Just as Intel’s (NASDAQ: INTC) CEO Brian Krzanich opens the regular staff meetings before a dramatically reduced IDF2015 Shenzhen conference, it is a good time to review how government and enterprises don’t see eye to eye when it comes to strategic business.

China’s Tianhe-2 supercomputer is world’s fastest supercomputer, at 33 PFLOPS demonstrated and 55 PFLOPS theoretical performance.

Remember the Tianhe-2 machine at Guangzhou Supercomputer Center, the current World’s number one according to Top 500 Supercomputer list? Unlike some other China supercomputers – Tianhe-2 is fully Intel based machine,  the world’s largest assembly of Intel Xeon CPUs and Xeon Phi accelerators.

Even after Intel ‘opened the kimono’ and gave a nearly 70%  discount on its processors and accelerators, it has given Intel, and therefore US technology sector a major foothold in China and Asian region as such. Over the course of past two years, we were involved in a lot of discussions with Intel staff who were not privy to see the financial impact of the deal — and even argued our undoubtedly solid information. We’re not here to report how things should be, or are in marketing and investor presentations to its numerous staff, but how things really are.

During 2015, the Tianhe-2 supercomputer was supposed to be doubled in its size, up to 110 PFLOPs peak, again using the very same Intel processors and accelerators. Since now these are mature products with lower real manufacturing cost for Intel, they could finally make some real money.

Well, it was not to be: our tweety bird from the window chirped to us that Uncle Sam has put this supercomputer centre, together with National University of Defense Technology in Changsha, the system’s creators, and Tianjin centre, among others, on so a so-called “Denial List”, which prevents any high technology from the USA to be sold to these sites. Our sources used even harsher words.

Knowing that these several sites alone are expected to order some 250+ PFLOPS of compute in the next few years (around 500,000 top-end Broadwell-EP Xeon E5v4 processors, or  approximately $1 billion high margin list price) and they were THE Intel friendly ones, this is quite a loss to Intel, thanks to Uncle Sam.

But, what’s worse strategic loss in time is that, based on this decision as an excuse, indigenous China high end processor architectures can now push the government to gradually remove any dependence on US. This means just one thing: an AMD or Intel x86 processor technology is increasingly becoming errata non grata. Should the Chinese government react in force, it will give the Chinese vendors the blank check support to go all the way a developing their Alpha, POWER and MIPS processors for both the government and the mainstream commercial use.

You may think they are not up to the mark, but remember how fast British ARM architecture became the dominant processing architecture in the world. And this group doesn’t need to worry about the antiquated x86 ISA, worry about satisfying the dumbed down shareholder masses, or overpaying their marketing and sales staff, as well as the fat check, golden parachute-protected CxOs.

They have taken the best that the USA has developed (some of key Alpha, GPGPU and MIPS architects left US over the course of past four years, a lot of them due to non-renewed visas) and discarded due to corporate shenanigans, and the continued developing it much farther than anyone expected both on hardware and software side.

Five years ago, ShenWei showed a CPU that performed faster than the fastest GPUs of the time. Now, fifth generation is approaching, slotting between Tesla and FirePro GPGPUs and next-gen Xeon Phi accelerators. However, this is not an accelerator or a GPGPU – this is a CPU.

So, thanks to Uncle Sam, China might not have a 110 PFLOPS Intel based supercomputer but it definitely will launch a 100 PFLOPS system based on upcoming 64-core, TFLOPS-class ShenWei Alpha, with true blue CPUs possibly faster per socket then even the next generation Xeon Phi or Volta/Pascal-based Teslas.  Next, of course 100 PFLOPS Chinese POWER8 or 9 — (thank you IBM) and then possibly even Loongson MIPS – -it may come back into the high end field with renewed government support because of this Uncle Sam move. All are clean, elegant, scalable high end RISC architectures.

So who are the winners and losers from this?

NUDT and Tianhe may be the losers for now, but only short term. They will simply speed up their HPC ARM plan.

Intel comes out the big loser from this and a lot: who will want to do a phased deployment large x86 machine in China now, and worry about future phases? Then comes Uncle Sam himself: they lost even that little bit of influence on the high end China HPC. How is that for “cutting your nose to spite your face?”.

VR WORLD’s  Analysis: US government moves accelerate the Chinese CPU roadmap while curtailing juiciest sales for Intel and other US vendors.

The post Uncle Sam Shocks Intel With a Ban on Xeon Supercomputers in China appeared first on VR World.

