Best offers
|
MediaSmart LX195 Server (1.6GHz Intel... | $249.99 HP Direct More info |
|
Dell PowerEdge 2800 Server Intel Xeon... | $600.00 ServerSupply.com More info |
|
ProLiant DL380 G3 2U Rack Mountable... | $680.00 ServerSupply.com More info |
|
Express IBM System x3550 M2 1U RM... | $1693.75 PCNation.com More info |
|
PowerEdge 2650 2U Rack Mountable... | $600.00 ServerSupply.com More info |
Reader's Voice: Building Your Own File Server
Whenever we review dedicated NAS devices, readers point out that do-it-yourself projects are often much less expensive and much more flexible. Jeff Deifik explains why he built his own file server and the hardware that went into the job. Read More
-
Hacking The HP EX470/475 MediaSmart Servers
HP's MediaSmart Servers are basically low-power PCs running Windows Home Server software from a corner of your house. Ed Tittel takes his apart and shows you how to upgrade (and benchmark) the machine's memory, cooling capacity, and CPU. Read More
-
Promise NS4300N: NAS For Small Offices
Promise is known for its RAID controllers, but how the company will fare long-term on the NAS market remains to be seen. We took a closer look at the first of its network storage offerings, the NS4300N Read More
Partners
The Games selection
adventure :
Scoobydoo: Episode 2
The sequel of Scooby and Sammy's adventures. Same principle as in the previous episode (available on this website). Click on "Instructions" to see...
|
crazy :
Xiao Xiao 7
A great fight scene from the animation movies Xiao Xiao.
|
Sponsored links
Asus ESC 1000 Powered By 960 Nvidia Cores
Next news- Email |
- Print |
- Comments (48) |
- Share
A new Asus supercomputer uses 960 Nvidia cores.
Talk about crazy: try building a supercomputer consisting of 960 Nvidia processing cores. That's what the madmen of Asus has done, a Frankenstein of the computer industry squashed into a desktop-sized form. In the end, Nvidia and Asus managed to squeeze out speeds of up to 1.1 teraflops--that's enough juice to run Crysis at its highest setting.
Will it be used for gaming? Probably after work hours, as the supercomputer is ideal for medical image manipulation, engineering, scientific research and other process-hogging applications. Under the hood, Asus' ESC 1000 uses Intel's LDA1366 Xeon W3580 microprocessor running at 3.33 GHz. Nvidia's 960 processing cores comprise of three Tesla c1060 computing processors and one Quadro FX 5800.
Additionally, the Asus supercomputer is housed in a 445-mm x 217.5-mm x 545-mm chassis. Other components include 24 GB of DDR3 1333 MHz DRAM, a 500 GB SATA II HDD, and a 1100W power supply. According to PC Advisor, the ESC 1000 will cost around $14,512 over a five-year period; the company would not provide pricing for individual units.
Asus said that the ESC 1000 will be available globally, and that the computers are now ready to ship.
Source : Tom's Hardware US
Sponsored links
Related articles
-
Calculating average performance per game will help us to see which titles most heavily favor high-bandwidth slots. Left 4 Dead appears to be the most significantly bandwidth-limited game, but we had previously noticed an oddity that almost disqualifies the game from further analysis. A combination of lower-bandwidth (P6T) and added PCIe latency (Eclipse Plus) could be what threw our expectations out-the-window for that title, but there’s no way of knowing the exact cause. World In Conflict appears more appropriate for analyzing the performance penalty of the P6T’s x4-width third slot, while also showing that two slots sharing sixteen pathways (Eclipse Plus) have the same performance as two x8 slots. If bandwidth limitations to the second and third graphics card are the biggest part of today’s story, we’d expect the problem to increase with higher resolutions and details. Is this the case? A performance difference of around 10% between the fastest and slowest PCIe configurations climbs to nearly 16% at the highest tested settings. Conclusion Tom’s Hardware doesn’t do many single-motherboard reviews, simply because it’s impossible to give every company that level of attention when faced with dozens of samples per year. Yet MSI provided an opportunity we couldn’t pass up: to access the gaming performance advantage of two shared x16 slots compared to two dedicated x8 slots. Unfortunately, we found no substantial advantage, though the Eclipse Plus remains a great motherboard in many ways that would only be completely revealed in a full review. The real revelation of today’s article is the level of handicap an x4 slot places on the third graphics card in a 3-way SLI configuration, even with PCIe 2.0 doubling its bandwidth. We knew it would be bad, but not this bad, and the results force us to cross the Asus P6T, the DFI X58-T3eH6, MSI’s previous Eclipse SLI, and several as-yet-untested products off our 3-way SLI list. If MSI’s intent was to show how much better the Eclipse Plus is for 3-way SLI than its x16-x16-x4 predecessor, we can give the company credit for that. What we didn’t see today, however, was whether or not three true x16 slots would provide any advantage at all over an x16-x8-x8 configuration. The reason, of course, is that the only Core i7 chipset doesn’t have the required 48 PCIe lanes, and adding a hub to “make” an x16-x16-x16 configuration leaves us with only as much bandwidth as the 32-lanes serving those three slots. We can imagine some “Home Supercomputer” applications where Nvidia’s CUDA technology might need to access various graphics processors at different times, where the flexibility of sixteen shared pathways might be advantageous compared to two sets of eight pathways especially if three GTX 295s are used. After all, a GTX 295 on a x8 slot would only have four lanes of bandwidth per GPU. But MSI doesn’t target high-end-workstations and above with this particular motherboard, leaving that market to the competing and yet-unexamined Asus P6T6 WS Revolution. Instead it follows the path of the also-yet-unexamined EVGA X58 Classified, marketing to gamers a board that has yet to reveal any gaming advantage. Perhaps a shootout of nForce 200-equipped X58 motherboards should be arranged, but until then, we’ll stick with the chipset-supported x16-x8-x8 configuration for our own 3-way SLI tests.
