all sorts of things throughout their domain. Mark my words: companies that miss this opportunity (to leverage CUDA and parallel processing) are going to find themselves looking like Blockbuster as compared to Netflix now.

The margins on Nvidia's "Quadro" business-oriented line of cards is likely higher than that on the consumer "GeForce" line, and certainly their Tesla dedicated supercomputing-desktop devices are going to be money makers as businesses figure out how to use these (i.e., find talented developers to help them write some seriously cool parallel-enabled algorithms and applications software). This may take a while yet, but I would say that within 3-5 years, MOST serious business applications will make some use of CUDA and/or GPUs for heavy analytical processing.

Bottom line: NVIDIA HAS SOME AWESOME GRAPHICS CARDS TO CONSIDER, and some nicely updated tools to go with them!

Nvidia CUDA Toolkit 3.1 - with Fermi card optimizations

The latest Nvidia "Fermi" GPUs (Graphical Processing Units) are making their way to the stores now by way of the latest Nvidia Graphics Cards that are definitely worth a look if it has been over a year since you upgraded your graphics card - the processing power per watt now is just unbelievable!  And, the latest Nvidia CUDA Toolkit 3.1 release has some features specific to the new Fermi cards and architecture that you may want to check into; just in case you are into GPU programming for fun.

Nvidia (NASDAQ:NVDA) has moved to a modern 40nm architecture for these new GPUs, which has allowed them to be much more power-efficient while cranking out tons of graphics horsepower for gaming and/or professional applications that make use of their stream-processors (aka, "CUDA cores") on the graphics card for high-performance computing (HPC) via massively-parallel-processed algorithms.  CUDA is NVIDIA’s parallel computing architecture that enables dramatic increases in computing performance by harnessing the power of the GPU (graphics processing unit) for applications including image and video processing, computational biology and chemistry, fluid dynamics simulation, CT image reconstruction, seismic analysis, ray tracing, and much more.

Get your NVidia Fermi Graphics Card
First, get hold of a new Fermi-based Nvidia CUDA Graphics card to develop and run your new CUDA applications on.  There are some really great cards out now that offer some really nice punch for the buck (aka, "price-to-performance ratio"), including these:

• Nvidia Geforce GTX 460 - a very reasonably priced (~ $200.00) super-powerful mainstream / desktop graphics card (targets gamers mainly) that smokes every other card on the market in this price range.  This card offers 336 CUDA processing cores and a Gigabyte of RAM to run your new Nvidia CUDA Toolkit 3.1 applications on.
• The brand new professional-class NVidia Quadro 4000 (NOT to be confused with the old Quadro FX 4000!) -- this ~$1000 card has 256 CUDA cores coupled to 2GB of GDDR5 RAM and is well suited to apps like CAD, Photoshop CS4 / CS5, and other CUDA-enabled professional apps. The card is quite power-efficient at only 142 watts max.

Now you can start putting some new CUDA abilities to work...

Nvidia CUDA Toolkit 3.1 Release Highlights

• GPUDirect(tm) gives 3rd party devices direct access to CUDA Memory
• Support for 16-way concurrency allows up to 16 different kernels to run at the same time on Fermi architecture GPUs
• Runtime / Driver interoperability enables applications to mix-n-match use of the CUDA Driver API with CUDA C Runtim and math libraries via buffer sharing and context migration
• New language features added to CUDA C / C++ include:
• Support for printf() in device code
• Support for function pointers and recursion make it easier to port many existing algorithms to Fermi GPUs
• Unified Visual Profiler now supports both CUDA C/C++ and OpenCL, and now includes support for CUDA Driver API tracing
• Math Libraries Performance Improvements, including:
• Improved performance of selected transcendental functions from the log, pow, erf, and gamma families
• Significant improvements in double-precision FFT performance on Fermi-architecture GPUs for 2^n transform sizes
• Streaming API now supported in CUBLAS for overlapping copy and compute operations
• CUFFT Real-to-complex (R2C) and complex-to-real (C2R) optimizations for 2^n data sizes
• Improved performance for GEMV and SYMV subroutines in CUBLAS
• Optimized double-precision implementations of divide and reciprocal routines for the Fermi architecture
• New and updated SDK code samples demonstrating how to use:
• Function pointers in CUDA C/C++ kernels
• OpenCL / Direct3D buffer sharing
• Hidden Markov Model in OpenCL
• Microsoft Excel GPGPU example showing how to run an Excel function on the GPU

Financial Opportunities - Nvidia (NASDAQ:NVDA) stock?
Since this blog also focuses on stock-market and investing opportunities, I have to contemplate whether the new Nvidia Fermi cards are going to drive substantial sales/revenue-gains and associated profit-gains for Nvidia corporation.  I can not help thinking that it is inevitable, especially when so many of the online retailers I went to in search of a new Nvidia GTX 460 card from were out of stock, backordered, and so forth.

