Nvidia's G300/GT300 GF100/Fermi Discussion

[DiGiTaL MoNkEY]

NVIDIA GT300 ''Fermi'' Detailed

NVIDIA's upcoming flagship graphics processor is going by a lot of codenames. While some call it the GF100, others GT300 (based on the present nomenclature), what is certain that the NVIDIA has given the architecture an internal name of "Fermi", after the Italian physicist Enrico Fermi, the inventor of the nuclear reactor. It doesn't come as a surprise, that the codename of the board itself is going to be called "reactor", according to some sources.

Based on information gathered so far about GT300/Fermi, here's what's packed into it:

•Transistor count of over 3 billion
•Built on the 40 nm TSMC process
•512 shader processors (which NVIDIA may refer to as "CUDA cores")
•32 cores per core cluster
•384-bit GDDR5 memory interface
•1 MB L1 cache memory, 768 KB L2 unified cache memory
•Up to 6 GB of total memory, 1.5 GB can be expected for the consumer graphics variant
•Half Speed IEEE 754 Double Precision floating point
•Native support for execution of C (CUDA), C++, Fortran, support for DirectCompute 11, DirectX 11, OpenGL 3.1, and OpenCL

.
.

Even though it shares the same first two letters with GT200 architecture [GeForce Tesla], GT300 is the first truly new architecture since SIMD [Single-Instruction Multiple Data] units first appeared in graphical processors.

GT300 architecture groups processing cores in sets of 32 - up from 24 in GT200 architecture. But the difference between the two is that GT300 parts ways with the SIMD architecture that dominate the GPU architecture of today. GT300 Cores rely on MIMD-similar functions [Multiple-Instruction Multiple Data] - all the units work in MPMD mode, executing simple and complex shader and computing operations on-the-go. We're not exactly sure should we continue to use the word "shader processor" or "shader core" as these units are now almost on equal terms as FPUs inside latest AMD and Intel CPUs.

GT300 itself packs 16 groups with 32 cores - yes, we're talking about 512 cores for the high-end part. This number itself raises the computing power of GT300 by more than 2x when compared to the GT200 core. Before the chip tapes-out, there is no way anybody can predict working clocks, but if the clocks remain the same as on GT200, we would have over double the amount of computing power.
If for instance, nVidia gets a 2 GHz clock for the 512 MIMD cores, we are talking about no less than 3TFLOPS with Single-Precision. Dual precision is highly-dependant on how efficient the MIMD-like units will be, but you can count on 6-15x improvement over GT200.

This is not the only change - cluster organization is no longer static. The Scratch Cache is much more granular and allows for larger interactivity between the cores inside the cluster. GPGPU e.g. GPU Computing applications should really benefit from this architectural choice. When it comes to gaming, the question is obviously - how good can GT300 be? Please do bear in mind that this 32-core cluster will be used in next-generation Tegra, Tesla, GeForce and Quadro cards.

This architectural change should result in dramatic increase in Dual-Precision performance, and if GT300 packs enough registers - performance of both Single-Precision and Dual-Precision data might surprise all the players in the industry. Given the timeline when nVidia begun work on GT300, it looks to us like GT200 architecture was a test for real things coming in 2009.

Just like the CPU, GT300 gives direct hardware access [HAL] for CUDA 3.0, DirectX 11, OpenGL 3.1 and OpenCL. You can also do direct programming on the GPU, but we're not exactly sure would development of such a solution that be financially feasible. But the point in question is that now you can do it. It looks like Tim Sweeney's prophecy is slowly, but certainly - coming to life.

Written by: http://brightsideofnews.com/news/200...s-a-cgpu!.aspx With information from: http://www.hardware-infos.com/news.php?news=2904

Nvidia will finally launch their much delayed 40nm DX10.1 discrete graphics cards; GeForce G210 based on GT218/D10M1 and GeForce GT220 based on GT216/D10M2. GeForce G210, a sub $40 entry level card has 24 shaders with DDR2 memories on 64-bit interface while GeForce GT 220, a sub $60 mainstream card has 48 shaders and DDR3 memories on 128-bit interface. Shortly after, Nvidia is planning to release another DX10.1 card based on the GT215/D10P1, the GeForce GT 240 with 96 shaders and GDDR5 memories on 128-bit interface.