-
The Truth Behind ASRock's X58 SuperComputer
ASRock was quick to respond to the overclocking issues encountered in our Core i7 Sub-$300 Motherboard Roundup, offering a replacement board and processor within hours of article publication. And how could we pass up the offer, when the replacement Core i7 processor ASRock offered was based on Intel’s improved D0 stepping? Yet, we wanted to make sure we were representing the situation as accurately as possible, so we took time to carefully consider our commitment to you before finally deciding the most balanced approach. We’ll begin today’s review update by presenting ASRock’s explanation, and finish by applying our previous overclocking configuration to a second motherboard we purchased at e-tail to see if a newer board revision and improved BIOS will protect overclocking enthusiasts. The only motherboard we’ve tested to properly support four double-thick graphics cards, ASRock's X58 SuperComputer’s name isn’t too far off from at least one of the product’s intended markets as an Nvidia Tesla Personal Supercomputer platform. Gaming fanatics can also appreciate the flexibility of supporting extra cards, since a 3-way SLI configuration still has space for a fourth card that can be dedicated to in-game physics calculations. We had much praise and few complaints in our full layout description, so our next page will re-examine the product’s overclocking capabilities using the latest BIOS and CPU core stepping.
-
Asus showed us its new concept of a true DIY notebook. The C90S is based on conventional desktop hardware such as Intel's 965 chipset, and it uses a desktop socket 775 processor. This may be any Core 2 Duo at up to FSB1066 speeds, and a maximum of 2.93 GHz core clock speed. The system was designed to be overclockable, but the 2.93 GHz limit remains, as the heat generated and power required is the maximum that the thermal solution and power supply support. Since the C90S isn't really quiet (four 40 mm fans) you probably wouldn't want to go much faster anyway. The system takes any 2.5" Serial ATA hard drive and up to two DDR2 SO-DIMMs. You will want to use two modules in order to enable dual-channel memory operation. For the graphics part you have several options, as the C90S supports MXM-1 and MXM2- graphics modules. MXM is Nvidia's Mobile PCI Express Module for graphics boards, and there are three different versions that differ mainly in size. MXM-1 is small and takes entry-level GeForce graphics; MXM-2 is for mainstream graphics such as the new mobile GeForce 8600; and MXM-3 is for high-end solutions. We wanted to know whether this new approach brings true DIY into the notebook space, which is why we asked Asus to disassemble the C90S sample. We took some pictures of the procedure, which took no longer than three minutes. Check out our slide show: How to (dis-)assemble the Asus C90S notebook in three minutes. The Asus C90S is based on a GeForce Go 8600 mobile processor on a MXM-2 module, but it can also take any other Nvidia graphics solution based on MXM-1 or MXM-2. This is the back side of the GeForce Go 8600.









Damn, at a desktop size, that is incredible.
After all this time, we finally conquer Crysis?
Each of those Nvidia cards (4) has four GB of RAM. Three Tesla cards and the Quadro. So that's 16 GB of RAM on cards alone and 24 GB of RAM for the system. That is insane.
This is going to be one hell of SuperPC for folding! I want one so bad now just to play around with and experiment.
So no "can it play crysis?" then?
I NEED BENCHMARKS.... MY HANDS THERE SHAKING
Crysis has been conquered since the 5870's came out
3 of them will beat any game at MAX settings
But can it run Windows?
Just think, in another year or two, they'll have assembled another system which will be twice as powerful if not more so, in the same sized package!
Why do you have to mislead readers with this "960 Nvidia Cores" BS?
It's not like it has 960 GPUs!
A GPU shader or "thread processor" is in no way the equivalent of the term "core" used for CPUs
uhh... do you mean 960 SP's?
Um my $900 PC will run Crysis on full so how is that a "super computer"?
Isn't one 4870 capable of 1.2 Teraflops, and the new 5870 of 2.4?
considering how much juice it pulls, those numbers are not anything to be impressed by... a 5870 has about 2.5flops, this machine has 4 cards and 1tflop
that's pretty stupendously craptacular if you ask me.