One of the most important advances in computing over the last decade has been the ability for programmers to properly harness the GPU for computing in their code. With this, and the massive increases in floating point power thanks to multi-core CPUs like the Intel Xeon E5v3, the amount of computer power available to users and institutions has spiked.

But there’s a problem. The link that binds the CPU and GPU, which in the vast majority of cases is PCI Express, is a bandwidth bottleneck. The full potential of the CPUs and GPUs cannot be realized since PCI Express has a maximum of 20GB/s achievable net bandwidth and up to 1 microsecond roundtrip latency.

But might AMD (NYSE: AMD) have a solution for this?

To read the full story, head to VR World, Bright Side of News* Asia-Pacific sister site.

The post Might AMD Hold The Solution For Connecting CPUs and GPUs? appeared first on VR World.

Qualcomm has been fairly quiet about their high-end ambitions after what is expected to follow the soon-to-launch Snapdragon 805 chipset. The Snapdragon 805 is Qualcomm’s chip that will likely ship in devices next quarter and is marketed by Qualcomm as their 4K chip with the Adreno 420 GPU.

Now, even though the Snapdragon 805 (APQ8084) is a very powerful chip, it lacks 64-bit capability and doesn’t have an integrated modem, requiring a separate modem like Qualcomm’s 20nm MDM9x35 to enable cellular capability. It also sports an improved Krait CPU with a Krait 450 CPU compared to the Snapdragon 801 and 800’s Krait 400. However, it still doesn’t quite satisfy Qualcomm’s need for a very powerful chip that looks into the future. The Snapdragon 805 is the chip for now.

That leads us to today’s announcement of the Snapdragon 810 and Snapdragon 808 chips. The Snapdragon 810 and Snapdragon 808 mark the first time that Qualcomm has ever announced a high-end 64-bit chip and the first time they have implemented ARM’s own cores into their SOC. Both the Snapdragon 810 and Snapdragon 808 will feature ARM’s A57 and A53 64-bit cores with the Snapdragon 810 being a 4 big + 4 LITTLE and the 808 being 2 big + 4 LITTLE in a big.LITTLE configuration. This marks a pretty big shift in the company’s attitude towards big.LITTLE when you consider that they were saying that their cores were good enough that they didn’t need to do a big.LITTLE configuration. And the fact that Qualcomm now has a high-end SoC that supports the ARMv8 64-bit instruction set architecture. Keeping in mind that they are already shipping the Snapdragon 410 to the middle of the market with an A53-based 64-bit CPU.

To me, and many others, it seemed like Qualcomm was poo pooing what Samsung and others were doing, and now they’re effectively doing the same with these two chips’ CPU cores. However, Qualcomm explained that they were reacting to market demand for 64-bit capable chips and that they wanted to deliver them to their customers as requested. They said that because their own custom 64-bit ARM CPU cores weren’t necessarily timed for what the market has demanded, they have adjusted and are delivering a 64-bit solution as quickly and effectively as they can with ARM’s cores. This can all be traced back to Apple’s release of the A7 64-bit chip in the iPhone 5S and how much they fundamentally shook the foundations of the mobile SoC world and what people were demanding.

In addition to the 64-bit ARM A57 and A53 cores, Qualcomm is integrating their modem directly into the SoC’s die, bringing the SoC and modem into one chip once again. Because they are integrating their 20nm MDM9x35 chip, they already know what they can expect to see from the modem side of things in terms of performance and power savings. Additionally, because Qualcomm has been working on 20nm with their modem, they could proof and tweak the process in anticipation for these Snapdragon chips. Because they both share the same modem, they will both be capable of Cat 6 LTE enabling LTE Advanced features globally with a single design. When paired with Qualcomm’s RF360 front-end solution, both chips will be capable of 3×20 MHz Carrier Aggregation, resulting in up to 300 Mbps download speeds in various spectrum scenarios.

The Snapdragon 810 will also be one of Qualcomm’s 4K-focused chips with the introduction of a 4K HEVC hardware video encoder as well as a 14-bit dual ISP (image signal processor) capable of 1.2 GP/s and supporting image sensors up to 55 MP. And because Qualcomm wants to focus on 4K so heavily with the Snapdragon 810, they once again bumped the GPU to an Adreno 430 which should bring 30% faster graphics performance when compared to the Adreno 420 in the Snapdragon 805. They also took a very serious look at the GPGPU compute capability of the Adreno chip and improved it by 100% all while reducing power consumption by 20% when compared to its predecessor. The Adreno 430 will also support OpenGL ES 3.1 which was recently announced as a standard, even though I suspect many of their currently available OpenGL ES 3.0 capable hardware could be updated to 3.1 via driver update as well. In addition to OpenGL ES 3.1 support, the Adreno 430 will also support hardware tessellation, geometry shaders and programmable blending. The Snapdragon 810 also will be Qualcomm’s first SoC to support LPDDR4 memory which means better performance and lower power consumption for high-end devices utilizing this SoC.