And, these cards are out there already... people lucky enough to have gotten hold of them already are essentially uniformly impressed and satisfied with the performance of the GTX 460 card.  I have read all sorts of reviews from buyers saying how these cards have set a new standard in desktop gaming performance (frame-rates, etc) while also being rather reasonable in their power consumption.  Nvidia allows for running two cards together (in SLI-mode) for even higher performance, and from all the tests and reviews I have read: wow... these are FAST!

So, it seems to be nearly a guarantee that Nvidia is going to move a LOT of these cards.  The question is: at what margin?  They are being VERY competitive and aggressive with their pricing model, which suggests that margins may not be TOO large, but I do not know.  I will assume they are being sold for a profit, and that with enough volume, their margins will also be pretty decent.

And, then there is the super-computing and professional market: THAT is what I am more interested in from an investing standpoint.  These cards are being used in the top of the line supercomputers and high-performance computing systems and clusters, where a single super-computer may use 100s or 1000s of these cards in it.  And, Nvidia's top Quadro 6000 graphics card lists for $6,000 -- targetting digital production firms (think: Adobe Photoshop and Premier e.g.) and engineering firms doing real-time 3D work and the like.  These firms WILL buy the new Fermi-based cards in order to gain efficiencies at their firms (since these cards are up to 8-times faster than the prior generation; meaning: much time saved when rendering, etc).

Sure, the economy is "slow" right now, but what better way for companies to gain efficiency for a reasonable sum?  Move some processing off to new super-powered Nvidia GPUs!  If your employees spend less time waiting for computing operations to complete, perhaps you can get by with less employees (note: none of us like the sound of that, but it IS what helps drive "productivity"'; I'd just prefer seeing and freed-up employee time being redirected toward more creativity and product design and improvement, etc).

Bottom line: NVIDIA HAS SOME AWESOME GRAPHICS CARDS TO CONSIDER, and some updated tools to go with them!

Stock Analysis: Nvidia (NASDAQ:NVDA) and Ford (NYSE:F) Company Performance

I just have to compare and contrast the operating results reported by two companies today (Nvidia and Ford Motors), and especially discuss a fundamental difference between these two companies' management philosophies and ability to adapt to changing markets. That difference is a large one, and an obvious one: market agility and anticipating the future - Nvidia does this well, Ford does not.

Both have faced similar challenges with regards to their products and markets lately: their markets have been challenged by competitors (for Nvidia, that mean AMD / ATI; for Ford, just think Toyota, Honda, etc.), consumers have been demanding different products (for Ford, that means higher mileage cars; for Nvidia, that means higher performing graphics processors and more and improved Notebook graphics chip functionality), and each company has seen their products caught in a period of transition.

But, where they differ dramatically is in how they reacted to these challenges, and how quickly they reacted to known changes as well as anticipating uncertainty.

Let me start by quickly presenting an excerpt from earnings reports for each company - first Ford, then Nvidia:

NEW YORK (CNNMoney.com) -- Ford Motor reported a $3 billion operating loss in the latest quarter, and said Friday it would reduce staff and capital spending in order to preserve its dwindling cash.

Ford said it would cut salaried employment costs by 10% - reducing compensation of its white collar workers by eliminating merit pay, bonuses and the company's matching contributions to their retirement accounts.

Now, Nvidiia's quarterly results release:

SAN FRANCISCO, Nov 6 (Reuters) - Graphics chipmaker Nvidia Corp reported quarterly results on Thursday that topped Wall Street's estimate, as the company held the line on expenses while sales fell, and shares surged 12 percent.

Nvidia (nasdaq: NVDA) said its fiscal third-quarter net earnings for the quarter ended Oct. 26 came in at $61.7 million, or 11 cents a share, down 74 percent from $235.7 million, or 38 cents a share, in the year-ago period.

But after excluding items related to the company's recently announced layoffs and other charges, Nvidia earned 20 cents a share, beating the average analyst estimate of 12 cents a share, according to Reuters Estimates.

Notice the key differences.

