Bookmark and Share


Nvidia Corp.’s code-named Maxwell graphics architecture has been discussed by the company for several years now and it is expected that it will debut sometime in April or March, 2014, around two years after the company unveiled its Kepler architecture. However, there are signs that Nvidia Maxwell may arrive a little earlier than expected.

Clevo, one of the world’s leading makers of high-performance laptops that manufactures such devices for well-known brands, recently leaked its roadmap and revealed Nvidia GeForce GTX and GT 8-series mobile graphics processing units (GPUs). The new high-end graphics adapters from Nvidia are projected to become broadly available in various versions already in February, 2014, in case the roadmap slides published by ThinkComputers are up-to-date. Clevo is preparing numerous new laptops with GeForce GTX 8-series mobile GPUs inside, including machines with multi-GPU SLI configuration.

Clevo’s roadmap reveals such graphics solutions as GeForce GTX 880M/8GB, GeForce GTX 870M/6GB, GeForce GTX 860M/4GB, GeForce GTX 860M/2GB, GeForce GTX 850M/2GB and GeForce GT 840M/2GB. The family looks to be very broad and ready to serve different requirements for performance.

While Nvidia is renowned for simple rebranding of mobile graphics solutions in the low-end, it is hard to imagine that such products like GeForce GTX 880M/8GB or GeForce GTX 870M/6GB are rebadged current-gen products with higher amount of memory. It is also unlikely that Nvidia will introduce a GK110-based graphics card for mobile applications. Keeping all the things in mind, it looks like at least several of the new GeForce GTX and GeForce GT 8-series will be based on brand-new GPUs featuring Maxwell graphics architecture.

Since Nvidia does not traditionally introduce high-end mobile GPUs ahead of advanced graphics cards for desktops, it is logical to expect the first members of Maxwell family to arrive in late January, 2014, or early February, 2014.

While Nvidia Corp. reveals some things about its next-generation products, it does not want to disclose the whole picture and it clearly does not provide any precise timeframes. While without providing further details, at the annual investor day conference this week the company did disclose some of its expectations for its next-gen graphics and compute architecture code-named Maxwell.

What we do know about the Maxwell family of chips so far from the official sources is that they will integrate general-purpose Denver ARMv8-compatible cores in addition to graphics stream processors and that they will be able to support unified virtual memory technology with microprocessors from Intel or AMD, a rather big deal for many applications. It is also logical to expect higher horsepower in general, which will boost video games, the main driver for Nvidia’s GeForce business.

Officially, Nvidia expects three major things from the Maxwell architecture: improved graphics capabilities, simplified programmability (probably thanks to ARMv8 and unified memory enhancements) as well as energy-efficiency.

“Number one for Maxwell, that is likely something that we are doing that breaks new ground in visual capability, something that is even more beautiful. […] Number two, it is likely that Maxwell breaks new ground in programmability, ease of programmability, because we want to expand the general purpose nature of the processor without sacrificing its speedup relative to a microprocessor. […] The last thing, the energy efficiency of Maxwell, it is going to crush Kepler. […] We know exactly how to measure it now and we know what it means to be good,” said Jen-Hsun Huang, chief executive officer of Nvidia, at a conference earlier this year.

Nvidia did not comment on the news-story.

Tags: Nvidia, Maxwell, Geforce, clevo


Comments currently: 23
Discussion started: 12/24/13 05:47:31 PM
Latest comment: 11/30/15 11:08:31 PM
Expand all threads | Collapse all threads


"What we do know about the Maxwell family of chips so far from the official sources is that they will integrate general-purpose Denver ARMv8-compatible cores"