One of my biggest complaints about many SoC vendors as they’ve started to try to attach themselves to 4K is the fact that none of these GPUs, no matter how powerful, are capable of 3D graphics in 4K. The amount of horsepower it takes to do 3d graphics in 4K is absolutely insane and is far outside of the realm of these mobile SoCs, for now. As such, we were given clarification that most gaming in 4K on these devices will be done at 1080P and upscaled to 4K on whatever display it’s being played on. So, whenever you hear 4K gaming being mentioned, that’s likely what you are going to be seeing, even though its really 1080P gaming being upscaled to 4K. After all, almost all of the mobile games are built for 720P or 1080P and not anything much higher.

In contrast to the Snapdragon 810, the Snapdragon 808 is a very similar chip to the Snapdragon 810 but it takes more of a smartphone focus rather than a 4K tablet one. Even though, either chip can be used for a smartphone or tablet, it would just be more cost prohibitive to put a Snapdragon 810 in a smartphone. The Snapdragon 808, as we stated earlier is a slightly differently configured chip with a dual ARM Cortex-A57 plus quad Cortex-A53 CPU configuration, so technically it’s a sexacore or hexacore chip rather than an octocore like the 810.

The Snapdragon 808 takes more of a focus on future smartphone resolutions with a targeted design for WQXGA resolution of 2560×1600 (close to the Oppo Find 7) even though that phone has a 2560×1440 resolution, it’s essentially the same resolution. It does, however, have an Adreno 418 GPU, which actually puts it below the Adreno 420 and Adreno 430 which are both targeted towards 4K resolutions. The Adreno 418 claims to deliver 20% faster 3d graphics than the Adreno 330 in the Snapdragon 800, giving a slight performance boost to smartphones even though they’ll be handling much higher resolutions. In fact, the 2560×1600 resolution is double the pixels of the current 1920×1080 on most smartphones, which in my eyes, should justify more than a 20% improvement over the current generation. The doubling of pixels is especially important when you consider exactly when this SoC is expected to be shipping in devices.

The Snapdragon 808 will also support 12-bit dual ISPs as opposed to the Snapdragon 810’s 14-bit dual ISPs, resulting in less performance available to the image sensors to utilize. It will also support LPDDR3 rather than LPDDR4 support in the Snapdragon 810.

Now, if you take into account the Snapdragon 810 platform as a whole, you would be looking at an 8-chip solution, assuming that you were to go with Qualcomm’s RF360 RF front-end (WTR 3925 and WTR 3905) as well as their PMICs (power management integrated circuits) and their 2×2 WLAN chip (QCA6174A) that enables 802.11ac and MU-MIMO. They also have an NFC chip (QCA1990) and an audio codec that round out their full Snapdragon 810 platform, which is if you want to go Qualcomm the whole way.

Based on what I see here, I would say that the Snapdragon 810 will likely be a pretty attractive part for many looking to win the spec wars against their competitors and that we’ll probably see it used in both phones and tablets. I say this primarily because of the fact that even though the Snapdragon 808 is a slightly dialed down version of the 810, I’m simply not convinced that it will be powerful enough to deliver a smooth experience at 2560×1600 and to enable what I’d expect to be upscaled gaming (in the short term). The Snapdragon 810 really looks to be a great part, but it will be interesting to see where it will fit in their stack once Qualcomm eventually shows off their own custom 64-bit CPU cores, the successor to Krait. Judging by Qualcomm’s own marketing they really seem to be focusing on the Snapdragon 810, and I can totally see why. I think we will see more devices launching with the 810 than we will with the 808, even though the 808 will be in a more affordable price range. A flagship chip is a for flagship devices and manufacturers want to make sure they are putting in the best possible at that time.

Both the Snapdragon 810 and Snapdragon 808 are expected to be sampling in the second half of 2014 and shipping in devices in the first half of 2015.

The post Qualcomm Announces 20nm Snapdragon 808 and Snapdragon 810 64-Bit Chips appeared first on VR World.

If I recall correctly, a while ago AMD claimed world’s first OpenCL demo, but it was done on a single core (and then scaled up to all four) on a Phenom II X4 CPU. If this video is correct, Nvidia gets the pole position for being the first company to demonstrate OpenCL working on a GPU, which is “usage as intended”.