Nvidia

Nvidia has been profitable in the past, and REMAINED profitable by acting quickly to reduce costs in light of the fact that demand was slowing and product-mix changes were taking place. In September, the company announced plans to cut 6.5 percent of its workforce when it was obvious their market was changing, and this quick reaction has helped them avoid losses.

You can argue that a graphics chip maker is inherently more agile than a car company because of the product lead-time and engineering/manufacturing cycle brevity compared to building cars, but on the other hand, that same logic must be applied to *competitors* that can also move just as fast to change their lineup of graphics cards and GPU offerings. So, Nvidia deserves credit for having management that is willing to act quickly, and decisively, in order to keep products inline with consumer expectations while keeping costs down as sales volume deteriorates a bit during a macroeconomic slowdown.

In addition, Nvidia shows foresight for future demand and growth in the parallel computing field with their CUDA offerings (this CUDA parallel processing cores feature is in nearly all their current GeForce and Quadro product lineup). I have been watching more and more commercial applications target this particular Nvidia platform advantage (which, I consider to be VERY large), and have seen applications like the upcoming Adobe Creative Suite CS4 even marketed as being best-with-Nvidia cards (and, Nvidia has a nice new high-end Quadro video card marketed specifically to Adobe CS4 users). This is great product planning, and will give them sales for many quarters to come.

And, if this were not enough reason to consider Nvidia a company with great foresight and momentum, consider how Apple Computer has just started offering the Nvidia mobile GeForce chipsets / GPUs in their new notebook lineup. As soon as I heard that news, it was even more obvious that Nvidia is making moves to future-proof their sales and grow their markets.

Ford

Ford has been losing money for a long time, and even while losing money, their management consistently acts slowly to reduce costs in light of the fact that demand was slowing and competitors were eating them alive. Ford Motors seems destined to live up to the image of the American Auto in general: outdated and behind the competition. Their entrenched management (just like General Motors) is a bunch of overpaid executives whose only tangible "plan" to fix things of late is to borrow (or be handed) more money from the United States Government and taxpayers.

I won't get into how lame this whole "rewarding failure" concept is with giving incentives to companies that fail is (instead of rewarding those that succeed and produce profits and jobs!), but it is highly indicative of the underlying problem with Ford and GM. They still, even after massive multi-Billion dollar repetitive quarterly losses, fail to act BEFORE the crisis gets worse, and are never able (or willing) to get ahead of the curve and show that they have any sort of visionary management abilities.

It is not like they have no knowledge of where the consumer is headed, where the economy is heading, and how their products stack up against the competition in regards to quality, features, mileage, and the like (start by reading Consumer Reports guys!). Nvidia has the same knowledge of what their competitors are doing (AMD's ATI division especially), and they must act to counter such competition quickly - and they do!

Ford looks like a slug by comparison to Nvidia, and repeatedly fails to deliver REAL change, and by the time (if ever it occurs) their management makes the DEEP CUTS and GROUNDBREAKING CHANGES required, they will have burned all their cash and find themselves once again knocking on Uncle Sam's door in hopes of more easy money. This door-knocking must be ignored, as it only encourages a repeat of their lame decisions, and will reinforce poor management "vision" while essentially penalizing those auto companies that DO have good vision for their companies' futures (by essentially subsidizing inferior products from Ford or any other competitor that is using government funds to prop-up their business).

Summary
Well, I think it is time to let some of the Silicon Valley management have their shot at Ford Motors management - or at least some of their foresight needs to go into play in Detroit. Funny thing is, the one Silicon Valley motor company I really find interesting (Tesla Motors) has breakthrough forward-looking products (like fully Electric sports cars), is creating new Auto-industry jobs, but yet has to compete with Detroit's entrenched industry without the massive handouts and support packages from the US government.

Instead, Tesla must secure private funding at market rates, and actually produce a product people want and set a price-point for its products where it can stay in business and perhaps even post a product. Gee, what an idea! All the more reason Detroit's antiquated auto industry needs to be allowed to simply DIE if they can not do the same -- it is time to reward success, and not failure. And, with regards to today's stock analysis, I plan to reward Nvidia's success with further purchase of their products (I have a Quadro in mind for before EOY 2008 yet) and their stock; but, by contrast, there is ZERO chance I will be buying a Ford product or company stock!

Intel Solid State Disk Drives - Awesome!

For years I have been awaiting the release of SSDs (Solid State Disks, aka Solid State Drives, aka RAM disks) that were affordable, offered great performance, require nearly zero power consumption to operate, and come from a "mainstream" technology company in a standard interface (SATA).