Initially, NVIDIA has planned for an enthusiast GPU codenamed GT212 but was canned as it was pointless against the more formidable EG Cypress. Instead, NVIDIA concentrate their effort on the next generation DX11 GT300 GPU and is determined to get it out this coming December. Interestingly, sources told us that AIC partners will get to design the GT300 cards themselves, which means, no more boring reference cards. Information we have gathered so far implies that GT300 has more than double the shader processors of the previous generation. Which means, more than 480 shader processors since GT200 has 240. GT300 card will be fitted with GDDR5 memories featuring a 384-bit memory interface.

NVIDIA seems pretty confident about this card snatching the performance crown back from AMD with a performance target of P2xxxx in 3DMark Vantage. However, it is too early to tell who is the winner in the next gen GPU performance war especially when the DX11 titles aren't out yet. Of course, AMD has their dual "Cypress" card, the "Hemlock" in place to stop NVIDIA advancement. Also, we heard faintly that there is a Radeon HD 5890 card somewhere in the pipeline too. Surely, this Christmas gonna be an interesting one.

http://vr-zone.com/articles/amd-vs-n....html?doc=7740

[Shinanigans]

I understood about 40% of that article but shitballs it sounds tasty

[theicemagic]

Hey... Now that's pretty nice. ^_^

[mcrow5]

Originally Posted by Shinanigans

I understood about 40% of that article but shitballs it sounds tasty

lol, yeah same here, its all mumbo jumbo to me, but i like the sound of the 2x performance increase with the same clocks

[anthony256]

It's about time they got their finger out of their ass and did something, the 8800GTX was the last true big jump and before that the 6 series and before that their original cards.

Other than that, it's all rehash's of old tech or strapped on stuff, this definitely sounds promising. Let's hope they've got some decent drivers for it and we don't need to wait till 2019 or something for "Big Bang 3".

[slyls1]

yes but will nvidia still use the old gddr3 for the memory? i mean they keep on using it while ATi is using gddr5 for awhile. Is there a reason for this?

[DiGiTaL MoNkEY]

Originally Posted by slyls1

yes but will nvidia still use the old gddr3 for the memory? i mean they keep on using it while ATi is using gddr5 for awhile. Is there a reason for this?

Not sure, most people are expecting them to move to GDDR5. But i guess that is dependent on their architecture put forth on the chip.

I’m curious to see some GDDR3 vs GDDR5 pricing comparisons.

[Shinanigans]

Originally Posted by anthony256

It's about time they got their finger out of their ass and did something, the 8800GTX was the last true big jump and before that the 6 series and before that their original cards.

Other than that, it's all rehash's of old tech or strapped on stuff, this definitely sounds promising. Let's hope they've got some decent drivers for it and we don't need to wait till 2019 or something for "Big Bang 3".

The GTX280 fair slaughters the 9800GTX though. Not as big as the 7900 > 8800, sure, but still a fairly decent improvement.

[DiGiTaL MoNkEY]

Originally Posted by Shinanigans

The GTX280 fair slaughters the 9800GTX though. Not as big as the 7900 > 8800, sure, but still a fairly decent improvement.

While i agree with that, if take into consideration high end cards for that generation, the 9800GX2. The GTX 280 isn't that far off, but we know that the GPU was mainly designed to be a GPGPU monster.

[ayles]

+1 for SLi 295s when theyre $399 a pop!

[Error-prone]

All hail nvidia!! Bow to the awesomeness of their power!!!

Sounds interesting for sure.

[MrSmoke]

hah glad i didnt get a GT200 based card
those cards sound nasty

[Shinanigans]

Originally Posted by DiGiTaL MoNkEY

While i agree with that, if take into consideration high end cards for that generation, the 9800GX2. The GTX 280 isn't that far off, but we know that the GPU was mainly designed to be a GPGPU monster.

Well if we're going to compare sandwiches, then we may as well compare the 9800GX2 to the GTX295

[DiGiTaL MoNkEY]

Originally Posted by Shinanigans

Well if we're going to compare sandwiches, then we may as well compare the 9800GX2 to the GTX295

Will i'll be comparing the GTX 295 to GTX 380 or whatever they call it.

[mshagg]

I wonder how much power the next round of monstrosities will cosume