Judging from the video, Nvidia showed Nbody simulation changing following parameters: point size, velocity damping, softening factor, time step, cluster scale, and velocity scale. The company used a laptop equipped with Quadro FX 570M graphics card, e.g. GeForce 8600M after a “GL” tune-up. As far as the official debut of OpenCL go, KHRONOS Group launched the API on Siggraph Asia 2008 in Singapore, from where both AMD & Nvidia videos came from.

But don’t expect OpenCL drivers until both companies supply drivers to Apple. It looks like Snow Leopard will be the beginning of OpenCL on a PC platform and Windows Vista is sitting on backburner… that’s what Microsoft get for botching OS development and sucking up to allmighty Chipzilla.

The post Youtube video shows OpenCL running on Nvidia GPU appeared first on VR World.

Fast forward to January 2009, and we have SP performance at 933 GFLOPS, while achievable DP performance dipped down to 78 GFLOPS. This figure is roughly half of what Nvidia boasted about at the time of launch, and sheer evidence that both manufacturers like to overstate the performance of actual parts. What makes things interesting is the fact that Tesla GPGPU boards aren’t even most powerful parts in the Nvidia line-up. That “honor” goes to newly introduced GTX285 and 295. In professional line-up, Quadro FX 5800 has more “oomph”, thanks to higher shader clock. but even FX5800 will remain below 100 GFLOPS in dual-precision operations… making this GPU “just” 2.5x faster than quad-core Xeon processor.

Then again, if you activate parallel execution, CPU will drop to sub-10 GFLOPS values, while the GPU will remain at 78 GFLOPS for DP and 933 GFLOPS in single precision. At the same time, ATI’s architectural concept of “emulating” the DP units by pairing more processing units in one cluster resulted in actual peak performance of 900 GFLOPS for the 4870 part (claimed performance: 1.2 TFLOPS) and 250 GFLOPS for the Dual-Precision formats. This is an impressive difference, showcasing ATI’s lead from the architectural standpoint. Extractable performance is a bit different, since some ISVs managed to extract that performance, such as ElcomSoft password cracker, while some hit different walls and could not get better performance.

The real dilemma now is to wait and see what kind of computing performance lies with upcoming 40nm GPUs.

The post Nvidia’s discloses its DP performance limitations appeared first on VR World.

Under a not-so-scientific moniker Need For Speed Seminar Series, this 13-week course will feature domestic alumni such as Mark Hasegawa-Johnson, Dan Roth, Narendra Ahuja, Stephen Boppart, John C. Hart, Tom Huang and Seth Hutchinson, and guests such as Keith Thulborn (UI Chicago),  Sam Blackman (Elemental), Nikola Bozinovic (MotionDSP), Mark Johns (Tapulous) and Tim Sweeney (Epic).

More information is available if you visit the official site for Need for Speed seminars from Parallel@Illinois.

The post University of Illinois streams its Parallel@Illinois seminars appeared first on VR World.

The answer is “have no fear, ElcomSoft is here”. This Russian company is mostly known inside the security elements of governments around the world, such as police, Interpol, secret agencies and all. Recently, ElcomSoft introduced Password Recovery suite with CUDA-enabled cracking of passwords for Outlook PST files, archived files or various password protected documents. First and foremost, I will admit first of forgetting a password for my old PST files from the days on INQ, and Password Recovery helped me a great deal (“promise I’ll never use stupid passwords which I forgot, scout’s honor” ).

Today, ElcomSoft announced that the company expanded its support from Nvidia-only to support both ATI and Nvidia with its Wireless Security Auditor 1.0. This application is intended to audit wireless network security by “attempting to recover the original WPA/WPA2 PSK password protecting Wi-Fi Communications”.
Using ATI HD2000 and above or GeForce 8 and above cards, ElcomSoft WSA is currently the fastest Wi-Fi password recovery tool on the market. In order to support both GPU manufacturers, ElcomSoft developed its own patent-pending technologies which expanded on the original CUDA-powered algorithms.

As you cam see, 4870 beats GTX280 hands-down...

There are no limits to how much GPUs can you actually empower, and ElcomSoft is the first company that demonstrated that 800 shaders inside Radeon 4870 can beat 240 shaders inside GeForce GTX280. We’re surprised to see that AMD’s PR department isn’t beating the drums over this achievement, especially after seeing ATI parts trailing Nvidia ones in apps such as Folding@Home, artifact-free video transcoding, Photoshop CS4, Premiere CS4 and so on…

The price of WSA 1.0 is $1199, but until March 1st, 2009 – ElcomSoft offers a 50% discount – yep, $599. If you are a network administrator that wants to know just how secure your network is, the answer is quite simple.