Well, you can't get more mainstream than Intel, and Intel has just recently announced / released their first SATA Flash Memory SSD products (production to start this month - September 2008):

• Intel® X18-M Mainstream SATA Solid-State Drive (80GB Capacity - 1.8" Form Factor)
• Intel® X25-M Mainstream SATA Solid-State Drive (160GB Capacity - 2.5" Form Factor)
• Intel® X25-E Extreme SATA Solid-State Drive (32 and 64 GB Capacities - 2.5" Form Factor)

You just have to see the specifications to believe them (see below), as these drives are just unreal! Check out that super-low power consumption, and those blazing fast speeds! Oh my god I am looking forward to building the most insanely fast disk-array ever!

And, I have already purchased, and I am purchasing more, Intel stock (NASDAQ:INTC) based on the potential I see for this product. If you have been following the news about the issues large data-centers in the USA are experiencing, with regards to power-density (and not being able to get enough physical electrical power to run all the servers in a room) and the related cooling issues (again, buildings housing server rooms are being tapped-out with regards to the air-conditioning and climate-control systems, and the power to run those systems), you will see why I am so excited.

Ignoring all other performance factors of these new flash-memory standard SATA-interface hard-drives, I still see massive potential for companies to swap-out their existing spinning-disks (i.e., traditional hard-drives) in there data-centers and replace them with these new Intel Solid State Drives. It's a "no brainer" in my opinion. And, I am just talking data-centers here, and for simple power-density and heat (or lack thereof) considerations.

I have yet to even describe the incredible performance gains that some applications, like databases and the like, are going to realize from these drives. Physical IO (Input-Ouput) has always been the Achiles heel of any high-end server, as disks (in general) can not even come close to keeping up with the speed of today's modern processors (like the latest Intel and/or AMD 64-bit and multi-core offerings). But, with these new Solid-State-SATA-Drives, things are about to change... the power of the CPU(s) is about to be unleashed! I am picturing incredible new database applications, analytics, and more.

And, if that wasn't enough, couple all this new super-speed-storage from Intel with technologies like NVIDIA's CUDA - which gives programmers the ability to exploit the incredible parallel processing power of the modern NVIDIA (NASDAQ:NVDA - another stock I am betting on for the coming years) GPUs through their CUDA development framework (essentially, Nvidia's extensions to C language). With CUDA, Nvidia has a product offering that is not just interesting, but substantially amazing, in its potential for massive financial analysis applications, data-analysis programs, and the like - by offloading compute-intensive (especially mathematical and matrix type) operations to the GPU. Nvidia only requires you own one of their GPUs from this list of CUDA-enabled Nvidia Products (GeForce 8 series ++ basically, or one of their "Tesla" servers). Oh, and what kind of processing power am I talking about here? Well, the new NVIDIA S1070 server (1U form factor too) has 4 Teraflops of performance (yes, 4 TRILLION OPERATIONS PER SECOND!). Holy $%*! So, I need only learn how to couple these new Intel SSDs and CPUs with this NVIDIA technology, and Skynet will be born! (a Terminator movie reference folks. lol.)

If that was not enough, I also guarantee that gamers (by the millions) are going to want these drives. Gaming afficionados have no problem dumping thousands on their "ultimate gaming computers" just to get a 10% speed-advantage over someone else. They will spend a thousand bucks on two new Nvidia GeForce SLI graphics cards for maximum frame-rates, and extra money on the fastest DDR RAM, and up until now, extra money for traditional RAID implementations (i.e., multiple hard drives with multiple spindles moving simultanesouly to serve up not just redundancy-enhanced data-storage, but speed given the right confiuration). Well, there is nothing even close to the speed of these new Intel SSDs when it comes to random reads, sequential reads, and nearly anything else. And, latency? Nearly non-existent!

