Slashdot Mirror
NVIDIA Ships Decent DX10 Graphics Card For Under $100
MojoKid writes "NVIDIA is launching a new mainstream graphics card today, aimed at consumers in the market for a relatively low-cost upgrade from an integrated graphics solution or older entry-level GPU. The new GeForce GT 240 features a GPU with 96 processor cores, 8 ROP units, and 32 texture filtering units. The GPU is manufactured using a 40nm process, features a GDDR5 memory controller (that's also compatible with GDDR3), and unlike NVIDIA's current high-end GPUs, the GT 240 is DirectX 10.1 compatible. For $100 or less, what's perhaps most interesting is that this graphics card actually puts up respectable frame rates with AA turned on and no external power needed beyond what a standard PCIe slot provides."
How does the GT240 compare to a 9400M?
Finally time for a standard PCI-E graphics solution? Death to integrated graphics!
While I understand that there is a psychological influence of the whole "under $100" mark, is it really that much different than the standard price reductions and increasing power of graphics cards over time?
I looked into this a little bit. It looks like it's more or less the same performance as my 512mb 8800 GT. Anyone else confirm that? So this is mainly just a power and price thing...
this is the best NVidia can do to try to answer the Radeon 5970 announcement tomorrow?
Nice chip. I'm waiting until you make a 40nm GPU that beats the 9800GT. 40mn is required because heat and noise are crucial to me. All of your fast 2xx series stuff is hot and power hungry, so I haven't moved.
Listen carefully: My magic price point is $200 or less. TPD must be no more than approximately 100W, ah la the 9800GT. I want 1GB (but I'll settle for 768) because 512MB is too small now. I have never cared about SLI and I won't start anytime soon. I *DO* care about heat and noise, so make these damn card builders use good cooling, which I define as "can tolerate less than perfect airflow (because fan filled holes = noise) using 1 large, quiet fan, at FULL load."
Do that and I'll upgrade. Don't and I'll look very hard at Larrabee...
- Loyal NVidia buyer
Wow, nice try. http://www.avsforum.com/avs-vb/showthread.php?t=1093445
All Radeons from 2000+ support VC1, H264 and MPEG2 hardware decoding. (Except the 2900, because ATI ruled that if you had that card, you probably had a fast enough processor to handle video anyway.)
I have a $40 Radeon in my HTPC, and it can decode Blu-rays with about 10% CPU usage. My Radeon 3650 with an Athlon 3000 is full capable of 1080p decoding.
This guy doesn't know what he's talking about.
First off, DX 10 and 10.1 have a lot more in common than DX 10.1 and 11, hence the version numbers. DX 10.1 was largely a more strict version of the DX 10 standard, for example requiring 4x FSAA filtering and 32-bit FP rendering. Well all DX 10 hardware supports that anyhow so no big deal. Still there were differences that required new hardware to fully support 10.1.
Now DX 11 has some new stuff and DX 10.1 cards are NOT compatible. Tessellation is one of those and yes earlier ATi cards do have a tessellator, but it's not DX11 compatible. However that's now all that's new. Another big one would be Shader Model 5.0. This adds various features such as double precision support and a new compute shader "basically a way of addressing the shader hardware for GPGPU stuff).
So older cards are NOT DX 11 capable. A notable absence in the ATi 4 series would be double precision support.
I should note that this doesn't mean that they can't use the DX 11 library, it just means they don't support DX 11 features. The break between 9 and 10 (where old hardware couldn't support 10 at all) appears to be the last for awhile. DX 10 hardware can use DX 10.1 and DX 11 APIs, but it doesn't support the new features.
However when someone calls something a "DX 11 card" what they mean is "A card that supports the full DX 11 feature set." Currently the only cards on the market meeting that designation are the ATi 5000 series. The ATi 4000 series are DX 10.1 cards.
For more info on what's new in DX 11 see http://msdn.microsoft.com/en-us/library/ee417843(VS.85).aspx#Full that's MS's page on it which will get as highly technical as you'd like.
1.7% yields of Fermi GPUs in first batch.
Wooden screws used in the non-working Fermi prototype card which Nvidia claimed was working.
Q2 2010 release date now for consumer Fermi GPUs instead of the promised Q4 2009 release.
20% clock miss on Fermi architecture.
And now they're releasing re-badged crap yet again.
When will it end?