The post Ultimate GPGPU hacking application is here! appeared first on VR World.

This specification covers all GPGPU-capable hardware, regardless of that hardware being in servers, workstations, desktops, notebooks or handhelds – if your GPU is able to compute, the manufacturer only needs to adopt OpenCL support in the driver and that’s that.

That should not be an issue, with AMD/ATI and Nvidia strongly standing behind the standard. Computing engineers on both sides bickered about Brooke+, CAL or CUDA in the past, but both makers are firmly behind OpenCL as the way for the future.

So far, companies that developed and ratified this initial spec include 3Dlabs, Activision Blizzard, AMD, Apple, ARM, BARCO, Broadcom, Codeplay, Electronic Arts, Ericsson, Freescale, HI, IBM, Intel Corporation, Imagination Technologies, Kestrel Institute, Motorola, Movidia, Nokia, NVIDIA, QNX, RapidMind, Samsung, Seaweed, TAKUMI, Texas Instruments and Umea University.

You can learn more at the official page of OpenCL API, but I wasn’t able to check the site as it was hammered down with numerous requests (this story was written at 10:05AM CET).

The post GPGPU is the future: Khronos releases OpenCL API appeared first on VR World.

As a part of AMD STREAM, ATI will release its own video transcoder for free. Given the limitations and performance that Badaboom has, I wonder did ATI decided to do something more in “formats supported” area.

On the Folding@Home front, there are great expectations from this upcoming driver, especially if you own Radeon 4800 series product. ATI worked hard to remove the driver overhead which keeps the CPU core to 100% per GPU, and as a result, folding smaller proteins should improve by some 20%.

Looking at a bigger picture, ATI’s GPGPU engineers are working night and day on the upcoming OpenCL platform, so the future of CAL and other proprietary standards is up in the air. It is obvious that any pre-4800 hardware suffers from serious drawbacks in the architecture, but 4000 and beyond has the potential to “rock on” in the GPGPU world. Radeon 4000 series finally brought several key features that Nvidia featured from GeForce 8 and beyond, so we are going to have a heated GPGPU battle in 2009.

One thing is certain, though. As soon as Catalyst 8.12 appear (according to our sources, look for Terry “Saint Catalyst Claus” Makedon coming down the chimney around December 10th), go to game.amd.com and download the drivers. There is a myriad of improvements, and you should check out Catalyst 8.12 driver with updated Folding@Home core, currently in version 1.2.1.

You can try the 1.2.1 client right now, but expect a final version showing up around the release of Catalyst 8.12. December is going to be an exciting month for ATI’s F@H community.

The post ATI and Folding working on F@H 1.2.1: Performance, stability updates appeared first on VR World.

The interview was focusing on Bjorn3D’s noteworthy Folding@Home effort (the team is on track to crack into Top100), thus Michael gave some interesting thoughts, such as this one.

There are a lot of very good guides out there that will walk users through the required steps to fold with multiple GPUs like the ones on HardOCP or overclockers.co.uk, just not in SLI mode yet.
NVIDIA SLI is a great extension to parallel processing and we’re looking at methods to take advantage of it with Folding@home. Stay tuned.

As you can read for yourself, there is a lot of things that Nvidia wants to do with Folding@Home. ATI is currently working on own reshuffling, but at the end of the day, whoever makes the best GPU for protein folding, always have a higher goal in your mind.
In my own opinion, Folding@Home is purest form of “pay it forward”, because GPUs helped scientists to accelerate their research measured in human generations. Couple of months ago, I spoke to Vijay and he said that GPU’s accelerated the protein research by some 30 years. Let’s hope that ATI and Nvidia can decrease that by additional 20-30-40 years to real-time levels.
Currently, most powerful ATI and Nvidia GPUs are folding in the range of 300-600 nanoseconds/day. Achieving a second per day is something I am looking forward to. In 10 years time, we’ll hopefully be at 1 hour of protein life/day on GPUs.
Just to help you calculate, 1TFLOPS of computing power would calculate 1 milisecond of protein life PER DAY. For one second of protein life per day, we need 1PFLOPS. Thus, even PFLOPS GPUs could not calculate real-time.
Anyways, you can read the interview here, and regardless of where you are, I am inviting you to join Folding@Home project. My team number is 69864, but you’re more than welcome to join any team, or even stay outside teams. But we need your idle computing power. If you feel up to contributing a little, consider Folding@Home.