From the Intel web-site...:

Technical specifications (Extreme SSD)

Model name	Intel® X25-E Extreme SATA Solid-State Drive
Capacity	32GB and 64GB
NAND Flash components	Intel® Single-Level Cell (SLC) NAND Flash Memory 10 Parallel Channel Architecture with 50nm SLC ONFI 1.0 NAND
Bandwidth	Sustained sequential read: up to 250 MB/s Sustained sequential write: up to 170 MB/s
Read Latency	75 microseconds
I/O Per Second (IOPS)	Random 4KB Reads: >35,000 IOPS Random 4KB Writes: >3,300 IOPS
Interface	SATA 1.5 Gb/s and 3.0 Gb/s
Form factor	2.5" industry standard hard drive form factor
Compatibility	SATA Revision 2.6 Compliant. Compatible with SATA 3 Gb/s with Native Command Queuing and SATA 1.5 Gb/s interface rates
Life expectancy	2 Million Hours Mean Time Before Failure (MTBF)
Power consumption	Active: 2.4W Typical (server workload¹) Idle (DIPM): 0.06 W Typical
Operating shock	1,000G / 0.5ms
Voltage	5V SATA supply rail
Operating temperature	0°C to +70°C
RoHS compliant	Meets the requirements of EU RoHS Compliance Directives
Product health monitoring	Self-Monitoring, Analysis and Reporting Technology (S.M.A.R.T.) commands, plus additional SSD monitoring

Technical specifications (Mainstream SSD)

Model Name	Intel® X18-M Mainstream SATA Solid-State Drive Intel® X25-M Mainstream SATA Solid-State Drive
Capacity	80GB and 160GB
NAND Flash Components	Intel® Multi-Level Cell (MLC) NAND Flash Memory 10 Channel Parallel Architecture with 50nm MLC ONFI 1.0 NAND
Bandwidth	Up to 250MB/s Read Speeds Up to 70MB/s Write Speeds
Read Latency	85 microseconds
Interface	SATA 1.5 Gb/s and 3.0 Gb/s
Form factor	X18-M: 1.8" Industry Standard Hard Drive Form Factor X25-M: 2.5" Industry Standard Hard Drive Form Factor
Compatibility	SATA Revision 2.6 Compliant. Compatible with SATA 3.0 Gb/s with Native Command Queuing and SATA 1.5 Gb/s interface rates
Life expectancy	1.2 million hours Mean Time Before Failure (MTBF)
Power consumption	Active: 150mW Typical (PC workload¹) Idle (DIPM): 0.06W Typical
Operating shock	1,000G / 0.5ms
Operating temperature	0°C to +70°C
RoHS Compliance	Meets the requirements of EU RoHS Compliance Directives
Product health monitoring	Self-Monitoring, Analysis and Reporting Technology (S.M.A.R.T.) commands plus additional SSD monitoring

I have looked at other proprietary SSD / RAMDISK solutions over the years, and they were always insanely expensive to say the least, and the performance just wasn't worth the investment. In the past, most drives were tiny by comparison (4GB, 8GB, and maybe 16GB) for insane prices (yes, thousands; even for tiny capacities).

But now, with these new mainstream Intel SSD drives, things are about to change in a big, industry-transformational, way (and, in a "green" way too - from massive power savings and reduced cooling requirements). You wait, and you will see. The end of the hard-drive as we know it is coming, and SSDs are going to become the "standard", and I have a feeling Intel is going to lead the way with Solid State Drives using their dominant technology brand and value-proposition.

I will be anxiously awaiting the shipment-date so I can get my hands on a few of these drives. The possibilities are just too wonderful to not try these drives out in my production systems! And, even for the higher price (which, I expect will fall considerably as early as next year when production ramps up), I should see instant returns on investment by replacing my disk-drives that I launch VMware Virtual Machines from, drives that I have SQL-Server database on, and so on. My electric bill is certainly going to show some downward movement!

If you want more detailed information, I found a nice article at TG Daily - Intel SSD Discussion.

UPDATE:
Intel announced a $595 pricetag for the mainstream 80GB SSD, which comes in 1.8-inch (X18-M) or 2.5-inch (X25-M) sizes for the same price; the 160GB version is coming later this year (Q4), and pricing has not yet been announced.

For this price, I will definitely grab a couple and put them to use in my primary desktop and notebook to see how they perform. If they meet expectations, I'll soon thereafter upgrade my servers.

I saw that PC Connection has a web-page listing the SKU for the 80GB mainstream model (Intel X25M) for $731.35 (though, the Manufacturer Part Number (SKU) being shown is SSDSA2MH080G1C5, which is supposedly a 5-pack, is obviously incorrect). I expect other vendors to have pricing online soon, and to hit Intel's target pricetag (or perhaps undercut a bit for the large volume dealers, since Intel's pricing was for 1000 or less drives - which suggests a volume disount above 1000 drives). Heck, some data-centers easily have over 1000 drives, if not 10,000... perhaps they should buy direct from Intel :)