And thus begins the Gaming Console on a PCI card wars, between Nvidia and AMD, look forward to APU type systems with CPUs shareing a big fat 512+ bit data data bus with the GPU, and linked to 8GB of GDDDR5 memory, all on a PCI based card! What is Intel to do, all isolated down there on the motherboard with its narrow data bus(64 puny bits wide), and slower memory! Nvidia will have ARM Holdings'(HSA Foundation) Access to HSA and uHMA, through its ARM ISA license, and will be producing some ARM 64 bit APU Monsters, While AMD will be forced to counter with some PCI based gaming APUs of the x86 KIND! These consoles on a card will be running Steam OS, or other gaming optimized Linux based Distros, and the gaming engines will be on the PCI cards also, shareing all the gaming cards high bandwidth/low latency fast memory goodness! For those that opt for dual gaming cards, the Gaming OSs and gaming engines, will auto configure to operate as a gaming cluster, with each PCI card offering a CPU/GPU gaming multicore CPU and GPU gaming cluster to the mix!
5 5 [Posted by: BigChiefRunAmok  | Date: 12/24/13 05:47:31 PM]
- collapse thread

You sir, have one serious dream. But as of now that is all that is in relations to Maxwell, APU's, and Unified Virtual Memory.
3 0 [Posted by: Bruce Regael  | Date: 12/25/13 12:24:09 AM]
They cannot make a proper multithreaded game engine, yet you are dreaming of cluster optimized games?
Except from that, even simple applications are not ready, yet, for the xeon phis.
On the other hand, nVidia is not in the huma hsa etc group. Do you think that amd or samsung or the other companies are so stupid to contribute in the R&D of that technology and let arm make a shitload of money out of it? DO NOT THINK SO.-
Merry Christmas to everyone and a happy new year.
2 1 [Posted by: Yorgos  | Date: 12/25/13 02:07:21 PM]
Anyone who liceses Arm Holdings' ISA, or ARM refrence designs, will benefit, as ARM Holdings' is the creator of the ARM ISA, and ARM refrence designs, and ARM Holdings is a founding member of the HSA Foundation, along with AMD. And Nivida(Licenses ARM designs) is welcome to join the HSA foundation any time Nividia chooses, as well as others! And you are wrong about the gaming engines not being able to use multithreading, as the SDKs, toolchains, and programming languages for multithreading have existed for some time, from ARM, Intel, AMD, and the Open source community for quite some time. Blender 3d has a CPU render mode that will peg all 4 cores and 8 processor threads on my core i7 based laptop at 100%, and it is an OpenSource software program, blender will also render using multiple networked computers(cluster rendering style), Blender is written in Python(API),GCC, uses OpenMP(Open Multi-Processing), as well as others, and comes with a gaming engine! Blender will also GPU render too.

Introduction from HSA webpage(first Part):

"The HSA (Heterogeneous System Architecture) Foundation is a not-for-profit consortium of SoC IP vendors, OEMs, academia, SoC vendors, OSVs and ISVs whose challenging the normal of how whole system architecture is structured for combing CPU’s, GPU’s, DSP’s, and other accelerators to bring about forward progress in computing’s foundation to make it dramatically easier to program heterogeneous parallel devices. HSA Foundation is driving this via Royalty Free Specifications and open source software: HSA runtimes and low level compilation tools based on open source technologies like LLVM and GCC."

Continue reading here!

The R&D for multithreading is in the SDK, that Intel provides (C++AMP), AMD provides Accelerated Parallel Processing (APP), Arm has many, Including GCC, openMP, etc. so ARM, x86, and others have had multithreading support SDKs, in linux, windows, Other OSs TOO, for a while, so do you homework before you post!!!

2 3 [Posted by: BigChiefRunAmok  | Date: 12/25/13 04:57:09 PM]
Games like bf3, crysis 3, far cry 3 and very few others are able to utilize multicore cpus. On the other hand most of the games can utilize 2 cores, while others are optimized for 3 cores (!?), yes three cores thanks to the xbox h/w. Nearly a decade later than the first desktop multicore cpu, yet there are serious problems in utilizing fully a cpu. I don't have to mention the problems that schedulers still have with amd's new architecture, patches came out nearly a year after the first launch of bd, and some problems with linux not able to properly assign threads to the threads that intel's cpus have. Intel cpus, at least for the Nahalem I've tested with various kernels.