The post Nvidia to bring SLI support for Folding@Home appeared first on VR World.

All of the images in gallery are 1000 pixels wide… if you would like to see wider or narrower galleries, it would be great if I could get some feedback and suggest a resolution
The reason for it is quite simple: unlike companies that are advertising personal supercomputers and charging them tens of thousand dollars (Cray CX1 and Tyan PSC come to mind) – Nvidia is targeting a price of less than 9,999.99 USD. Yep, that’s correct, 4+ TFLOPS supercomputer for less than 10 grand.
At the time of writing, I haven’t been able to found are those Tesla cards using older 65nm or new, power-saving 55nm chips. If we would judge by chips that power new Quadros, 55nm should be under the hood, but then again, Nvidia has a large nVentory of older 65nm chips, so the company might be inclined to clean them out.
Somehow, I would bet that this is not the only supercomputer announcement from companies in Red-Green-Blue area. After all, certain motherboards from ASUS and GigaByte could fit six single-slot cards… can you say FireStream and 6 TFLOPS?
To end this article, here is some food for thought: Nvidia got inspiration to create this system after seeing tens of enthusiasts build their own supercomputers. 11 years ago, overclockers and gamers jump-started the IT industry we know today by overclocking and modding their own systems.
Can this underground movement turn desktop computers into handhelds and supercomputers onto desktops? Only time will tell. After all, what CPU, GPU or motherboard manufacturer does not encourage overclocking or offer an overclocking/gaming product line? After all, all of these products are based on overclocking Celeron 300A to 450 MHz, Dual-Celeron Abit BP6 and so on. Makes me wonder who will remember those crazy doctors that created FASTRA 10 years from now.

UPDATE Nov 18, 01:23AM CET: For some odd reason, gallery didn’t work – uploaded the pictures again. It should work now.

The post UPDATED: Nvidia set to launch Tesla powered Personal Supercomputer appeared first on VR World.

The post GPGPU Revolution: OpenCL to launch next week appeared first on VR World.

Nvidia's Quadro CX is getting a twin brother...

This is also the reason why Quadro CX and the upcoming 4800/5800 line-up feature a model geared to 32-bit and a model geared towards 64-bit users. If you want all the latest and greatest that 64-bit world can offer you, you will go for the Quadro FX 5800. If you need broadest compatibility, you will go with the CX/4800, and the problem is solved.

Thus, CX/4800 will be nothing else but a souped up GTX260 cards with 1536 MB of memory, while the FX5800 will be that monster with 4GB of memory.

All in all, I have to agree with Tim Sweeney when he announced his disappointment with Microsoft Vista – we had the ideal chance to bring 64-bit operating system to life and get rid of this duality that only drives application engineers to the ground. I guess we’ll have to wait for Windows 7 to finally wave goodbye to graphics cards that can’t have 2GB of memory or more…

The post Nvidia plans to bridge the 32-bit and 64-bit divide appeared first on VR World.

The reason for GeForce dominance lies in the purple bar - scratch cache

Then again, problem in gaming with X1K and later R600 and RV670 was the relative lack of texture units (TMUs), and problem with GPGPU continued to be – local share. You now might be wondering what will happen if you don’t put that “scratch cache” in the GPU. What happens is that your CPU will be constantly polled, and this drags the performance down to the gutter.
We heard a lot of technical details about that particular issue, and the difference in scaling between dual-core CPU and a quad-core one. All in all, quite interesting stuff. But there is one large point to be made: the reason why Nvidia is so successful with CUDA is the fact that Nvidia offered what companies needed (scratch cache, CUDA, math libraries), while ATI suffers from selecting Brook+ to be their bread and butter until OpenCL comes along.
RV770 saves the day…or not?
The RV770 GPU, more known as Radeon 4800 series is a vast improvement over previous generations. GPGPU-wise, most important thing is introduction of local share, since every 10 shaders got their “slice of the pie”. But GPGPU is more complex field that just “here is the feature, we can all use it now”.
Our sources repeatedly criticized Brook+, claiming it is not in sync with AMD’s own CTI and Stream SDK’s. Brook+ allegedly breaks “with new drivers, with old drivers”, “whatever can go wrong, it can” and so on.
ATI’s hardware now has local share, but that support has to be hard-coded into Brooke+. AMD recently released 1.21 Beta Stream SDK featuring local share, but that same support has to come inside Brooke+ as well.
The Solution: Q1’09
So, we have shown you the problem, and now the time is for the solution. ATI can’t fix the performance issue on previous-gen hardware, but it will solve multitude of issues on Radeon 4800 boards. The team at Stanford is taking some necessary steps to re-do the workflow and introduce local memory share. This could take months, so realistic goal is to have a new client coming in Q1’09.
Once that Radeon 4870 gets fully utilized, those 800 shaders and 70% of theoretical value (700-800 GFLOPS instead of 1-1.2 TFLOPS) should be good enough for reaching the level of GTX280.