Isn't one 4870 capable of 1.2 Teraflops, and the new 5870 of 2.4?
no...thats flops..not teraflops
960 cores = (3) GTX 280 class cards... whoop de doo. A single HD4870 can put out what? 1.2 Teraflops
And people should be surprised that 3 GTX 280's can play Crysis on high?
ill just wait for a single card to blow all those cards out of the water - with video cards going the way there going it wont be long
no...thats flops..not teraflops
http://techreport.com/articles.x/14990
"In fact, this is the first teraflop-capable GPU, with a theoretical peak of a cool one teraflops in the Radeon HD 4850 and up to 1.2 teraflops in the Radeon HD 4870."
mrddr6 let me put it this way. Can it run crysis in dx10 on all very high at 2560 x 1600 with 16xQ AA and still score an average of 60 fps or higher? They are talking about a system that can pull 60 fps on ANY settings with even the largest size monitors we have. Although his math is way off, it's 3.3 TF not 1.1, check your math. Although I find it silly to talk about crysis on a system like this, one that is clearly designed for professional/business use not gaming. Why would you use tesla and quadro for gaming! You should be using this as a render farm/simulator (physics, ray tracing, etc.).
But can it run Windows?
LOL @ ProDigit80, you mean windows Vista?
But can it run Windows?
But can it run Windows?
LOL EPIC POST!
FYI: Yes, after all it has a Xeon Nehalem which is x86-64. No Mac for this tho.
Um my $900 PC will run Crysis on full so how is that a "super computer"?
O RLY? What res and detail? 1600*1200? med res? No AA/AF?
$14.5K? definitely not for your average joe to play games. never tried online gaming. do they limit what you can use?
Can the Radeon run CUDA ?
This really isn't anything new. People have been building setups like this with Asus' P6T7 WS SuperComputer motherboard and ASRock's X58 SuperComputer board.
My rough calculations tell me that the components cost around $10,000 per machine. I expect they're using Linux and since Linux and OpenCL are either open source or free so those costs are relatively minimal. Assuming electricity costs of 12 cents per kWh and assuming this machine's PSU is kept maxxed out 50% of the time for five years constantly (pretty reasonable for a departmental supercomputer imo) then that'd be about $3000. So all in all, if their "5 year costs" include electricity then this is a pretty good deal.
Cost estimate details:
(500+3000+1300*3+1100+250+400*2+100+200+100)
(Mobo+Quadro+Tesla*3+CPU+PSU+RAM+HDD+Case+Misc)
So? ATI's HD5850 gets 2.09 Teraflops and the HD5870 gets 2.72 Teraflops. Why is 1.1 teraflops news?
Source:http://www.guru3d.com/article/radeon-hd-5850-review-crossfire/2
LOL @ ProDigit80, you mean windows Vista?
Insufficient RAM for Vista
Why don't they just ditch the dual slot coolers (use water or something that only require single slot cooling) and cram 7 GPUs in there?
Hate these dual slot cards, even Geforce range with pre-installed waterblocks usually require twice the number of slots they should.
So? ATI's HD5850 gets 2.09 Teraflops and the HD5870 gets 2.72 Teraflops. Why is 1.1 teraflops news?Source:http://www.guru3d.com/article/radeon-hd-5850-review-crossfire/2
ATI's HD5850 doesn't have 4GB of RAM per card though. That's the problem. ATI don't offer an equivalent to nVidia's Tesla line. They don't have a workstation class card out that uses the same chips as in the 5000 line either. Lastly, their GPGPU support isn't too great just yet. Nvidia's already got CUDA pretty much down and absorbed by a lot of GPGPU programmers. Their support of OpenCL is also way ahead of ATI's. OpenCL stuff already runs much faster on my laptop with a Core 2 Duo and a 512MB GeForce 9600 GT m than it does on my desktop workstation with Xeon chips and dual 1GB 4870s.
Just for reference an Nvidia Tesla card is capable of 933 GFlops, in single precision and 78 in double precision mode. A Radeon 5870 is 2.7 TFlops in single precision and 544 GFlops in double precision. However, the tesla cards are exclusively for processing data only, there are no graphics ports and it won't make Crysis run faster unless Crytek decides to offload CPU computations to the unified shader cores, these are typically caled cores by the Nvidia CUDA crowd as they aren't just for doing shader operations anymore. They can do real computations for scientific computing, oil and gas exploration (seismic data processing and visualization), GIS, computational fluid dynamics, MATLAB. There are over 500 different applications that currently take advantage of CUDA, ATI, which will probably use OpenCL for its programmability has a lot of catching up to do despite currently having a technological lead in it's latest consumer card. For an example of a CUDA based personal supercomputer using 4 nvidia Geforce 295GTX's check out http://www.manifold.net/info/news.shtml they have a serious rig that gives 5 TFlops in single precision.
damn this is some fast system. i have a 280sli system and it decodes fairly fast comparing to a processor. for now i will keep what i have, it's not like i have that much money to spend anyways
But can it run Windows?
It's using a Xeon processor. You've have to either be a complete idiot, or brain dead to think a Xeon processor can't run Windows.