I am sorry to burst your bubble, but I am sure that you don't know what hsafoundation really does. As a matter of fact, on the next paragraph that you copied from the site it mentions:
HSA Foundation members are building a heterogeneous compute software ecosystem which is rooted on open royalty free industry standards.
There is a big difference between licensing an IP core with hUMA and HSA capabilities and building SOFTWARE that is free and not-for-profit just to help others make more programs for it.
So as far as this thing goes until now, nVidia and everyother IP or ASIC vendor if it doesn't provide either a hsa capable h/w according to the specifications needed or if that vendor doesn't contribute to the hsa foundation SOFTWARE development no ARM or AMD or SAMSUNG is going to write S/W that will utilize kepler or maxwell h/w along with arm or x86 cpus.
0 0 [Posted by: Yorgos  | Date: 12/26/13 06:52:16 AM]
AMD's hUMA(marketing term) is just a term that AMD uses instead of the generic term: Uniform memory access, where compute devices share the same memory. No one company owns the idea of shareing memory address space between different processing units, as this has been the norm that supercomputing has utilized for some time with different types of processing units that share the same memory address space! And anyone could write a VM virtual machine to run HSAIL (HSA Intermediate Language), or JAVA IL (Java VM) in OpenCL, and OpenCL will run on AMD or Nivida GPUs! A VM (JAVA based, HASIL based, Dalvik based(ANDROID), could easily be written to convert/compile any IL(intermediate language) bytecode into OpenCL to run on a GPU. So OpenCL does make any GPU that has OpenCL ability(drivers) Heterogeneous for GPGPU compute! So if you have a GPU with OpenCL you are already on the way to HSA!

The article states:
"What we do know about the Maxwell family of chips so far from the official sources is that they will integrate general-purpose Denver ARMv8-compatible cores in addition to graphics stream processors and that they will be able to support unified virtual memory technology."

A Nivida "APU" based on ARM ARMv8-compatible cores! If is walks like an APU (ARM 64 bit CPU based), and has Nivida graphics stream processors (GPU), then Nivida can call it what ever Nivida wants, but it is Nvidias' version of what AMD calls an APU. AMD does not own address space shareing between compute units, and anyone is free to implement their own unified memory address space based CPU/GPU based systems.

OpenCL royalty-free
0 1 [Posted by: BigChiefRunAmok  | Date: 12/26/13 05:25:17 PM]

But will it play crysis? I'll get my rain coat......

On a more serious note I look forward to the next gen series die shrink from both AMD and Nvidia.
Should see better speed increases as they automatically get power savings with shrink so can boost speed instead(we hope). Last gen cards were all about efficiency which was great but I would like to see a return to bigger performance increases instead of the usual 15-20% we are use to.
1 0 [Posted by: ozegamer  | Date: 12/25/13 04:14:22 AM]
- collapse thread

Do not look for even those 15-20%, let alone larger improvments, as Dennard scaling and Moors Law is running up against the laws of physics, and thermal budgets.

From earlier post:

With each process node shrink, the circuits become smaller, and yes they use less power, and produce less heat, but these smaller circuits are packed closer together and the power savings/heat produced does not scale at the same rate as the circuit count per unit area! these densely packed circuts can actually result in more heat per unit area produced, than the available(less atoms with each process node shrink) substrate atoms to carry away the thermal load, resulting in a hotter running chip. These thermal budget scaling issues were present with the process node shrink from 32nm to 22 nm, and they will be even more pronounced going from 22nm to 14nm, as there needs to be enough atoms surrounding these circuits to absorbe and transfer the heat, this can, and is interfering, with the 14nm process node development/engineering, and may result in the smaller 14nm size circuits having to be spaced farther apart, to maintain some form of thermal equilibrium, and thereby, resulting in less die area space savings, compared to the earlier 32nm and 22nm process node shrinks.

The days of processing node shrinks, allowing 15-20%, or larger clock increases, and much more die area, are coming to a close! It's back to more cores on larger die areas, until some new technology is invented, and the way things look, even those with Intel sized budgets, will be hard pressed, for the R&D funds, to go beyond the silicon based industral fabrication process.
3 3 [Posted by: BigChiefRunAmok  | Date: 12/25/13 05:31:49 PM]

I am going to have an x99 chipset from Intel and a GTX 980 graphics card from NVidia. A lot of nines.
1 1 [Posted by: Teemu Ruskeepää  | Date: 12/26/13 04:18:53 AM]