Next story update will bring some views and opinions from AMD folk.

The post AMD’s Folding performance explained, future development revealed appeared first on VR World.

The motherboard for the ultimate workstation

The board supports both SLI and CrossFire in their respective maximum configurations (3 or 4 GPUs), but what makes this board really interesting is the fact that you could connect 12 LCD displays on it, or create a GPGPU/rendering/scientific/folding farm in a single case. Putting six ATI Radeon 4850 graphics cards would enable roughly 6TFLOPS of computing power. In case of Nvidia, you would have to pick up GeForce 9800GT (Palit has single-slot 1GB card) and have less theoretical computing power, but in terms of folding, you would be looking at 30-35.000 PPD system (at a cost of two GTX260 cards).
This is really impressive engineering feat from ASUS, with the only disappointment being usage of RealTek GbE controller. For a workstation motherboard, I would much happier if Marvell was on-board.
Storage-wise, you can install no less than eight SATA devices and not a single IDE device, since ASUS stayed in “Revolution” theme and killed of the PATA connector. Also, I found that a shared PS/2 port was also pretty neat solution, even though real revolution would be killing both PS/2 slots. This way, you still have one legacy part: PS/2.

There is one shared PS/2 port, for either keyboard or a mouse

Funny part of this story is that if anybody would have a time machine and go back to IBM engineers in 1986-7 frame and told them that only remain of their failed standard is going to be a keyboard/mouse connector, and that PS/2 connector will outlive PATA, I guess they would call you… crazy? Lunatic? Infidel?

P.S. If you’re wondering… yes, the answer is true. There are no technical issues that would prevent you from installing 3-Way SLI and 4-Way CrossFireX setup, consisting out of three GTX280 and two 4870X2 cards. Only problem is that you would have to have a watercooling setup, since you are limited to single-slot cooling solutions. I guess Asetek, CoolIT or somebody similar could come up with a solution for this “problem”.

The post ASUS kills PATA and PCI standards! appeared first on VR World.

Nvidia's client may not look as nice as ATI one, but it's the efficiency that counts...

Fast forward to GTX280 launch and the Vijay Pande team debuted the Folding@Home client for Nvidia chips as well. Nvidia and ATI lead a short marketing war who can fold better and things went quiet… apparently, for a reason.
The reason why things went quiet is probably the “inconvenient truth”: ATI showed up with Radeon 4800 series and demolished Nvidia’s dominance in the segment, with GTX260 and 280 going through radical price drops in order to stay competitive. However, ATI’s Radeon 4800 series has one field where the card is losing against 5-10x cheaper cards: Folding@Home.
The 10x argument lies in comparison between current ATI’s flagship, the  Radeon 4870X2 and Nvidia’s GeForce 9600GSO. This $50 card can easily out-fold ATI Radeon 4870X2, which retails for more than 500 USD/450EUR in respective markets.
In the past weeks, I’ve conducted a series of tests with various graphics cards (all that I own or could put my hands on), and the results were quite depressing if you own an ATI card. I’ve asked some of my contacts in AMD why the performance is so bad and the answers were ranging from “we wanted to make best gamer’s card, not a card for Folding” to sad silence. It seems to me that the difference lies in shader type and clock: ATI’s R6xx and RV7xx architecture lies around big fat units and lot of tiny ones (64+256 in case of Radeon 3800, 80+720 in case of Radeon 4800), and the clock is much lower than in case with GeForce cards. At the same time, Nvidia went the other route and came up with large number of “fat” units, while the company didn’t even count the “thin” (MADD) ones.
When we compare the GTX280 and 4870X2, comparisons are just astounding: in a period of a month, EVGA’s GTX280 SSC achieved an average of 6,802 points per day, while ATI Radeon 4870X2 managed puny 3,870 ppd. At the same time, I’ve witnessed higher PPD scores achieved even by two-year old GeForce 8800GTS 640 MB, which was quite a surprise. Around two weeks ago, I started following PPD numbers using FahMon on a large number of systems that mostly bear the same configuration: dua-core processor or more, 2GB system memory or more and the graphics cards. In all cases, with the help of my friends, I’ve managed to check FahMon and KakaoStats for rougly 25 cards and came to a surprising result.
With the recent update to the GPU2 client and new Fah_Core11.exe (ATI uses v1.17, Nvidia v1.15), the community witnessed further fall in number of completed packets per day. If you’re not familiar with Folding@Home packets, every package features certain number of mathematical simulations for tested protein – in case of Nvidia, packet consists out of 25 million, while ATI’s one features 10 million operations. However, due do different type of mathematical operations, Nvidia’s packet usually will result in 480 points, while ATI’s 10 million will return 548 points (or recently introduced ATI packets with 338 points).
Like I previously wrote, the table below is not the result of one packet score and Excel calculation, but rather continuous number crunching over the course of several weeks, with one week used for measurement.

