BioShock Infinite Is The Latest Game Showing Why Linux Gamers Choose NVIDIA

[glxextxexlg]

AMD and its opensource & closedsource in parallel driver development model is a complete scam and utter crap. Its a lie and some in the linux community still refuse to understand this. A professional needs good driver support. Drivers are not your instant messaging app that you can easily change with something else. Good driver support is as essential as good hardware. People that still support AMD on their false and half-assed driver support model are a bunch of zealots that don't use computers to make a living or for entertainment but for their zealotism.

[Ansla]

Unfortunately AMD has no modern (28 nm, GCN) dGPU with less than 25 Watt.
NVIDIA has the GT 720 and GT 730.

That is quite understandable. Any CPU this days has a built in GPU and the above cards would be a regression compared to an AMD A10 APU and only a minor improvement over the Intel GPUs and other APUs.

[pal666]

AMD and its opensource & closedsource in parallel driver development model is a complete scam and utter crap. Its a lie and some in the linux community still refuse to understand this. A professional needs good driver support. Drivers are not your instant messaging app that you can easily change with something else. Good driver support is as essential as good hardware. People that still support AMD on their false and half-assed driver support model are a bunch of zealots that don't use computers to make a living or for entertainment but for their zealotism.

well, nvidia does not have linux drivers at all(no, windows blob in half-assed wrapper does not count), and intel has comparable driver with slower hardware.

[tuke81]

I am not really sure other than Ubuntu says there isn't a point in it. Seems like it would be harder to split it out than just put the entire Nvidia driver in the PPA's.

http://www.phoronix.com/scan.php?pag...tem&px=MTczMzE

Afaik there are openCl libs on the ppa driver, might be some problem with symbolic linking. Files are located /usr/lib/nvidia-346 and /usr/lib32/nvidia-346 folders and I think you might need to install corresponding nvidia-libopencl1-346 package from ppa to get symbolic linking right.

[vsteel]

Afaik there are openCl libs on the ppa driver, might be some problem with symbolic linking. Files are located /usr/lib/nvidia-346 and /usr/lib32/nvidia-346 folders and I think you might need to install corresponding nvidia-libopencl1-346 package from ppa to get symbolic linking right.

There are openCL libs in the Ubuntu PPAs but I couldn't ever get them to work. I followed numerous guides to get things to work and the only thing I found that would fix everything was to install the driver from Nvidia itself. I read (take with a grain of salt) on the Folding at Home forums that the openCL version that Ubuntu has is different than the one straight from Nvidia. I don't know how or if it is just in linking. Something is broken with the openCL version you can install in Ubuntu or for me at least.

/shrug

[drSeehas]

That is quite understandable. Any CPU this days has a built in GPU and the above cards would be a regression compared to an AMD A10 APU and only a minor improvement over the Intel GPUs and other APUs.

What about the AMD socket AM3+ CPUs?

NVIDIA seems to be quite happy with their GEFORCE GT 720 and GT 730. Both are available also as GDDR5 versions. I don't think an AMD A10 APU is competitive with a GDDR5 GT 730.

[Ansla]

What about the AMD socket AM3+ CPUs?

Those are all 95W+, I don't think a TDP of 30W for the R7 240 compared to the 19W of GT 720 will make much of a difference. BTW, the GT 730 is 38W.

NVIDIA seems to be quite happy with their GEFORCE GT 720 and GT 730. Both are available also as GDDR5 versions. I don't think an AMD A10 APU is competitive with a GDDR5 GT 730.

The GDDR5 variant, uses a 64bit bus and 5000MHz frequency. That should make its bandwidth slightly more than that of a dual channel (128bit) 2400MHz DDR3 setup. So I expect on average it would give roughly the same performance as the A10-7850K, as the 7850K has more processing power.

[drSeehas]

Those are all 95W+, ...

No!
There are 8-core Opterons (OS3365OLW8KHK and OS3380OLW8KHK) which have a TDP of 65 W.
There is even the 3320 EE with 25 W TDP.

I don't think a TDP of 30W for the R7 240 compared to the 19W of GT 720 will make much of a difference. BTW, the GT 730 is 38W.

From where did you get these numbers?
And the 30 W R7 240 comes with GDDR5? I doubt it.
The 25 W limit comes from the PCI Express specification. Some Mainboards can't provide more.

The GDDR5 variant, uses a 64bit bus and 5000MHz frequency. That should make its bandwidth slightly more than that of a dual channel (128bit) 2400MHz DDR3 setup. ...

64 bit GDDR5 @ 5000 MHz are roughly 40 GB/s.
128 bit DDR3 @ 2133 MHz (2400 is OC) are roughly 34 GB/s shared with the CPU.

[MoonMoon]

The 25 W limit comes from the PCI Express specification. Some Mainboards can't provide more.

Power supply through the PCIe slot is dependent on the number of lanes. x1, x4 and x8 are limited to 25W (x1 initially to 10W, but the card can inform the board that it wants to have 25W), x16 is limited to 75W (initally 25W, can demand to get 75W, but graphics cards only).

[Kano]

@drSeehas

The GT 720 / GT 730 use Kepler (GK) not Maxwell (GM1) or Maxwell Gen 2 (GM2). I would not buy those, they are for HTPC and not for gaming but if you want HDMI 2.0 you have to wait for new GM2 models.