This looks like another disruptive tech to farther shake up the PC Market. Ever-one is looking for an answer to the declining sales. The prime reason for every manufacture of x86 hardware is the failure of the x86 monopoly O/S. Steam/Linux is aimed at the gaming sector, Android is aimed at the mobile sector, Apple is trying to eat away at the edges and all the While Chrome/Linux etc are growing stronger. The video sector is moving away from the Intel dominance.
2 2 [Posted by: tedstoy  | Date: 12/26/13 08:33:19 AM]
- collapse thread

In what universe is this happening? In the universe I live in, Windows is about as dominant as ever (although Windows 8 is a disaster), Chrome/Linux are not getting anywhere, Mac OS X' market share is stagnant (and lower than that of the catastrophic Windows 8 OS) and Intel's market share in graphics has been steadily increasing (from a very low point, which makes your statement "the video sector is moving away from the Intel dominance" all the more strange) at the expense of Nvidia and AMD in the last few years. Sure, Android is a huge success but in a completely new market (the mobile market) and it has been ported to x86 and runs perfectly fine on it.
1 0 [Posted by: lol123  | Date: 12/27/13 08:31:00 AM]
Really with Google, facebook, and others(Many enterprises) running their servers and online services on linux, and just about every device from refrigerators, dvd players, TVs, running on ARM chips and an embedded linux distro, as well as phones and tablets(android/chrome)! Tell that windows fib to the HPC supercomputer folks, and they will laugh you out of the building! You were probably not alive when Intel was taken to court over its trade practices, and OEM arm twisting with respect to its x86 competition! Do not delude yourself into thinking that any sane mobile device OEM will ever base their entire product portfolio on just Intel's CPU SKUs! The City of Munich has transitioned to linux, and expect other cities to partner with Munich, and share the development costs, to convert to Linux, as the city specific Linux based application ecosystem now exists(Developed By Munich) to run any city, Munich has built it and others will come, and benifit from a vendor lock-in free Europe, and world. Welcome to the world of the OS as a vital infrastructure, owned by no one, and maintained by all, just like the autobahns, and interstate highway system! And expect AMD and Nvidia to produce "APU" type complete gaming console systems on a PCI card, running gaming Linux distros on GDDR5 memory, and with optimized gaming engines living in an on CPU/GPU Die RAM. What will become of WINTEL down there on that narrow BUS, slow memory motherboard world.
1 3 [Posted by: BigChiefRunAmok  | Date: 12/27/13 12:22:57 PM]
haha.. nope. and with more ARM chips going tits up and out of business x86 will live on foever.

unless you are going to pay hundreds of billions to recode x86 to arm/risc.. will never happen. it is to cheap to maintain x86 and to spendy to switch.
0 0 [Posted by: amdzorz  | Date: 12/30/13 12:09:19 PM]
Calxeda is just a small portion of the ARM based ecosystem, and they are going through a restructuring, and if they go under, their ARM CPU interconnect fabric IP is sure to find a buyer! AMD and Nvidia will be making ARM based server chips, as well as many others! And the SDKs and cross compliers exist to cross compile any of the few remaining software stacks to ARM! The software ecosystem built around Arm Holdings' ARM ISA has existed for many decades, and Linux has been running on ARM as the OS of choice, as most ARM based products, refrigerators, DVD, TV, SSDs, and hard drives, are using ARM CPUs, running embedded Linux! The internet of things already has an ISA, and it is the ARM ISA, not x86! Billions of ARM CPUs running on every conceivable device, long before anyone even tried, to gimp the power hungry x86 ISA on to the mobile and embedded devices market. Its just a matter of time before the ARMv8 server chips hit the market. Apple is already there with their ARM ISA based Apple CPUs, the all Apple A7 Cyclone microarchitecture that runs the ARM 64 bit instruction set, which will lead to the first full Laptop based ARM multicore CPUs, and complete the process that was started with the ARM based "Chromebooks". The mobile world runs on ARM based chips, and the big search engines will be using ARM server chips soon. ARM based products/software are already there, and do not expect the "to spendy" to be a factor in any thing but using x86, beacuse ARM is the low power low price king, and has been for decades. Those chromebooks are selling like hot cakes, and many are loading them with full linux distros. M$ is the only one spending billions to recode from x86(How is that working out so far!), as everyone else has been on ARM from the beginning the mobile devices market.
0 0 [Posted by: BigChiefRunAmok  | Date: 12/30/13 03:02:28 PM]