Improvised Top 20 Folding@Home GPUs:

1. Nvidia GeForce GTX280 1GB (EVGA SSC)
2. Nvidia GeForce GTX260-216 898MB (EVGA SSC)
3. Nvidia GeForce GTX260 898MB (EVGA Superclocked)
4. Nvidia GeForce 9800GTX+ 512MB (ASUS TOP)
5. Nvidia Quadro FX 4600 SDI 768MB (PNY)
6. Nvidia GeForce 9800GTX 512MB (ASUS TOP)
7. Nvidia GeForce 8800GTX 768MB (Zotac AMP! Edition)
8. Nvidia GeForce 8800Ultra 768MB (XFX XXX Edition)
9. Nvidia GeForce 8800GTS 512MB (Gainward)
10. Nvidia GeForce 8800GT 512MB (Gainward)
11. Nvidia GeForce 9600GSO 768MB (EVGA)
12. Nvidia GeForce 8800GTS 640MB (LeadTek)
13. ATI Radeon 4870X2 2GB (PowerColor)
14. ATI Radeon 4870 512MB (PALIT)
15. Nvidia GeForce 9600GT 256MB (Zotac)
16. ATI Radeon 4850 512MB (PALIT)
17. ATI Radeon 3870 512MB (Sapphire Atomic)
18. ATI FireGL V8600 1GB (ATI)
19. Nvidia GeForce 8600GTS 256MB (XFX XXX Edition)
20. ATI Radeon 3850 256MB (Sapphire)

This is not a complete table by no means, since I am missing several new GPUs. But in this one, as you can see for yourself – results are quite dramatic for the red team. Two year old GeForce GPUs demolished otherwise-brilliant Radeon series, and it is incredible that even GeForce 9600 will outfold Radeon 4850. This is a rude wake-up call for guys at Markham, because this is just unbelievable.
Personally, I am running a combination of AMD Spider platform (9850BE + 790GX + ATI Radeon 4870X2) and hybrid Intel’s V8-Skulltrail platform with Quadro FX 4600 SDI.
Of course, everything can be changed with a simple driver update. I don’t understand what happened with AMD/ATI, company that lead the field of GPGPU computing for so long – why should AMD work on optimizing Folding@Home client… I am aware that AMD poached Mike Houston from Stanford to work on Brooke+ and now OpenCL APIs, but surely the performance didn’t went downhill from the influence of just one person. Or just maybe…
Overall, I hope that Catalyst 8.11 or 8.12 will bring more performance for ATI cards, since I do not believe that it would be so hard to optimize drivers for GPGPU/GPU Computing usage. For now, in Folding@Home, ATI is complete washout.

For the end of this article, if you find that your GPU cycles could be used for something good, I invite you to read the following article and join F@H family, regardless of what client (CPU or GPU) or team you choose in the end. Intel, AMD, ATI, Nvidia, Windows, Linux or Mac OS – it does not matter, just join – If you want, of course.

The post Nvidia’s $50 card destroys ATI’s $500 one or “Why ATI sucks in Folding?” appeared first on VR World.

Sapphire

Sadly, Sapphire was not allowed to make any changes, so FireStream 9250 continues to come with only 1GB of GDDR3 memory, while the most GPGPU scientists we talked about are talking about the need for massive amount of memory. Nvidia’s respond was 1st generation Tesla with 1.5 GB and the latest one with massive 4GB of GDDR3 memory.
We certainly hope that AMD will release FireStream with 2-4GB of memory, given their track record in professional graphics space. In any case, I congratulate to Sapphire for releasing the card.

The post AMD releasing professional cards to partners – Sapphire first appeared first on VR World.

All in all, the dawn of GPGPU computing is finally ending, and we’re heading forward into GPU accelerated world. Welcome.

The post Adobe ships GPU-accelerated Creative Studio 4, Flash Player 10 appeared first on VR World.