I don't know why are you arguing that much. NVidia will simply bring some HPC functionality from their workstation class graphics down to the mainstream desktop. With the litecoin frenzy on newer Radeons it became apparent that the mainstream user buys calculation power, and NVidia's way of cheapening on memory bus width and reduced floating point precision calculation cost them sales. That's why they hurry in the new architecture, to replace the 6 and 7 series castrates.
2 0 [Posted by: Ananke  | Date: 12/27/13 01:30:52 PM]
- collapse thread

Yes the HPC world, as well as the old mainframe world, where VM hosted OSs have been around since the 1960s, and the HPC/Server/ Cloud computing clusters now rule, with their advanced interconnect fabrics. HSA has been in the mainframe/HPC world for quite a few years now, and the ARM cpu interconnect IP of calxeda may be purchased by others to produce PCI card sized server clusters for the home/SMB market, as well as home based GPU accelerated gaming clusters via Nvidia or AMD. Nvidia's Maxwell Architecture with ARMv8 is its logical next step, in the competition with AMD x86 based console gaming APUs, as they both spawn the home cluster on a PCI card based gaming, and home server cluster PCI card based systems. With CPUs and the APU type systems migrating more than just northbridge and southbridge chipset functions, and GPU functions onto the chips die space, expect integrated server/hpc type interconnects to follow, all while shareing the extra wide GPU style data buses, and fast memory! Expect motherboard makers to begin offering many more products with many PCIe x16 slots and the power traces/conditioning circuits as well as wide buses and/or 10gb ethernet to support the home clusters on a PCI card, complete systems. The cloud computing industry has been using PCI protocol tunnled over 10gb/faster ethernet protocol for some time also. Nvidia is now forced to deal with HPC clusters in the home, as a result of AMD's and the HSA foundation's members developing HSA GPGPU open standards, and other groups open sourcing bitcoin software that the bitcoin community uses. Many of the students that compete in the HPC cluster competitions will be the ones founding the new home cluster Gaming/server/Bitcoin companies that will form based around the open server hardware/software utilized in the HPC competitions.
2 2 [Posted by: BigChiefRunAmok  | Date: 12/27/13 04:04:54 PM]
calxeda is bankrupt. and liquidated. there shitty product was not even close to being good enough to replace x86 isa.
0 0 [Posted by: amdzorz  | Date: 12/30/13 12:10:33 PM]
Hay Spendy, spend some time researching! with less time stamping your feet, and gnashing your teeth! Calxeda, is one ARM licensee, among the many thousands of other ARM licensees, who build billions of ARM CPUs every quarter! What's that Spendy, you broke your wittle rattle.
0 0 [Posted by: BigChiefRunAmok  | Date: 12/30/13 03:24:55 PM]


Add your Comment

Related news

Latest News

Wednesday, November 5, 2014

10:48 pm | LG’s Unique Ultra-Wide Curved 34” Display Finally Hits the Market. LG 34UC97 Available in the U.S. and the U.K.

Wednesday, October 8, 2014

12:52 pm | Lisa Su Appointed as New CEO of Advanced Micro Devices. Rory Read Steps Down, Lisa Su Becomes New CEO of AMD

Thursday, August 28, 2014

4:22 am | AMD Has No Plans to Reconsider Recommended Prices of Radeon R9 Graphics Cards. AMD Will Not Lower Recommended Prices of Radeon R9 Graphics Solutions

Wednesday, August 27, 2014

1:09 pm | Samsung Begins to Produce 2.13GHz 64GB DDR4 Memory Modules. Samsung Uses TSV DRAMs for 64GB DDR4 RDIMMs

Tuesday, August 26, 2014

10:41 am | AMD Quietly Reveals Third Iteration of GCN Architecture with Tonga GPU. AMD Unleashes Radeon R9 285 Graphics Cards, Tonga GPU, GCN 1.2 Architecture